Department of Crop Sciences 

 

 

Head of department: 

Univ.Prof. Dr.nat.techn. Dipl.‐Ing. Hans‐Peter Kaul 

 

Supervisor: 

Univ.Prof. Dr.sc.agr. Dipl.‐Ing.sc.agr. Astrid Forneck  

Co‐supervisor: 

Univ. Prof. Dr.rer.nat. Dipl.‐Chem. Antje Potthast 

 

 

 

ANALYSIS OF COMPLEX CARBOHYDRATE MIXTURES 

 

 

 

Dissertation 

for obtaining doctoral degree of 

University of Natural Resources and Life Sciences, Vienna 

 

 

Submitted by 

Manuel Becker (M.Sc.) 

 

 

November 2018, Vienna 

II 

Analysis of complex carbohydrate mixtures 

Manuel Becker, Department of Crop Sciences, 

University of Natural Resources and Life Sciences,  

Vienna (BOKU, Wien) 

 

Abstract    

Carbohydrates are widely distributed both  in animal and plant  tissues and  fulfill various biological 

functions  and  are  involved  in  physiological  processes.  Moreover,  carbohydrates  are  substantial 

resources  used  in  both  the  food  and  non‐food  industry.  Qualitative,  quantitative,  and  structural 

information of carbohydrates are essential for the characterization of plant tissue status as well as to 

control and monitor processing, quality and the prediction of product properties. Hence, the demand 

of a growing  industry and  research  sector  increases with new application areas  for  carbohydrates 

demanding a more precise carbohydrate characterization. 

However, analyzing carbohydrates  is a challenge. Carbohydrates occur highly diverse with regard to 

their  structure,  size,  and  functionality  ranging  from  simple  monosaccharides  to  highly  complex 

polysaccharides.  The  complexity  of  carbohydrate  continues within  their  species  (e.g.,  hexoses)  as 

highly similar compounds with equal molecular weight, only differing  in  their stereochemistry. The 

presence of an equilibrium mixture of up to five tautomeric forms per reducing sugar (e.g., glucose) 

isolated from natural products also complicates the analysis. The challenge of carbohydrate analysis 

lies  in  the  complexity  of  a  carbohydrate  mixture,  depending  on  the  number  of  different 

monosaccharides or mono‐, di‐ and tri‐saccharides, as well as by occurrence of carbohydrates in very 

different  concentrations  and  the  presence  of  different  sample matrices.  This  study  aims  at  the 

analysis  of  complex  carbohydrate  mixtures  with  regard  to  carbohydrate  identity,  quantity,  and 

information on the molecular structure and side reactions by GC‐MS‐based analysis methods. 

The present  study about  carbohydrate analysis of complex mixtures  involved  the evaluation of  six 

different  derivatization  approaches  at  ten  reference  compounds  (C2‐C6)  by  GC‐MS.  The  most 

appropriate  derivatization methods were  applied  to  32 monosaccharides  and  13  disaccharides  to 

receive  further  information. Additionally, carbohydrate degradation products of pulp mill effluents 

and  book  paper  extracts  were  studied.  Information  on  the  carbohydrates  was  combined  with 

characterization of lignocellulosic material in form of polysaccharides, paper, and pulps. The study of 

lignocellulosic  side‐reactions  describes  three  different  analysis methods:  (1)  the  determination  of 

carbohydrate  composition  by  parallel  hydrolysis  of  sulfuric  acid  and  acid  methanolysis  of  21 

polysaccharides  and  21  cellulosic  samples  by GC‐MS.  (2)  A  Zemplen‐deacetylation  based  analysis 

method  for  the  quantification  of  bound  acetyl  groups  in  polysaccharides  and  monomeric 

III 

carbohydrates.  (3) The  volatile organic  acid emission potential of  storage‐materials present  in  the 

collection  of  the  drawings  and  prints  of  Karl  Friedrich  Schinkel  in  Berlin  and  their  impact  on  the 

cellulose integrity of two indicator papers. 

The  evaluation  of  two  single‐step  and  four  two‐step  derivatization  approaches  for  liquid  GC‐MS 

analysis  showed  that  the  sequential ethoximation and  trimethylsilylation  is advantageous  to other 

approaches.  The  benefits  are  a  low  number  of  peaks  obtained  per  reducing  carbohydrate,  good 

chromatographic  resolution,  low  limits  of  detection  and  quantitation,  low  relative  standard 

deviations, a high  informational value of mass spectra, and high robustness towards matrix effects. 

Consequently,  a  deeper  insight  into  the O‐ethoximation  followed  by  silylation  approach  revealed 

chromatographic and mass spectrometric properties of 46 carbohydrates. Based on these results, an 

oxime peak  identifier  is proposed, which  involves  the elution order and  the  retention  time shift of 

the  syn/anti‐peak  to  increase  the  reliability  of  the  identification  of  reducing  carbohydrates.  The 

analysis  of  carbohydrate  composition  also  comprises  the  quantitative  and  qualitative  monomer 

distribution  of  polysaccharides  and  cellulosic materials.  Also,  the  degradation  effects  of  electron 

beam  irradiation  treatments  or  aging  processes  on  the  carbohydrate  composition  of  cellulosic 

samples  are  investigated.  The measurement  of  the  degree  of  acetylation  showed  the  analysis  of 

bound acetyl groups at polysaccharides without being affected by the presence of free or adsorbed 

acetic  acid  and  acetates.  The  analysis  of  methyl  acetate  in  the  vapor  phase  by  SPME‐GC‐MS 

combined with  4‐O‐(13C2‐acetyl)  vanillin  for  internal  standardization, which  generates  isotopically 

labeled methyl  13C2‐acetate  in  situ, overcomes  the contamination problem and eliminates possible 

influences, e.g., discrimination effects. 

A  quantitative  analysis  of  formic  acid  and  acetic  acid  emission  of  storage  materials  by  static 

headspace GC‐MS with selected‐ion monitoring  (SHS‐GC/SIM‐MS) showed significant differences of 

acetic  acid  concentration  among  storage  samples  and  a  formic  acid  emitting  fabric  sample.  The 

indicator‐paper based degradation caused by emissions from Schinkel exhibit materials revealed an 

oxidation  impact  on  both  indicator  papers  and  a  stronger  formic  acid‐induced  hydrolysis  in 

association with a higher loss of keto groups compared to acetic acid. 

 

 

 

 

 

 

 

IV 

Kurzfassung 

Kohlenhydrate sind sowohl  in  tierischen als auch  in pflanzlichen Geweben weit verbreitet, erfüllen 

verschiedene  biologische  Funktionen  und  sind  an  physiologischen  Prozessen  beteiligt.  Darüber 

hinaus  sind  Kohlenhydrate wichtige  Ressourcen,  die  sowohl  in  der  Lebensmittel‐  als  auch  in  der 

Verarbeitungsindustrie verwendet werden. Qualitative, quantitative und strukturelle  Informationen 

über Kohlenhydrate sind essentiell für die Charakterisierung des Status von Pflanzengeweben sowie 

für  die  Kontrolle  und  Überwachung  der  Verarbeitung,  der  Qualität  und  der  Vorhersage  von 

Produkteigenschaften. Die Anforderungen  eines wachsenden  Industrie‐  und  Forschungssektors  an 

eine  präzisere  Kohlenhydratcharakterisierung  steigen  daher  mit  der  Etablierung  neuer 

Anwendungsgebiete für Kohlenhydrate.  

Die Analyse von Kohlenhydraten  ist jedoch eine Herausforderung. Kohlenhydrate treten hinsichtlich 

ihrer  Struktur,  Größe  und  Funktionalität  sehr  unterschiedlich  auf  und  reichen  von  einfachen 

Monosacchariden bis zu hochkomplexen Polysacchariden. Die Komplexität von Kohlenhydraten setzt 

sich  innerhalb  ihrer  C‐Gruppe  (z.  B.  Hexosen)  als  sehr  ähnliche  Verbindungen  mit  gleichem 

Molekulargewicht  fort, die sich  in  ihrer Stereochemie unterscheiden. Das Auftreten von bis zu  fünf 

tautomeren  Formen  pro  reduzierendem  Zucker  (z.  B. Glukose)  bei  der  Extraktion  aus  natürlichen 

Produkten erschwert die Analyse. Die Herausforderung der Kohlenhydratanalyse beruht zudem auf 

der  Komplexität  einer  Kohlenhydratmischung,  die  zum  einen  abhängig  von  der  Anzahl  der 

verschiedenen Mono‐,  Di‐  und  Trisaccharide  sein  kann  und  zum  anderen  vom  Auftreten  dieser 

Kohlenhydrate  in  sehr  unterschiedlichen  Konzentrationen,  sowie  der  Anwesenheit  von 

verschiedenen  Probenmatrizen  beeinflusst  wird.  Diese  Studie  zielt  darauf  ab,  komplexe 

Kohlenhydratmischungen  in Bezug auf Identität und Konzentration der vorhandenen Kohlenhydrate 

mittels GC‐MS‐basierter Verfahren zu analysieren und  Informationen über die Molekülstruktur und 

deren Nebenreaktionen bereit zu stellen. 

Die vorliegende Dissertation über komplexe Kohlenhydratmischungen umfasst die Evaluierung von 

sechs  verschiedenen  Derivatisierungsmethoden  an  zehn  verschiedenen  Kohlenhydraten  (C2‐C6) 

mittels GC‐MS. Die am besten geeignete Derivatisierungsmethode wird mit 32 Monosacchariden und 

13  Disacchariden  nochmals  genauer  untersucht.  Darüber  hinaus  werden  Abbauprodukte  von 

Kohlehydraten  aus  Zellstoffabwässern  bzw.  Buchpapierextrakten  und  Lignocellulose  in  Form  von 

Polysacchariden,  Papier  und  Zellstoffen  charakterisiert.  Die  Analyse  von  Nebenreaktionen  an  der 

Struktur  von  Lignocellulose  wird  anhand  dreier  verschiedener  Verfahren  untersucht:  (1)  die 

Bestimmung der Kohlenhydratzusammensetzung durch parallele Hydrolyse von Schwefelsäure und 

Methanolyse  von  21  Polysacchariden  und  21  Zelluloseproben  mittels  GC‐MS.  (2)  Ein 

Analyseverfahren  basierend  auf  der  Zemplén‐Deacetylierung  zur Quantifizierung  von  gebundenen 

Acetylgruppen  in  Polysacchariden  und  Kohlenhydraten.  (3)  Die  Untersuchung  des 

V 

Emissionspotentials flüchtiger organischer Säuren von Zeichnungen bzw. Drucken der Sammlung Karl 

Friedrich  Schinkel  im  Kupferstichkabinett  (Berlin)  und  vorhandenen  Lagermaterialien  und  deren 

Auswirkung auf die Zelluloseintegrität.  

Die Auswertung von zwei einstufigen und vier zweistufigen Derivatisierungsansätzen für die GC‐MS‐

Analyse  zeigte,  dass  die  Ethoximierung  und  anschließende  Trimethylsilylierung  anderen 

Derivatisierungsmethoden  überlegen  ist.  Die  Vorteile  sind  eine  geringe  Anzahl  von  Peaks  pro 

reduzierendem  Kohlenhydrat,  eine  gute  chromatographische  Auflösung,  niedrige  Nachweis‐  und 

Bestimmungsgrenzen,  geringe  relative  Standardabweichungen,  ein  großer  Informationswert  der 

Massenspektren und eine hohe Robustheit gegenüber Matrixeffekten. Aus diesem Grund wurde eine 

intensivere Untersuchung der Ethoximierung und anschließender Silylierung von 46 Kohlenhydraten 

hinsichtlich  chromatographischer  und  massenspektrometrischer  Eigenschaften  durchgeführt.  Auf 

Basis dieser Ergebnisse wurde ein Parameter zur  Identifizierung der Oxim‐Peaks vorgeschlagen, um 

die  Zuverlässigkeit  der  Identifizierung  von  reduzierenden  Kohlenhydraten  zu  verbessern.  Dieser 

begründet sich  in der Reihenfolge der Retention und der Verschiebung der Retentionszeit zwischen 

dem  syn/anti‐Peak.  Die  Analyse  der  Kohlehydratzusammensetzung  umfasst  auch  die  quantitative 

und  qualitative  Zusammensetzung  von  Monomeren  in  Polysacchariden  und  cellulosehaltigen 

Materialien bzw. untersucht den Einfluss von Elektronenbestrahlungen und Alterungsprozessen auf 

Degradationseffekte  dieser  Proben.  Die Messung  des  Acetylierungsgrades  zeigte  die  Analyse  von 

gebundenen  Acetylgruppen  an  Polysacchariden,  ohne  durch  die  Anwesenheit  von  freier  oder 

adsorbierter Essigsäure bzw. Acetaten beeinflusst zu werden. Die Analyse von Methylacetat  in der 

Dampfphase  durch  SPME‐GC‐MS  kombiniert  mit  4‐O‐(13C2‐Acetyl)‐Vanillin  zur  internen 

Standardisierung,  die  in  situ  isotopenmarkiertes  Methyl‐13C2‐Acetat  erzeugt,  löst  das 

Kontaminationsproblem und eliminiert mögliche Einflüsse wie Diskriminierungseffekte.  

Eine quantitative Analyse der Ameisensäure‐ und Essigsäureemission von Speichermaterialien durch 

eine  Headspace‐GC‐MS‐Methode  mit  ausgewähltem  Ionen‐Monitoring  (SHS‐GC/SIM‐MS)  zeigte 

signifikante  Unterschiede  von  Essigsäurekonzentration  in  historischen  Materialien  und  einer 

Ameisensäure‐emittierenden  Textilprobe.  Die  ausgasenden  Materialien  in  der  Schinkelsammlung 

zeigen Degradationseffekte in Form einer oxidativen Wirkung auf Indikatorpapiere und im Vergleich 

zu  Essigsäure  eine  stärkere Ameisensäure‐induzierte Hydrolyse  in Verbindung mit  einem  höheren 

Verlust an Keto‐Gruppen. 

 

 

 

 

VI 

List of publications 

The following papers are part of this thesis: 

 

I. Bogolitsyna,  A.;  Becker, M.; Dupont,  A.‐L.;  Borgards,  A.;  Rosenau,  T.;  Potthast,  A.  (2011): 

Determination of  carbohydrate‐  and  lignin‐derived  components  in  complex  effluents  from 

cellulose  processing  by  capillary  electrophoresis  with  electrospray  ionization‐mass 

spectrometric detection. Journal of Chromatography A, vol. 1218, pp. 8561‐8566. 

 

II. Bogolitsyna,  A.;  Becker,  M.;  Borgards,  A.;  Liebner,  F.;  Rosenau,  T.;  Potthast,  A.  (2012): 

Degradation products of lignocellulosics in pulp mill effluents – comparison and evaluation of 

different gas chromatographic techniques for a comprehensive analysis. Holzforschung, vol. 

66, pp 917‐925. 

 

III. Becker,  M.;  Zweckmair,  T.;  Forneck,  A.;  Rosenau,  T.;  Potthast,  A.;  Liebner,  F.  (2013): 

Evaluation  of  different  derivatisation  approaches  for  gas  chromatographic‐mass 

spectrometric  analysis  of  carbohydrates  in  complex  matrices  of  biological  and  synthetic 

origin. Journal of Chromatography A, vol. 1281, pp. 115‐126. 

 

IV. Becker, M.; Liebner, F.; Rosenau, T.; Potthast, A. (2013): Ethoximation‐silylation approach for 

mono‐ and disaccharide analysis and  characterization of  their  identification parameters by 

GC‐MS. Talanta, vol. 115, pp. 642‐651. 

 

V. Zweckmair,  T.1,  Becker, M.1,  Ahn,  K.,  Hettegger,  H.,  Kosma,  P.,  Rosenau,  T.,  Potthast,  A. 

(2014):  A  novel method  to  analyze  the  degree  of  acetylation  in  biopolymers,  Journal  of 

Chromatography A, 1372, pp. 212‐220. 1 These authors contributed equally to this work. 

 

VI. Becker, M.; Meyer, F.; Jeong, M.‐J.; Ahn, K.; Henniges, U.; Potthast, A. (2016): The museum in 

a  test  tube – Adding a  third dimension  to  the evaluation of  the  impact of volatile organic 

acids on paper. Polymer Degradation and Stability, vol 130, pp. 109‐117. 

 

VII. Manuel  Becker,  Kyujin  Ahn, Markus  Bacher,  Chunlin  Xu,  Anna  Sundberg,  Stefan Willför, 

Thomas  Rosenau,  Antje  Potthast:  Comparative  hydrolysis  analysis  of  pulps  and  papers: 

carbohydrate compositions and uncovering the supramolecular structure of cellulose. To be 

submitted. 

VII 

List of other related publications 

 

Patel,  I.; Opietnik, M.;  Böhmdorfer,  S.;  Becker, M.;  Potthast,  A.;  Saito,  T.;  Isogai,  A.;  Rosenau,  T. 

(2010):  Side  reactions  of  4‐acetamido‐TEMPO  as  the  catalyst  in  cellulose  oxidation  systems. 

Holzforschung, 64, (5), 549. 

 

Opietnik, M.; Nabihah Binti  Syed  Jaafar,  S.; Becker, M.; Bohmdorfer,  S.; Hofinger, A.; Rosenau,  T. 

(2012): Ascorbigen ‐ Occurrence, Synthesis, and Analytics. Mini‐Reviews in Organic Chemistry, 9, (4), 

411‐417. 

 

Theuretzbacher, F.; Bauer, A.; Lizasoain, J.; Becker, M.; Rosenau, T.; Potthast, A.; Friedl, A.; Piringer, 

G.; Gronauer, A.  (2013): Potential of different Sorghum bicolor  (L. Moench) varieties  for combined 

ethanol and biogas production in the Pannonian climate of Austria. Energy, 55, (0), 107‐113. 

 

Griesser, M.; Weingart, G.; Schoedl‐Hummel, K.; Neumann, N.; Becker, M.; Varmuza, K.; Liebner, F.; 

Schuhmacher, R.; Forneck, A. (2015): Severe drought stress is affecting selected primary metabolites, 

polyphenols, and volatile metabolites in grapevine leaves (Vitis vinifera cv. Pinot noir). Plant Physiol. 

Biochem, 88, 17‐26. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VIII 

Table of Contents 

Abstract   ................................................................................................................................................. II 

Kurzfassung ............................................................................................................................................ IV 

List of publications.................................................................................................................................. VI 

List of other related publications .......................................................................................................... VII 

Table of Contents ................................................................................................................................. VIII 

Abbreviations .......................................................................................................................................... X 

 

1  Introduction ............................................................................................................................. 1 

1.1  Characteristics of carbohydrates ............................................................................................. 2 

1.2  Functions of carbohydrates ..................................................................................................... 7 

1.3  Complex carbohydrate mixtures ............................................................................................. 9 

1.4  Carbohydrate analysis ‐ State of the art ................................................................................ 10 

1.4.1  Non‐chromatographic analysis techniques ........................................................................... 10 

1.4.2  Chromatographic analysis techniques ................................................................................... 12 

1.4.3  GC versus LC ........................................................................................................................... 15 

1.5  Derivatization of carbohydrates for GC‐MS analysis ............................................................. 17 

1.5.1  Single‐Step derivatization procedures ................................................................................... 19 

1.5.2  Multi‐step derivatization procedures .................................................................................... 28 

 

2  Material and Methods .......................................................................................................... 35 

2.1  Derivatization methods .......................................................................................................... 35 

2.1.1  Ethyloximation‐trimethylsilylation (EtOx‐TMS) ..................................................................... 35 

2.1.2  Other derivatization methods ................................................................................................ 35 

2.1.3  O‐isopropylidenation (ISP) ..................................................................................................... 36 

2.1.4  Oximation reactions followed by trimethylsilylation (TMS) or trifluoroacetylation (TFA) .... 36 

2.2  Hydrolysis reactions for polysaccharide analysis ................................................................... 36 

2.2.1  Sulfuric acid hydrolysis ........................................................................................................... 36 

2.2.2  Acid methanolysis .................................................................................................................. 37 

2.2.3  Per‐trimethylsilylation (TMS) of monosaccharides obtained from hydrolysis reactions ...... 37 

2.2.3.1  GC‐MS analysis of TMS‐derivatized hydrolysis products ....................................................... 37 

2.2.4  Solid‐State NMR ..................................................................................................................... 38 

2.2.5  GPC analysis of cellulose samples .......................................................................................... 38 

2.3  Analysis of the degree of acetylation in biopolymers ............................................................ 39 

IX 

2.3.1  Method 1: Direct liquid‐phase analysis ................................................................................. 39 

2.3.1.1  Sample preparation ............................................................................................................... 39 

2.3.1.2  GC–MS conditions for the analysis of the liquid phase ......................................................... 39 

2.3.2  Method 2: Analysis via gas phase .......................................................................................... 39 

2.3.2.1  Sample preparation ............................................................................................................... 39 

2.3.2.2  GC–MS conditions for the analysis by SPME ......................................................................... 40 

2.4  Evaluation of the impact of volatile organic acids on paper ................................................. 41 

2.4.1  Analysis of formic acid and acetic acid emission potential by static headspace GC‐MS with 

  selected‐ion monitoring (SHS‐GC/SIMMS) ............................................................................ 41 

2.4.2  Artificial aging ........................................................................................................................ 41 

2.4.3  Molar mass and carbonyl content measurement .................................................................. 42 

 

3  Results and discussion .......................................................................................................... 43 

3.1  Analysis of carbohydrates, lignocellulosic side‐reactions, and degradation products .......... 43 

3.1.1  Analysis of lignocellulosic effluent streams and aged paper extracts (Paper I & II) .............. 43 

3.1.1.2  Study of complex mixtures of lignocellulose degradation products by GC‐MS methods ..... 44 

3.1.1.3  Comparison of CE‐MS and GC‐MS methods for analyzing complex mixtures of  lignocellulose 

  degradation products ............................................................................................................ 45 

3.1.2  Analysis of complex mixtures of carbohydrates (Paper III & IV) ............................................ 46 

3.1.2.1  Derivatization methods (Paper III) ......................................................................................... 46 

3.1.3  Analysis of lignocellulosic side‐reactions (Paper V, VI and VII) .............................................. 50 

3.1.3.1  Applications ........................................................................................................................... 50 

3.1.3.1.1 Carbohydrate compositions of polysaccharides (Paper V to be submitted) ......................... 50 

3.1.3.1.2 The degree of Acetylation (Paper VI) ..................................................................................... 53 

3.1.3.1.3 Volatile organic acids in the paper (Paper VII) ....................................................................... 54 

 

4  Conclusion ............................................................................................................................. 57 

 

Acknowledgements ............................................................................................................................... 60 

References ............................................................................................................................................. 61 

List of Figures ......................................................................................................................................... 72 

Appendix  ............................................................................................................................................... 73 

 

 

X 

Abbreviations 

AE     Anion‐exchange 

BeOx    O‐benzyl oxime 

BSTFA     N,O‐bis(trimethylsilyl)trifluoroacetamide 

CE    Capillary electrophoresis 

CI     Chemical ionization 

CP     Cross‐polarization 

DMAP    4‐(dimethylamino)pyridine 

EE    Ethyl acetate 

EI    Electron ionization 

ELSD     Evaporative light‐scattering detection 

ESI     Electrospray ionization 

EtOx    O‐ethyl oxime 

eV     Electron volt 

FID     Flame ionization detector 

FPD    Flame photometric detector 

GC     Gas chromatography 

GC‐FID    Gas chromatography coupled with flame ionization detection 

GC‐MS    Gas chromatography coupled with mass spectrometry 

GFC     Gel filtration chromatography 

GPC     Gel permeation chromatography 

HILIC     Hydrophilic interaction liquid chromatography 

HPLC     High performance liquid chromatography 

IUPAC     International Union of Pure and Applied Chemistry  

LC     Liquid chromatography (see HPLC) 

LC‐MS     Liquid chromatography coupled with mass spectrometry 

LOD     Limit of detection 

LOQ     Limit of quantification 

M+•    Molecular ion 

MALDI‐TOF   Matrix‐assisted laser desorption ionization coupled with time of flight mass  

    spectrometry 

MeOx    O‐methyl oxime 

MS     Mass spectrometry 

MSTFA    N‐methyl‐N‐(trimethylsilyl)trifluoroacetamide 

XI 

MW     Molecular weight 

m/z     Mass‐to‐charge 

NMR     Nuclear magnetic resonance 

NPD     Nitrogen phosphorus detector 

Ox    O‐hydroxyl oxime 

PAD     Pulsed amperometric detection 

PCA     Principal component analysis 

Py‐GC     Pyrolysis gas chromatography 

RI     Refractive Index 

SEC    Size exclusion chromatography 

SIM    Selected ion monitoring 

SPME    Solid phase micro extraction 

TIC    Total ion current 

TMCS     Trimethylchlorosilane 

TMS     Trimethylsilyl 

TOF    Time‐of‐flight 

UV     Ultra‐violet 

1 

1 Introduction 

Carbohydrates are the most abundant organic compounds in nature and the most versatile materials 

available [1]. In nature, they are synthesized predominantly by photosynthesis and consist of carbon, 

hydrogen, and oxygen [2]. Carbohydrates are widely distributed both in animal and plant tissues and 

fulfill  various  biological  functions  as  structural  materials,  energy  storage,  and  regulators  in 

physiological processes. Moreover, carbohydrates are  substantial  resources used  in both  food and 

non‐food  industry. Besides  their use as a  foodstuff,  their use also  includes  food gums, stabilizer or 

quality marker. In the non‐food sector, carbohydrates are applied as feedstock  in the production of 

textiles,  paper,  plastics,  and  downstream  industries  as well  as  raw material  for modification  and 

transformation  to  substances,  which  are  primary  sources  for  detergents,  emulsifiers,  foams  or 

vitamins  [3].  The  amount  of  carbohydrates  and  their  accessibility  in  plant  tissues  plays  also  an 

important  role  for  the use as  renewable energy. The demands of a growing  industry and  research 

sector and the trend towards more precise carbohydrate data  increase with the occurrence of new 

possible  applications  for  carbohydrates.  The  information  about  the  qualitative  and  quantitative 

distribution of carbohydrates in foodstuff, feedstock, and processed materials is essential to control 

and monitor processing, quality and prediction of product properties. 

However,  analyzing  carbohydrates  is  a  challenge.  Carbohydrates  are  uncharged  and  colorless  [4]. 

They  occur  very  diversely  regarding  structure,  size,  functionality,  and  ranging  from  simple 

monosaccharides to highly complex polysaccharides [5]. Monosaccharides with the same number of 

carbon atoms (e.g., hexoses) can form isomers, which are analytically similar compounds with equal 

molecular weight,  only  differing  in  their  stereochemistry  [6].  Furthermore,  reducing  sugars  (e.g., 

glucose) exist as an equilibrium mixture of up  to  five  tautomeric  forms,  α‐ &  β‐pyranose,  α‐ &  β‐

furanose and  the open‐chain  form  [7]. The number of different mono‐, di‐ and  trisaccharides  in a 

carbohydrate mixture and their occurrence in very different concentrations affect the complexity of a 

carbohydrate  blend.  The  challenge  of  carbohydrate  analysis  lays  in  the  complexity  of  the 

carbohydrate mixture. Complex carbohydrate blends can be extracts or samples from plants tissues 

(e.g., fruits), honey or hydrolysis products of polysaccharides (e.g., paper). 

A  wide  range  of  analytical  techniques  exists  for  separation  and  identification  of  different 

carbohydrate  species. They  range  from physical,  chemical, and enzymatic measurements  to highly 

sensitive  chromatographic  methods.  The  primary  chromatographic  methods  for  analyzing 

carbohydrate  compounds  comprise  high‐performance  liquid  chromatography  (HPLC)‐,  gas 

chromatography (GC)‐ and capillary electrophoresis (CE)‐based methods [3]. The gas chromatograph 

coupled to mass spectrometry (GC‐MS) system has been chosen for this study, providing a very high 

separation  performance  combined  with  highly  selective  and  sensitive  ways  of  detection  for 

carbohydrates [8]. However, the high polarity, pronounced hydrophilicity with a strong tendency to 

2 

hydrogen bonding, and near‐zero volatility of carbohydrates denote the drawbacks of applying this 

method.  Therefore,  all  carbohydrates  require  suitable derivatization  to  convert  them  into  volatile 

products prior to GC‐MS analysis [9]. 

Various derivatization methods and strategies exist to produce carbohydrate derivatives for GC‐MS 

analysis. Derivatization of carbohydrates is usually conducted prior GC injection, while the strategies 

are  differentiated  in  single‐step  and  two‐  or  multi‐step  strategies.  The  primary  single‐step 

derivatization strategy applied for analyzing polyalcohols and non‐reducing carbohydrates by GC‐MS 

comprises  alkylation  (particular methylation),  acetylation  (acetates  or  trifluoroacetates),  silylation 

and the formation of cyclic derivatives, e.g., by isopropylidination [9‐11]. However, most of the single 

step derivatization procedures retain up to five tautomers per reducing carbohydrate. Two‐ or multi‐

step derivatization  strategies have  been developed  to overcome  this drawback by  converting  the 

carbonyl group into a specific derivative (e.g., oxime) to inhibit the formation of different tautomers 

by intermolecular conversion prior trimethylsilyl or trifluoroacetyl derivatization [7, 10]. Each method 

has  its advantages and disadvantages  regarding  reaction  time, handling, costs, and  labor. Previous 

studies  about  carbohydrates  analysis based  only  on  a  few  reference  standards,  sample origins  or 

compared  derivatization  methods  [12,  13].  There  is  also  a  lack  of  identification  data  for 

carbohydrates beyond the well‐known glucose, fructose, and sucrose. 

This  study  aims  to  evaluate  the  most  appropriate  analysis  approaches  for  the  investigation  of 

complex carbohydrate mixtures. The evaluation includes single and two‐step derivatization methods 

for carbohydrates and associated degradation products with regard to their performance for GC‐MS 

and  CE‐MS  analysis.  Evaluation  criteria  are  derivatization  efficiency,  chromatographic  resolution 

characteristics,  informational  value,  and  pattern  diversity  of mass  spectra,  the  reliability  of  peak 

assignment,  reproducibility  of  quantification  results,  and  robustness  towards matrix  effects.  The 

study  also  includes  the  analysis  of  polysaccharides  on  their monomer  composition  by  hydrolysis 

approaches,  their  degree  of  acetylation,  and  the  degradation‐caused  release  of  volatile  organic 

compounds (VOCs). 

 

 

1.1 Characteristics of carbohydrates 

Carbohydrates  are  polyhydroxy  carbonyls, which  include  a wide  range  of molecules,  occurring  in 

single or multiple units as  low molecular weight mono‐ and disaccharides,  intermediate molecular 

weight oligosaccharides up to high molecular weight polysaccharides [14, 15].  

The  classification of monosaccharides  takes place  according  to  their number of  carbons,  available 

ketone or aldehyde groups and  the  corresponding  structural  configuration. The number of carbon 

atoms in the molecules’ carbon chain determines the naming of monosaccharides, starting at trioses 

3 

(C3), tetroses (C4), pentoses (C5), hexoses (C6), heptoses (C7), et cetera [16]. Monosaccharides, which 

contain  an  aldehyde  group  at  carbon‐1  (C‐1)  are  aldoses  (e.g.,  glucose). Whereas  ketoses  (e.g., 

fructose) contain a carbonyl group often located at carbon‐2 (C‐2) but can also occupy any position in 

a carbohydrate chain, except the terminal position. According to this naming, there are aldo‐tetroses, 

aldo‐pentoses,  aldo‐hexoses,  et  cetera  and  keto‐tetroses,  keto‐pentoses,  keto‐hexoses,  et  cetera. 

The carbon atom of  the carbonyl or aldehyde groups  is  the reactive center of  the monosaccharide 

chain, called the anomeric carbon atom [15, 17]. 

The  high  variety  of  monosaccharides  bases  on  the  different  stereochemical  configurations.  The 

elongation  of  the monosaccharides’  carbon  chain  starting  at  triose  by  insertion  of  one  or more 

hydroxymethyl  groups  (–CHOH)  generates  a new  chiral  center  in  the  carbohydrate molecule  [14]. 

When drawn  in the Fischer projection, carbohydrates showing the hydroxyl group at the reference 

carbon atom oriented to the right belong to the D‐chiral family; the hydroxyl group oriented to the 

left means  the  carbohydrate  belongs  to  the  L‐chiral  family  [14,  15].  So, D‐carbohydrates  are  the 

mirror  images  of  L‐carbohydrates.  However,  they  show  specific  optical  rotation  [14,  15].  The  D‐

configuration is the present form of naturally occurring carbohydrates. L‐carbohydrates rarely appear 

in  nature,  except  L‐arabinose  and  L‐galactose,  which  exist  as  monomers  in  many  carbohydrate 

polymers  [14].  L‐sugars  are  handled  differently  in  biological  systems  and  are  not metabolized  by 

humans  [15].  Epimers  and  diastereoisomers  describe  other  stereochemical  configurations  of 

carbohydrates. Epimers denote for a pair of carbohydrates, which differ in the configuration of only 

one asymmetric atom (chiral center), e.g., D‐glucose and D‐mannose are epimers at the C‐2 position; 

D‐glucose and D‐galactose are epimers at the C‐4 position. Carbohydrate pairs differing in more than 

one asymmetric carbon atoms (chiral centers) are diastereoisomers [15, 16]. 

The  stereochemical  configuration  of  aldohexoses  includes  four  chiral  carbons,  which  allow  the 

formation of 16 different sugars with an aldehyde end, and eight different keto‐hexoses. So,  there 

are 24 different C6‐carbohydrates, 12 belonging to the D‐series, and 12 belonging to the L‐series [14]. 

Considering  that  L‐sugars  rarely occur  in nature,  12 hexoses possess  the  same  empirical  formula, 

C6H12O6, and own  therefore  the  same molecular weight of 180.16 g*mol−1. The 12  stereochemical 

isomers only differ in their orientation of hydroxyl groups within the molecule [6]. 

Monosaccharides  can  form  a  cyclic  hemiacetal  or  hemiketal  structure  by  an  intramolecular 

nucleophilic addition of an OH‐group at the ketone or aldehyde , resulting in either a five‐membered 

or  six‐membered  ring  [16,  18].  A  six‐membered  cyclic  hemiacetal  is  called  pyranose,  a  five‐

membered  cyclic  hemiacetal  a  furanose.  Accordingly,  the  six‐membered  ring  of  glucose  is  called 

glucopyranose and the five‐membered ring of fructose fructofuranose [16]. The cyclization reaction 

of monosaccharides  in  solution  is  reversible  and  in hence  in equilibrium with e.g.  the open  chain 

form, which  is dependent on temperature and pH [18]. The chemical reactivity and functionality of 

4 

different carbohydrates are directly related to the presence of the hemiacetal, hemiketal, acetal, and 

ketal functional groups [15]. Cyclization of carbohydrates produce two new discrete  isomeric forms 

as  the  hydroxyl  group  on  the  anomeric  carbon  atom,  and  formerly  achiral  center  has  now  two 

possible  orientations,  designated  as  α  and  β  [14,  15].  These  two  configurations  have  different 

chemical properties and are called anomers, which differ only at C‐1 for aldoses and C‐2 for ketoses 

[15]. 

Cyclic hemiacetals are particularly stable; usually they are more stable than their open‐chain forms. 

So the solid, crystalline  form of carbohydrates can consist of molecules with specific anomeric ring 

form [16]. When a reducing carbohydrate dissolves  in water, different tautomeric forms occur, due 

to molecular  rearrangements  causing  ring  openings  and  subsequent  ring  closures.  This  process, 

described  as mutarotation,  alter  the optical  rotation of  carbohydrates  and produce  the  α‐  and  β‐

pyranose and α‐ and β‐furanose forms (Fig. 1) [19]. These isomers together with the open chain form 

of  a  monosaccharide  constitute  a  specific  equilibrium  mixture.  Glucose  undergoes  a  “simple” 

mutarotation, since only the α‐ and β‐pyranose forms are available in significant amounts. α‐ and β‐

furanose,  as  well  as  the  open‐chain  form,  only  occur  in  traces  (Fig.  1),  while  a  “complex” 

mutarotation process describes the formation of three or more tautomers, e.g., for xylose [19].  

 

 

 

Figure 1: D‐glucose  forms  five different epimers  in  solution with a  specific equilibrium distribution 

(brackets). According to [14], modified. 

 

5 

The distribution of  xylose  isomers  at 20°C  is  similar  to  glucose;  the  available  forms  are mostly  α‐

pyranose  (34.8%)  and  β‐pyranose  (65.2%).  Though,  in  contrast  to  glucose,  xylose  produces 

significantly higher  amounts of  almost  all possible  forms  at 31°C:  α‐pyranose  (36.5%),  β‐pyranose 

(58.5%),  α‐furanose  (6.4%),  β‐furanose  (13.5%)  and  the  open‐chain  form  (0.05%)  [14].  The 

temperature  increase  leads to the formation of the higher energy forms α‐pyranose, furanose, and 

the open‐chain  [19]. Each  tautomeric  form has different  chemical and physical properties  such as 

solubility, optical  rotation,  relative  sweetness, chemical  reactivity, et cetera  [19]. The  formation of 

two  to  five  tautomers  alone by  each of  the 12 hexoses  increases  the number of possible  further 

components and complicates their analytical separation. 

The  formation of disaccharides occurs by  reaction of  the hydroxyl group  located on  the anomeric 

carbon atom of a monosaccharide with one  the hydroxyl groups of another monosaccharide, also 

referred  to  as  a  glycosidic  linkage  [14,  19].  In  contrast  to  the  formation  of  a  hemiacetal  by  a 

nucleophilic  addition  reaction,  the  formation  of  an  acetal  from  a  hemiacetal  is  a  nucleophilic 

substitution reaction here the original carbonyl oxygen  leaves the molecule as water molecule [18]. 

Disaccharides  can be  composed of  two  identical  (homogeneous) or  two different  (heterogeneous) 

monomers  [14]. When  considering  a  homogenous  glucose‐based  disaccharide,  the  possibility  of 

different  linkage  conformations  can  cause  the  formation  of  11  different  disaccharides:  (1)  α‐D‐

glucose  can  react with  the hydroxyl  group  at C‐2, C‐3, C‐4 or C‐6 of  the other  glucose monomer, 

resulting in four reducing disaccharides. (2) β‐D‐Glucose can react with the same hydroxyl group at C‐

2, C‐3, C‐4 or C‐6, resulting in four other possible reducing disaccharides. (3) The α‐ and β‐hemiacetal 

hydroxyl  groups  (at  C‐1)  of  two  glucose monomers  can  react with  each  other,  resulting  in  three 

nonreducing disaccharides:  α,α‐trehalose,  β,β‐trehalose, and  α,β‐trehalose. Although  the  structure 

of the 11 glucose‐based disaccharides differs only in the type of glycosidic linkage between the two 

identical  monomers,  each  disaccharide  has  unique  chemical  and  physical  properties  [14].  The 

glycosidic  linkage,  the  α‐  or  β‐form,  and  the  ring  size  are  essential  characteristics  to  distinguish 

among  disaccharides  [19]. Disaccharides  occur  either  as  reducing  carbohydrates, which  contain  a 

reactive hemiacetal or hemiketal  functional  group or  as non‐reducing  carbohydrates with no  free 

hemiacetal hydroxyl group (Fig. 2).  

6 

O

OH

O

OH

CH2OH

OH

Sucrose ( ‐D‐glucopyranosyl‐(1‐2)‐ ‐D‐fructofuranoside)

O

HO

CH2OH

OH

CH2OHO

OH

OH

CH2OH

OH

O

OH

OH

OH

CH2OH

O

Cellobiose ( ‐D‐glucopyranosyl‐(1‐4)‐D‐glucopyranose)  

Figure  2: Monosaccharide  linkage  of  the  non‐reducing  disaccharide  sucrose  (α‐D‐glucopyranosyl‐

(12)‐β‐D‐fructofuranoside) and the reducing disaccharide cellobiose (β‐D‐glucopyranosyl‐(14)‐D‐

glucopyranose).  

 

Non‐reducing  carbohydrates  form  a  glycosidic  linkage between  the  two  anomeric  carbons of  two 

monomers [16, 19]. Reducing disaccharides will alsoform an equilibrium mixture of α and β, furanose 

and  pyranose,  and  open‐chain  forms.  Hence,  non‐reducing  carbohydrates  (e.g.,  sucrose  and 

trehalose) cannot be oxidized with Tollens' or Fehling's reagents, due to missing aldehyde functions 

[19]. Monosaccharides are building units of common polysaccharides, e.g. starch consists of α‐(1,4) 

linked glucose units [19], cellulose consists of β‐(1,4) linked glucose units [20]. 

Oligosaccharides  contain  three  to  10  monosaccharide  units  connected  by  a  glycosidic  bond, 

according  to  IUPAC  [21].  Carbohydrates, with more  than  ten monosaccharide  residues  are  called 

glycans or polysaccharides [21]. Most of the polysaccharides contain between 100 and several 1000 

monosaccharides  and occur  as  linear or branched  chains with  a  reducing  and non‐reducing  chain 

character  [19].  Homopolysaccharides,  e.g.  glycans,  consist  of  only  one  kind  of monosaccharides, 

while  heteropolysaccharides  consist  of  two  or  more  different  kinds  of  monosaccharides  [14]. 

Homopolysaccharides  can  differ  by  the  type(s)  of  glycosidic  linkages,  (α‐  or  β‐configuration  and 

carbon positions) between  the monomers,  as described  for  the disaccharides  [14].  The monomer 

linkages  in  polysaccharides  can  be  homogenous  with  α‐  or  β‐configuration  to  a  single  position. 

Heterogenous  linkages  show  a mixture  of  α‐  and  β‐configurations  and  can  differ  in  the  carbon 

positions  [22].  Heteropolysaccharides  show  the  same  kind  of  linkage  diversity  as 

homopolysaccharides, but they can differ in types and sequences of monosaccharide units as well as 

in  different  types  and  sequences  of  glycosidic  linkages  [14].  Beside  neutral  polysaccharides  (e.g., 

amylose,  amylopectin,  cellulose),  which  compose  only  of  sugar  units,  there  are  also  anionic 

polysaccharides, which  contain  sugar acids  in  their  structure, e.g.,  the galacturonic acid  in pectins 

[14]. For all these reasons, polysaccharides may have an almost unlimited diversity of their structure. 

Monosaccharides with an reducing end can relatively readily interconvert by alkaline isomerization. A 

treatment of pH 11 at 25°C  results  in an equilibrium mixture of  the  starting  carbohydrate and  its 

7 

corresponding C2‐epimer, e.g., D‐glucose can convert  into D‐fructose and D‐mannose  [16, 22]. The 

process of epimerization of two aldoses and the formation of a ketose by enediol rearrangement  is 

called the Lobry de Bruyn‐Alberda van Eckenstein transformation [16, 22]. The  isomerization can be 

induced by base or enzyme and at a much slower rate under acid or neutral conditions [14]. Under 

stronger alkaline conditions (pH>14) and higher temperatures, the combination of epimerization and 

enediol  rearrangements will  take place  along  the whole  carbohydrate  chain  and  result  in  a more 

complex mixture of sugars [16, 22]. Conditions occurring during Kraft pulping also lead to an alkaline 

degradation of polysaccharides and, thus, forming various hydroxyl acids  in addition to formic acid, 

acetic acid, and small amounts of dicarboxylic acids [23]. D‐glucose, for example, can convert into all 

of  the possible aldohexoses and ketohexoses  (D‐allose, D‐galactose, et cetera) and over 100 other 

byproducts [22]. Other derivatives of saccharides are sugar phosphates, deoxy sugars, amino sugars, 

and sugar alcohols. 

 

 

1.2 Functions of carbohydrates 

Carbohydrates are widely distributed  in nature, both  in animal and plant tissues and fulfill different 

roles  [14]. They have various biological  functions: as  structural  support  for cell walls of plants and 

microorganisms and the exoskeletons of  insects and other arthropods (e.g., cellulose, chitin, xylans, 

mannans) [14, 22, 24]. They act as energy storage (e.g., starch, fructans, glycogen) and have different 

regulatory  roles  in physiological processes, e.g., photosynthesis,  sink metabolism, nitrogen uptake, 

defense  reactions,  secondary metabolism,  and  hormonal  balance  [25,  26].  In many  dicotyledons, 

sucrose  is  the  form  of  long‐distance  transport  in  the  phloem  [27],  complemented  by  sorbitol  in 

Rosaceous trees (apples, pears, stonefruits) [28].  

Carbohydrates enable cell‐cell  interaction, receptor binding,  infection, and  immunity responses and 

other signaling effects by biological recognition of carbohydrate moieties [14, 22]. Carbohydrates can 

induce  signaling  effects  directly  by  the molecules  or  indirectly  by  affecting  gene  expression  [29]. 

Oligosaccharides conjugated to protein or lipids are essential components of cell membranes and can 

directly  interact  in the cell to cell recognition and signaling, based on their oligosaccharide moieties 

[14].  Studies  showed  that monosaccharides  (e.g.,  glucose,  fructose,  and mannose),  sugar  alcohols 

(e.g.,  galactinol),  and  disaccharides  (e.g.,  sucrose,  trehalose,  and  sucrose  analogs) might  regulate 

plant metabolism by affecting gene expression [29]. 

Carbohydrates  can  act  as  a marker  and  key‐component  to  reflect  the  plant’s  energy  status  and 

indicating biotic or abiotic stress factors [30]. Abiotic stress factors often induce an accumulation of 

specific carbohydrates, which either protect organisms from temporal changes  in the environment, 

such as varying temperatures, pH, and water supply, or by adaptation and fixation of organisms to a 

8 

specific environmental niche  [22]. Cold  treatments  result  in  the accumulation of glucose,  fructose, 

and sucrose in combination with other osmolytes to protect plant cells by osmotic adjustment or by 

stabilizing membranes and proteins  [31‐33]. Arabinoxylans have also been postulated  to  inhibiting 

intercellular ice formation to enhance winter survival of cereals [14]. On the other hand, heat stress 

induced by local application of high temperature at Vitis vinifera Cabernet Sauvignon clusters results 

in the accumulation of galactinol in berries, which mainly acts as a galactosyl donor for biosynthesis 

of RFOs (raffinose family oligosaccharides)  in plants [29, 34]. There are different reports of drought 

stress effects on the carbohydrate  level at various plant organs. The ribose concentration  increased 

with  increasing water deficit  in  grapevine  leaves  [35].  In berry pulp,  the  level of myo‐inositol and 

sucrose  increase  under  water‐deficit  stress,  which  probably  also  acts  as  osmoprotectants  and 

precursors  for  the  formation of  raffinose  series  sugars  to enhanced drought  stress  tolerance  [36]. 

Carbohydrates  can also act as protective  substances against biotic  stress  caused by  insects,  fungi, 

viruses  or  other microorganisms  [14,  22].  Specific  cell wall  polysaccharides  of  plants  operate  as 

elicitors  of  antibiotics  (phytoalexins),  e.g.,  α‐1,4‐dodecagalacturonide  fragments  of  pectic 

polysaccharides  in  soybean  induce  the  synthesis  of  a  protein  that  inhibits  insect  and  microbial 

proteinases  [14].  Some  general  features  in  the  stress mechanisms  can  be  activated  by  biotic  and 

abiotic stimuli. Therefore, galactinol and RFOs are known to act as signaling units as well as real ROS 

(reactive oxygen species) scavengers [37].  

Beside the functional effect of carbohydrates  in organisms, carbohydrates are substantial resources 

and  feedstocks  used  both  in  food  and  non‐food  industry.  A  distinction  among  digestible  and 

indigestible  carbohydrates  occurs  in  food  industry.  The  digestive  tract  directly  resorb  only 

monosaccharides, e.g., D‐glucose and D‐fructose. Therefore, higher saccharides need to be digested 

before absorption and utilization. The disaccharides sucrose and  lactose, maltooligosaccharides and 

maltodextrins, as well as  the polysaccharide  starch, are  readily digested by humans and used as a 

source  of  calories  and  carbon  [2].  All  other  polysaccharides  are  non‐digestible  [2].  The  most 

abundant  naturally  occurring  carbohydrate  in  food  products  is  starch,  followed  by  pectin, 

hemicellulose  and  cell  wall  materials  [38].  The  non‐digestible  polysaccharide  fraction  acts  as 

stabilizers and dietary fiber in food systems. They can significantly influence functional properties as 

viscosity,  the  stability  of  emulsions  and  foams,  freeze‐thaw  stability,  water‐holding  capacity, 

browning, aroma, flavor, and enable a variety of desirable textures from crispness to smooth or soft 

gels [2, 38].  

Common ingredients with functional properties are food gums, such as carboxymethyl cellulose, guar 

gum, gum Arabic, locust bean gum, methylcellulose, modified pectin, xanthan or naturally occurring 

cell‐wall polysaccharides, such as pectin, cellulose, hemicelluloses and β‐glucans [2]. The stabilization 

of emulsions and avoidance of  ice crystal growth  in  ice cream  is achieved by adding guar gum and 

9 

locust bean gum to the food system [38]. On the other hand, carbohydrate can act as quality markers 

in  the  food  sector,  such  as  lactulose,  maltulose  and  difructose  anhydrides.  Their  formation  by 

isomerization  from  their  corresponding  monomers  under  high  temperature  indicates  heat 

treatments of milk, honey, and coffee, respectively [39‐41].  

In  the non‐food  sector, carbohydrates occur as  feedstock  in  the production of  textiles, paper, and 

plastics. In downstream industries, such as the building and construction industry, carbohydrates are 

used  as  insulation  and  filling material;  the  automobile production  and  the  furniture  industry uses 

processed carbohydrates as upholstery material for furniture and mattresses, et cetera [3, 42]. Also, 

monosaccharides  are  very  versatile  compounds  for many  applications,  such  as modification  and 

chemical transformation to substances, which are primary sources for detergents, emulsifiers, foams, 

and vitamins [3]. Carbohydrates also act as feedstock for bioplastics (e.g., polyurethane, polylactide, 

polyalkenoate), nonionic surfactants (alkyl polyglucosides), natural glues, and adhesives, as a source 

for biofuels (e.g., bioethanol), organic solvents, and fine chemicals, such as (hydroxymethyl)furfural 

and 2,5‐dimethylfuran [43]. 

Equivalent to the wide‐ranging fields of carbohydrate utilization, the analysis of carbohydrates has to 

cover  the demands and  requirements of  the different areas by  the  implementation of appropriate 

methods.  The  determination  of  the  carbohydrate  composition  and  the  amounts  of  specific 

carbohydrates  requires  quantitative  carbohydrate  analysis methods  to  produce  reliable  and  valid 

data to meet the requirements of science and industry. 

 

 

1.3 Complex carbohydrate mixtures 

Carbohydrates  arise  as  simple  or  complex  mixtures.  The  number  of  involved  carbohydrates 

significantly influences the complexity of a carbohydrate‐containing sample. Firstly, a diverse mixture 

of  carbohydrates  can  be  composed  of  different monosaccharides  (pentoses  and  hexoses)  or  can 

consist of mono‐, di‐ and trisaccharides. Secondly, the occurrence of carbohydrates in very different 

concentrations  affects  the  complexity.  The  complete  hydrolysis  of  polysaccharides  can  result  in  a 

mixture  of  different monosaccharides, while  the  analysis  of  soluble  plant  or  fruit  extracts  usually 

present glucose, fructose, and sucrose as mono‐ and disaccharides in high concentrations and other 

carbohydrate  species  in  much  lower  concentrations.  Further  examples  of  complex  plant‐based 

carbohydrate mixtures are extracts or samples of leaves, roots or the whole plant, honey, and paper. 

The  composition  of  carbohydrates  in  plant  extracts  and  honey  mainly  ranges  from  mono  to 

trisaccharides,  while  polysaccharides  occurring  in  cell  walls  comprise  celluloses,  xyloglucans, 

heteroxylans, mannans, and pectins or consist of starch and fructans, which act as reserve material in 

different plant organs [44]. Plant‐based polysaccharides can either have a simple composition of only 

10 

one monomer,  e.g.,  glucans  or  can  compose  of  an  incredibly  complex  construction  of  different 

monosaccharides  and  uronic  acids, which  apparently  differ  between  different  species  and  by  the 

function  of  the  cell  type  [44,  45].  Hydrolytic,  enzymatic,  and  mechanical  processes  can  cleave 

polysaccharides  into  monosaccharides  and  thereby,  enable  the  carbohydrate  analysis  of 

polysaccharides. Polysaccharides are rarely  found  in pure  form, except polymers  in storage organs; 

they often occur as a complex mixture or are available in purified form through processing. 

Processing of carbohydrate‐containing material in paper industry produces pulp bleaching effluents, 

which contain myriad carbohydrates and lignin‐derived compounds in an aqueous matrix with a high 

concentration  of  inorganic  salts  [46].  These  effluents  contain  a  high  number  of  carbohydrate 

degradation  products,  e.g.,  hydroxy  monocarboxylic  acids  and  dicarboxylic  acids  [47].  A  non‐

carbohydrate  fraction  consisting of  fatty acids,  resin acids,  sterols, and  constituents of  tall oil also 

arises during the pulping process and complicates the analysis of this kind of carbohydrate mixtures 

[48]. The  sample origin  interferes with  the analysis of  the carbohydrate composition by additional 

compounds,  which  increase  the  complexity  of  the  sample  matrix  and  commonly  interferes  the 

analysis procedures [49]. 

The  identification  and  quantification  of  carbohydrates  in  a  complex mixture,  either  consisting  of 

monosaccharides  with  equal  molecular  weight  (e.g.,  hexoses)  or  carbohydrates  with  different 

molecular  weights  (mono‐,  di‐  and  trisaccharides)  in  one  sample  or  both  is  a  challenge  for 

chromatographic separation and characterization. 

 

 

1.4 Carbohydrate analysis ‐ State of the art 

The knowledge about the qualitative and quantitative carbohydrate composition in fruits, processing 

materials, and other natural matrices  is essential for controlling and monitoring product properties, 

structure  elucidation,  or  understanding  metabolic  processes.  To  fulfill  the  simple  demands  of 

carbohydrate  analysis,  non‐chromatographic  or  traditional  methods  have  been  developed  to 

measure, e.g., total carbohydrate content or the amount of reducing carbohydrates by physical and 

chemical methods [50]. The increasing demands on a more detailed carbohydrate analysis led to the 

development  of  chromatographic  analysis  techniques,  ensuring  the  identification  of  specific 

carbohydrates,  highly  sensitive  quantification  and  the  possibility  of  determining  unknown 

compounds. 

 

1.4.1 Non‐chromatographic analysis techniques 

Physical  and  chemical  methods  can  measure  the  concentration  of  carbohydrate  solutions. 

Refractometry  is based on the refractive  index of a carbohydrate solution, polarimetry depends on 

11 

the optical rotation of the carbohydrate molecule under polarized light, while hydrometry measures 

the amounts of solutes  in a solution by the density of  liquids [50].  In contrast to physical methods, 

chemical analysis procedures can detect more specific structural properties, e.g., reducing groups or 

distinguishing  among  aldoses  and  ketoses  [50].  Chemical  methods,  so  called  classical  methods, 

determine reducing carbohydrates by reduction of alkaline metal (particular copper) salt solutions to 

the corresponding  free metals or oxides  [50]. These methods  include  the reaction or  titration with 

Fehling’s  solution  (solution  A:  CuSO4,  solution  B:  NaOH  and  Na‐K‐tartrate)  and  slightly  changed 

modifications [50]. The reaction with ferricyanide ([Fe(CN)6]3−) solution and the iodometric titration, 

allow  the determination of aldoses  in  the presence of ketoses  [50]. Chemical methods  to quantify 

carbohydrates in solution also involve colorimetric measurements. Examples are (1) the neocuproine 

method for reducing carbohydrate concentration (orange complex with an absorption maximum at 

457 nm),  (2)  the anthrone method  (blue/green color  in  the presence of carbohydrates  larger  than 

hexoses)  or  (3)  the  Nelson‐Somogyi  method  (molybdenum  blue  measured  at  820  nm).  (4)  The 

phenol‐sulfuric  method,  which  yellow  color  intensity  is  proportional  to  the  total  carbohydrate 

concentration of a solution [50]. Despite the simple application of these methods, the drawbacks are 

their  dependence  on  temperature,  pH  and  carbohydrate  concentration,  time‐consuming  chemical 

reactions and the requirement of a pure carbohydrate solution because contamination with organic 

(proteins, amino acids) and inorganic (metal ions) matter can influence the results [50]. 

Although physical and chemical methods are common practice in many laboratories [50], nowadays, 

there  are  demands  for  more  information  beyond  the  determination  of  the  total  amount  of 

carbohydrates  in  a  sample.  The  research  and  industry  sector  is  exploring  and  processing 

carbohydrate‐based or carbohydrate‐containing materials. It requires information about the identity, 

structural distribution and quantitative evaluation of carbohydrate compounds in a mixture of them 

[50].  Thus,  chromatographic methods with  higher  sensitivity  and  enhanced  selectivity  have  been 

developed  to  meet  the  emerging  needs  of  carbohydrate  analysis.  High‐performance  liquid 

chromatography  (HPLC)  and  gas  chromatography  (GC)  largely  replaced  earlier  chromatographic 

methods  like  paper  chromatography  (PC)  and  thin‐layer  chromatography  (TLC).  However,  older 

methods are still applied, when only qualitative or semi‐qualitative results are required. The use of 

enzymatic  methods  for  determination  and  quantification  of  carbohydrates  are  based  on  two 

different approaches: (1) the breakdown of a substrate and subsequent measurement of the reaction 

products by  chemical or physical methods  as mentioned  above;  (2)  the quantification of  reaction 

products  by  determination  of  the  reaction  rate  or  the  end‐product  of  an  enzyme‐substrate 

interaction  performed  with  electrochemical  or  spectrophotometric  techniques  [50,  51].  The  end 

products of an enzyme‐substrate interaction can be measured directly (e.g., by H2O2 or NADPH) or by 

applying an indicator or dye to form a colored complex, allowing a spectrophotometric measurement 

12 

[52]. These methods are  sensitive,  rapid and  in  theory  specific  to a  certain  carbohydrate, but  the 

specificity depends on enzyme purity  [50]. Most of  these methods are available  in kit  form, which 

facilitates  that  the  correct  concentrations and  reaction  conditions are  compiled  [2]. A well‐known 

enzymatic method for glucose quantification  is the enzymatic conversation of glucose to glucose‐6‐

phosphate and further to 6‐phosphogluconate forming NADPH  in the presence of NADP, which can 

be quantified at 334 nm [51]. The enzyme glucose oxidase produces hydrogen peroxide (H2O2) that 

can be measured by electrochemical oxidation at a platinum electrode [53] or by combination with 

the enzyme peroxidase and a colorless  indicator, which forms a colored complex, known as glucose 

oxidase‐peroxidase method [2]. 

 

1.4.2 Chromatographic analysis techniques 

The primary chromatographic methods for analysis of carbohydrate compounds consist of HPLC‐ or 

GC‐based methods  and  to  a  lesser  extent,  capillary  electrophoresis  (CE)‐based methods  [3].  The 

principle  of  the  three  techniques  comprises  the  separation  of  a  carbohydrate  mixture  and  the 

subsequent detection of  the separated compounds. Dependent on  the sample matrix,  the analysis 

methods  often  involve  an  intense  sample  pretreatment  to  extract  the  target  compounds  from 

interfering  compounds  and  to  clean  them  up  [50].  Beside  these  primary  carbohydrate  analysis 

methods, other techniques as nuclear magnetic resonance spectroscopy (NMR) and matrix‐assisted 

laser desorption ionization coupled with time of flight mass spectrometry (MALDI‐TOF) are available 

and increasingly being used, particularly for qualitative polysaccharide analysis. 

The widely used  separation modes  for  carbohydrate analysis by HPLC are  the  chromatography on 

normal  or  reversed  phase,  followed  by  ion‐exchange  chromatography  particular  with  anionic 

exchange resins and size exclusion chromatography (SEC) [2, 5]. The separation on normal phase  is 

based on stationary phases with  trifunctional amino propylsilane bound  to spherical silica particles 

also  called  hydrophilic  interaction  liquid  chromatography  (HILIC)  [5].  In  case  of  reverse  phase 

separation,  the  column  contains  a hydrophilic  stationary phase made by  adding, e.g.,  longer  alkyl 

chains  (C‐18  column)  or  phenyl  groups  (phenyl  column)  to  the  silica  gel  [2].  Ion‐exchange 

chromatography  is  usually  carried  out  as  anion‐exchange  (AE)  chromatography,  where  a  highly 

alkaline mobile  phase  produces  carbohydrate  anions  by  ionization  of  hydroxyl  groups with  slight 

differences  in  pKa  values,  acting  as  an  efficient  separation  between  low  molecular‐weight 

carbohydrates  [2, 50]. Alternatively,  the cation exchange or  ligand exchange chromatography uses 

different bonding strengths between cis‐glycols of carbohydrates with e.g. Ca2+‐ or Ag+‐loading on the 

column to separate compounds [5, 51]. The SEC covers the applications gel filtration chromatography 

(GFC) and gel permeation chromatography  (GPC), while GFC applies an aqueous mobile phase and 

GPC an organic phase. The separation technique is based on differences in the hydrodynamic volume 

13 

of the carbohydrates and use of resins with particular pore size and structure, which  influence the 

speed of elution of the compounds [5]. 

The  most  common  detector  technique  applied  to  quantify  carbohydrates  subsequent  to  a 

chromatographic  separation without  derivatization  is  refractive  index  (RI)  detection,  followed  by 

pulsed  amperometric  detection  (PAD),  evaporative  light‐scattering  detection  (ELSD)  and  mass 

spectrometry  (MS)  [1, 54]. The RI detector still represents the most widely used detection method 

for carbohydrates  [50]. The advantage of  the universal  technique prevails  for application with  low 

sensitivity  in  the  order  of  1  mg  [51],  but  interfering  solutes  in  the  non‐specific  refractometric 

detection can complicate peak assignment and fail for gradient elution and trace analysis [5, 50]. The 

triple‐pulsed electrochemical detector called PAD (pulsed‐amperometric detector) [2]  is also widely 

applied in carbohydrate analysis, allowing the specific electrochemical detection of carbohydrates by 

direct oxidation of the hydroxyl groups at a gold or platinum electrode  in a highly alkaline medium, 

making  it  suitable  to  being  coupled with  an  AE‐HPLC  [5,  50].  The  ELSD  is  becoming  increasingly 

popular in carbohydrate analysis because the detector provides approximately the same response for 

every  carbohydrate  or  compound  that  is  less  volatile  than  the mobile  phase  [50],  however  it  is 

directly  associated  to  the  concentration  of  the  carbohydrate  [5].  Nebulization  and  evaporation 

techniques  prior  detection  of  non‐volatile  carbohydrates  by  light  scattering  increasingly  replace 

mobile  phase methods,  allow  gradient  elution,  and  are  not  sensitive  to  temperature  changes  or 

variations compared to the mobile phase flow [50]. Among the group of detectors that analyze non‐

derivatized carbohydrates, the PAD provides the highest sensitivity (LODs < 10 µg/L) compared to RI 

and ELSD detectors  [5]. Carbohydrates are  colorless, absorb at  small wavelengths, and  contain no 

ultra‐violet (UV) detectable chromophores. For these reasons, carbohydrates require a pre‐ or post‐

column derivatization to apply UV or fluorescence detectors  in HPLC, which  increase the sensitivity 

markedly but invalidate the most outstanding advantage of HPLC: the avoidance of derivatization [2, 

50]. The pre‐ or post‐column derivatization of hydroxyl or carbonyl groups enable the application of 

photometric  (e.g.,  anthrone  or  phenol  in  sulfuric  acid),  fluorescence  (e.g.,  benzamidine)  or 

electrochemical  detection  of  appropriately  labeled  carbohydrates  [5,  55].  Although  the  detectors 

mentioned  above  show  good  results, MS  is  being  used  increasingly with HPLC,  due  to  enhancing 

qualitative and quantitative carbohydrate analysis and due to the possibility of structure elucidation 

of  unknown  saccharides,  both  at  high  sensitivity  on  picogram  level  [54].  The  coupling  of  liquid 

chromatography  (e.g., AE‐HPLC or  SEC‐HPLC) with  electrospray  ionization mass  spectrometry  (ESI‐

MS) volatilize and ionize compounds prior MS, enabling the analysis of carbohydrate at trace level in 

complex media  [56] or of higher molecular weight  (< 2.000 Da) with a  single quadrupole analyzer 

[57].  

14 

The  capillary  electrophoresis  (CE) uses high  electric  fields  to migrate  ionic  compounds  through  a 

capillary, filled with buffer or gel, to separate the carbohydrates by charge and size, respectively [50]. 

The detection of carbohydrates takes place near the end of the capillary, processed by all types of 

detectors  applied  at  liquid  chromatography.  Spectrometric  detection  (e.g.,  absorbance  or 

fluorescence)  is  the  most  applied  technique,  followed  by  electrochemical  detection  (e.g., 

conductivity) and mass spectrometry (e.g., electrospray ionization, ESI) [1]. Most carbohydrates have 

no charge,  so carbohydrates are converted  into  charged compounds  to move  them  in  the electric 

field; the mobility of carbohydrates is enhanced by complexation using alkaline borate buffer or by a 

highly alkaline electrolyte solution (pH > 12.5) [1, 4].  

The fact, that carbohydrates contain no chromophore or fluorophore moiety requires a pre‐ or on‐

column derivatization of the carbohydrate molecule by suitable chromophores (absorbing  light at a 

specific wavelength)  or  fluorophores  (absorbing  light  and  re‐emit  light  at  a  specific wavelength), 

respectively,  to allow their detection  [1, 50]. The direct UV detection can be performed at 195 nm 

[58]  or  indirect  UV  detection,  carried  out  by  adding  a  strongly  UV‐absorbing  reagent  as  carrier 

electrolyte,  whereby  a  carbohydrate  induces  a  decrease  in  UV  absorbance  [1].  However,  these 

universal  detection modes  are  lacking  selectivity  and  sensitivity  for  carbohydrates  [4].  The  laser‐

induced fluorescence detection of carbohydrate derivatives is more precise due to the measurement 

of their specific wavelength absorbance (e.g., 6‐aminoquinoline at 245 nm) or emission [5, 59]. 

The carbohydrate analysis by gas chromatography (GC) necessitates the derivatization of the highly 

polar and low volatile carbohydrate molecules to increase volatility and stability, e.g., by forming the 

corresponding  methyl,  acetate  or  trimethylsilyl  derivatives  [54].  Nowadays,  the  separation  of 

carbohydrates  is conducted using  fused  silica capillary columns, coated with dimethyl‐polysiloxane 

and diphenyl‐or cyanopropyl‐phenyl‐polysiloxane as the stationary phase and helium or hydrogen as 

mobile phase [60]. The routine detection of carbohydrate derivatives is performed almost entirely by 

flame  ionization detection  (FID) with growing use of  the MS. The MS enables  the  identification of 

carbohydrates through determination of molecular mass and molecule fragmentation patterns [54]. 

The FID  induces  the  formation of  ions by burning samples  in a small hydrogen/air  flame  [61]. This 

process induces the formation of ions, recognized as a small electric current between two electrodes 

which intensity is proportional to the rate of ionization and thereby to the concentration of analytes 

present in the FID detector [61, 62]. An FID can recognize almost every organic compound, only the 

carbohydrate degradation products formaldehyde and formic acid, are non‐detectable [61]. Routine 

quantification analysis of carbohydrates usually relies on GC‐FID, while a GC‐MS method previously 

identified  the  target  compounds  [5].  Chapter  1.5  describes  comprehensively  the  principle  of  gas 

chromatography coupled mass spectrometry (GC‐MS).  

 

15 

1.4.3 GC versus LC 

GC  and  LC  are  the  most  widely  used  chromatographic  techniques  for  carbohydrate  analysis  in 

academic and  industrial  laboratories. The comparison of the respective properties  in context to the 

field  of  application  leads  to  a  selection  of  the  respective  technique.  In  some  cases,  selecting  the 

method  is simple. GC‐MS  is appropriate for very volatile molecules, while LC‐MS  is suitable for very 

polar molecules  and molecules with  high molecular weight  (Fig.  3)  [62].  In  other  cases,  e.g.,  the 

selection between GC‐MS and LC‐MS is more complicated when further information of the analytes 

are  necessary,  due  to  some  general  differences  between  the  two  techniques.  By  comparing,  the 

techniques of GC‐MS and LC‐MS emerge their advantages and disadvantages. 

 

 

Figure 3: The approximate analyte polarity and molecular mass  (Mr) ranges  for GC‐MS, LC‐MS, and 

MALDI  indicating  some  desired  effects  of  chemical  derivatization.  A:  derivatization  for GC‐MS,  B: 

derivatization for ESI and MALDI (EI, CI, ES, MALDI; LD). Source: Reproduced from Zaikin and Halket, 

2009, with permission. 

 

The differences mainly  concern  the operating  temperatures,  column dimensions,  retention mode, 

and  the  available  detection  principles.  LC  analysis  is  predominantly  operated  at  ambient 

temperature, while  GC  analysis  is  conducted  at  elevated  temperature,  either  held  at  a  constant 

temperature  (isothermal) or variable  temperatures determined by a defined  temperature program 

[63].  The  analysis  at  ambient  temperature  favors  the  LC  for  the  analysis  of  thermally  sensitive 

substances. Many LC  instruments have a column heater to ensure defined operation temperatures 

but with a much  lower  temperature  range  compared  to GC. The polyimide  coating of  fused  silica 

columns  limits  the maximum  operating  temperature  of GC  instruments, which  can  be  overcome 

using  stainless  steel‐clad  fused  silica  to  increase  column  temperatures  [64].  Although  it  sounds 

16 

simple that LC uses a liquid mobile phase and GC a gas as carrier phase, this mobile phase is a critical 

factor in the considered methods. LC systems transport any compound, which is soluble in the liquid 

phase [63]. Here, the type of mobile phase  is manifold and comprises solvents of different polarity, 

various pH, and different viscosity.  In contrast to LC, the GC system uses only a highly purified gas, 

which limits this technique to volatile compounds [64]. In addition to the temperature, the polarity of 

the  column  has  a  significant  influence  on  the  separation  efficiency  [64].  The  expansion  of  GC 

application  on  non‐volatile  compounds  requires  at  least  one  derivatization  step  to  enhance  their 

volatility,  which  complicates  sample  preparation  and  limits  the  analysis  of  components  with 

excessive  molecular  weight  [62,  64].  In  contrast  to  GC,  LC  needs  no  derivatization  to  conduct 

carbohydrate analysis, but this results in a lower selectivity of the separations [65]. 

The higher viscosity of liquids compared to gases cause an increased column back pressure in LC and 

results  in shorter columns with wider diameters compared to substantially  longer GC columns with 

thinner diameter [63]. Compared to LC systems, the more extended column of GC systems leads to a 

superior resolution capability and higher compounds sensitivity [64].  

Conventional MS detectors are  capable of being  coupled with  LC and GC, but apply a destructive 

detection principle [63]. LC detection usually bases on nondestructive detection (RI, UV, photodiode 

array  detectors  or  conductivity)  and  allow  a  preparative  separation  step  to  gain  the  analytes  for 

further analysis, except  for MS  [63].  In contrast  to LC, GC detection  is mainly based on destructive 

principles  (FID,  NPD,  and  FPD)  [63].  GC  instruments  also  have  lower  costs  for  acquisition, 

maintenance, and carrier gases compared to LC  instruments and their extensive use of solvents for 

analysis [63]. The coupling with a MS detector is more expensive but allows identifying the structure 

of  the  compounds  and  increases  the  limits  of  analyte  detection  [63].  These  differences  lead  to 

definite advantages and disadvantages and hence prefer each method for different applications.  

LC  systems  have  some  advantages:  dissolving  polar  carbohydrates  in  the  liquid  phase  facilitates 

sample  preparation  and  enable  the  analysis  of  low  molecular  weight  monosaccharides  and 

polysaccharides with high molecular weight. In complex carbohydrate mixtures, e.g., from biological 

origin, carbohydrates occur simultaneously with equal molecular weight, which differ only  in  littlle 

with  regrad  to molecular  structure.  The  analysis  of  these mixtures  requires  a  system  with  high 

separation efficiency and high sensitivity  to  identify and quantify  the significant carbohydrates and 

also  carbohydrates  in  the background, which may  act  as  a biomarker with  controlling  function of 

physiological processes. The higher separation capability and high sensitivity, combined with a  fast 

and highly  accurate quantitative  analysis  lead  to  a preference  for GC  systems over  LC  systems  to 

analyze complex carbohydrate mixtures.  

The  low volatility of carbohydrate  limits the analysis by GC mainly at mono‐, di‐ and tri‐saccharides 

[5].  A  study  evaluated  more  than  40  publications  to  compare  the  results  of  simultaneous 

17 

carbohydrate and acid analysis by HCPLC and GC systems  [65]. The study showed  that GC systems 

provide  a  better  selectivity,  higher  sensitivity,  and  allow  the  separation  and  quantitation  of  the 

substantially higher amount of compounds on the same column with the same detector at lower cost 

in contrast to HPLC [65]. 

 

 

1.5 Derivatization of carbohydrates for GC‐MS analysis 

The GC‐MS technique requires volatile components to conduct the separation of analytes and mass 

spectrometric analysis of the analytes. Thus, a derivatization procedure  is necessary for compounds 

with  a high polarity  and  a high molecular weight  to overcome  the  related  very  low  volatility  and 

thermal instability [10]. The application of high temperatures at low volatility compounds in GC, e.g., 

carbohydrates would lead to a thermal decomposition before analysis [10]. Functional groups cause 

the high polarity of compounds, in case of carbohydrates and associated degradation products, these 

are predominantly hydroxyl groups and carboxyl groups, especially sugar acids and  low molar mass 

carboxylic  acids  [10, 66]. The hydroxyl  and  carboxyl  groups  contain  active hydrogens, which  form 

intermolecular hydrogen bonds and thus, reduce the volatility [10]. Hydrogen atoms can also cause 

strong interactions with the stationary phase and induce chromatographic peak tailing, which lower 

the chromatographic  resolution and  reduce  the signal‐to‐noise  ratios of  the corresponding analyte 

peaks  [62]. Carbonyl groups of aldehydes and ketones do not cause  fundamental difficulties  in  the 

GC‐MS analysis  [66]. However,  the derivatization of carbonyl groups can contribute  to  the analyte 

stability,  reduce  the  polarity  due  to  the  reduction  of  their  hydrogen  bonding  capacity  and  thus, 

improve  the  peak  geometry  by  reduction  of  interactions  with  the  stationary  phase  [66].  Most 

importantly, the derivatization of the carbonyl group in ketoses and aldoses hinders the formation of 

open chain forms and thereby decrease the number of isomers of reducing carbohydrates. There are 

many  other  reasons  derivatizations  beside  increasing  the  volatility,  reduction  of  polarity,  and 

improving  stability.  Derivatization  procedures  are  also  conducted  to  improve  chromatographic 

properties  to  separate  closely  related  compounds,  improve  the  sensitivity  and  selectivity  of 

compound  detection  and  to  increase  the  structural  information  content  of  mass  spectra  for 

improved  compound  identification  [10,  66].  Nevertheless,  derivatization  procedures  have  some 

disadvantages:  the applications are  time‐consuming, may  lose analytes during sample preparation, 

can  cause undesirable  side  reactions between  sample  components and  can  lead  to discrimination 

effects.  Applying  derivatization  procedures  often  require  concentration  and  drying  prior  step 

derivatization.  These  pre‐derivatization  steps  also  lead  to  a  change  from  an  aqueous  to  the  non‐

aqueous solvent system, which might affect the sample  integrity condensation reactions of reactive 

components. 

18 

The  classical derivatization methods  for  carbohydrates  involve  the  substitution of active‐hydrogen 

atoms  in polar groups with  less or non‐polar  substituents  to  increase  compound  volatility  [6, 67]. 

Although the selection of derivatization reagents for carbohydrates is mainly based on the exchange 

of active hydrogens in the molecule, there are more demands on derivatization reagents than merely 

substitute functional groups in carbohydrate molecules. Furthermore, the kind of reagents, catalysts, 

solvents, and by‐products involved in derivatization reaction influence whether the reaction mixture 

can  be  directly  injected  onto  GC  column  or  if  a  cleaning  step  is  necessary  prior  injection  [11]. 

Therefore, a derivatization reagent should produce a stable derivative and should not cause a  loss, 

any  rearrangements or  structural  alterations of  carbohydrate  compounds during  formation of  the 

derivative [11]. 

Derivatization of carbohydrates  for GC‐MS analysis  is conducted prior  injection with the distinction 

between  single‐step,  two‐  or  multi‐step  derivatization  strategies.  The  single‐step  derivatization 

methods are alkylation  (e.g., methylation), acetylation  (e.g.,  trifluoroacetylation), silylation and  the 

formation  of  cyclic  derivatives  are widely  used  in  the  analysis  of  polyalcohols  and  non‐reducing 

carbohydrates by GC‐MS [9‐11]. However, the application of single‐step derivatization procedures at 

reducing  carbohydrates  challenges  the user:  the  theoretical existence of up  to  five  tautomers per 

carbohydrate, e.g., an aldohexose can form α‐ and β‐pyranose, α‐ and β‐furanose, and an open‐chain 

isomer, which can retain after some derivatization strategies [68, 69]. For this reason, two‐ or multi‐

step  derivatization  strategies  have  been  developed  to  reduce  the  number  of  isomer  peaks  for 

carbohydrate  analysis  by  GC‐MS  and  lead  to  an  improvement  of  chromatographic  resolution  at 

constant or increased informational value of mass spectra [67]. Two‐step carbohydrate derivatization 

strategies usually convert the carbonyl group into a specific derivative as the first step, which inhibits 

the  intermolecular  conversion  between  the  different  tautomers  of  a  carbohydrate  [10].  Possible 

reactions aiming at the carbonyl group of carbohydrates are (1) the formation of the corresponding 

primary  alcohols  (alditols)  through  reduction  and  (2)  the  oximation  by  using  pure  or  substituted 

hydroxylamine.  The  second  reaction  step  derivatizes  the  available  hydroxyl  groups  of  the 

carbohydrate by procedures described as single‐step derivatization methods and forms, e.g., in case 

of (trifluoro)acetylation the respective alditol acetates or oxime acetates. A significant advantage of 

two‐step  derivatization  strategies  is  the  formation  of  only  one  or  two  derivatives  for  each 

carbohydrate, which  enhances  the  chromatographic  behavior. However,  the  application  faces  the 

user with new problems, such as higher workload by processing additional derivatization steps, and, 

as reported for alditol acetates, that different carbohydrates can form the same derivative [67]. 

 

 

 

19 

1.5.1 Single‐Step derivatization procedures 

The  most  commonly  used  derivatives  in  single‐step  derivatization  procedures  for  carbohydrate 

determination  are  methyl  ethers,  acetates,  trifluoroacetates,  and  trimethylsilyl  ethers  [9].  The 

description  of  the  following  single‐  and  multi‐step  derivatization  methods  include  the  history, 

reaction, applied derivatization reagents, recent usage as well as advantages and drawbacks. 

 

1.5.1.1 Alkylation (especially methylation) 

The  alkylation  (or  arylation)  is  a  standard derivatization method  for  the  replacement of  an  active 

hydrogen atom  in a hydroxyl group  (R‐OH) by an alkyl  (e.g., methyl) or aryl  (e.g., perfluorobenzyl) 

group, which result in the formation of ethers (R‐O‐R´) [10]. The derivatization method also replaces 

hydrogen atoms both at primary amines (R‐NH2) of amino sugars and carboxyl groups (R‐COOH) of 

sugar acids and react the latter to esters (R‐COO‐R) [10].  

The alkylation eliminates interfering effects of hydroxyl or carboxyl groups and is commonly applied 

to  carbohydrates,  polyols,  and  acids  in  the  presence  of  various  catalysts  [10,  66].  The  procedure 

reduces  the  polarity  and  forms  products  with  higher  thermal  stability  compared  to  the  starting 

compounds [11, 70]. Within alkylation procedures, the per‐methylation of carbohydrates is the most 

widely applied method in GC‐MS analysis, because the methylation leads only to a small increase in 

molecular mass  of  the  analyte  [10,  70].  The  per‐methylation  combined with  a  hydrolysis method 

followed by a second derivatization step (e.g., deuterated alkyl, alkyl, aralkyl or as their fluorinated 

analogs) allow the linkage analysis of polysaccharides for structure elucidation [10, 70]. 

In  1903,  Purdie  and  Irvine  reported  the  first  derivatization  of  non‐anomeric  hydroxyl  groups  of 

carbohydrates  into alkyl ethers, accomplished with methyl  iodide (CH3I) and silver oxide (Ag2O) as a 

catalyst  in methanol  (Fig. 8)  [71].  In 1915, Haworth described  the combination of dimethyl  sulfate 

and sodium hydroxide (NaOH) as derivatization reagents with silver oxide to produce methyl ethers 

of carbohydrates [72]. Unfortunately, carbohydrates are susceptible to oxidation  in the presence of 

silver oxide and higher  temperatures, which  limits  the method  in GC‐MS analysis  to  low molecular 

ethyl or alkyl carbohydrate derivatives [73]. Despite this disadvantage, the two methylation reactions 

were used for almost 50 years without modification [74]. The introduction of dimethylformamide as 

reaction solvent by Kuhn et al. in 1955 and replacement of Ag2O as a catalyst by barium oxide (BaO) 

allowed the application  to a broader variety of carbohydrates  [75, 76]. However, the drawbacks of 

extremely  long reaction times  (several days or even weeks) to reach a complete permethylation of 

carbohydrate  and  undesirable  side  reactions  remained  from  the  previous  two  methylation 

procedures [73]. First, the per‐methylation method of Hakomori in 1964 led to a breakthrough in the 

handling  and  the  implementation  of  methylation  as  a  routine  analysis  method  [77].  Hakomori 

introduced  the  methylsulfinyl  carbanion  (CH3SOCH2‐),  also  called  "dimsyl"  ion,  by  reaction  of 

20 

dimethylsulfoxide  (DMSO)  and  sodium  hydride  (NaH)  [10,  73].  As  a  strong  base,  the  dimsyl  ion 

deprotonates all active hydrogens of carbohydrates, and methyl iodide subsequently methylates the 

carbohydrate.  [10,  73].  The Hakomori  procedure  has  replaced  the  older methylations  procedures 

immediately  and  was  adopted  as  superior  methylation  reagent  for  carbohydrates  and 

polysaccharides  [70,  73,  74]. Nowadays,  a  reagent mixture  for  the methylation  of  carbohydrates 

consists of a simple combination of finely milled NaOH and CH3I in DMSO at room temperature [73, 

78].  The  very  hygroscopic  characteristic  of  solid  powdered  sodium  hydroxide,  if  added  in  excess, 

allows the capture of traces of water from the sample [10]. 

 

 

Figure 4: Etherification reaction of glucose using methyl iodide and silver oxide as a catalyst to form 

permethylated glucose (2,3,4,6‐tetra‐O‐methyl‐D‐glucopyranose) [71]. 

 

Recent studies propose a  large number of reagents  for the preparation of alkyl derivatives: among 

them are alkyl  (e.g., methyl, ethyl, propyl) and benzyl bromides or  iodides and their substituted or 

fluorinated analogs [10]. The methylation procedure of carbohydrates is most often performed with 

methyl  iodide (CH3I), followed by dimethyl sulfate (CH3O‐SO2‐OCH3) and dimethyl carbonate (CH3O‐

CO‐OCH3)  [62].  Reagents  such  as  methyl  triflate  (CF3O‐SO2‐OCH3)  and  methylfluorophosphonate 

(FSO2‐OCH3)  show  a higher  reactivity, but  their higher  toxicity  reduces  the  frequency of use  [62]. 

Other common derivatization reagents to conduct alkylation are boron trifluoride (BF3)  in methanol 

or butanol, dialkyl acetals, and diazoalkales; but they are not suitable for carbohydrate analysis due 

to inefficient methylation of hydroxyl groups [11]. BF3 in methanol or butanol is most commonly used 

for  acids  to  form  their  corresponding  methyl  or  butyl  ester,  while  dialkyl  acetals,  such  as 

dimethylformamide (DMF), react quickly with carboxylic acids, phenols, and thiols [11]. 

The methylation with the reagent diazomethane (DAM, N2CH2) is the fastest and cleanest method to 

form methyl esters, and  leads to the most reproducible results  ‐ but  it  is carcinogenic, highly toxic, 

and an explosive gas and the synthesis of DAM has to be done shortly before derivatization [11, 79]. 

The use of trimethylsilyl‐diazomethane (TMS‐DAM) overcomes the drawbacks of DAM, because it is 

less toxic, non‐explosive, tolerates water, and commercially available [80‐82]. The acetylated alditol 

section (1.6.2.1) describes the application of permethylation for carbohydrates as part of the partially 

methylated alditol acetates (PMAAs) derivatives. 

21 

The  advantages  of  alkylation  procedures  comprise  enhanced  volatility,  improved  gas 

chromatographic properties,  and  the detection  limit of derivatives  at picomole  level  [10, 70]. The 

significant  advantage  of  the  methylation  reaction  compared  to  other  alkylation  and  arylation 

derivatizations is based on the low mass increment of 14 mass units per substituted hydrogen atom, 

which allow analysis of a high number of hydroxyl groups containing compounds, e.g., oligomer, via 

GC‐MS  [70].  The  low  mass  increment  engaged  the  derivatization  method  for  the  analysis  of 

polysaccharides  consisting  of  up  to  eleven  monosaccharide  units  by  application  onto  high 

temperature qualified capillary columns [70]. Some reagents, e.g., diazomethane, allow a rapid and 

straightforward conversion of carbohydrates  to methyl derivatives  in  less  than one minute even  in 

aqueous  systems  without  the  need  of  an  excessive  clean‐up  [10,  70].  Some  researchers  prefer 

methylation  regents  over  silylation  reagents  to  examine  the  carbohydrate  analysis,  due  to  the 

increased sensitivity of methyl ethers and the fact that the fragmentation spectra of per‐methylated 

carbohydrates reveal detailed information of the monosaccharides’ linkage and branching position in 

polysaccharides [70]. 

Significant  disadvantages  of  alkyl  derivatives,  e.g., methyl  ethers  of  carbohydrates,  are  that  the 

reaction  forms different  tautomeric  isomers, which  result  in complex chromatograms and  that  the 

derivatization procedure with  some  reagents  requires  time‐consuming preparation  steps  [67].  So, 

the need of at least one purification step during the preparation of per methyl ethers can complicate 

the quantitation of some derivatives  [70]. A drawback of alkylation and  (arylation)  reagents  is also 

that  they  are  often  toxic  and  require  severe  reaction  conditions,  ranging  from  strongly  acidic  to 

strongly  basic  and  associated  with  high  reaction  temperatures,  which  could  lead  to  thermal 

degradation of non‐derivatized carbohydrates  [11, 67]. Another reported problems associated with 

the methylation procedure  is  the occurrence of overmethylation, which  leads  to  ions  in  the mass 

spectra  that  are  30  mass  units  higher  than  those  of  fully  methylated  carbohydrate  [70]. 

Overmethylation occurs because the permethylation reagent can form small amounts of iodomethyl 

ether, which  competes with methyl  iodide  for  the  reaction  of  hydroxyl  groups  and  result  in  the 

formation  of  a methoxymethyl moiety  instead  of  a methyl  group, which  causes  the  higher mass 

increment [83]. 

 

1.5.1.2 Acylation 

The  acylation  process  defines  the  introduction  of  an  acyl  moiety  (R‐C=O)  into  a  molecule  via 

substitution  of  active  hydrogen  [10].  Almost  all  active  hydrogen  atoms  of  functional  groups 

(e.g., ‐OH, ‐SH and ‐NH) can substituted by acyl groups, except carboxylic, sulphonic, and phosphoric 

acids  [10]. Any carboxylic acid  residue can be used as derivatization  reagent, e.g., acid anhydrides, 

acyl halides, and acyl amides [10]. Acylation strongly increases the volatility of polar compounds and 

22 

raises  the  chromatographic mobility of derivatives  [10]. Hence,  the application of  fluorinated acid 

residues has  proven  to produce  highly  volatile derivatives  and  lead  to  good  gas  chromatographic 

properties  [10].  Some  acyl  amides,  e.g.,  bis(trifluoroacetamide)  (BTFA)  and  N‐methyl‐

bis(trifluoroacetamide)  (MBTFA)  are  to  be  preferred  over  acid  anhydrides  (e.g.,  acetic  anhydride, 

trifluoroacetic  anhydride)  and  acyl  halides.  Acyl  amides  are  very  active  and  do  not  produce 

undesirable by‐products (e.g., HCl) [10]. Acid by‐products have to removed during reaction injection 

by  carrying out  the  reaction  in  solvents  such  as pyridine, which  catches  acid by‐products or by  a 

cleaning step after the reaction [11, 62]. Due to these reasons, perfluoro acylating reagents are the 

most widely used for derivatization reactions. The reagents lead to a substitution of active hydrogen 

atoms by trifluoroactetyl (TFA) or higher perfluoroacetyl groups [10]. In general, the temperatures of 

acylation reactions range from 20°C to 150°C for 15 minutes to 3 hours [10, 62]. The derivatization 

process of carbohydrates takes place at elevated temperatures (60 – 80°C) for 2‐3 hours [10]. Basic 

compounds  are  usually  involved  in  acylation  reactions  to  bind  HCl  or  carboxylic  acids  produced 

during the reaction [10]. 

The application of carbohydrate acetates for GC analysis was  introduced  in 1961, while Vilkas et al. 

reported in 1966 a high sensitivity and good resolution of trifluoroacetylated carbohydrates [84‐86]. 

The  most  common  reagents  for  alkylation  derivatization  are  fluorinated  anhydrides, 

fluoracylimidazoles,  N‐methyl‐bis(trifluoroacetamide)  (MBTFA),  pentafluoropropanol  (PFPOH)  and 

pentafluorobenzoyl chloride (PFBCI) [11]. Within these reagents, MBTFA is particularly recommended 

for  carbohydrate  analysis  [87, 88],  although  this  reagent  shows only  slow  reactivity with hydroxyl 

groups and thiols in contrast to high reactivity at primary and secondary amines. However, the major 

benefit beside the mild reaction conditions are relatively inert and non‐acidic by‐products during the 

derivatization reaction [89]. These by‐products do not cause any damage to the capillary column of 

the  GC  in  contrary  to  fluorinated  anhydrides, where  the  by‐products  have  to  be  removed  prior 

analysis [10, 11]. 

 

 

Figure  5:  Acylation  of  a  hydroxyl  group  using  N‐methyl‐bis(trifluoroacetamide  (MBTFA)  reagent, 

according to [10]. 

 

Advantages  of  acetylation  derivatives  are  their  chemical  and  thermal  stability,  while  N‐acetyl 

derivatives are more stable against hydrolysis reactions compared to corresponding N‐trifluoroacetyl 

derivatives  [10, 67]. Trifluoroacetates are  the most volatile groups  for carbohydrate derivatization, 

23 

while the volatility of derivatives decreases further from trimethylsilyl ethers to acetates [67, 90]. The 

high  volatility  of  trifluoroacetates  allows  lower  analysis  temperatures  and  the  application  of 

carbohydrates with a wide  range of molecular weight  [67]. Both acetates and  trifluoroacetates of 

carbohydrates produce  complex  EI mass  spectra, but  their derivatives usually  show no M+• peaks 

[91]. The drawbacks of acylated derivatives are the occurrence of different tautomeric carbohydrate 

isomers during derivatization  reaction and  the high moisture‐sensitivity of derivatization  reagents. 

[62, 67]. Thus, any traces of water have to be avoided.  

 

1.5.1.3 Silylation 

The  silylation process defines  the  introduction of a  trimethylsilyl group  into a molecule, mainly by 

substitution of an active hydrogen atom or casually by replacing the metal component of a salt [10, 

11].  Silylation  reagents  allow  the derivatization of nearly  all  functional  groups  that  contain  active 

hydrogens,  such  as  alcohols, phenols,  sugars,  amines,  thiols,  steroids,  and  carboxylic  acids, which 

otherwise can cause problems in the GC analysis [92, 93]. The reactivity of functional groups towards 

silylation  decreases  in  the  following  order:  alcohol  >  phenol  >  carboxyl  >  amine  >  amide,  and 

decreases for alcohols in the following order: primary > secondary > tertiary [9].  

Thus,  the  silylation  reagents  react with  both  alcohols  and  acids  to  form  trimethylsilyl  ethers  and 

trimethylsilyl  esters  respectively  [11].  Silylation  reagents  decrease  the  polarity  of  molecules  by 

reducing  hydrogen  bonding  and  thereby  increase  the  volatility  and  stability  of  the  silylated 

derivatives [94, 95]. Silylated derivatives show an excellent chromatographic behavior by eluting as 

narrow and symmetrical peaks [11, 95]. 

The  first  silylation  derivatization  of  carbohydrates was  published  by  Sweeley  et  al.  in  1963.  The 

derivatization  used  hexamethyldisilazane  (HMDS)  as  silylation  regent  in  pyridine,  catalyzed  by 

trimethylchlorosilane  (TMCS)  [96]. HMDS  is a weak TMS donor and reacts only with easily silylated 

hydroxyl groups [11]. The authors also reported on the application of separating carbohydrates by GC 

for  the  first  time  [96]. This was  the  starting point  for  the  research on  silylation of  carbohydrates, 

followed by a large number of books, publications, and reviews concerning this topic [68, 93, 95, 97‐

102].  

Common  derivatization  reagents  are  hexamethyldisilzane  (HMDS),  trimethylchlorosilane  (TMCS), 

(trimethylsilyl)‐imidazole  (TMSI),  N,O‐bis(trimethylsilyl)‐acetamid  (BSA),  N,O‐bis(trimethylsilyl)‐

trifluoracetamid  (BSTFA), N‐methyl‐N‐(trimethylsilyl)‐trifluoracetamid  (MSTFA),  and  the  two  rarely 

applied  reagents  N‐trimethylsilyl‐diethylamin  (TMS‐DEA)  and  N‐methyl‐N‐t‐butyl‐(dimethylsilyl)‐

trifluoroacetamide (MTBSTFA) [11, 62].  

The reactivity of silylation reagents is affected by the solvent system (e.g., pyridine) and the addition 

of catalysts, which can accelerate the reaction of hindered  functional groups  in secondary alcohols 

24 

and  amines  [11].  Pyridine  is  the most widely  used  solvent  for  silylation,  due  to  its  ability  as  acid 

scavenger and function as a catalyst [11]. Silylation reagents are often used in mixture with 1 to 10 % 

TMCS or TMSI, which act as a catalyst in the reaction, as a solvent for the derivation reaction as well 

as  an  injection  solvent  for  gas  chromatography  [11,  62].  The  operational  conditions  of  silylation 

reactions depend on the applied reagent and the target molecule. Commonly these reactions were 

carried out at  reaction  temperatures between 60°C and 80°C and  reaction  times up  to 30 minutes 

[62]. Hindered functional groups, e.g., secondary alcohols and amines can require long‐term heating 

for up to 16 hours [103]. 

N,O‐Bis(trimethylsilyl)trifluoroacetamide (BSTFA) was introduced by Stalling et al. [104] and became 

a standard derivatization  reagent due  to  the ability  to  react with all common protic sites. The  two 

derivatization  by‐products  trimethylsilyl‐trifluoroacetamide  and  trifluoroacetamide  generated  by  a 

further derivatization step, are highly volatile, an advantage over alternative silylation reagents [93]. 

 

 

Figure  6:  Silylation  reaction  of  a  hydroxyl  group  using  N,O‐bis(trimethylsilyl)trifluoroacetamide 

(BSTFA), according to Gerhke and Leimer [105] and Tallent and Kleiman [106]. 

 

TMCS is often added to BSTFA as a catalyst in a proportion of 1% to 10% to increase the TMS donor 

strength, but produces hydrochloric  acid  as  a by‐product  [9].  TMCS  can  also  act  as derivatization 

reagent, often  applied  in  combination with pyridine  as basic  auxiliary  and HCl  trap, but  also with 

other  trialkylsilyl  halides  [107].  The  frequent  addition  of  pyridine  to  BSTFA  and  other  silylation 

reagents based on the already mentioned catalytic capability (Fig. 11) and the  increased stability of 

the silylated products in its presence [108, 109]. The reagent 4‐(dimethylamino)pyridine (DMAP) also 

acts  as  a  catalyst  (Fig.  12)  in  derivatization  reactions  and  shows  in  combination with  pyridine  a 

reduction of side reactions during silylation or acylation reactions at ambient temperatures [110]. 

 

 

Figure 7: The catalytic function of pyridine for silylation of hydroxyl groups, according to Held et al. 

[111]. 

 

25 

N

NCH3H3C

+

N

N

CH3H3C

Si(CH3)3 Cl

+

N

N

CH3H3C

R‐OH, Pyridine

N

+

ClH

4‐(Dimethylamino)‐pyridine

Cl(H3C)3Si

Trimethyl‐chlorosilane

OR Si(CH3)3

Pyridinehydrochloride

4‐(Dimethylamino)‐pyridine  

Figure 8: The catalytic reaction of 4‐dimethylamino pyridine (DMAP) for silylation of hydroxyl groups, 

according to Chaudhary and Hernandez [112]. 

 

The  gas  chromatographic  properties  of  carbohydrates  as  TMS  derivatives  have  been  studied 

intensively  at  aldopentoses  [98],  aldohexoses  [102],  ketohexoses  [68],  and  disaccharides  [93]. De 

Jongh  et  al.  [97]  extensively  examined  the  EI mass  spectra  of  silylated  carbohydrates  and  their 

fragmentation mechanism, reviewed by Kochetkov and Chizov [113]. 

Recent  studies  showed  the  conversion of  carbohydrates  into  corresponding TMS derivatives  for  a 

wide range of samples. These  involve the  identification and quantification of neutral carbohydrates 

in environmental  samples  [54];  the  simultaneous analysis of carbohydrates,  sugar acids, and  sugar 

alcohols [114‐117]; analysis of hydrolysis products of oligo‐ and polysaccharides [118‐125]; and the 

analysis of degradation products of carbohydrates [126, 127]. The silylation procedure also showed 

good  results  for  the analysis of phenols  [128, 129] and amino acids  [104, 105, 130]. The  silylation 

procedure  improved  the modification of  the original method  for production of  PMAA  to perform 

linkage analysis of polysaccharides, which involves a methanolysis step to produce methyl glycosides 

instead of the borohydride reduction step and a trimethylsilylation step in place of acetylation [131]. 

Silylation derivatives have many advantages such as higher volatility,  lower polarity, and  thermally 

higher  stability  compared  to methyl  ethers  or  acetates  [67].  The  silylation  procedure  generates 

derivatives  with  excellent  chromatographic  properties  and  mass  spectral  characteristics  with 

distinctive fragmentation pattern, which is necessary for the elucidation of molecular structures [10]. 

A  large number of different derivatization reagents are available, allowing the application to a wide 

range of areas [67]. 

A significant disadvantage of the silylation procedure  is that reducing sugars can usually form up to 

five  isomers per monosaccharide due to anomer formation and pyranose‐furanose  interconversion: 

an  α‐  and  β‐pyranoside,  an  α‐  and  β‐furanoside,  and  an  open‐chain  form.  Each  isomer  shows 

different physical properties and elutes separately in GC analysis [68, 69]. In the worst case, the eight 

available hexoses may form up to 40 configurational  isomers and can appear as a separate peak  in 

the chromatograms. Although the quantification of a monosaccharide is possible by a well‐separated 

isomer  peak,  a  high  number  of  isomer  peaks  increase  the  risk  of  peak  overlapping,  rise  the 

complexity  of  chromatograms,  and  impede  a  definite  carbohydrate  identification  especially  in 

26 

complex mixtures [131, 132]. The mass spectra of isomeric TMS derivatives at the same species, e.g., 

hexoses,  contain  no  specific  m/z  ions,  only  the  relative  abundance  of  the  available  ions  can 

discriminate among isomers [70].  

A further disadvantage  is that silylation reagents are moisture sensitive, and therefore the samples 

have  to  be  completely  dry  [67].  The  per‐trimethylsilylation  reaction  of  carbohydrates  is  a more 

straightforward procedure than a corresponding per‐methylation, but the mass increment of a TMS 

group with 72 g/mol is significantly higher than that of a methyl group (14 g/mol) [70]. A combination 

of HMDS/TMCS  in pyridine  form ammonium chloride, which precipitates and can contaminate  the 

capillary column, if no cleaning step takes place prior GC injection [93]. Alternative silylation reagents 

such as BSTFA can solve this drawback and allow direct injection of the reaction mixture into the gas 

chromatograph without further cleaning steps and no fear of column contamination [93]. BSTFA also 

reacts faster and more completely than BSA, due to the presence of the trifluoroacetyl group [11]. 

Tert‐butyldimethylsilyl derivatives can be up to 10,000 times more resistant to hydrolysis than their 

corresponding TMS and TBDMCS derivatives [11].  

 

1.5.1.4 O‐Isopropylidenation 

The O‐isopropylidenation,  also  called  acetonidation,  uses  acetone  as  a  derivatizing  agent  to  form 

cyclic  derivatives  in  the  presence  of H2SO4  and  copper  sulfate  as  catalysts  [133]. One  of  the  first 

studies about the formation of O‐isopropylidene (ISP) derivatives of carbohydrates was conducted by 

Freudenberg et al. in 1928 [134]. The mass spectra and fragmentation pathways of O‐isopropylidene 

derivatives of pentoses and hexoses were discussed by De Jongh and Biermann in 1964 [135], further 

examined by Morgenlie [136, 137]. 

The  O‐isopropylidenation  of  carbohydrates  occurs  with  anhydrous  acetone  at  37°C  mixed  with 

sulfuric  acid  and  anhydrous  copper  sulfate  as  catalysts  for  20  hours  [133].  In  general,  the  O‐

isopropylidenation  produces  five‐membered  cyclic  ring  systems  at  cis‐vicinal  diol moieties, which 

carry  two  adjacent  hydroxyl  groups  in  syn‐configuration  for  the  reaction with  acetone  [22].  The 

furanose  ring  formation of D‐glucose contains  two of  these diols moieties at positions 1 and 2, as 

well as at position 5 and 6. Thus,  the O‐isopropylidenation of glucose produces  the corresponding 

1,2:5,6‐di‐O‐isopropylidene‐D‐glucofuranoside  (Fig.  13),  one  of  the  best‐known  isopropylidene 

derivatives  [22].  In  contrast,  D‐fructose  gives  two  O‐isopropylidene  derivatives:  2,3:4,5‐di‐O‐

isopropylidene‐D‐fructopyranoside and 1,2:4,5‐di‐O‐isopropylidene‐D‐fructopyranoside [22]. 

 

27 

 

Figure  9:  O‐Isopropylidenation  of  α‐D‐glucofuranose  form  1,2:5,6‐di‐O‐isopropylidene‐D‐

glucofuranose, according to [22, 133, 135]. 

 

Different modifications of  the original methods are available, e.g.,  replacing copper  sulfate by zinc 

chloride,  reaction  temperatures  at  20°C  or  shorter  reaction  times  [22,  137].  The  fragmentation 

pattern and chromatographic separation characteristics of  ISP aldoses [136],  ISP hexuloses, and  ISP 

pentuloses [137] reported studies in the 1980s. A study about the analysis of a carbohydrate mixture 

obtained  from  a  formose  reaction  (a  triose  aldol  condensation  reaction)  also  applied  an  O‐

isopropylidenation reaction [137]. 

The advantages of O‐isopropylidene derivatives over acetates and  trimethylsilyl ethers are  (1)  the 

derivatives are  very  sensitive  in mass  spectrometry and produce unique mass  spectra and  (2)  the 

derivatization  form  structural  different  monosaccharide  with  no  appearance  of  configurational 

isomers, resulting in less‐complex gas chromatograms [135‐137]. So, the chromatographic properties 

of  O‐isopropylidene  derivatives  allowed  the  complete  separation  of  a  carbohydrate  mixture 

containing  fucose,  arabinose,  xylose,  rhamnose,  galactose,  glucose,  and  mannose  [136].  The 

proportional relation between peak area and molar concentration allowed the use of the parent O‐

isopropylidene  derivative  of  aldoses  for  the  quantitative  analysis  of  carbohydrate mixtures  [136]. 

Another significant advantage of ISP derivatives is the introduction of a very low mass increment per 

hydroxyl group (20 g mol‐1 ) compared to trifluoroacetylation (96 g mol‐1) and trimethylsilylation (72 g 

mol‐1) [138].  

Despite  these  advantages,  O‐isopropylidene  derivatives  have  not  proven  to  be  suitable  for  the 

analysis of complex mixtures of monosaccharides, explained by the  long reaction time (> 10 hours) 

for the ISP derivatization of carbohydrates and the dependence of the reaction conditions for some 

monosaccharides  on  the  sample  composition  [136].  The  isopropylidenation  process  forms 

thermodynamically stable structures, but acid hydrolysis conditions can easily remove isopropylidene 

groups [22]. The O‐isopropylidenation of common monosaccharides usually leads to the formation of 

one  (e.g.,  glucose)  or  two  (e.g.,  fructose)  primary  products,  but  the  treatment with  acetone  and 

28 

sulfuric acid can  lead  to  the  formation of additional small peaks. However,  the small peaks do not 

interfere  with  the  identification  of  O‐isopropylidene  carbohydrate  derivatives.  They  show  higher 

retention  times  than  2,3‐5,6‐di‐O‐isopropylidene‐D‐mannofuranose, which  is  the O‐isopropylidene 

derivative with the highest retention within the common aldoses [136]. 

 

1.5.2 Multi‐step derivatization procedures 

The  formation  of  carbohydrate  anomers  can  prevented  by  (1)  reducing  neutral  carbohydrates  to 

alditols, (2) forming aldonitriles or (3) oximes prior trimethylsilyl or trifluoroacetyl derivatization [7]. 

These  derivatization  procedures  consist  of  two  or  more  reaction  steps  to  produce  the  desired 

derivatives. The  trimethylsilyl oximes and ethers are possibly  the most used derivatives  for  the GC 

analysis of carbohydrates [13]. 

 

1.5.2.1 Alditol acetates  

The formation of alditol acetates involves the reduction and acetylation of monomeric carbohydrates 

in successive steps, which eliminate the possibility of anomeric derivatives and increase the volatility 

the carbohydrate derivative [9]. The alditol acetate method was first described in 1961 by Gunner et 

al.  [84]  and modified by  various  researchers,  such  as  Sawardeker  et  al.  [139]  and Blakeney  et  al. 

[140].  

The original derivatization procedure (Fig. 14) involves sodium borohydride (NaBH4) reduction of the 

carbohydrate to an alditol as the first step [84]. The formed alditols are present as open chain, which 

is a significant difference  to many other single step derivatization methods, where  the equilibrium 

mixture between  the  ring  forms  and  the open  chain  form  are  retained  after derivatization  [141]. 

According  to  the original method,  several evaporations with acidic methanol volatilize  the borate, 

formed during  reduction, as methyl borate and  removes  it  [84]. This  step  is necessary as hydroxyl 

groups containing borates lead to interaction with the acetylation reagent and form complexes [84, 

141].  The  third  step  of  the  original method  carries  out  the  acetylation  of  the  hydroxyl  groups  of 

carbohydrates  by  acetic  anhydride  (Ac2O)  in  the  presence  of  pyridine  [84,  141].  The  further 

procedure includes the analysis of derivatives by GC [84]. 

 

29 

Figure 10: Formation of alditol acetate derivatives, according to the original method described by 

Gunner et al. [84] and applied by Urbani et al. [142]. 

 

Different  modifications  of  the  original  alditol  derivatization  method  solve  problems  of  the 

application,  such as adding DMSO  to  increase  the  stability of  reduction mixture or  the addition of 

methylimidazole as a catalyst, which allows the acetylation without removal of water or borate [140, 

143].  Sodium  acetate  as  catalyst  improved  the  acetylation  reaction  with  acetic  anhydride  prior 

derivatization reaction [141]. The high moisture content of the samples complicates the exchange of 

acetic anhydride by trifluoroacetylation reagents [144]. Alditols, which the sample already contains, 

can be distinguish from aldoses or ketoses after derivatization by replacing sodium borohydride with 

sodium borodeuteride, whereby the newly formed alditol carry one deuterium in the molecule [145]. 

Other  modifications  append  hydrolysis  procedures  for  the  analysis  of  polysaccharides  prior  the 

reduction step [141]. The reduction of carbohydrates eliminates the anomeric center in the molecule 

by  forming primary alcohols with an open chain structure and after acetylation of  the alditol, only 

one chromatographic peak is showed per aldose [67]. However, the reduction step of ketoses forms 

secondary alcohol and introduces another asymmetric center into the carbohydrate molecule, which 

can cause a significant loss of information for certain kinds of carbohydrate mixtures [9].  

A drawback of the method  is that the reduction of specific carbohydrates forms the same alditol or 

more than one alditol. (1) Different aldoses produce the same alditol, e.g., both D‐arabinose and D‐

lyxose produce D‐arabitol, while D‐glucose and D‐gulose produce D‐glucitol (sorbitol) and L‐glucitol, 

which are not separable by GC with regular capillary columns. (2) The reduction of specific aldoses 

and  related  ketoses  produce  the  same  alditol,  which  prevents  the  determination  of  the  origin 

carbohydrate, e.g., both glucose and fructose produce glucitol. (3) The reduction converts all ketoses 

into approximately equal amounts of the corresponding R‐ and S‐alditol  isomer, which can result  in 

two  individual  peaks  of  the  reducing  carbohydrate  origin,  e.g.,  fructose  form  the  two  C2‐epimers 

glucitol and mannitol [9, 101, 141, 146]. The alditol acetate method is widely used for the analysis of 

carbohydrates, e.g.,  in bacteria  [147], potatoes  [148] and  the analysis of monomer composition of 

polysaccharides  from  plant  cell‐walls  [140,  149]  and  wood  pulps  [150].  Further  studies  of 

30 

polysaccharides  applied  the  alditol  acetate method  for  the  analysis  of  linkage  positions  between 

monosaccharide  rings  in  carbohydrate  polymers  [151].  In  the  first  step  of  the  analysis  of  linkage 

positions,  all  hydroxyl  groups  in  a  polysaccharide  are  per‐methylated,  then  the  molecule  is 

hydrolyzed  to obtain monosaccharides with additional  free hydroxyl groups at  the  sites of  linkage 

[151]. The received monosaccharides are reduced to alditols and finally acetylated, e.g., with acetic 

anhydride  to  produce methylated  alditol  acetates,  also  known  as  PMAAs  [70,  151].  The  GC‐MS 

analysis  of  the  monosaccharide  derivatives  indicates  the  position  of  methylated  and  acetylated 

hydroxyl groups and allows conclusions of the original linkage positions in the polymer [151]. 

The primary advantage of the alditol acetates method is the formation of only one peak per aldose, 

which reduces the complexity of chromatograms and enhances the limit of detection for quantitative 

measurements  of  these  carbohydrates  [141].  The  derivatives  are  extraordinarily  stable  and  allow 

post derivatization and cleaning procedures [141]. The method enhances the separation of complex 

carbohydrate mixtures  and  allows  a  subsequent  hydrolysis  reaction  of  polysaccharides  for  sugar 

composition analysis [67]. 

The significant disadvantages of the methods are the production of the same alditol from different 

carbohydrates and  that  the  reduction of each ketose  results  in a mixture of  two alditols  [67]. The 

reduction  process  eliminates  the  problem  of  anomeric  derivatives  at  aldoses  but  creates  new 

allocation  problems  in  ketoses  for  the  identification  and  quantification  of  carbohydrates  [9]. 

Fortunately,  the  fixed  proportion  of  fructose  conversion  to  glucitol  and  mannitol  allows  a 

reproductive quantification of this carbohydrates in samples [148]. Carbohydrate derivatives such as 

uronic  acids,  ulosonic  acids  (e.g.,  Kdo),  or  4‐aminosugars  react  in  the  presence  of NaBH4  to  non‐

volatile sodium salts and require a modification either by dehydration to  lactones or by converting 

the carboxyl group to an alcohol prior reduction [141]. Drawbacks of the method are also that a large 

number of processing steps for derivatization and cleaning steps for removing of by‐products before 

analysis, which makes  the  procedure  time‐consuming  and  increases  the workload  [67].  Similar  to 

other derivatization reagents, acylation reagents are moisture sensitive [67, 141]. Studies observed 

that  the  alditol  acetate  method  is  not  suitable  for  trace  analysis,  due  to  the  appearance  of 

contaminations  from  side  reactions  during  the  derivatization  process  and  their  occurrence  as 

background  peaks  in  the  chromatogram  [141].  The  development  of  other  derivatization methods 

such as aldononitrile acetate or O‐methyloxime acetates overcomes  these  limitations and produce 

different  derivatives  for  aldoses  and  ketoses,  which  allow  the  correct  determination  of  the 

carbohydrate origin [146, 152].  

 

 

 

31 

1.5.2.2 Per‐acetylated aldononitriles (PAAN) 

The elimination of the carbohydrates’ carbonyl group (anomeric center) avoids formation of various 

carbohydrate isomers during derivatization. One possibility is the formation of aldose oximes and the 

subsequent  dehydration  to  their  per‐acetylated  aldononitriles  (PAAN),  also  called  aldononitrile 

acetates. 

The first PAAN derivatives of a carbohydrate were probably prepared by Lance and Jones in 1967 by 

producing and separating PAAN derivatives of methyl ethers from the hydrolyzates of D‐xylans [153]. 

The  formation of PAAN derivatives  is only possible  for aldoses because  the derivatization process 

requires the unique 1‐position of carbonyl group at aldoses to produce a non‐isomeric nitrile group 

[9, 67]. 

According to the original method (Fig. 15), the first step involves the conversion of a carbohydrate’s 

aldehyde group  to  the  corresponding oxime by addition of hydroxylamine hydrochloride  (NH2OH  ∙ 

HCl) and pyridine as  solvent and catalyst  for 30 min at 90°C  [152, 153].  In  the  second  step, acetic 

anhydride  is  added  in  excess  to  simultaneously  dehydrate  the  oxime  to  nitrile  and  acetylate  all 

available hydroxyl groups of carbohydrates and water molecules at a reaction temperature of 90°C 

for 30 min [67, 152]. 

 

 

Figure 11: Derivatization reaction of the formation of per‐acetylated aldononitriles (PAAN) from α‐D‐

glucopyranose (aldoses), according to [152], modified. 

 

The added catalyst  in  the  first  step  is usually acting  throughout  in  the  second  reaction  step  [152]. 

Ketoses  react  to  their  corresponding oximes  in  the  first  step, but not  to PAANs due  to  the above 

mentioned structural demands for nitrile formation [67]. Applying other derivatization methods can 

convert  the  hydroxyl  groups  of  ketoximes,  e.g.,  by  conversation  to  trimethylsilyl  derivatives  [67]. 

Modifications of the original method based on replacing pyridine by 1‐methylimidazole as a catalyst 

[154], 4‐(dimethylamino)pyridine (DMAP) [155] or sodium acetate decrease the reaction time of the 

acetylation  process  to  less  than  10 minutes  [152].  Another  modification  uses  hydroxylamine‐O‐

sulfonic acid  to convert aldononitriles by  silylation  to  the corresponding  silylaldolnitrile derivatives 

[7].  

32 

PAAN  derivatives  have  been  used  to  analyze  soluble  carbohydrates  of  various  origins,  e.g.,  plant 

tissue  [156],  capsular  and  extracellular  polysaccharides  from  bacteria  [157],  and  neutral 

monosaccharides  of  polysaccharide  gums  in  food  [158].  The  quantitative  analysis  of  eight 

monosaccharides  aldononitrile  acetates  derivatives  obtained  from  a  Lyceum  barbarum  L.  extract 

reported satisfying results [159]. 

The primary advantages of PAAN derivatives are that equal to the alditol acetates method only one 

unique  product  arises  for  each  aldose,  but  allows  a  faster  derivatization  procedure  and  that  the 

occurrence  of  water  does  not  influence  the  reaction  [67,  152].  Compared  to  TMS  derivatives, 

aldononitrile acetates derivatives showed higher stability and better recovery rate [67]. 

Drawbacks  of  the  PAAN  derivatization method  are  that  the  reaction  is  unable  to  convert  other 

carbohydrates than aldols because of ketoses form only oxime derivatives [9, 67]. Although studies 

showed  the simultaneous quantitation of aldoses and ketoses  [159, 160],  the PAAN method  is not 

suitable for quantitative analysis of real samples, where aldoses and ketoses are present at the same 

time  [67]. Studies  reported  that  the  reaction of  the oxime with acetic anhydride  could  result  in a 

mixture  of  the  nitrile  and  glycosylamine  [152].  However,  glycosylamine  occurs  regardless  of  the 

reaction conditions and does not interfere with the analysis [152]. So, this method does not entirely 

solve the demands of obtaining a single derivative for each monosaccharide [9]. 

 

1.5.2.3 TMS‐Oximes 

The TMS‐oximation reaction of carbohydrates  involves a  two‐step derivatization strategy,  involving 

(1)  the  conversion  of  the  carbonyl  group  into  oximes  and  the  subsequent  (2)  derivatization  of 

hydroxyl groups with a silylating reagent (Fig. 16) [138]. 

 

 

Figure 12: General reaction scheme for the derivatization of carbonyls to oximes. 

 

The  oximation  reaction  results  in  two  peaks,  the  corresponding  syn  (Z)  and  anti  (E)  forms  of  the 

oxime  per  reducing  sugar,  which  remain  after  the  silylation  procedure  (Fig.  17)  [138].  The  per‐

trimethylsilylated carbohydrates show up to five peaks per compound [68, 102] and the isomer ratio 

of carbohydrates to have to equilibrate prior TMS derivatization [161]. Non‐reducing carbohydrates 

(e.g.,  sucrose,  trehalose,  raffinose)  cannot  be  oximated  because  of  the  absent  aldehyde  or 

hemiacetal function and show only one peak, the per‐trimethylsilyl derivative [67]. 

33 

  

Figure  13: Derivatization  of  glucose:  (1)  converting  glucose  in  the  corresponding  syn‐  and  anti‐O‐

methyloxime  isomers by methoxyamine hydrochloride (CH3ONH2  ∙ HCl) and subsequently trimethyl‐

silylation. 

 

Sweeley et al.  introduced  the oxime derivatives of carbohydrates  for GC‐MS analysis  in 1963  [96]. 

The original derivatization method of aldoximes by unsubstituted hydroxylamine was later modified 

by O‐alkyl‐hydroxylamine to form O‐alkyloxime [146, 162].  

The oximation  reaction  converts  the  carbonyl group  (C=O) of  reducing  carbohydrates  to an oxime 

(C=N‐O‐R). The derivatization of  the stereoisomeric center  inhibits  the cyclization and  results  in an 

open structure molecule [69]. Different oximation reagents are available: O‐hydroxylamine (R:  ‐OH) 

[163], O‐methylamine (R: ‐CH3) [69], O‐ethylamine (R: ‐C2H5) [49] and O‐benzylamine (R: ‐C6H6) [69]. 

The application of the oximation reagents occurs as the respective hydrochloride salts, because they 

enable a higher oximation performance due to the higher nucleophilicity of the amine and the higher 

informational value of the respective mass spectra compared to the unsubstituted forms [163, 164]. 

The  oximation  reaction  converts  the  anomeric  carbon  atom  of  the  carbonyl  group  into  a 

stereoisomeric center that forms the corresponding syn and anti isomers [138]. The identification of 

the  syn  and  anti  isomers  supported  by NMR  provided  information  about  the  peak  ratio.  The  syn 

formation is the predominant isomer of aldopentoses and aldohexoses, representing the major peak; 

while the minor peak with approximately 10‐30% of the corresponding major peak area represents 

the anti  isomer, but ketohexoses and 2‐deoxyhexoses form almost similar areas of the syn and anti 

peak [165‐167]. 

Previous  studies  investigated  the  oximation  followed  by  trimethylsilylation  derivatization  of 

carbohydrates only at a small number of reference standards or modifications of the derivatization 

method,  respectively:  Ox‐TMS  vs. MeOx‐TMS  [12,  13],  TMS  vs. MeOx‐TMS  [115].  Other  studies 

compared O‐methyl, O‐ethyl and O‐benzyl oxime  trimethylsilylated keto and aldo acids  [168] or a 

34 

formose mixture  of  carbohydrates  [69].  A  study mentioned  that  O‐alkoxyamines  are  superior  to 

hydroxyamines for identification purposes, but the authors published no related data [164]. 

The  primary  advantage  of  the  reducing  sugars  conversion  to  an  oxime  prior  hydroxyl  group 

derivatization is the formation of only two isomers instead of up to five isomers compared to other 

derivatization methods  [67,  146,  164].  The  oximation  decreases  the  number  of  chromatographic 

peaks, which  reduces  the  complexity  of  the  chromatogram.  The  occurrence  of  two  isomers  per 

reducing  carbohydrate  also  enhanced  the  sensitivity of  analysis  compared  to  single‐step  silylation 

procedure based on  the  increased  signal  intensity  [164, 169].  In  contrast  to  the alditol acetate or 

aldononitrile  acetate  method  the  oximation  procedure  different  syn‐  and  anti‐isomers  of  both 

aldoses  and  ketoses  [67].  Further  advantages  are  the  increasing  volatility  of  carbonyl‐containing 

compounds  and  the  ability  to  protect  keto  acids  (e.g.,  2‐oxoacids  and  4‐oxoacids)  from 

decarboxylation [146, 168, 170].  

The disadvantages of the oximation derivatization are the  instability of aldoxime derivatives at high 

temperatures, and hydrolysis condition as well as the moisture sensitivity of the silylation reagents, 

which  require  an  entirely  dried  sample  [67].  High  temperatures  can  cause  decomposition  of  the 

oxime  into  the  corresponding  nitriles  [67].  The  use  of N‐alkoxy‐oximes  instead  of  aldoximes  can 

overcome  this  disadvantage  since  they  are  thermally  more  stable.  So,  N‐alkoxy‐oximes  enable 

additional chemical modifications, e.g.,  the  silylation of  the oxime derivative without  the  fear  that 

analytes decompose into nitriles during the derivatization reaction or GC injection, respectively [66]. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

35 

2 Material and Methods 

2.1 Derivatization methods  

2.1.1 Ethyloximation‐trimethylsilylation (EtOx‐TMS) 

The carbohydrate samples were lyophilized and dissolved in 200 µl of anhydrous pyridine, containing 

40 mg/ml O‐ethylhydroxylamine  hydrochloride  and  1 mg/ml methyl  β‐D‐galactopyranoside  as  an 

internal standard. The mixture was heated for one hour at 70°C. Subsequently, 200 µl of a solution 

containing  1.5 mg/ml  4‐(dimethylamino)pyridine  (DMAP)  in  anhydrous pyridine was  added  to  the 

mixture,  followed  by  200  µl  of  N,O‐bis(trimethylsilyl)trifluoroacetamide  (BSTFA),  containing  10% 

trimethylsilyl  chloride  (TMCS).  The  mixture  was  heated  for  two  hours  at  70°C.  The  derivatized 

samples were kept at ‐20°C until analysis.  

Prior  injection,  the derivatized  samples were diluted with ethyl acetate  (600 µl) and  filtered  (PTFE 

membrane, 0.45 µm, 13 mm diameter). Aliquots of 0.2 µl were introduced into the splitless injector 

with an Agilent GC Sampler 120 (Agilent Technologies, USA). All standards, chemicals, and reagents 

were purchased  from Sigma‐Aldrich‐Fluka  (Sigma‐Aldrich, Schnelldorf, Germany);  they were of GC‐

grade purity and used without further purification.  

The GC‐MS analysis was performed on an Agilent 7890A gas chromatograph coupled with an Agilent 

5975C mass selective detector. The analysis conditions were as  followed: column: HP‐5MS  (30m × 

0.25mm  x  0.25µm;  J&W  Scientific,  Folsom,  CA, USA);  carrier  gas:  helium,  injector:  280°C;  column 

flow: 0.9 ml/min; purge flow: 32.4 ml/min, 0.6 min; oven program: 50°C (2 min), 5°C/min, 280°C (20 

min); MS: EI mode, 70 eV, source pressure: 1.13∙10‐7 Pa, source temperature: 230°C. The scan range 

was set from 50 to 950 Da.  

 

2.1.2 Other derivatization methods 

The GC‐MS analysis of the products of the following derivatization procedures were carried out using 

the same GC‐MS analysis conditions as applied to the ethoximation‐silylation approach, mentioned 

above. Only the scan range for ISP was set from 35 to 950 Da.  

 

2.1.2.1 Per‐trimethylsilylation (TMS) 

A defined amount of freeze‐dried grapevine leaves (10 mg), a crude product of the formose reaction 

(50 µl) or reference compound (1 µg to 100 µg) was used for analysis. The sample was dissolved  in 

200 µl  pyridine, which  contained  75 µg/ml methyl‐β‐D‐galactopyranoside  as  internal  standard  (IS) 

and 300 µg/ml of DMAP as a silylation catalyst. The derivatization was accomplished by adding 200 µl 

of BSTFA and heating the mixture to 70°C for two hours. The derivatized samples were kept at ‐20°C 

until further analysis. 

36 

2.1.3 O‐isopropylidenation (ISP) 

An  aliquot  of  0.1 ml  of  an  internal  standard  solution with  13C‐glucose  in  50%  aqueous methanol 

(2 µg/µl) was added  to  the sample or  reference compound. Subsequently,  the mixture was  freeze‐

dried  and  stored  in  vacuum over P4O10  at  room  temperature  for  two days. Then,  a  spatula  tip of 

anhydrous copper‐(II)‐sulfate was added to the sample prior dissolution  in 625 µl dry acetone. The 

derivatization  procedure was  started  by  adding  7.5  µl  of  concentrated H2SO4  to  the  sample.  The 

reaction was stopped by adding 80 mg of anhydrous sodium carbonate. 

 

2.1.4 Oximation reactions followed by trimethylsilylation (TMS) or trifluoroacetylation (TFA) 

Benzyloximation  (BO):  Similar  to  O‐ethoximation;  200 µl  of  a  solution  containing 

40 mg/ml O‐benzylhydroxylamine hydrochloride in pyridine was added for the production of BO‐TMS 

derivatives; the half amount of O‐benzylhydroxylamine hydrochloride (20 mg) was added for BO‐TFA. 

The reaction mixtures were kept at 75°C for 40 min. 

Trimethylsilylation (TMS) of oximated samples (EO‐TMS, BO‐TMS): BSTFA (200 µl) was added to the 

solution of oximated derivatives containing the sample or reference, the  internal standard, and the 

catalyst  (see  the EtOx‐TMS method). The mixture was kept at 70°C  for  two hours. After cooling  to 

ambient temperature, ethyl acetate was added to the solution and stored at ‐20°C. 

Trifluoroacetylation (TFA) of oximated samples (EO‐TFA, BO‐TFA): 50 µl (EO‐TFA) or 80 µl (BO‐TFA) of 

N‐methyl‐bis(trifluoroacetamide)  (MBTFA)  were  added  to  the  solution  of  oximated  derivatives 

containing  the  sample  or  reference,  the  internal  standard,  and  the  catalyst  (see  the  EtOx‐TMS 

method).  The  mixture  was  kept  at  70°C  for  two  hours.  After  cooling  to  ambient  temperature, 

pyridine was added to the solution and cooled down to 4°C. 

The derivatized samples were diluted with ethyl acetate (TMS: 880 µl; BO‐TMS: 665 µl; ISP: 567.5 µl) 

or pyridine  (EO‐TFA and BO‐TFA: 665 µl)  to a  final volume of 1.2 ml and  filtered  (PTFE membrane, 

0.45 µm, 13 mm diameter) prior injection.  

 

 

2.2 Hydrolysis reactions for polysaccharide analysis 

2.2.1 Sulfuric acid hydrolysis 

The sulfuric acid hydrolysis was conducted with sample amounts of 40 mg (±1 mg) per polysaccharide 

and cellulose sample  for analysis. The sulfuric acid hydrolysis was conducted using a  two‐step acid 

treatment with different concentrations according to Bose et al. (2009), modified for GC‐MS analysis 

[171]. During  the primary hydrolysis  step, 1.5 ml of 72% H2SO4 was added  to  the  sample at  room 

temperature  for  two  hours.  For  the  second  hydrolysis  step,  2 ml H2O was  added  to  the  primary 

hydrolysis mixture to obtain a 40% H2SO4 concentration and subsequently heated in an oven at 80°C 

37 

for one hour. The hydrolysis solution was cooled down in an ice bath and stored at 4°C overnight to 

polymerize the lignin fraction. Then, 7 ml of an internal standard solution (150 mg Sorbitol / 100 ml 

H2O) were  added  to  the  hydrolysis  solution.  An  aliquot  of  1.5 ml was  neutralized  using  sodium 

carbonate  (Na2CO3)  until  no  more  bubbles  appeared  (approx.  290  mg).  The  liquid  was  filtered 

(0.45 µm, 13 mm diameter) into a new GC vial and the pH was checked with pH‐paper and adjusted 

to pH 7 with one or two drop(s) of acetic acid. 

 

2.2.2 Acid methanolysis 

The samples for acid methanolysis were freeze‐dried prior treatment, then 1‐2 mg of polysaccharides 

or  10 mg  (±2 mg)  of  cellulose  samples were  used  for  analysis.  The  dried  sample materials were 

depolymerized  through  acid  methanolysis  by  addition  of  2 ml  2 M  HCl  in  anhydrous  methanol 

according to Sundberg et al. (1996) [121]. The samples were kept at 100°C for three hours. When the 

reaction mixtures reached ambient temperatures, the samples were neutralized with 100 µl pyridine. 

Afterwards, 1 ml of an  internal standard solution (0.1 mg sorbitol/ml methanol) was added to each 

sample. A calibration solution (1 ml) containing 0.1 mg/ml of sugar monomers and uronic acids was 

also subjected to the acid methanolysis under same conditions. The sugar monomers consisted of D‐

(‐)‐arabinose, D‐(+)‐galactose, D‐(+)‐glucose, D‐(+)‐mannose, L‐(+)‐rhamnose (6‐deoxy‐mannose) and 

D‐(+)‐xylose. The uronic acids consisted of D‐(+)‐galacturonic acid monohydrate (GalA), D‐glucuronic 

acid  (GlcA)),  and  4‐O‐methyl  glucuronic  acid  (4‐O‐MeGlcA).  Acid  methanolysis  samples  and 

calibration mixtures were evaporated in a water bath (50°C) under nitrogen stream until dryness and 

further dried in a vacuum desiccator at room temperature for 30 min. 

 

2.2.3 Per‐trimethylsilylation (TMS) of monosaccharides obtained from hydrolysis reactions 

The  lyophilized  hydrolysates  obtained  from  sulfuric  acid  hydrolysis  or  acid methanolysis  reaction, 

calibration mixtures and  reference  compounds were dissolved  in 200 µl pyridine and  incubated at 

room  temperature  for  30 min  to  ensure  proper  isomer  equilibration.  Subsequently,  200  µl  of  a 

solution  of  1.5 mg/ml  DMAP  as  a  silylation  catalyst  in  pyridine  was  added  to  the mixture.  The 

derivatization was  accomplished  by  adding  200 µl BSTFA  (containing  10%  TMCS)  and  heating  the 

mixture  to 70°C  for  two hours according  to Becker et al.  [172]. The derivatized samples were kept 

at ‐20°C until analysis. 

 

2.2.3.1 GC‐MS analysis of TMS‐derivatized hydrolysis products 

The derivatized sulfuric acid hydrolysis or acid methanolysis products were diluted with ethyl acetate 

(600 µl) and  filtered prior  injection. Aliquots of 0.2 µl derivatized sample were  introduced  into  the 

38 

splitless injector using an Agilent GC Sampler 120. The GC‐MS analysis was performed on an Agilent 

7890A gas chromatograph coupled with an Agilent 5975C mass selective detector. 

The general GC‐MS analysis conditions were as followed: Column HP‐5MS (30 m × 0.25 mm x 25 µm; 

J&W Scientific, Folsom, CA, USA); carrier gas: helium, MS: EI mode, 70 eV, source pressure: 1.13∙10‐7 

Pa, purge  flow: 36.3 ml/min, 0.6 min;  source  temperature: 230°C.  Scan  range was  set  from 43  to 

950 Da.  

Parameters  for  analysis  of  acid methanolysis  products:  injector  temperature:  140°C  (30°C/min  to 

260°C); column  flow: 0.9 ml/min; oven program: 140°C  (1 min), 4°C/min  to 210°C,  then 30°C/min, 

260°C (5 min); inlet pressure 78.361 kPa.  

Parameters for analysis of sulfuric acid hydrolysis products: injector temperature: 150°C (30°C/min to 

260°C); column  flow: 0.9 ml/min; oven program: 120°C  (2 min), 5°C/min  to 230°C,  then 20°C/min, 

260°C (10 min); inlet pressure 78.361 kPa. 

 

2.2.4 Solid‐State NMR 

 The  Solid‐State NMR  experiments were  performed  on  a  Bruker  Avance  III HD  400  spectrometer 

(resonance  frequency of  1H of 400.13 MHz, and  13C of 100.61 MHz,  respectively), equipped with a 

4 mm dual broadband CP/MAS probe. The samples were swollen in deionized water overnight before 

measurement. 13C spectra were acquired by using the TOSS (total sideband suppression) sequence at 

ambient temperature with a spinning rate of 5 kHz, a cross‐polarization (CP), a contact time of 2 ms, 

a recycle delay of 2 s, SPINAL 64 1H decoupling and an acquisition time of 43 ms. Chemical shifts were 

referenced externally against the carbonyl signal of glycine with δ = 176.03 ppm. The acquired FIDs 

were apodized with an exponential  function  (lb = 1 Hz) before Fourier  transformation. Peak  fitting 

was performed with the Dmfit program [173]. Deconvolution fitting and assignment of fractions were 

conducted according to the model and method of Larsson et al. (1997) [174]. 

 

2.2.5 GPC analysis of cellulose samples 

The characterization of samples was carried out by measuring weight‐average molecular mass (Mw). 

The paper samples were dissolved  in N,N‐dimethylacetamide containing 9%  lithium chloride  (w/v). 

The measurement was performed on a GPC system consisted of a multi‐angle  laser  light scattering 

detector  (Wyatt  Dawn  DSP with  argon  ion  laser  [λ0  =  488  nm]),  and  a  refractive  index  detector 

(Shodex RI‐71). The  separation was carried out on a  set of  four PLgel mixed‐ALS columns  (20 µm, 

7.5×300 mm, Varian/Agilent)). N,N‐dimethylacetamide  containing  0.9%  lithium  chloride  (w/v) was 

used as mobile phase. The system was operated at a flow rate of 1.0 ml/min with an injection volume 

of 100 µl. Data evaluation was performed with standard Chromeleon 4, Astra 4.73, and GRAMS/32 

software packages. 

39 

2.3 Analysis of the degree of acetylation in biopolymers 

2.3.1 Method 1: Direct liquid‐phase analysis 

2.3.1.1 Sample preparation 

Paper or pulp  samples were  cut  into  small pieces  (approx. 1–2 mm2). 50 mg of  the  carbohydrate 

polymer sample was transferred  into a 1.5 ml vial. The vials were sealed with 1.3 mm silicon/PTFE 

septum  crimp  caps. 700 µl  sodium methanolate  solution  containing one µl/ml  toluene was added 

through the septum. After 5 min reaction time, the sample was ready for analysis. 

 

2.3.1.2 GC–MS conditions for the analysis of the liquid phase 

The GC–MS analysis was performed on an Agilent 7890A gas chromatograph equipped with a Gerstel 

PTV inlet, coupled with an Agilent 5975C mass selective detector. 

The GC‐MS analysis conditions were as  followed: column: HP‐5MS  (30m × 0.25 mm  i.d. × 0.25 µm 

film  thickness;  J&W Scientific, Folsom, CA, USA); glass  liners packed with quartz wool  for  the PTV‐

inlet were obtained from Gerstel (Mülheim an der Ruhr, Germany); carrier gas: helium; column flow: 

1.5 ml/min; purge flow: 50.0 ml/min, 0 min; PTV inlet temperature program: 40°C (2 min), 20°C/min 

to  100°C  (0  min),  then  25°C/min  to  270°C  (6  min);  oven  temperature  program:  30°C  (2  min), 

10°C/min  to 50°C  (0 min),  then 25°C/min  to 270°C  (2 min), MS: EI mode, 70 eV,  source pressure: 

1.13*10−7 Pa at 230°C. Quadrupole temperature: 150°C, transfer  line: 200°C; MS data acquisition  in 

scan and single ion‐mode (SIM): scan range: 45 – 400 amu; SIM: m/z 60 (acetic acid), m/z 74 (methyl 

acetate) and m/z 65  (toluene) at 100 ms dwell time each. Aliquots of 0.3 µl were  injected  into the 

PTV inlet with an Agilent GC Sampler 120. The GC‐MS as well as the autosampler were controlled by 

Agilent MSD Chemstation E.02.01 (Agilent Technologies, Santa Clara, CA, USA). 

 

2.3.2 Method 2: Analysis via gas phase 

2.3.2.1 Sample preparation 

Approximately  2 mg  of  cellulose  or  paper  sample was  transferred  into  a  1.5 ml  vial  for  analysis. 

4‐O‐(13C2‐acetyl) vanillin was selected to generate isotopically labeled methyl13C2‐acetate in situ as an 

internal standard. The synthesis was carried out according  to  [175]. The  internal standard solution 

consisted of 10 mg of  the  13C2‐labeled  internal  standard dissolved  in 4 ml of anhydrous methanol. 

50 µl of anhydrous methanol and 20 µl of the internal standard solution were added to each sample. 

300 µl of  sodium methanolate was  transferred  to every  sample before  the vials were  sealed with 

1.3 mm silicon/PTFE septa crimp caps. The samples were  immediately ready for analysis by GC‐MS. 

Hemicellulosic samples were prepared in analogy to paper and cellulose samples except using 150 µl 

of anhydrous methanol, 40 µl of  the  internal standard solution and 180 µl of sodium methanolate 

solution. 

40 

2.3.2.2 GC–MS conditions for the analysis by SPME 

The SMPE‐GC–MS analysis was carried out on an Agilent 7890A gas chromatograph coupled with an 

Agilent 5975C  triple axis mass  selective detector  (MSD). The GC was equipped with  two  inlets, an 

Agilent multimode as well as an Agilent split‐/splitless inlet. Bleed and temperature optimized septa 

(Agilent part no. 5183‐4757) were used for covering both  inlets. The analysis of methyl acetate was 

applied  on  a  VF‐WaxMS  column  (30 m  ×  0.25 mm  i.d.  ×  0.25  µm  film  thickness;  J&W  Scientific, 

Folsom, CA, USA), connected to the multimode inlet on one end and the MSD on the other.  

A screening of common SPME  fiber  types was conducted and  revealed  that  the CAR/PDMS  (black) 

fiber enabled the highest number of injections. The following optimized SPME‐parameters have been 

applied  (briefly  explained):  agitator  temperature:  35°C  (temperature  of  the  agitator  used  to 

equilibrate  the  sample  vial);  incubation  time: 20 min  (time used  for equilibrating  the  sample  at  a 

specific agitator temperature without enrichment); the enrichment time has been adopted acording 

to the methyl acetate concentration of the sample to prevent the fiber from being overloaded (time 

used to expose the fiber into the sample headspace for analyte enrichment); desorption time: 12 min 

(time  used  to  expose  the  fiber  into  the  inlet  to  facilitate  analyte  desorption);  conditioning  time: 

10 min (time used to expose the fiber at a specific conditioning temperature for preparing the fiber 

for the subsequent analyte enrichment). 

The split‐/splitless inlet was used for the fiber conditioning before enrichment. A 5 m × 0.25 mm i.d. 

fused silica capillary was connected on one end to the split‐/splitless inlet to ensure that a moderate 

helium carrier gas pressure was build‐up. The other end just protruded into the GC‐column oven. The 

multimode‐inlet was  operated  under  the  following  conditions:  constant  column  flow:  0.9 ml/min 

using helium carrier gas, purge  flow: 60.0 ml/min  (2.54 min);  injector  temperature: 90°C constant. 

Oven temperature gradient profile: 30°C  (1 min), 8°C/min to 50°C  (0 min), then 10°C/min to 200°C 

(2 min). The MSD was operated in EI‐mode at 70 eV ionization energy and 1.13 × 10−7 Pa. Ion source 

temperature: 230°C, quadrupole: 150°C,  transfer  line: 200°C. The data was  acquired  in  SIM mode 

selecting 74 m/z for detection of the methyl acetate analyte and 76 m/z for detection of its carbon‐

labeled derivate at 50 ms dwell  time each. The  fully automatized operation of all SPME steps was 

realized  with  a  CTC‐PALxt  autosampler,  which  was  controlled  by  Chronos  software  v.3.5  (Axel 

Semrau, Spockhövel, Germany). SPME fibers were obtained from Supelco (Bellefonte, PA, USA). 

 

 

 

 

 

 

41 

2.4 Evaluation of the impact of volatile organic acids on paper 

2.4.1 Analysis of formic acid and acetic acid emission potential by static headspace GC‐MS 

with selected‐ion monitoring (SHS‐GC/SIMMS) 

Samples of Schinkel exhibit materials were cut  into small pieces (approx. 1‐2 mm2) and, depending 

on  the  calibration  range of  the  respective VOC, between 100 mg  to 250 mg of  the  samples were 

transferred  in a ten ml crimp‐cap vial. The vials were sealed with a crimp cap containing UltraClean 

silicon/PTFE septa (45° shore A, 3mm). The addition of 2.5 ml of a toluene‐d8 in mixture with veratrol 

(1 µl/ml) was used as the  internal standard for headspace measurements. Static headspace GC‐MS 

analysis of formic acid and acetic acid was performed on a 7890A gas chromatograph coupled with a 

5975C mass selective detector  (Agilent Technologies). Samples were  introduced by an autosampler 

(Agilent GC Sampler 120). Agitator settings:  incubation time 1800 s,  incubation temperature 120°C, 

and syringe temperature 130°C. Injection: 250 µl, split ratio 5:1, 240°C. Separation: A fused silica HP‐

5MS column (30 m x 0.25 mm x 0.25 µm; J&W Scientific, Folsom, CA, USA) and helium as a carrier gas 

were used. Column flow: 1.2 ml/min. Oven program: 30°C for 5 min, then 5°C per min to 70°C, and 

15°C per min to 280°C, then hold  for 5 min. The MS was operated  in  the electron  impact mode at 

70 eV  (230°C, 1.13 × 10−7 Pa).  Selected  ion monitoring  (SIM) mode was applied  for  formic acid at 

m/z 46, acetic acid at m/z 60, and toluene‐d8 at m/z 98 to detect peak area. 

 

2.4.2 Artificial aging 

The  degradation  effects  of  Schinkel  exhibit materials were  evaluated  by  artificially  aging  of  each 

material with Whatman No. 1, composed of a minimum α‐cellulose content of 98%, or historical rag 

paper, according to the  test proposed by Strlic et al.  [176].  In short, a defined amount of  indicator 

paper  (Whatman No. 1) or historical  rag paper,  is mounted  in  the volume of  tightly closed 100 ml 

Schott bottles, containing at the bottom a defined mass of the suspicious materials that need to be 

checked  for  its volatile emissions. Deviating  from Strlic et al.,  the aging was  further  carried out  in 

small (10 ml) crimp cap vials sealed with UltraClean™ silicon/PTFE septa (45° shore A, 3 mm) because 

of  the  small  amount  of  the  available  original  samples  and  the  fact  that  the  100 ml  Schott  flasks 

showed  measurable  gas  leaking  at  elevated  temperatures.  Sample  weights  of  (91.7  ±  0.3)  mg 

Whatman No.1 or historical rag paper were hinged in the air volume and (183.3 ± 0.3) mg of original 

samples to be tested were placed at the bottom of the same vial. Samples of the storage materials to 

house  the  collection  and papers were preconditioned  in  a desiccator  at  room  temperature  (25°C, 

relative  humidity:  52.9%)  before  aging  for  24  hours  using  saturated magnesium  nitrate  solution. 

Samples were  aged  at  100°C  for  five  days.  After  aging,  the  paper  samples were  cooled  at  room 

temperature for one hour before opening the vial. 

 

42 

2.4.3 Molar mass and carbonyl content measurement 

The hydrolysis and oxidation characterization of paper caused of different materials were carried out 

by measuring molar mass  and  carbonyl  group  content.  A  CCOA  (carbazole‐9‐carbonyl‐oxy‐amine) 

labeling for measuring the carbonyl group content was performed according to earlier reports [177, 

178].  After  labeling,  the  paper  samples  were  dissolved  in  N,N‐dimethylacetamide  containing  9% 

lithium chloride (w/v) after the solvent exchange. The measurement was performed on a GPC system 

consisting of a  fluorescence detector  (TSP FL2000), a multiple‐angle  laser  light  scattering detector 

(Wyatt Dawn DSP with argon  ion  laser, =488 nm), and a refractive  index detector  (Shodex RI‐71). 

The separation was carried out on a set of  four PLgel mixed‐A LS columns  (20 mm, 7.5 x 300 mm, 

Varian/Agilent).  N,N‐dimethylacetamide  containing  0.9%  lithium  chloride  (w/v)  was  used  for  as 

mobile phase. The  system was operated at a  flow  rate of 1.0 ml/min with an  injection volume of 

100 ml. The fluorescence of the CCOA label was detected at an excitation wavelength set at 290 nm, 

and  the  emission wavelength was  set  at  340  nm.  Data  evaluation was  performed with  standard 

Chromeleon 4, Astra 4.73, and GRAMS/32 software packages. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

43 

3 Results and discussion 

3.1 Analysis of carbohydrates, lignocellulosic side‐reactions, and degradation 

products 

3.1.1 Analysis of lignocellulosic effluent streams and aged paper extracts (Paper I & II) 

The degradation of cellulose, hemicellulose, and  lignin during pulp and paper processing produce a 

wide  variety  of  degradation  products  caused  by  multiple  fragmentations,  rearrangement,  and 

condensation  reactions.  These  lignocellulosic  effluent  streams  consist  of  different  low‐molecular‐

weight  carbohydrates,  lignin‐derived  compounds,  and highly  concentrated  salts,  all  in  an  aqueous 

solution.  Knowledge  about  the  composition  of  the  degradation  compounds  is  essential  due  to 

environmental  concerns  of  the  high  total  organic  carbon  (TOC)  level  of  effluents,  to  optimize 

processing steps and effluent treatments by examination of degradation mechanism, and because of 

degradation  products  are  potentially  valuable  compounds  as  resources  for  future  biorefinery 

scenarios. 

The  analytical  challenge  of  analyzing  lignocellulosic  effluent  streams  involves  the  simultaneous 

determination of various carbohydrate‐derived reaction products. The degradation of celluloses and 

hemicelluloses leads to the formation of aliphatic mono‐ and dicarboxylic acids, hydroxyacids, as well 

as  aromatic  compounds  from  lignin  degradation  and  chromophores  from  carbohydrates.  The 

challenge of analyzing pulp bleaching effluents also consists  in the presence of highly concentrated 

salts  in this aqueous solution and the  low concentrations of target compounds. Similar degradation 

reactions  during  pulp  processing  also  occurred  under  ambient  conditions  over  a more  extended 

period of natural or artificial accelerated paper aging. The composition of degradation products  in 

aged  paper  extracts  contains,  similar  to  lignocellulosic  effluent  streams,  structurally  similar 

degradation products of carbohydrate‐derived reaction products. 

 

3.1.1.1 Study of complex mixtures of lignocellulose degradation products by CE‐MS 

The  study  of  complex  lignocellulose  degradation  products  in  pulp  bleaching  effluents  and 

degradation products of paper aging in Paper I analyzed the target compounds by a combination of 

capillary  electrophoresis  (CE) with  electrospray  ionization‐mass  spectrometry  (ESI‐MS).  The CE‐MS 

method allows the simultaneous analysis of lignin and carbohydrate‐derived compounds without the 

need of a preliminary derivatization process  in contrast  to GC analysis  (for volatility) or LC analysis 

(for UV  label attachment). The direct analysis  is  the primary advantage, which enables a rapid and 

straightforward  analysis method  and  avoids  the  effect  of  losing  compounds, may  be  possible  by 

pretreatment steps. 

44 

The study revealed that the analyzed effluent samples contain a large number of aliphatic carboxylic 

acids,  comprising  short‐chain mono‐  and  di‐acids,  short  chain  hydroxy  acids,  and medium‐chain 

monoacids.  The  examination  also  detected  vanillic,  syringic  and  cinnamic  acids  as  degradation 

products of lignin origin. The aged paper extract contained sugar acids and LMM mono‐ and di‐acids 

predominantly  originated  from  far‐reaching  carbohydrate  degradation.  The  analysis  revealed  only 

vanillic acid as a compound of lignin origin in the aged paper extract, due to the low concentration of 

other phenolic compounds. 

The study showed that the CE‐MS analysis allows fast and efficient separation of aliphatic carboxylic 

acids  and phenolic  compounds  in  a  single  run without  the need of  solvent  exchange.  The  ESI‐MS 

analysis of unknown compounds provides indications of the chemical nature of the compounds. The 

method  allows  the  investigation  of  aqueous  or water‐soluble  samples  and  requires  only  sample 

amounts in the range of nanoliters. 

Nevertheless,  the high concentration of  inorganic salts  in pulp effluent samples hinders  the CE‐MS 

analysis, which results  in  imperfect separation of some compounds and shifting of migration times. 

Short  chain  carboxylic  acids  (formic  acid,  acetic  acids)  could not be  analyzed  in  the paper extract 

sample due to limits of the MS detector related to their low molecular weight. The complex mixture 

of compounds in the studied samples complicated further fragmentation of analyte ions for structure 

information and increased the required analysis time. 

The  analysis  of  lignocellulosic  degradation  products  by  CE‐MS  is  a  fast  and  derivatization‐free 

approach, which only requires a concentration step of the sample. However, the significant weakness 

of  the  analysis  approach  is  that  no  fragmentation  database  exists,  in  case  of missing  reference 

standards  the  verification  of  compounds  could  not  be  performed.  The  GC‐MS  approach  can 

overcome the limitation of compound identification, by providing access to mass spectra databases, 

but requires a derivatization step prior analysis. 

 

3.1.1.2 Study of complex mixtures of lignocellulose degradation products by GC‐MS methods 

The application of different derivatization methods and GC‐analysis systems (GC‐MS, Py‐GC‐MS, and 

HS‐GC‐MS) in the study of bleaching effluents in Paper II revealed a comprehensive overview of the 

compounds  in  the  samples.  Bleaching  effluent  samples  mainly  consist  of  the  carbohydrate 

degradation products 3,4‐dihydroxybutyric acid and glycolic acid, as well as smaller amounts of LMW 

carboxylic acids. Also, a large number of fatty acids, dimethoxybenzenes, methoxybenzoic acids, and 

lignin‐derived compounds were present in the bleaching effluent sample. 

However,  every  method  showed  discrimination  effects  during  analysis,  due  to  their  different 

sensitivity  to  components  of  a  particular  class  of  substance.  The  use  of  a  single method  would 

discriminate several compounds by concentration or detectability.  

45 

44 45

38

20

7

20

14

99

5

34

2

12

12 2 2 2

12

EtOx‐TMS DAM Py TMS‐DAM HS0

10

20

30

40

50

Num

ber

of c

ompo

unds

GC‐MS approaches

 LMM acids

 Fatty acids

 Lignin‐derivatives

 Carbohydrates

 Extractives

 Aldehydes

The  analysis  of  bleaching  effluent  by  EtOx‐TMS  or  DAM  derivatization  and  subsequent  GC‐MS 

analysis  as  a  standalone method  provide  a  general  overview  of  composition,  but  not  a  detailed 

picture. The EtOx‐TMS derivatization  followed by GC‐MS analysis  represents  the method of choice 

for  the  analysis  of  LMW  (chain  length  ≤  C6)  carboxylic  acids,  carbohydrates  and  sugar‐derived 

compounds  (Table  4  in  Paper  II).  The  combination  of  TMAH‐derivatization  and  Py‐GC‐MS  analysis 

provides  the most  information  about  lignin‐derived  compounds  and  long‐chain  carboxylic  acids, 

while  HS‐GC‐MS  detect  volatiles,  which  mainly  comprises  aliphatic  aldehydes.  Only  a  combined 

approach  of  different  GC‐MS  analysis  methods  allows  an  almost  complete  identification  of 

lignocellulose degradation products in bleaching effluent samples. 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 14: Number of compounds  identified within the different compound classes of the bleaching 

effluents analytes, according to various GC‐MS derivatization approaches. 

 

 

3.1.1.3 Comparison of CE‐MS and GC‐MS methods for analyzing complex mixtures of lignocellulose 

degradation products 

CE‐MS  and GC‐MS  are  two  approaches with different methodologies,  also  reflected  in  the  results 

achieved. On the one hand, the fast and straightforward CE‐MS analysis method with the possibility 

of  direct  analysis  and  on  the  other  hand  the  GC‐MS  analysis,  which  requires  almost  always  a 

derivatization  step  prior  analysis  and  allows  the  access  to mass  spectra  databases  for  compound 

identification.  GC  analysis  approaches  showed  particular  advantages  in  the  analysis  of  bleaching 

effluents  since  the  high  salt  concentration  in  the  samples  forms  a  deposite  during  derivatization, 

which allows separation from the injection mixture. In turn, GC methods have the disadvantages that 

preliminary  sample  treatments,  e.g.,  lyophilization  to  ensure water‐free  samples  or  derivatization 

46 

steps  to  produce  volatile  analytes,  require  extensive  and  time‐consuming  work.  Also,  a  loss  of 

components could occur during sample preparation. The effect of compound discriminations  in the 

various GC analysis methods exists, and the sample analysis must take place according to the desired 

target components. However, comparing the results of the bleaching effluents the  identification of 

components shows the  limitation of the CE‐MS method and the advantage of the GC‐MS approach. 

The  results  of  CE‐MS  analysis  showed  a  very  complex  electropherogram with  a  high  number  of 

components, but the data allowed only the identification of 20 compounds (Table 2 in Paper I) in the 

bleaching effluent sample, due to the lack of reference compounds and available database systems. 

In contrast,  the GC‐MS method reveals seven  identified compounds by HS‐GC‐MS and between 42 

and 60 compounds (Table 1 in Paper II) depended on the different GC‐based analysis approaches. In 

total, GC‐MS approaches  identified 110 components  in the bleaching effluents sample, showing the 

high GC separation efficiency. 

 

3.1.2 Analysis of complex mixtures of carbohydrates (Paper III & IV) 

The  analysis of  carbohydrates  requires  the elimination of  various problems, due  to  the molecular 

structure of  carbohydrates. The  carbohydrate  related peculiarities are extensively discussed  in  the 

previous  chapters.  Also,  complex  carbohydrate  mixtures  can  show  very  different  amounts  of 

carbohydrates.  The  carbohydrate  analysis  by  gas  chromatography  coupled  to mass  spectrometry 

provides  the necessary  resolution and sensitivity  to study  the occurrence of carbohydrates  in such 

complex mixtures in contrast to CE and HPLC. 

 

3.1.2.1 Derivatization methods (Paper III) 

A  GC‐based  analysis  requires  the  derivatization  of  carbohydrates  to  produce  volatile  derivatives. 

However, derivatization methods influence the analysis to a far greater extent than just the increase 

of volatility. The studied derivatization approaches  included  the single‐step derivatization methods 

per‐trimethylsilylation and O‐isopropylidenation as well as  two‐step derivatization methods, which 

comprise  sequential  oximation  (with  O‐ethyl‐  or  O‐benzylhydroxylamine)  followed  by 

trimethylsilylation or trifluoroacetylation. The six different derivatization methods occurred at eight 

monosaccharides  (C5–C6),  glycolaldehyde,  and  dihydroxyacetone.  The  evaluation  was  performed 

according to their derivatization efficiency, chromatographic characteristics (e.g., number of isomers 

formed per compound and resolution), analytical sensitivity (limit of detection), informational value 

of mass spectra, reproducibility of quantification results and robustness towards matrix effects.  

The  ISP‐derivatization  showed  an  excellent  signal  intensity  and  complex  mass  spectra.  The 

chromatographic resolution is good for almost all carbohydrates, since ISP produced only one peak, 

except  for  ribose  and  fructose.  A  drawback  of  the  ISP method  for  carbohydrate  analysis  is  that 

47 

compounds  with  no  vicinal  cis‐diol  moiety  in  the  molecular  structure  are  not  derivatized,  e.g., 

glycolaldehyde.  The  main  disadvantages  are  the  high  sensitivity  towards  water  and  that  the 

derivatization procedure requires a long time to neutralize the reaction mixture. These circumstances 

exclude the ISP derivatization as high throughput analysis method of carbohydrate‐rich matrices. 

The  per‐trimethylsilylation  showed  the  lowest  LOD  values  and  the  slightest  influence  of  matrix 

effects. Although the obtained mass spectra are less informational compared to other derivatization 

methods, they allow an unambiguous assignment of pyranose and furanose  isomers. However, the 

formation of up to five isomers per carbohydrate led to a significant peak overlapping and result in a 

higher number of coeluting compounds. 

 

Table 1: Evaluation of derivatisation approaches for GC‐MS‐based carbohydrate analysis of complex 

sample matrices. 

Derivatization approach 

Ran

king 

Chromatographic properties 

MS infor‐

mational value   Quantitative 

characteristics 

Matrix effect 

Method 

laboriousness 

Peak  Complete detection of reference compounds 

LOD  RSD count per compound 

separation 

EO‐TMS  1                 

TMS  2                 

BO‐TMS  3                 

EO‐TFA  4                 

BO‐TFA  5                 

ISP  6                    

  Rating of characteristics:  Good  moderate  negative  

 

The oximation procedure reduced  the number of  isomers per carbohydrate by the conversation of 

the carbonyl group into the corresponding oxime derivative. EtOx‐TMS and BeOx‐TMS derivatizations 

revealed well‐resolved  chromatograms and  informational mass  spectra. The benzyl group of BeOx 

caused  a  higher  mass  increment  compared  to  EtOx,  which  increased  the  retention  times 

substantially. Trifluoroacetylation  instead of trimethylsilylation after the oximation reaction showed 

some severe drawbacks, e.g., the decomposition of ketohexoses by traces of trifluoroacetic acid and 

the low informational content of mass spectra obtained by BeOx‐TFA derivatives. The study showed 

that  the  BeOx‐TFA method  should  only  be  appled  to  samples with  no  or  low  complex matrices, 

where no interfering components influence the analysis of target compounds.  

In the course of the study, the evaluation of matrix effects during carbohydrate analysis resulted  in 

the development of a matrix  impact value. The calculation based on the recovery rates of standard 

addition, calculated by peak height and peak area calibrations, to quantify the effects of the matrix 

on the results of the different derivatization methods. 

48 

The  results  showed  the possibility  to  analyze  carbohydrates by direct derivatization, which  allows 

derivatization  of  samples  in  the  presence  of  a  complex matrix.  The  study  also  proved  that  the 

EtOx‐TMS  derivatization  approach  is  superior  to  other  approaches,  due  to  an  excellent 

chromatographic resolution and a low number of isomers. Although demanding higher labor efforts 

compared  to a single step derivatization procedure,  the high  informational values of mass spectra, 

low LOD, LOQ and RSD values and the robustness towards matrix effects improves this derivatization 

procedure upon other. 

 

 

Figure  15:  Gas  chromatograms  with  MS  detection  of  a  mixture  of  46  carbohydrates  and 

carbohydrate‐derived compounds converted to different derivatization products. 

 

 

3.1.2.2 Deeper insight into the EtOx‐TMS approach (Paper IV) 

The  evaluation  six  different  derivatization  approaches  revealed  that  the  EtOx‐TMS  derivatization 

procedure  is  the most  appropriate  derivatization method  for  carbohydrates  in  complex mixtures 

within the investigated approaches. Hence, a subsequent study (paper IV) investigated the EtOx‐TMS 

derivatization  in more  detail.  Therefore,  the  examination  of  chromatographic  and  mass  spectra 

characteristics  of  46  EtOx‐TMS  derivatized  carbohydrates  was  carried  out,  involving  32 

monosaccharides, 13 disaccharides, and one trisaccharide. As described in the previous chapter, the 

49 

Glycolaldehyde

Methylglyoxal

Glyceraldehyde

Threose

Erythrulose

Erythrose

2‐Deoxy‐D‐ribose

Digitoxose

Xylose

Lyxose

Ribulose

Rhamnose

Fucose

2‐Deoxy‐D‐glucose

Psicose

Tagatose

Allose

Mannose

Gulose

Galactose

Talose

Altrose

Glucose

Sedoheptulose

Glucoheptose

L‐Glycero‐D‐manno‐heptose

Xylobiose

Lactose

Cellobiose

Maltulose

Maltose

Turanose

Leucrose

Gentiobiose

Palatinose

Melibiose ‐‐

10

20

30

40

50

Reten

tion tim

e ‐ syn isomer pea

k 

or pea

k 1 [min]

‐40

‐30

‐20

‐10

0

10

20

30

40

50

 Retention time ‐ syn isomer peak or peak 1 [min]

 Time shift between syn and anti isomer peak or peak 1 and peak 2 [sec]

Time shift betwee

n syn

 and anti isomer pea

k 

or pea

k 1 and pea

k 2 [sec]

oximation reaction of the carbonyl group  in carbohydrates eliminates the occurrence of furanosidic 

and  pyranosidic  isomers.  The  reaction  forms  two  oximes,  the  syn  (Z)  and  anti  (E)  isomer, which 

remains after silylation and produces two peaks in the chromatogram.  

The mass spectra of EtOx‐TMS derivatized carbohydrates provided high  informational content, but 

showed congruent mass spectra regarding the fragmentation pattern. The mass spectra information 

allows  the  differentiation  between  aldose  and  ketose,  but  a more  precise  carbohydrate‐specific 

identification  is  challenging  without  additional  chromatographic  characteristics.  The 

chromatographic characteristics of the two occurring syn and anti isomers are their peak area ratio, 

elution order and time shift between the two peaks. For monosaccharides, the syn/anti ratio of peak 

area showed high values  for aldoses  (C3  to C6)  in contrast  to  low values  for ketoses  (C4  to C6). The 

study  observed  an  elution  order  of  syn  peak  before  anti  peak  at  ketotetroses,  ketopentoses, 

aldohexoses, and aldoheptoses.  In contrary  to aldotriose, aldotetroses, aldopentoses, ketohexoses, 

and  ketoheptoses, were  the  the  anti  peak  eluted  before  the  syn  peak.  The molecular  shape  of 

monosaccharides also led to an influence on compound retention, by which a more planar molecular 

form resulted in higher retention times. The study of disaccharides revealed that the respective type 

of  glycosidic  linkage  and  the  contained monosaccharide moieties  influenced  the  chromatographic 

characteristics. The position of the glycosidic linkage showed an effect on the elution order of the syn 

peak, which appeared first at 1,4‐linked disaccharides before 1,3‐, 1,5‐ and 1,6‐linked disaccharides. 

The  molecular  shape  of  the  disaccharide  explains  the  influence  of  the  glycosidic  bond  on  the 

retention  time.  Garcia‐Raso  et  al.  proposed  that  the  higher  conformational  flexibility  of  1,6‐

disaccharides cause higher retentions [99]. 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 16: Retention  time of  syn peaks  (■) and  retention  time  shift  (●) between  the  syn peak and 

corresponding anti peak or peak 1 and peak 2 of different EtOx‐TMS derivatized carbohydrates. 

50 

The monosaccharide moieties of disaccharides influence the syn/anti peak area and anti peak order. 

So,  galactose‐,  glucose‐  or  xylose‐containing  disaccharides  showed  high  syn/anti  peak  area  ratios 

compared  to  fructose‐containing  disaccharides.  For  that,  the  anti  peak  of  fructose‐containing 

disaccharides  eluted  before  the  syn  peak,  while  the  syn  peak  of  non‐fructose  containing 

disaccharides eluted before the anti peak. 

The  elution  order  and  the  time  shift  between  the  two  oxime  peaks  are  carbohydrate‐specific 

identification criteria. The  incorporation of  the  two chromatographic parameters  into a compound 

identification index, the oxime peak identifier (OPI), allows a reliable identification of carbohydrates. 

The OPI based, in contrast to conventional retention indices, on the second peak of the syn/anti pair 

as dynamic  standard and  a  static  internal  standard as an anchor point.  In  contrast  to  the  relative 

retention  index,  the  OPI  differs  more  precise  between  individual  carbohydrates,  increases  the 

resolution  between  carbohydrate  index  values  and  solves  the  problem  of  small  retention  time 

differences between stereoisomers. 

The EtOx‐TMS derivatization method and subsequent GC‐MS analysis combined with  the proposed 

oxime  peak  identifier  (OPI)  evaluation  overcomes  several  problems  of  the  GC‐MS  based 

carbohydrate analysis and enables reliable and easy identification of carbohydrates. 

 

3.1.3 Analysis of lignocellulosic side‐reactions (Paper V, VI and VII) 

Lignocellulosic materials  represent a very complex matrix, as  they  include cellulose, hemicellulose, 

and  lignin.  The  evaluation  of  chemical  and  physical  properties  of  lignocellulosic  materials 

necessitates an analysis  regarding  lignocellulosic  side‐reactions. These  lignocellulosic  side‐reactions 

involve  the  investigation  of  polysaccharides,  containing  acetyl  groups,  as  well  as  oxidation‐  and 

hydrolysis‐reactions. The characterization of chemical and physical properties is essential to provide 

extensive information to a broad variety of industrial branches: to ensure a consistent quality of the 

produced  products  or  to  validate  treatments  for  improving  fiber  properties,  to  name  only  a  few 

examples.  The  following  applications  show  the  analysis of  side‐reactions  at polysaccharides, pulp, 

and paper samples. 

 

3.1.3.1 Applications 

3.1.3.1.1 Carbohydrate compositions of polysaccharides (Paper V to be submitted) 

Polysaccharides can contain a large number of different monosaccharides and vary in the occurrence 

of them. Information about the monomer composition of fresh and aged pulps, pure polysaccharides 

and  cellulose  samples  are  essential  for  their  characterization  and  further  processing.  The 

composition  and  accessibility  of  carbohydrates  in  cellulosic  samples  can  be  affected  by  electron 

beam irradiation (e‐beam) treatments or aging processes. Different hydrolysis methods are available 

51 

to analyze the carbohydrate composition of polysaccharides and cellulosic sample. However, a single 

hydrolysis method alone cannot be capable simultaneously for the analysis of pentoses, uronic acids, 

and total glucose content. Strong acid hydrolysis leads to a release of reducing monosaccharides but 

destroys  uronic  acids.  Sundberg  et  al.  reported  in  1996  that  mild  acid  hydrolysis  reduces  the 

degradation of fragile hemicelluloses and allow the possibility to quantify uronic acids, but in case of 

crystalline cellulose structures, the method reaches its limit [121]. 

Paper V describes the parallel hydrolysis of 21 polysaccharides and 21 cellulosic samples by sulfuric 

acid and acid methanolysis,  followed by GC‐MS analysis of  the  corresponding hydrolysis products. 

The acid methanolysis represents a hydrolysis method under mild conditions and allows the analysis 

of monosaccharide and sugar acid composition of hemicelluloses, pectins and an amorphous fraction 

of cellulose  samples. The acid methanolysis also converts  the  released monosaccharides  into  their 

corresponding  methyl  glycosides  and  stabilize  the  carboxyl  groups  of  uronic  acids  with  methyl 

groups. Silylation derivatization followed the acid methanolysis procedure according to Sundberg et 

al. [121] to increase the volatility of hydrolysis products, subsequently analyzed by a modified GC‐MS 

analysis. A two‐step sulfuric acid hydrolysis according to Bose et al. [171] allows the determination of 

total glucose content, subsequently silylated and analyzed via GC‐MS.  

The acid methanolysis of polysaccharides showed a recovery rate of 54.96% – 102.19% of released 

monosaccharides and sugar acids. Only Inulin showed a very low recovery rate of 4.29%, because the 

polysaccharide  consists  almost  entirely  of  fructose,  which  is  also  available  in  stachyose. 

Unfortunately, fructose decomposes during acid methanolysis, while glucose as the terminal end of 

the  inulin  polymer  remained.  The  primary monosaccharides  identified with  acid methanolysis  in 

cellulosic samples were glucose, xylose, and mannose, followed by arabinose and galactose (Figure 3 

in  Paper  V).  The  hydrolysis  of  these  samples  showed  recovery  rates  below  22.71%  after 

depolymerization.  The  study observed differences  in  the  chemical  composition between historical 

and modern rag paper, indicating a diverse origin of the fiber material. 

The  sulfuric  acid  hydrolysis  of  polysaccharides  revealed  a  specific  carbohydrate  pattern  for  each 

polymer  according  to  their  corresponding monosaccharide  distribution  (Figure  5  in  Paper V).  The 

strong hydrolysis of cellulosic samples showed higher recovery rates compared to acid methanolysis, 

because  the  sulfuric  acid  hydrolysis  affected  the  amorphous  as well  as  the  crystalline  fraction  of 

cellulosic samples. There are several reasons sulfuric acid hydrolysis could not achieve full recovery 

rate in this study. The sulfuric acid hydrolysis causes decarboxylation of uronic acids, and can lead to 

an information loss during analysis due to the destruction of low‐content existing carbohydrates. No 

signals of disaccharides and trisaccharides occurred during GC‐MS analysis. The analysis of hydrolysis 

reaction mixture by a HPTLC based method observed cellobiose and higher gluco‐oligosaccharides, 

which indicated an incomplete hydrolysis of the cellulosic samples [179]. However, this could not be 

52 

verified  by  GC‐MS,  probably  due  to  non‐GC‐analyzable  oligomers.  The  analysis  did  not  consider 

fractions  of  fats,  proteins,  lignin,  and  minerals,  which  is  mainly  noticeable  in  wheat  bran.  The 

presence of non‐GC‐analyzable compounds explains the high recovery rates of cellulose samples and 

the low recovery rates of specific polysaccharides at once.  

The  proposed method  has  been  applied  to  samples  treated  with  an  electronic  beam  (e‐beam): 

eucalyptus paper pulp, hemp paper pulp, and  rag paper or artificially aged  rag paper. The analysis 

detected similarities at monosaccharide level between the e‐beam and aged rag paper sample.  

 

no treatment

e‐beam

no treatment

e‐beam 

no treatment

e‐beam 

aged

0

50

100

150

200

250

300

0

50

100

150

200

250

300

Released carbohydrates [m

g/g sample]

     Acid  methanolysis

        Total carbohydrates

        Glucose                  

        Xylose                    

Eucalyptus

paper pulp

Rag paper

(modern)

Hemp

paper pulp 

Figure  17:  Released  carbohydrates  by  acid  methanolysis  (total  amount,  glucose  and  xylose)  of 

untreated, electron beam (e‐beam) treatment (eucalyptus and hemp paper pulps: 120 kGy, and rag 

paper: 60 kGy) and artificial aged cellulose samples. 

 

 

The  applied  treatments  (aging  or  e‐beam)  decreased  the  total monosaccharide  yield  obtained  by 

sulfuric  acid  hydrolysis.  The monosaccharide  yield  obtained  by  acid methanolysis  increased  in  all 

samples  after  the  respective  treatment.  Sulfuric  acid  hydrolysis  showed  decreasing  amounts  of 

glucose  and  acid  methanolysis  obtained  equal  or  higher  amounts  of  glucose  from  the  treated 

samples compared to the untreated samples. The treatments  increase the accessibility of cellulose‐

containing  samples  and  especially  the  e‐beam  treatment  can make  cellulose‐containing materials 

more interesting as raw material for biomass to fuel conversion. 

The  comparative  analysis  enables  the  subtraction  of  the  glucose  amount  determined  with  acid 

methanolysis from the amount of glucose determined with sulfuric acid hydrolysis (CG2H). The result 

leads to a quantification of the recalcitrance cellulose fraction within cellulosic samples, referred as 

53 

the  index of recalcitrance cellulose fraction (CRCI). While the 13C CP‐MAS NMR method distinguishes 

between structural alterations based on  the different cellulose  fractions  (Iα,  Iβ and paracrystalline), 

the  CG2H‐method  derived  information  about  the  structure  elucidation  from  the  carbohydrate 

composition and the quantity of  individual monosaccharide. Both methods can detect and uncover 

structural rearrangements from the crystalline to the amorphous fraction. The NMR method  is fast, 

but  requires  expensive  equipment,  while  the  CG2H‐method  comes  along  with  less  expensive 

equipment,  but  is more  time‐consuming  and  requires more  labor  and  analytical  resources.  The 

analysis of artificial aging or electron beam  irradiation  treated cellulose samples uncovered similar 

tendencies  in the structural rearrangements  from crystalline to amorphous cellulose  fraction and a 

general loss of total carbohydrates within all samples. Despite a similar behavior of the treatments at 

samples’ carbohydrate level, the study indicated that e‐beam treatment caused a significantly higher 

effect of chain scission at rag paper in contrast to the ageing treatment.  

The results showed that the proposed approach represents a useful analytical tool for determination 

of  structure elucidation and carbohydrate characterization  for a wide  range of polysaccharide and 

cellulosic samples.  

 

3.1.3.1.2 The degree of Acetylation (Paper VI) 

The presence of acetyl groups in polysaccharides is an essential factor, which significantly influences 

the  properties  of  polysaccharides.  The  formation  of  acetyl  groups  naturally  takes  place  in  the 

polysaccharides hemicelluloses, pectin, and  lignin. The acetylation can also perform  in the presence 

of acetic acid, whether intentionally or unintentionally.  

Acetyl  groups  affect  various  functions  of  polysaccharides,  e.g.,  the  solubility  in water  and  certain 

organic acids, and especially  the surface properties, such as hydrophobicity, water  interaction, and 

the  reactivity  of  pulp,  fibers,  and  paper.  However,  the  presence  of  free  acetates  as  acetic  acid 

(adsorbed or residues released by processing steps), acetic acid salts or residues of acetylation agents 

complicates the analysis of bounded acetyl groups. Acetyl groups containing polysaccharides can also 

frequently  emit  acetic  acid,  e.g.,  from  hemicelluloses.  Although  the  pulping  process  leads  to 

deacetylation and  therefore only a  few acetyl groups are present  in hemicelluloses of paper, small 

traces of acetic acid can be formed due to degradation processes. Recent analysis methods cannot 

distinguish between bound and free acetates, since they are relying on the cleavage of the esters and 

analyze the released acetic acid. 

Paper VI describes a Zemplén‐deacetylation based analysis method  for the quantification of bound 

acetate groups in polysaccharides and carbohydrate‐related materials by GC‐MS, not affected by the 

presence of  free or adsorbed acetic acid and acetates. The Zemplén‐deacetylation describes a  fast 

transesterification reaction between covalently specific bound acetyl groups and anhydrous sodium 

54 

methanolate in methanol, resulting in the formation of methyl acetate. The sodium methanolate acts 

as a catalyst; the formation of the corresponding methyl acetate consumes the methanol. However, 

free acids can neutralize  sodium methanolate and  thus,  the deacetylation  reaction  requires higher 

amounts  beyond  catalytic  quantities.  Paper  VI  describes  two  different  introduction  techniques,  a 

direct liquid‐phase analysis and the analysis of the methyl acetate containing vapor phase via SPME. 

The  liquid‐phase based GC‐MS method represents a fast and straightforward method, but revealed 

some weaknesses, as (1) the accumulation of aggressive sodium methanolate in the liner and column 

head, and (2) the internal standard toluene does not fulfill the common requirements, since it is not 

taking part of  the actual deacetylation process. So,  internal standardization cannot correct entirely 

influences  as moisture  content,  adsorptive  interactions  of  the methyl  acetate  and  discrimination 

effects by a time‐depended loss of methyl acetate. The analysis of the vapor phase by SPME‐GC‐MS 

in  combination  with  4‐O‐(13C2‐acetyl)  vanillin  for  internal  standardization,  which  generates 

isotopically labeled methyl 13C2‐acetate in situ, overcome the contamination problem and eliminates 

possible  influences  and  discrimination  effects.  The  degree  of  acetylation  can  vary  considerably 

between  the  individual  samples.  This  circumstance  requires  an  adaptation  of  the  protocol  and 

calibration for samples containing trace amounts or higher amounts of acetyl groups. 

The  analysis  of  cotton  linters  and Whatman  paper  showed  only  trace  amounts  of  acetyl  groups 

resulting  in a very  low degree of acetylation. The absence of hemicelluloses  in  the  sample and an 

alkaline  purification  step,  which  removes  remaining  acetyl  groups  in  the  samples  explain  the 

observed trace amounts. Book paper and rag paper contained higher amounts of acetyl groups, and 

so, the corresponding degree of acetylation is noticeably higher. Analyzing hemicellulose‐containing 

samples (spruce, birch, and beech origin) revealed a significant increase in the degree of acetylation. 

The results showed that the proposed method enables an accurate and precise quantification of the 

degree of acetylation. 

 

3.1.3.1.3 Volatile organic acids in the paper (Paper VII) 

Volatile  organic  compounds  (VOCs)  can  affect  the  consistency  of  paper‐based  collections,  by 

accelerating the hydrolytic depolymerization, which leads to a significant loss of physical stability or 

changing  optical  properties.  Possible  VOCs‐emitting  sources  in  museum  collections  are  storage, 

presentation or construction materials as well as furnishings and the objects themselves. Among the 

VOCs, formic acid and acidic acid are of particular importance on paper degradation. The knowledge 

about  VOCs  emission  by  storage‐materials  and  their  impact  on  paper  degradation  can  help  to 

improve the storage environment of paper‐based objects. 

Paper  VII  describes  an  evaluation  of  the  emission  and  damaging  potential  of  storage‐materials 

present  in  the  collection  of  the  drawings  and  prints  of  Karl  Friedrich  Schinkel  stored  in  the 

55 

Kupferstichkabinett  (Museum of Prints and Drawings) of  the Staatliche Museen zu Berlin  (National 

Museums in Berlin). 

The study of storage materials combines (1) the quantitative analysis of formic acid and acetic acid 

emission  potential  by  static  headspace  GC‐MS  (SHS‐GC‐MS)  with  selected‐ion  monitoring  (SHS‐

GC/SIM‐MS) and (2) the evaluation of the emission impact on the carbonyl group content and molar 

mass by an  indicator‐paper based degradation analysis. The carbonyl group content and  the molar 

mass distribution best describe  cellulose  integrity. The  indicator‐paper based degradation  analysis 

applied a modified Strlic‐Test introduced by Strlic et al. [176]. The degradation analysis includes the 

exposure of Whatman No. 1 and rag paper to various to different storage materials under artificially 

aging  and  subsequently  analyzing  the  two  papers  according  to  their molar mass  distribution  and 

profile of carbonyl groups. 

The  emission  analysis  of  storage  materials  showed  substantial  differences  in  acetic  acid 

concentration among the samples, identifying the leather sample as releasing the highest amount of 

acetic acid, and the cabinet shelves and walls as a significant source of acetic acid. Fabric materials 

emitted  formic  acid  in  low  amounts.  The  analyzed  VOCs  originate  from  processed materials  and 

wood with acetyl group‐containing hemicelluloses. 

 

 

Figure 18: Samples of storage materials in headspace vials for emission analysis. 

 

The  indicator‐paper based degradation analysis reveals two groups of Schinkel exhibit materials,  in 

the first group, the released VOCs led to a relatively low degradation in both indicator papers, while 

the  second group caused a higher degradation  in both papers and  led  to a  significant decrease  in 

molar mass.  The  formic  acid  releasing  fabric  caused  the  highest  number  of  chain  scission  after 

artificial aging. The determination of  the carbonyl group profile allows  the differentiation between 

oxidation processes along the cellulose backbone (keto groups) and reducing end groups (aldehydes) 

by mathematical calculation. The results showed an oxidation  impact on both  indicator papers, but 

the presence of formic acid led to stronger hydrolysis and induced a loss of keto groups, because of 

its higher pKa value and  reductive behavior, compared  to acetic acid.  In  total,  the emitted volatile 

organic  acids  correlated  with  the  observed  degradation  processes  in  both  indicator  papers.  The 

56 

different degradation behavior of the two‐indicator paper occurred due to differences in the specific 

polymer composition, pre‐treatments, and additives. Whatman No.1 paper shows the ability to act as 

a universal  indicator, due to the sensitive reaction towards hydrolytic degradation, while rag paper 

reflects the reality, due to containing additional matrix compounds. 

The  study observed degradation  effects of  acetic  acids  and  especially  for  formic  acid.  The  results 

exclude  various materials  from  the  suitability  of  storing  paper‐based  collections  and  confirm  the 

need for a better understanding of storage materials to create a safe environment for paper‐based 

cultural heritage. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

57 

4 Conclusion 

Carbohydrates are essential compounds to fulfill various functions in physiological processes and are 

substantial resources used both  in food and non‐food  industry. They occur highly diverse regarding 

structure,  size,  functionality,  and  range  from  simple  monosaccharides  to  highly  complex 

polysaccharides. 

The  purpose  of  the  current  study  was  to  evaluate  GC‐MS  based  analysis methods  for  complex 

carbohydrate mixtures concerning qualitative and qualitative  composition  in  conjunction with  side 

reactions of  the  carbohydrate molecule.  The  complex mixtures occurred  as  soluble  extracts, pure 

polysaccharides, processed materials,  and degradation products of  carbohydrates.  The  analysis of 

the compounds requires the derivatization of  low molecular carbohydrates prior GC‐MS analysis or 

an additional hydrolysis step, which  is necessary  to convert oligo‐ and polysaccharides  into GC‐MS 

measurable units as well as an identification procedure for the carbohydrate analysis. 

There are numerous derivatization methods  to convert carbohydrates  in derivatives, which can be 

analyzed by gas chromatography. However, some of these methods have reaction conditions that are 

not suitable for a modern laboratory with optimized workflows and high sample throughput. Despite 

specific advantages of the different derivatization methods, some have significant disadvantages e.g., 

extremely  long  reaction  times,  several  derivatization  steps  and  the  requirement  of  neutralization 

steps,  respectively.  The  described  ExOx‐TMS  derivatization method  was  evaluated  based  on  ten 

carbohydrates  and  compared  to  five other derivatization methods. The  subsequent  analysis of 46 

carbohydrates  using  EtOx‐TMS  provides  comprehensive  data  for  the  chromatographic  and mass 

spectrometric characterization of various mono‐ and disaccharides. Despite  two reaction steps and 

the need to prepare water‐free samples, the method shows significant advantages in contrast to the 

other derivatization methods conducted in the study. Basic features include a low number of peaks, 

the  suitability  for  a  variety  of  sample  origins,  the  possibility  of  a  high  sample  throughput  in 

combination with derivatization robots, and the reduced sample preparation compared to previous 

methods. 

The evaluation of the different derivatization methods was carried out with regard to their suitability 

to conduct the reaction in the presence of the sample matrix. The derivatization in the sample matrix 

has the advantage that target components do not undergo an extraction procedure and thus avoid a 

possible  loss  of  components  or  discrimination  effects.  It  has  been  observed  that  the  amount  of 

derivatized target components are least affected in the presence of the sample matrix by conducting 

the  approaches  EtOx‐TMS  and  TMS.  Therefore,  these methods  are  suitable  for  carrying  out  the 

derivatization  reaction  on  samples without  the  need  of  previous  extraction  steps.  The  study was 

carried  out  on  the  following  samples:  formose  reaction,  pulp  mill  effluent,  book  paper,  and 

lyophilized  and  finely  ground  grapevine  leafs.  The method  revealed  the  advantage,  that  the  salt 

58 

precipitates  in the presence of the derivatization reaction mixture and cannot  interfere further the 

analysis, which is especially important for pulp mill effluents. 

There  is still a need to  increase the sample throughput at the derivatization process. Derivatization 

robots can partially solve the problem or minimize  the workload associated with the derivatization 

reaction.  

The identification of carbohydrates with equal molecular weight based on retention times is difficult 

due  to  the  low  elution  differences.  Especially,  routine  examinations  without mass  spectrometry 

depend on a  reliable  identification. The presented EtOx‐TMS‐derivatization prior GC‐MS analysis  in 

combination with the new oxime peak‐base retention index (oxime peak identifier) show significant 

improvements  in the retention time‐based  identification of carbohydrates. Using both oxime peaks 

with  an  internal  standard  increases  the  accuracy  and  can  help  to  enhance  positive  carbohydrate 

identification, even in highly complex biological samples. The procedure has already been applied in 

other studies for the determination of carbohydrates at different types of samples, e.g., extracts of 

books,  vine  roots,  and  sugar millet  plants. Recent  studies  also  applied  the  procedure  to  evaluate 

different thermal digestion processes of rice straw for ethanol recovery. 

However,  it must  be  taken  into  account,  if  the  analysis  base  only  on  one  isomer  peak,  then  the 

equilibrium between the isomer peaks is essential for the quantification process. Another important 

fact to mention  is that carbohydrates produce ubiquitous mass  ions and very similar mass spectra. 

For  this  reason,  the use of mass  spectroscopy  for  carbohydrate  identification  is  still  tricky despite 

various derivatization procedures.  

It  is recommended to conduct further  investigations to  improve the quantification process of EtOx‐

TMS‐derivatized carbohydrates after GC‐MS analysis. Possible approaches to improve the calibration 

and the quantification process could be made by providing a better separation of the syn‐ and anti‐

peak  to  conduct  the  quantification  on  both  peak  areas  as well  as  an  automated  software‐aided 

analysis  of  the  GC  chromatogram  by  considering MS  data.  The  use  of  carbohydrate  isotopes  for 

internal  standardization  could  further  improve  the  standard  derivation  and  reproducibility  of  the 

derivatization approaches. 

The  investigations of exhibition and storage materials of  the Schinkel collection  in  the Kupferstich‐

Kabinett  in Berlin showed that the shelves as well as the shelf walls consisted of pressed chipboard 

material  and  released  a  considerable  amount of  acetic  acid, which  can  lead  to  the destruction of 

cellulose‐based Schinkel exhibits. Likewise, the  leather and fabric sample of the magazine cassettes 

showed significant levels of absorbed acetic acid or low level of formic acid. The Kupferstich‐Kabinett 

responded by replacing the shelves and shelf walls with metal compounds, renewing the magazine 

cassettes  and  by  venting  the magazines  to  reduce  the  high  levels  of  acetic  acid  in  the  storage 

environment. 

59 

The  analysis  of  polysaccharide  consisting  carbohydrate  samples  was  conducted  by  the  parallel 

hydrolysis  of  sulfuric  acid  and  acid methanolysis  (CG2H‐method)  combined  with  a  GC‐MS‐based 

analysis of the hydrolysis products. The comparative analysis enables the subtraction of the glucose 

amount determined with  acid methanolysis  from  the  amount of  glucose determined with  sulfuric 

acid  hydrolysis.  The  CG2H‐method  allows  a  more  detailed  view  on  the  carbohydrate  profile  of 

samples and is preferred to discover background information of structural rearrangements compared 

to methods, which measure only the crystallinity index of paper samples. The information, obtained 

by  the proposed CG2H‐approach, can be used  to  improve  the efficiency of current  technologies  to 

increase the carbohydrate accessibility of cellulose‐containing samples. The study revealed that the 

e‐beam  treatment  reduce  the  crystallinity‐fraction  fraction  of  cellulose  towards  the  amorphous 

fraction, which makes  the  treatment  attractive  as  an  application  for  renewable  resources.  The  e‐

beam  irradiation can  improve carbohydrate yield by enhancing polysaccharide accessibility towards 

enzyme or hydrolysis methods. So, a e‐beam  treatment makes cellulose containing materials more 

interesting as raw material for biomass to fuel conversion.  

More  research  is  necessary  to  investigate  the  structure  elucidation  of  polymers  and  to  uncover 

internal relationships between the crystalline and amorphous cellulose fraction of cellulosic samples. 

Further efforts are needed to understand the effects of various treatments on cellulosic samples to 

increase the accessibility of carbohydrates for utilization as renewable resources and to extend the 

technique to samples of different biological origin. 

The  analysis  of  the  degree  of  acetylation  based  on  the  Zemplin  reaction method  has  significant 

advantages compared to many previous methods, it is not affected by free or in the sample material 

contained  acetic  acid.  The  analysis  was  carried  out  on  materials  ranging  from  powdered 

polysaccharides to various types of paper. Due to the high sensitivity of the analysis, it can be applied 

to the very small sample amounts. The method is easy to apply and enable in combination with short 

reaction times a high sample throughput. The simple reaction conditions and the SPME‐based GC‐MS 

analysis method allow the use of the analysis approach in many laborites without high investments in 

additional lab equipment. 

 

 

 

 

 

 

 

 

 

 

60 

Acknowledgements  

First, I would like to thank my supervisors Antje Potthast and Astrid Forneck for teaching me 

scientific  work,  and  for  introducing  me  to  many  different  research  areas  and  analysis 

methods. 

 

Thanks to Thomas Rosenau and Falk Liebner for always having the door open for questions, 

their availability and beautiful evenings. 

 

My colleagues, Stefan Böhmdorfer, Thomas Zweckmair, Sonja Schiesser, Martina Opietnik, 

Georg Pour, Gerald Ebner, Johannes Hell, Michaela Griesser, and many other people for their 

helpfulness and the many great moments that have enriched the work at the BOKU so much. 

 

I would like to thank all those who have reminded me with their questions about the status 

of this thesis that I should finally complete this. 

 

Special thanks to Martin Pour‐Nikfardjam for his motivating words and proofreading. 

 

I want  to  leave my  huge  thanks  to my  family,  parents  and  friends  for  your  support  and 

the good persuasion. 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

61 

References

1.  Klampfl, C.W., M. Himmelsbach, and W. Buchberger, Analysis of Simple Carbohydrates by 

Capillary Electrophoresis and Capillary Electrophoresis–Mass Spectrometry, in Capillary 

Electrophoresis of Carbohydrates: From Monosaccharides to Complex Polysaccharides, N. 

Volpi, Editor. 2011, Humana Press: Totowa, NJ. p. 1‐21. 

2.  BeMiller, J.N., Carbohydrate Analysis, in Food Analysis. 2010, Springer US: Boston, MA. p. 

147‐177. 

3.  Stevens, C.V., Industrial products from carbohydrates, wood and fibres, in Renewable 

Bioresources. 2004. 

4.  Breadmore, M., E. Hilder, and A. Kazarian, Fluorophores and Chromophores for the 

Separation of Carbohydrates by Capillary Electrophoresis, in Capillary Electrophoresis of 

Carbohydrates. 2011, Springer. p. 23‐51. 

5.  Herrero, M., A. Cifuentes, E. Ibáñez, and M.D. del Castillo, Advanced Analysis of 

Carbohydrates in Foods, in Methods of Analysis of Food Components and Additives, Second 

Edition. 2011, CRC Press. p. 135‐164. 

6.  Sanz, M.L. and I. Martinez‐Castro, Recent developments in sample preparation for 

chromatographic analysis of carbohydrates. Journal of Chromatography A, 2007. 1153(1‐2): 

p. 74‐89. 

7.  Scott, R.P.W., Carbohydrates: Derivatization for GC Analysis, in Encyclopedia of 

Chromatography, Third Edition. 2009, Taylor & Francis. p. 306. 

8.  Sanz, M.L., J. Sanz, and I. Martinez‐Castro, Gas chromatographic‐mass spectrometric method 

for the qualitative and quantitative determination of disaccharides and trisaccharides in 

honey. Journal of Chromatography A, 2004. 1059(1‐2): p. 143‐8. 

9.  Knapp, D.R., Handbook of Analytical Derivatization Reactions. 1979, New York: Wiley. 

10.  Zaikin, V. and J. Halket, A Handbook of Derivatives for Mass Spectrometry. 2009, Chichester: 

IM Publications. 517. 

11.  Orata, F., Derivatization reactions and reagents for gas chromatography analysis, in 

Advanced Gas Chromatography ‐ Progress in Agricultural, Biomedical and Industrial 

Applications. 2012, Dr. Mustafa Ali Mohd, InTech, Available from: 

http://www.intechopen.com/books/advanced‐gaschromatography‐progress‐in‐agricultural‐

biomedical‐and‐industrial‐applications/derivatization‐reactions‐andreagents‐for‐gas‐

chromatography‐analysis. 

12.  MolnarPerl, I. and K. Horvath, Simultaneous quantitation of mono‐, di‐ and trisaccharides as 

their TMS ether oxime derivatives by GC‐MS .1. In model solutions. Chromatographia, 1997. 

45: p. 321‐327. 

13.  Rojas‐Escudero, E., A.L. Alarcon‐Jimenez, P. Elizalde‐Galvan, and F. Rojo‐Callejas, 

Optimization of carbohydrate silylation for gas chromatography. Journal of Chromatography 

A, 2004. 1027(1‐2): p. 117‐120. 

14.  Izydorczyk, M., Understanding the Chemistry of Food Carbohydrates, in Food Carbohydrates. 

2005, CRC Press. 

15.  Wrolstad, R.E., Classifying, Identifying, Naming, and Drawing Sugars and Sugar Derivatives, in 

Food Carbohydrate Chemistry, R.E. Wrolstad, Editor. 2012a, John Wiley & Sons. p. 1‐21. 

16.  Wade, L.G., Organic Chemistry. 2013: Pearson. 

17.  Miljkovic, M., Carbohydrates: synthesis, mechanisms, and stereoelectronic effects. 2009: 

Springer Science & Business Media. 

62 

18.  Soderberg, T., Organic Chemistry with a Biological Emphasis Volume II. 2012. 

19.  Wrolstad, R.E., Reactions of Sugars, in Food Carbohydrate Chemistry, R.E. Wrolstad, Editor. 

2012b, John Wiley & Sons. p. 35‐47. 

20.  French, A.D., Glucose, not cellobiose, is the repeating unit of cellulose and why that is 

important. Cellulose, 2017. 24(11): p. 4605‐4609. 

21.  McNaught, A.D. and A. Wilkinson, Compendium of chemical terminology. Vol. 1669. 1997, 

Oxford: Blackwell Scientific Publications. 

22.  Robyt, J.F., Essentials of carbohydrate chemistry. 1998: Springer Science & Business Media. 

23.  Sjostrom, E., Chapter 7 ‐ WOOD PULPING, in Wood Chemistry (Second Edition), E. Sjostrom, 

Editor. 1993, Academic Press: San Diego. p. 114‐164. 

24.  Horacio, P. and G. Martinez‐Noel, Sucrose signaling in plants: A world yet to be explored. 

Plant Signaling & Behavior, 2013. 8(3): p. e23316. 

25.  Roitsch, T., Source‐sink regulation by sugar and stress. Current Opinion in Plant Biology, 

1999. 2(3): p. 198‐206. 

26.  Smeekens, S., J. Ma, J. Hanson, and F. Rolland, Sugar signals and molecular networks 

controlling plant growth. Current Opinion in Plant Biology, 2010. 13(3): p. 274‐279. 

27.  Coombe, B. The grape berry as a sink. in VI International Symposium on Growth Regulators in 

Fruit Production 239. 1988. 

28.  Moing, A., Sugar alcohols as carbohydrate reserves in some higher plants. Developments in 

Crop Science, 2000. 26: p. 337‐358. 

29.  Lecourieux, F., C. Kappel, D. Lecourieux, A. Serrano, E. Torres, P. Arce‐Johnson, and S. Delrot, 

An update on sugar transport and signalling in grapevine. Journal of Experimental Botany, 

2014. 65(3): p. 821‐832. 

30.  Lastdrager, J., J. Hanson, and S. Smeekens, Sugar signals and the control of plant growth and 

development. Journal of Experimental Botany, 2014. 65(3): p. 799‐807. 

31.  Cook, D., S. Fowler, O. Fiehn, and M.F. Thomashow, A prominent role for the CBF cold 

response pathway in configuring the low‐temperature metabolome of Arabidopsis. 

Proceedings of the National Academy of Sciences of the United States of America, 2004. 

101(42): p. 15243‐15248. 

32.  Hare, P.D., W.A. Cress, and J. Van Staden, Dissecting the roles of osmolyte accumulation 

during stress. Plant, Cell & Environment, 1998. 21(6): p. 535‐553. 

33.  Kaplan, F., J. Kopka, D.Y. Sung, W. Zhao, M. Popp, R. Porat, and C.L. Guy, Transcript and 

metabolite profiling during cold acclimation of Arabidopsis reveals an intricate relationship of 

cold‐regulated gene expression with modifications in metabolite content. The Plant Journal, 

2007. 50(6): p. 967‐981. 

34.  Pillet, J., A. Egert, P. Pieri, F. Lecourieux, C. Kappel, J. Charon, E. Gomès, F. Keller, S. Delrot, 

and D. Lecourieux, VvGOLS1 and VvHsfA2 are Involved in the Heat Stress Responses in 

Grapevine Berries. Plant and Cell Physiology, 2012. 53(10): p. 1776‐1792. 

35.  Griesser, M., G. Weingart, K. Schoedl‐Hummel, N. Neumann, M. Becker, K. Varmuza, F. 

Liebner, R. Schuhmacher, and A. Forneck, Severe drought stress is affecting selected primary 

metabolites, polyphenols, and volatile metabolites in grapevine leaves (Vitis vinifera cv. Pinot 

noir). Plant Physiology and Biochemistry, 2015. 88: p. 17‐26. 

36.  Grimplet, J., M.D. Wheatley, H.B. Jouira, L.G. Deluc, G.R. Cramer, and J.C. Cushman, 

Proteomic and selected metabolite analysis of grape berry tissues under well‐watered and 

water‐deficit stress conditions. Proteomics, 2009. 9(9): p. 2503‐2528. 

63 

37.  Valluru, R. and W. Van den Ende, Myo‐inositol and beyond – Emerging networks under stress. 

Plant Science, 2011. 181(4): p. 387‐400. 

38.  Brummer, Y. and S.W. Cui, Understanding Carbohydrate Analysis, in Food Carbohydrates. 

2005, CRC Press. 

39.  Montilla, A., F.J. Moreno, and A. Olano, A Reliable Gas Capillary Chromatographic 

Determination of Lactulose in Dairy Samples. Chromatographia, 2005. 62(5): p. 311‐314. 

40.  Montilla, A., A.I. Ruiz‐Matute, M.L. Sanz, I. Martínez‐Castro, and M.D. del Castillo, Difructose 

anhydrides as quality markers of honey and coffee. Food Research International, 2006. 39(7): 

p. 801‐806. 

41.  Morales, V., A. Olano, and N. Corzo, Ratio of Maltose to Maltulose and Furosine as Quality 

Parameters for Infant Formula. Journal of Agricultural and Food Chemistry, 2004. 52(22): p. 

6732‐6736. 

42.  Rymowicz, W., W. Kopec, and C.V. Stevens, Primary Production of Raw Materials, in 

Renewable Bioresources. 2004. 

43.  Rudnik, E., Compostable polymer materials. 1st ed. 2008, Oxford ; Boston: Elsevier. xii, 211 p. 

44.  Smith, B.G., L.D. Melton, and R.E. Wrolstad, Plant Cell Wall Polysaccharides, in Food 

Carbohydrate Chemistry. 2012, John Wiley & Sons Inc. p. 135‐146. 

45.  Thomas, S., S. Paul, L. Pothan, and B. Deepa, Natural fibres: structure, properties and 

applications, in Cellulose Fibers: Bio‐and Nano‐Polymer Composites. 2011, Springer. p. 3‐42. 

46.  Bogolitsyna, A., M. Becker, A.‐L. Dupont, A. Borgards, T. Rosenau, and A. Potthast, 

Determination of carbohydrate‐ and lignin‐derived components in complex effluents from 

cellulose processing by capillary electrophoresis with electrospray ionization‐mass 

spectrometric detection. Journal of Chromatography A, 2011. 1218(47): p. 8561‐8566. 

47.  Holmbom, B., A Procedure for Analysis of Toxic Compounds in Pulp and Paper‐Mill Waste‐

Waters. Paperi Ja Puu‐Paper and Timber, 1980. 62(9): p. 523‐531. 

48.  Niemelä, K., Gas‐Liquid Chromatography‐Mass Spectrometry Studies on Pine Kraft Black 

Liquors .3. The Liberation of Carboxylic‐Acids in the Initial Phase of Pulping. Journal of 

Chromatography, 1988. 446: p. 247‐252. 

49.  Koek, M.M., B. Muilwijk, M.J. van der Werf, and T. Hankemeier, Microbial metabolomics with 

gas chromatography/mass spectrometry. Analytical Chemistry, 2006. 78(4): p. 1272‐81. 

50.  Peris‐Tortajada, M., Carbohydrates and starch, in Handbook of Food Analysis, Thirth Edition. 

2015, CRC Press. p. 509‐533. 

51.  Wrolstad, R.E., Food Carbohydrate Chemistry. 2012: John Wiley & Sons Inc. 

52.  Powers, J.R., Application of Enzymes in Food Analysis, in Food Analysis. 2010, Springer US: 

Boston, MA. p. 283‐299. 

53.  Henshall, A., Analysis of starch and other complex carbohydrates by liquid chromatography. 

Cereal foods world (USA), 1996. 

54.  Medeiros, P.M. and B.R. Simoneit, Analysis of sugars in environmental samples by gas 

chromatography‐mass spectrometry. Journal of Chromatography A, 2007. 1141(2): p. 271‐8. 

55.  Alvarez, J.G., Carbohydrates: HPLC Analysis, in Encyclopedia of Chromatography, Third 

Edition. 2009, Taylor & Francis. p. 307‐309. 

56.  Cheng, C., H.‐R. Tsai, and K.‐C. Chang, On‐line cut‐off technique and organic modifier addition 

aided signal enhancement for trace analysis of carbohydrates in cellulase hydrolysate by ion 

exclusion chromatography–electrospray ionization mass spectrometry. Journal of 

Chromatography A, 2006. 1119(1): p. 188‐196. 

64 

57.  Bruggink, C., R. Maurer, H. Herrmann, S. Cavalli, and F. Hoefler, Analysis of carbohydrates by 

anion exchange chromatography and mass spectrometry. Journal of Chromatography A, 

2005. 1085(1): p. 104‐109. 

58.  Jager, A.V., F.G. Tonin, and M.F. Tavares, Comparative evaluation of extraction procedures 

and method validation for determination of carbohydrates in cereals and dairy products by 

capillary electrophoresis. Journal of separation science, 2007. 30(4): p. 586‐594. 

59.  Schmitz, O., Carbohydrates: CE Analysis, in Encyclopedia of Chromatography, Third Edition. 

2009, Taylor & Francis. p. 303‐305. 

60.  Folkes, D. and M. Jordan, Mono‐and disaccharides: analytical aspects, in Carbohydrates in 

Food ‐ 2nd ed., A.‐C. Eliasson, Editor. 2006, CRC Press. 

61.  Andersson, J.T., Detectors, in Practical Gas Chromatography: A Comprehensive Reference, K. 

Dettmer‐Wilde and W. Engewald, Editors. 2014, Springer: Berlin, Heidelberg. p. 205‐248. 

62.  Bouchonnet, S., Introduction to GC‐MS Coupling. 2013: CRC Press. 

63.  Deepak, B. What are the Differences between GC and HPLC? 2014; Available from: http://lab‐

training.com/2014/04/02/what‐are‐the‐differences‐between‐gc‐and‐hplc/. 

64.  McNair, H.M. and J.M. Miller, Basic gas chromatography. 2011: John Wiley & Sons. 

65.  Molnar‐Perl, I., Simultaneous quantitation of acids and sugars by chromatography: gas or 

high‐performance liquid chromatography? Journal of Chromatography A, 1999. 845(1‐2): p. 

181‐195. 

66.  Drozd, J., Chemical derivatization in gas chromatography. Vol. 19. 1986: Elsevier. 

67.  Ruiz‐Matute, A.I., O. Hernandez‐Hernandez, S. Rodriguez‐Sanchez, M.L. Sanz, and I. Martinez‐

Castro, Derivatization of carbohydrates for GC and GC‐MS analyses. Journal of 

Chromatography B, 2011. 879(17‐18): p. 1226‐40. 

68.  García‐Raso, A., M. Fernández‐Díaz, M.I. Páez, J. Sanz, and I. Martínez‐Castro, Gas 

chromatographic retention of carbohydrate trimethylsilyl ethers. III. Ketohexoses. Journal of 

Chromatography A, 1989. 471(C): p. 205‐216. 

69.  Andrews, M.A., Capillary gas‐chromatographic analysis of monosaccharides: improvements 

and comparisons using trifluoroacetylation and trimethylsilylation of sugar O‐benzyl‐ and O‐

methyl‐oximes. Carbohydrate Research, 1989. 194: p. 1‐19. 

70.  Harvey, D.J., Derivatization of carbohydrates for analysis by chromatography; electrophoresis 

and mass spectrometry. Journal of Chromatography B, 2011. 879(17): p. 1196‐1225. 

71.  Purdie, T. and J.C. Irvine, C.‐The alkylation of sugars. Journal of the Chemical Society, 

Transactions, 1903. 83(0): p. 1021‐1037. 

72.  Haworth, W.N., III.—A new method of preparing alkylated sugars. Journal of the Chemical 

Society, Transactions, 1915. 107: p. 8‐16. 

73.  Levery, S.B., Use of Permethylation with GC‐MS for Linkage and Sequence Analysis of 

Oligosaccharides: Historical Perspectives and Recent Developments, in Glycopeptides and 

Related Compounds: Synthesis, Analysis, and Applications, D.G. Large and C.D. Warren, 

Editors. 1997, CRC Press. 

74.  Carpita, N.C. and E.M. Shea, Linkage structure of carbohydrates by gas chromatography‐mass 

spectrometry (GC‐MS) of partially methylated alditol acetates. Analysis of Carbohydrates by 

GLC and MS: p. 157‐216. 

75.  Kuhn, R., H. Trischmann, and I. Löw, Zur permethylierung von zuckern und glykosiden. 

Angewandte Chemie, 1955. 67(1): p. 32‐32. 

76.  Kuhn, R., H.H. Baer, and A. Seeliger, Zur Methylierung Von N‐Acetylglucosamin‐Derivaten. 

Justus Liebigs Annalen der Chemie, 1958. 611(1): p. 236‐241. 

65 

77.  Hakomori, S.‐i., A rapid permethylation of glycolipid, and polysaccharide catalyzed by 

methylsulfinyl carbanion in dimethyl sulfoxide. The Journal of Biochemistry, 1964. 55(2): p. 

205‐208. 

78.  Ciucanu, I. and F. Kerek, A simple and rapid method for the permethylation of carbohydrates. 

Carbohydrate research, 1984. 131(2): p. 209‐217. 

79.  Hirschlag, H. and R. Kӧster, Development of a solid‐phase extraction and derivatization 

method for polar carboxylic acids from aqueous extracts of inorganic multi‐component 

incineration residues. Fresenius J. Anal. Chem., 1998. 263: p. 274‐280. 

80.  Seyferth, D., A.W. Dow, H. Menzel, and T.C. Flood, Trimethylsilyldiazomethane and 

Trimethylsilylcarbene. Journal of the American Chemical Society, 1968. 90(4): p. 1080‐1082. 

81.  Kühnel, E., D.D.P. Laffan, G.C. Lloyd‐Jones, T. Martínez del Campo, I.R. Shepperson, and J.L. 

Slaughter, Mechanism of methyl esterification of carboxylic acids by 

thrimethylsilyldiazomethane. Angew. Chem., 2007. 46: p. 7075‐7078. 

82.  Presser, A. and A. Hufner, Trimethylsilyldiazomethane ‐ A mild and efficient reagent for the 

methylation of carboxylic acids and alcohols in natural products. Monatshefte für Chemie, 

2004. 135(8): p. 1015‐1022. 

83.  Robinson, S., A. Routledge, and J. Thomas‐Oates, Characterisation and proposed origin of 

mass spectrometric ions observed 30 Th above the ionised molecules of per‐O‐methylated 

carbohydrates. Rapid communications in mass spectrometry, 2005. 19(24): p. 3681‐3688. 

84.  Gunner, S.W., J.K. Jones, and M.B. Perry, THE GAS‐LIQUID PARTITION CHROMATOGRAPHY OF 

CARBOHYDRATE DERIVATIVES: PART I THE SEPARATION OF GLYCITOL AND GLYCOSE 

ACETATES. Canadian Journal of Chemistry, 1961. 39(10): p. 1892‐1899. 

85.  VandenHeuvel, W. and E. Horning, Gas chromatographic separations of sugars and related 

compounds as acetyl derivatives. Biochemical and biophysical research communications, 

1961. 4(6): p. 399‐403. 

86.  Vilkas, M., M.G. Boussac, and M.M.‐C. Bonnard, Chromatographie en phase vapeur de sucres 

a l'etat de trifluoracetates. Tetrahedron Letters, 1966. 7(14): p. 1441‐1446. 

87.  Matsuhisa, M., Y. Yamasaki, Y. Shiba, I. Nakahara, A. Kuroda, T. Tomita, M. Iida, M. Ikeda, Y. 

Kajimoto, and M. Kubota, Important role of the hepatic vagus nerve in glucose uptake and 

production by the liver. Metabolism, 2000. 49(1): p. 11‐16. 

88.  Hannestad, U. and A. Lundblad, Accurate and precise isotope dilution mass spectrometry 

method for determining glucose in whole blood. Clinical Chemistry, 1997. 43(5): p. 794‐800. 

89.  Leloux, M.S., E.G. De Jong, and R.A. Maes, Improved screening method for beta‐blockers in 

urine using solid‐phase extraction and capillary gas chromatography—mass spectrometry. 

Journal of Chromatography B: Biomedical Sciences and Applications, 1989. 488(2): p. 357‐

367. 

90.  Pritchard, D.G. and W. Niedermeier, Sensitive gas chromatographic determination of the 

monosaccharide composition of glycoproteins using electron capture detection. Journal of 

Chromatography A, 1978. 152(2): p. 487‐494. 

91.  Chizhov, O.S., B.A. Dmitriev, B.M. Zolotarev, A.Y. Chernyak, and N.K. Kochetov, Mass spectra 

of alditol trifluoroacetates. Organic Mass Spectrometry, 1969. 2(10): p. 947‐952. 

92.  Poole, C.F., W.F. Sye, S. Singhawangcha, F. Hsu, A. Zlatkis, A. Arfwidsson, and J. Vessman, 

New Electron‐Capturing Pentafluorophenyldialkylchlorosilanes as Versatile Derivatizing 

Reagents for Gas‐Chromatography. Journal of Chromatography A, 1980. 199(Oct): p. 123‐

142. 

66 

93.  Evershed, R.P., Advances in silylation, in Handbook of derivatives for chromatography, 2nd 

edn, K. Blau and J.M. Halket, Editors. 1993, Wiley: New York. p. 51–108. 

94.  Kataoka, H., 2.1.2. Gas chromatography of amines as various derivatives, in Journal of 

Chromatography Library, M.‐P. Ibolya, Editor. 2005, Elsevier. p. 364‐404. 

95.  Pierce, A.E., Silylation of Organic Compounds: A Technique for Gasphase Analysis. 1968: 

Pierce Chemical Company. 

96.  Sweeley, C.C., R. Bentley, M. Makita, and W.W. Wells, Gas‐Liquid Chromatography of 

Trimethylsilyl Derivatives of Sugars and Related Substances. Journal of the American 

Chemical Society, 1963. 85(16): p. 2497‐&. 

97.  Dejongh, D.C., T. Radford, J.D. Hribar, Hanessia.S, M. Bieber, G. Dawson, and C.C. Sweeley, 

Analysis of Trimethylsilyl Derivatives of Carbohydrates by Gas Chromatography and Mass 

Spectrometry. Journal of the American Chemical Society, 1969. 91(7): p. 1728‐&. 

98.  García‐Raso, A., I. Martínez‐Castro, M.I. Páez, J. Sanz, J. García‐Raso, and F. Saura‐Calixto, Gas 

chromatographic behaviour of carbohydrate trimethylsilyl ethers. I. Aldopentoses. Journal of 

Chromatography A, 1987. 398(C): p. 9‐20. 

99.  Garcia‐Raso, A., M.I. Paez, I. Martinez‐Castro, J. Sanz, and M.M. Calvo, Gas chromatographic 

retention of carbohydrate trimethylsilyl ethers. IV. Disaccharides. Journal of Chromatography 

A, 1992. 607(2): p. 221‐225. 

100.  Haverkamp, J., J.P. Kamerling, and J.F.G. Vliegenthart, Gas‐liquid chromatography of 

trimethylsilyl disaccharides. Journal of Chromatography A, 1971. 59(2): p. 281‐287. 

101.  Kakehi, K. and S. Honda, Silyl ethers of carbohydrates. Analysis of Carbohydrates by GLC and 

MS, 1989: p. 43‐85. 

102.  Martínez‐Castro, I., M.I. Páez, J. Sanz, and A. García‐Raso, Gas chromatographic behaviour of 

carbohydrate trimethylsilyl ethers. II. Aldohexoses. Journal of Chromatography A, 1989. 462: 

p. 49‐60. 

103.  Butts, W.C., Two‐column gas chromatography of trimethylsilyl derivatives of biochemically 

significant compounds. Analytical biochemistry, 1972. 46(1): p. 187‐199. 

104.  Stalling, D.L., C.W. Gehrke, and R.W. Zumwalt, A New Silylation Reagent for Amino Acids Bis 

(Trimethylsilyl) Trifluoracetamide (Bstfa). Biochemical and Biophysical Research 

Communications, 1968. 31(4): p. 616‐&. 

105.  Gehrke, C.W. and K. Leimer, Trimethylsilylation of amino acids derivatization and 

chromatography. Journal of Chromatography A, 1971. 57: p. 219‐238. 

106.  Tallent, W. and R. Kleiman, Bis (trimethylsilyl) acetamide in the silylation of lipolysis products 

for gas‐liquid chromatography. Journal of lipid research, 1968. 9(1): p. 146‐148. 

107.  Makita, M. and W.W. Wells, Quantitative analysis of fecal bile acids by gas‐liquid 

chromatography. Analytical Biochemistry, 1963. 5(6): p. 523‐530. 

108.  Zhang, Z., A. Hibberd, and J. Zhou, Optimisation of derivatisation for the analysis of 

estrogenic compounds in water by solid‐phase extraction gas chromatography–mass 

spectrometry. Analytica Chimica Acta, 2006. 577(1): p. 52‐61. 

109.  Mason, P.S. and E.D. Smith, A Quantitative Study of Reagents and Procedures for Synthesis of 

Trimethylsilyl Derivatives. Journal of Gas Chromatography, 1966. 4(11): p. 398‐&. 

110.  Anumula, K.R. and P.B. Taylor, A Comprehensive Procedure for Preparation of Partially 

Methylated Alditol Acetates from Glycoprotein Carbohydrates. Analytical Biochemistry, 1992. 

203(1): p. 101‐108. 

111.  Held, I., A. Villinger, and H. Zipse, The stability of acylpyridinium cations and their relation to 

the catalytic activity of pyridine bases. Synthesis, 2005(9): p. 1425‐1430. 

67 

112.  Chaudhary, S. and O. Hernandez, 4‐Dimethylaminopyridine: an efficient and selective catalyst 

for the silylation of alcohols. Tetrahedron Letters, 1979. 20(2): p. 99‐102. 

113.  Kochetkov, N.K. and C.O. S., Mass spectrometry of carbohydrates, in Methods in 

Carbohydrate Chemistry. Vol. VI: General Carbohydrate Methods., R.L. Whistler and J.N. 

BeMiller, Editors. 1972, Academic Press: New York. p. 540‐554. 

114.  Adams, M.A., Z.L. Chen, P. Landman, and T.D. Colmer, Simultaneous determination by 

capillary gas chromatography of organic acids, sugars, and sugar alcohols in plant tissue 

extracts as their trimethylsilyl derivatives. Analytical Biochemistry, 1999. 266(1): p. 77‐84. 

115.  Bartolozzi, F., G. Bertazza, D. Bassi, and G. Cristoferi, Simultaneous determination of soluble 

sugars and organic acids as their trimethylsilyl derivatives in apricot fruits by gas‐liquid 

chromatography. Journal of Chromatography A, 1997. 758(1): p. 99‐107. 

116.  Katona, Z.F., P. Sass, and I. Molnar‐Perl, Simultaneous determination of sugars, sugar 

alcohols, acids and amino acids in apricots by gas chromatography‐mass spectrometry. 

Journal of Chromatography A, 1999. 847(1‐2): p. 91‐102. 

117.  Fuzfai, Z., Z.F. Katona, E. Kovacs, and I. Molnar‐Perl, Simultaneous identification and 

quantification of the sugar, sugar alcohol, and carboxylic acid contents of sour cherry, apple, 

and ber fruits, as their trimethylsilyl derivatives, by gas chromatography‐mass spectrometry. 

Journal of Agricultural and Food Chemistry, 2004. 52(25): p. 7444‐7452. 

118.  Willför, S., A. Pranovich, T. Tamminen, J. Puls, C. Laine, A. Suurnakki, B. Saake, K. Uotila, H. 

Simolin, J. Hemming, and B. Holmbom, Carbohydrate analysis of plant materials with uronic 

acid‐containing polysaccharides‐A comparison between different hydrolysis and subsequent 

chromatographic analytical techniques. Industrial Crops and Products, 2009. 29(2‐3): p. 571‐

580. 

119.  Mejanelle, P., J. Bleton, A. Tchapla, and S. Goursaud, Chapter 24 Gas chromatography‐mass 

spectrometric analysis of monosaccharides after methanolysis and trimethylsilylation. 

Potential for the characterization of substances of vegetal origin: Application to the study of 

museum objects, in Journal of Chromatography Library, R. Ziad El, Editor. 2002, Elsevier. p. 

845‐902. 

120.  Sundberg, A., A.V. Pranovich, and B. Holmbom, Chemical characterization of various types of 

mechanical pulp fines. Journal of Pulp and Paper Science, 2003. 29(5): p. 6. 

121.  Sundberg, A., K. Sundberg, C. Lillandt, and B. Holmbom, Determination of hemicelluloses and 

pectins in wood and pulp fibres by acid methanolysis and gas chromatography. Nordic Pulp 

and Paper Research Journal, 1996. 11(4): p. 216‐219+226. 

122.  Bertaud, F., A. Sundberg, and B. Holmbom, Evaluation of acid methanolysis for analysis of 

wood hemicelluloses and pectins. Carbohydrate Polymers, 2002. 48(3): p. 319‐324. 

123.  Savary, B. and A. Nunez, Gas chromatography–mass spectrometry method for determining 

the methanol and acetic acid contents of pectin using headspace solid‐phase microextraction 

and stable isotope dilution. Journal of Chromatography A, 2003. 1017(1‐2): p. 151‐159. 

124.  Willför, S., A. Sundberg, A. Pranovich, and B. Holmbom, Polysaccharides in some industrially 

important hardwood species. Wood Science and Technology, 2005. 39(8): p. 601‐617. 

125.  Huang, Y., L. Indrarti, J. Azuma, and K. Okamura, Simultaneous Determination of Xylose and 

Uronic‐Acid in Beech Xylan by Methanolysis. Mokuzai Gakkaishi, 1992. 38(12): p. 1168‐1171. 

126.  Horvath, K. and I. Molnar‐Perl, Simultaneous GC‐MS quantitation of o‐phosphoric, aliphatic 

and aromatic carboxylic acids, proline, hydroxymethylfurfurol and sugars as their TMS 

derivatives: In honeys. Chromatographia, 1998. 48(1‐2): p. 120‐126. 

68 

127.  Senila, L., A. Gog, M. Senila, C. Roman, and L. Silaghi‐Dumitrescu, Development of a GC‐MS 

method for 5‐hydroxymethylfurfural determination in wood after steam‐explosion 

pretreatment. Revista de Chimie, 2012. 63(6): p. 559‐563. 

128.  Proestos, C., I. Boziaris, G.‐J. Nychas, and M. Komaitis, Analysis of flavonoids and phenolic 

acids in Greek aromatic plants: Investigation of their antioxidant capacity and antimicrobial 

activity. Food Chemistry, 2006. 95(4): p. 664‐671. 

129.  Fuzfai, Z. and I. Molnarperl, Gas chromatographic–mass spectrometric fragmentation study 

of flavonoids as their trimethylsilyl derivatives: Analysis of flavonoids, sugars, carboxylic and 

amino acids in model systems and in citrus fruits. Journal of Chromatography A, 2007. 

1149(1): p. 88‐101. 

130.  Molnar‐Perl, I. and Z.F. Katona, GC‐MS of amino acids as their trimethylsilyl/t‐

butyldimethylsilyl derivatives: in model solutions III. Chromatographia, 2000. 51(1): p. S228‐

S236. 

131.  Laine, C., T. Tamminen, A. Vikkula, and T. Vuorinen, Methylation analysis as a tool for 

structural analysis of wood polysaccharides. Holzforschung, 2002. 56(6): p. 607‐614. 

132.  Amelung, W., M.V. Cheshire, and G. Guggenberger, Determination of neutral and acidic 

sugars in soil by capillary gas‐liquid chromatography after trifluoroacetic acid hydrolysis. Soil 

Biology & Biochemistry, 1996. 28(12): p. 1631‐1639. 

133.  Levene, P.A. and E.T. Stiller, Acetone Derivatives of D‐Ribose. Journal of Biological Chemistry, 

1933. 102(1): p. 187‐201. 

134.  Freudenberg, K., W. Dürr, and H. von Hochstetter, Zur Kenntnis der Aceton‐Zucker, XIII.: Die 

Hydrolyse einiger Disaccharide, Glucoside und Aceton‐Zucker. Berichte der deutschen 

chemischen Gesellschaft (A and B Series), 1928. 61(8): p. 1735‐1743. 

135.  De Jongh, D.C. and K. Biemann, Mass spectra of O‐isopropylidene derivatives of pentoses and 

hexoses. Journal of the American Chemical Society, 1964. 86(1): p. 67‐74. 

136.  Morgenlie, S., Analysis of mixtures of the common aldoses by gas chromatography‐mass 

spectrometry of their O‐isopropylidene derivatives. Carbohydrate Research, 1975. 41(1): p. 

285‐289. 

137.  Morgenlie, S., Gas chromatography‐mass spectrometry of hexuloses and pentuloses as their 

O‐isopropylidene derivatives: analysis of product mixtures from triose aldol‐condensations. 

Carbohydrate Research, 1980. 80(2): p. 215‐222. 

138.  Blau, K. and J.M. Halket, Handbook of Derivatives for Chromatography. 1993, Wiley, New 

York. 

139.  Sawardeker, J.S., J.H. Sloneker, and A. Jeanes, Quantitative Determination of 

Monosaccharides as Their Alditol Acetates by Gas Liquid Chromatography. Analytical 

Chemistry, 1965. 37(12): p. 1602‐1604. 

140.  Blakeney, A.B., P.J. Harris, R.J. Henry, and B.A. Stone, A simple and rapid preparation of 

alditol acetates for monosaccharide analysis. Carbohydrate Research, 1983. 113(2): p. 291‐

299. 

141.  Fox, A., S.L. Morgan, and J. Gilbart, Preparation of alditol acetates and their analysis by gas 

chromatography (GC) and mass spectrometry (MS). Analysis of Carbohydrates by GLC and 

MS, 1989: p. 87‐117. 

142.  Urbani, R., S. Paola, G. Pletikapić, T.M. Radić, V. Svetličić, and V. Žutić, Diatom 

Polysaccharides: Extracellular Production, Isolation and Molecular Characterization, in The 

Complex World of Polysaccharides. 2012, InTech. 

69 

143.  McGinnis, G.D., Preparation of aldononitrile acetates using N‐methylimidazole as catalyst 

and solvent. Carbohydrate Research, 1982. 108(2): p. 284‐292. 

144.  Matsui, M., M. Okada, T. lmanari, and Z. Tamura, Gas chromatography of trifluoroacetyl 

derivatives of alditols and trimethylsilyl derivatives of aldonolactones. Chemical and 

Pharmaceutical Bulletin, 1968. 16(7): p. 1383‐1387. 

145.  Walla, M.D., P.Y. Lau, S.L. Morgan, A. Fox, and A. Brown, Capillary gas chromatography‐mass 

spectrometry of carbohydrate components of legionellae and other bacteria. Journal of 

Chromatography A, 1984. 288: p. 399‐413. 

146.  Neeser, J.R. and T.F. Schweitzer, Analysis of carbohydrates as o‐alkyloxime derivatives by gas‐

liquid chromatography (GLC), in Analysis of Carbohydrates by GLC and MS. 1989, CRC Boca 

Raton, FL. p. 143‐156. 

147.  Steinberg, P. and A. Fox, Automated derivatization instrument: Preparation of alditol acetates 

for analysis of bacterial carbohydrates using gas chromatography/mass spectrometry. 

Analytical chemistry, 1999. 71(9): p. 1914‐1917. 

148.  Brunton, N., T. Gormley, and B. Murray, Use of the alditol acetate derivatisation for the 

analysis of reducing sugars in potato tubers. Food chemistry, 2007. 104(1): p. 398‐402. 

149.  Englyst, H.N. and J.H. Cummings, Simplified method for the measurement of total non‐starch 

polysaccharides by gas‐liquid chromatography of constituent sugars as alditol acetates. 

Analyst, 1984. 109(7): p. 937‐942. 

150.  Crowell, E.P. and B.B. Burnett, Determination of the carbohydrate composition of wood pulps 

by gas chromatography of the alditol acetates. Analytical Chemistry, 1967. 39(1): p. 121‐124. 

151.  Björndal, H., B. Lindberg, and S. Svensson, Mass spectrometry of partially methylated alditol 

acetates. Carbohydrate Research, 1967. 5(4): p. 433‐440. 

152.  McGinnis, G.D. and C.J. Biermann, Analysis of monosaccharides as per‐O‐acetylated 

aldononitrile (PAAN) derivatives by gas–liquid chromatography (GLC), in Analysis of 

carbohydrates by GLC and MS. 1989, CRC Press, Inc. Boca Raton, Florida. p. 119‐125. 

153.  Lance, D. and J. Jones, Gas chromatography of derivatives of the methyl ethers of d‐xylose. 

Canadian Journal of Chemistry, 1967. 45(17): p. 1995‐1998. 

154.  Chen, C.C. and G.D. McGinnis, The use of 1‐methylimidazole as a solvent and catalyst for the 

preparation of aldononitrile acetates of aldoses. Carbohydrate Research, 1981. 90(1): p. 127‐

130. 

155.  Guerrant, G.O. and C.W. Moss, Determination of monosaccharides as aldononitrile, O‐

methyloxime, alditol, and cyclitol acetate derivatives by gas chromatography. Analytical 

Chemistry, 1984. 56(4): p. 633‐638. 

156.  Seymour, F.R., S.L. Unruh, and D.A. Nehlich, Quantitation of free sugars in plant tissue by Glc 

of their peracetylated aldononitrile and ketoxime derivatives. Carbohydrate research, 1989. 

191(2): p. 175‐189. 

157.  Jackson, L.K., M.E. Slodki, R.D. Plattner, K.A. Burton, and M.C. Cadmus, Capsular and 

extracellular polysaccharides from Rhizobium microsymbionts of Acacia decurrens. 

Carbohydrate Research, 1982. 110(2): p. 267‐276. 

158.  Lawrence, J.F. and J.R. Iyengar, Gas chromatographic determination of polysaccharide gums 

in foods after hydrolysis and derivatization. Journal of Chromatography A, 1985. 350: p. 237‐

244. 

159.  Ye, F., X. Yan, J. Xu, and H. Chen, Determination of aldoses and ketoses by GC‐MS using 

differential derivatisation. Phytochemical Analysis, 2006. 17(6): p. 379‐383. 

70 

160.  Zhang, W., H. He, and X. Zhang, Determination of neutral sugars in soil by capillary gas 

chromatography after derivatization to aldononitrile acetates. Soil Biology and Biochemistry, 

2007. 39(10): p. 2665‐2669. 

161.  Lehrfeld, J., Separation of Some Perbenzoylated Carbohydrates by High‐Performance Liquid‐

Chromatography. Journal of Chromatography A, 1976. 120(1): p. 141‐147. 

162.  Laine, R.A. and C.C. Sweeley, O‐methyl oximes of sugars. Analysis as O‐trimethylsilyl 

derivatives by gas‐liquid chromatography and mass spectrometry. Carbohydrate Research, 

1973. 27(1): p. 199‐213. 

163.  Mason, B.S. and H.T. Slover, A gas chromatographic method for the determination of sugars 

in foods. Journal of Agricultural and Food Chemistry, 1958. 6(6): p. 551‐554. 

164.  Fiehn, O., J. Kopka, R.N. Trethewey, and L. Willmitzer, Identification of uncommon plant 

metabolites based on calculation of elemental compositions using gas chromatography and 

quadrupole mass spectrometry. Analytical Chemistry, 2000. 72(15): p. 3573‐3580. 

165.  Funcke, W. and C. von Sonntag, Syn and Anti Forms of Some Monosaccharide O‐Methyl 

Oximes ‐ C‐13‐Nmr and Glc Study. Carbohydrate Research, 1979. 69(Mar): p. 247‐251. 

166.  Snyder, J.R., An Nmr Investigation of the Aldopentose Oximes. Carbohydrate Research, 1990. 

198(1): p. 1‐13. 

167.  Petersson, G., Gas‐chromatographic analysis of sugars and related hydroxy acids as acyclic 

oxime and ester trimethylsilyl derivatives. Carbohydrate Research, 1974. 33(1): p. 47‐61. 

168.  Chalmers, R.A. and R.W.E. Watts, Derivatives for Identification and Quantitative‐

Determination of Some Keto and Aldo‐Carboxylic Acids by Gas‐Liquid Chromatography. 

Analyst, 1972. 97(1161): p. 951‐957. 

169.  Rumpel, C. and M.F. Dignac, Chromatographic analysis of monosaccharides in a forest soil 

profile: Analysis by gas chromatography after trifluoroacetic acid hydrolysis and reduction‐

acetylation. Soil Biology & Biochemistry, 2006. 38(6): p. 1478‐1481. 

170.  Schweer, H., Gas Chromatography‐Mass Spectrometry of Aldoses as O‐Methoxime, O‐2‐

Methyl‐2‐Propoxime and O‐Normal‐Butoxime Pertrifluoroacetyl Derivatives on Ov‐225 with 

Methylpropane as Ionization Agent .1. Pentoses. Journal of Chromatography, 1982. 236(2): p. 

355‐360. 

171.  Bose, S.K., V.A. Barber, E.F. Alves, D.J. Kiemle, A.J. Stipanovic, and R.C. Francis, An improved 

method for the hydrolysis of hardwood carbohydrates to monomers. Carbohydrate Polymers, 

2009. 78(3): p. 396‐401. 

172.  Becker, M., T. Zweckmair, A. Forneck, T. Rosenau, A. Potthast, and F. Liebner, Evaluation of 

different derivatisation approaches for gas chromatographic‐mass spectrometric analysis of 

carbohydrates in complex matrices of biological and synthetic origin. Journal of 

Chromatography A, 2013. 1281: p. 115‐26. 

173.  Massiot, D., F. Fayon, M. Capron, I. King, S. Le Calvé, B. Alonso, J.‐O. Durand, B. Bujoli, Z. Gan, 

and G. Hoatson, Modelling one‐ and two‐dimensional solid‐state NMR spectra. Magnetic 

Resonance in Chemistry, 2002. 40(1): p. 70‐76. 

174.  Larsson, P.T., K. Wickholm, and T. Iversen, A CP/MAS13C NMR investigation of molecular 

ordering in celluloses. Carbohydrate Research, 1997. 302(1–2): p. 19‐25. 

175.  Kelly, D.R., S.C. Baker, D.S. King, D.S. de Silva, G. Lord, and J.P. Taylor, Studies of nitrile oxide 

cycloadditions, and the phenolic oxidative coupling of vanillin aldoxime by Geobacillus sp. 

DDS012 from Italian rye grass silage. Organic & Biomolecular Chemistry, 2008. 6(4): p. 787‐

796. 

71 

176.  Strlič, M., I.K. Cigić, A. Možir, D. Thickett, G. De Bruin, J. Kolar, and M. Cassar, Test for 

Compatibility with Organic Heritage materials ‐ A Proposed Procedure. e‐Preservation 

Science, 2010. 7: p. 78‐86  

177.  Rohrling, J., A. Potthast, T. Rosenau, T. Lange, G. Ebner, H. Sixta, and P. Kosma, A novel 

method for the determination of carbonyl groups in cellulosics by fluorescence labeling. 1. 

Method development. Biomacromolecules, 2002. 3(5): p. 959‐68. 

178.  Rohrling, J., A. Potthast, T. Rosenau, T. Lange, A. Borgards, H. Sixta, and P. Kosma, A novel 

method for the determination of carbonyl groups in cellulosics by fluorescence labeling. 2. 

Validation and applications. Biomacromolecules, 2002. 3(5): p. 969‐75. 

179.  Oberlerchner Josua, T., S. Böhmdorfer, T. Rosenau, and A. Potthast, A matrix‐resistant HPTLC 

method to quantify monosaccharides in wood‐based lignocellulose biorefinery streams, in 

Holzforschung. 2018. p. 645. 

 

72 

List of Figures 

Figure 1: D‐glucose forms five different epimers in solution with a specific equilibrium distribution 

(brackets). According to [14], modified. ....................................................................................... 4 

Figure 2: Monosaccharide linkage of the non‐reducing disaccharide sucrose (α‐D‐glucopyranosyl‐

(12)‐β‐D‐fructofuranoside) and the reducing disaccharide cellobiose (β‐D‐glucopyranosyl‐

(14)‐D‐glucopyranose). .............................................................................................................. 6 

Figure 3: The approximate analyte polarity and molecular mass (Mr) ranges for GC‐MS, LC‐MS, and 

MALDI indicating some desired effects of chemical derivatization. A: derivatization for GC‐MS, 

B: derivatization for ESI and MALDI (EI, CI, ES, MALDI; LD). Source: Reproduced from Zaikin and 

Halket, 2009, with permission. ................................................................................................... 15 

Figure 4: Etherification reaction of glucose using methyl iodide and silver oxide as a catalyst to form 

permethylated glucose (2,3,4,6‐tetra‐O‐methyl‐D‐glucopyranose) [71]. ................................... 20 

Figure 5: Acylation of a hydroxyl group using N‐methyl‐bis(trifluoroacetamide (MBTFA) reagent, 

according to [10]. ........................................................................................................................ 22 

Figure 6: Silylation reaction of a hydroxyl group using N,O‐bis(trimethylsilyl)trifluoroacetamide 

(BSTFA), according to Gerhke and Leimer [105] and Tallent and Kleiman [106]. ....................... 24 

Figure 7: The catalytic function of pyridine for silylation of hydroxyl groups, according to Held et al. 

[111]. ........................................................................................................................................... 24 

Figure 8: The catalytic reaction of 4‐dimethylamino pyridine (DMAP) for silylation of hydroxyl groups, 

according to Chaudhary and Hernandez [112]. .......................................................................... 25 

Figure 9: O‐Isopropylidenation of α‐D‐glucofuranose form 1,2:5,6‐di‐O‐isopropylidene‐D‐

glucofuranose, according to [22, 133, 135]. ............................................................................... 27 

Figure 10: Formation of alditol acetate derivatives, according to the original method described by 

Gunner et al. [84] and applied by Urbani et al. [142]. ................................................................ 29 

Figure 11: Derivatization reaction of the formation of per‐acetylated aldononitriles (PAAN) from α‐D‐

glucopyranose (aldoses), according to [152], modified. ............................................................. 31 

Figure 12: General reaction scheme for the derivatization of carbonyls to oximes. ............................ 32 

Figure 13: Derivatization of glucose: (1) converting glucose in the corresponding syn‐ and anti‐O‐

methyloxime isomers by methoxyamine hydrochloride (CH3ONH2 ∙ HCl) and subsequently 

trimethyl‐silylation. ..................................................................................................................... 33 

Figure 14: Number of compounds identified within the different compound classes of the bleaching 

effluents analytes, according to various GC‐MS derivatization approaches. ............................. 45 

Figure 15: Gas chromatograms with MS detection of a mixture of 46 carbohydrates and 

carbohydrate‐derived compounds converted to different derivatization products. ................. 48 

Figure 16: Retention time of syn peaks (■) and retention time shift (●) between the syn peak and 

corresponding anti peak or peak 1 and peak 2 of different EtOx‐TMS derivatized 

carbohydrates. ............................................................................................................................ 49 

Figure 17: Released carbohydrates by acid methanolysis (total amount, glucose and xylose) of 

untreated, electron beam (e‐beam) treatment (eucalyptus and hemp paper pulps: 120 kGy, 

and rag paper: 60 kGy) and artificial aged cellulose samples. .................................................... 52 

Figure 18: Samples of storage materials in headspace vials for emission analysis. .............................. 55 

 

 

73 

Appendix 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

74 

    

 Paper I   Determination of carbohydrate‐ and lignin‐derived 

components in complex effluents from cellulose processing by 

capillary electrophoresis with electrospray ionization‐mass 

spectrometric detection. 

 

 Journal of Chromatography A, vol. 1218, pp. 8561‐8566. 

 

   Bogolitsyna, A.; Becker, M.; Dupont, A.‐L.; Borgards, A.; Rosenau, T.; Potthast, A. (2011)        

Dee

ATa

Vb

c

a

ARR2AA

KCsCLPP

1

timlaTsf

mba

0d

Journal of Chromatography A, 1218 (2011) 8561– 8566

Contents lists available at SciVerse ScienceDirect

Journal of Chromatography A

jou rn al h om epage: www.elsev ier .com/ locat e/chroma

etermination of carbohydrate- and lignin-derived components in complexffluents from cellulose processing by capillary electrophoresis withlectrospray ionization-mass spectrometric detection

nna Bogolitsynaa, Manuel Beckera, Anne-Laurence Dupontb, Andrea Borgardsc,homas Rosenaua, Antje Potthasta,∗

Department of Chemistry and Christian Doppler Laboratory “Advanced cellulose chemistry and analytics”, University of Natural Resources and Life Sciences, Muthgasse 18, A-1190ienna, AustriaCentre de recherches sur la conservation des collections, Muséum national d’Histoire naturelle, CNRS USR 3224, 36, rue Geoffroy-Saint-Hilaire, F-75005 Paris, FranceProcess Innovation, Lenzing AG, A-4860 Lenzing, Austria

r t i c l e i n f o

rticle history:eceived 30 May 2011eceived in revised form2 September 2011ccepted 23 September 2011vailable online 29 September 2011

eywords:apillary electrophoresis-mass

a b s t r a c t

Degradation products from lignocellulosic materials receive increasing attention due to the continuouslygrowing interest in their utilization. The inherent structural variance of lignocellulosics combined withthe intricacy of lignocellulosic processing (e.g. pulping of wood and bleaching of cellulosic pulps) and thecomplexity of degradation processes occurring therein result in rather complex mixtures in the processstreams and effluents that contain a large quantity of structurally different degradation products. This istrue for most processing steps, but also for degradation reactions occurring during aging of lignocellulosicmaterials, such as paper, cellulosic tissue or textiles. In order to render such mixtures better analyticallyaccessible than hitherto possible a CE-ESI-MS method was established for the simultaneous determina-

pectrometryarboxylic acidsignocellulosesulp bleaching effluentaper aging

tion of aliphatic carboxylic acids from the degradation of (hemi)celluloses and aromatic compounds fromlignin degradation. CE and ESI-MS parameters have been optimized towards sensitivity and good repro-ducibility. The method was tested in two real-world scenarios: the determination of major componentsin effluents from bleaching stages in the pulp and paper industry, and the analysis of degradation prod-ucts in extracts of naturally aged papers. The advantages and drawbacks of this approach are criticallydiscussed.

© 2011 Elsevier B.V. All rights reserved.

. Introduction

The analysis of lignocellulose degradation and identification ofhe products formed from these materials are still posing analyt-cal challenges, especially due to the complexity of the reaction

ixtures. This applies to nearly all byproduct streams in lignocel-ulose processing, no matter whether related to “classical” pulpnd paper processing or to more recent biorefinery scenarios.

he degradation products in industrial effluents after differentteps of pulp bleaching have large similarities to the productsound in extracts of aged paper. The mixtures of degradation

Abbreviations: CE, capillary electrophoresis; ESI-MS, electrospray ionization-ass spectrometry; GC–MS, gas chromatography–mass spectrometry; BGE,

ackground electrolyte; LMM, low molecular mass; MT, migration time; PA, peakrea.∗ Corresponding author. Tel.: +43 1 47654 6071; fax: +43 1 47654 6059.

E-mail address: antje.potthast@boku.ac.at (A. Potthast).

021-9673/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2011.09.063

products of low-molecular weight carbohydrates under stronglyalkaline, alkaline-oxidative or acidic-oxidative conditions arecomplex, being formed by superposition of multiple fragmentation,rearrangement, and condensation reactions. These carbohydrate-derived products come along with fragmentation products of(residual) lignin. In the pulp and paper industries, such degrada-tion products contribute significantly to the spent liquor streams,and hence also to the corresponding organic effluent load (totalorganic carbon). Degradation reactions similar to those enforcedduring the pulp bleaching stages by drastic reaction environmentsare proceeding also under ambient conditions over longer timesupon natural or artificially accelerated paper aging. Also here, car-bohydrates and lignins are fragmented and degraded into verycomplex compound mixtures. Degradation processes in paper aredetermined by numerous factors including the material of the

paper (fiber type, sizing, fillers, etc.) as well as storage condi-tions (temperature, relative humidity, acids, pollutants, etc.). As aresult of such degradation, mechanical and optical properties ofpaper can suffer dramatically [1]. More detailed information on the

8 matog

dgad

aadrei(Tv[l[

eoaIamT[

slbvpc[imta

anmno(pdp

2

2

wS

ai((aatpt

562 A. Bogolitsyna et al. / J. Chro

egradation mechanisms and the reaction products formed is ofreat significance for a suitable conservation treatment of papernd in summary for the general understanding of lignocelluloseegradation.

For the separation of degradation products of (hemi)celluloses,n analytical methodology is required which shows high sensitivitynd robustness as well as the ability to simultaneously determineegradation products of both carbohydrates and lignin. The task isendered even more complicated by the complexity in case of efflu-nt mixtures, the matrix and the rather low concentrations of itsndividual components. Gas chromatography/mass spectrometryGC–MS) is often applied for the analysis of such complex mixtures.his method requires preliminary sample derivatization to produceolatile analytes, e.g. by esterification, etherification, silylation, etc.2–4]. This is not only a time-consuming procedure, but might effectosses during sample preparation, and cause discrimination effects5–7].

In the recent years capillary electrophoresis (CE) has beenstablished as a fast and suitable method for both the analysisf carbohydrates, their degradation products (aliphatic carboxyliccids) and the determination of phenolic and aromatic compounds.t provides short analysis times and, at the same time, high sep-ration efficiencies without a preliminary derivatization step. Inost cases UV-detection is applied after separation by CE [8–11].

his method was used to study effluents of pulping woody biomass12–14], which offers complex matrices.

Hyphenation to MS provides more information on the actualtructures of the analytes. CE-MS was applied for the analysis ofow molecular mass carboxylic acids in atmospheric particles [7],everages [5,15], biological fluids [16,17], urban atmosphere andehicle emissions [6]. CE-MS was used further for the analysis ofhenolic compounds (with similar structures to the lignin-derivedompounds) in biomass pyrolysis and burning [18], virgin olive oil19,20], walnut [21] and atmospheric aerosols [22]. Pulp bleach-ng effluents and celluloses aging extracts have been studied by CE

uch less extensively, since due to dilution effects of the processhe overall concentrations – at similarly complex compositions –re significantly lower.

In this study, we would like to communicate our efforts towards CE-MS-based analytical technique for the simultaneous determi-ation of carbohydrate-derived reaction products, such as aliphaticono- and di-carboxylic acids and hydroxy-acids, as well as phe-

olic lignin-fragments in effluents of pulp bleaching and extractsf aged paper artifacts. Together with our parallel study onhemi)cellulose and lignin degradation products in artificially agedapers [23], this is the first report on a MS-hyphenated method forirect analysis of the complex pulp bleaching effluents and agedaper extracts.

. Materials and methods

.1. Chemicals

All chemicals were of the highest purity available and were usedithout further purification. Ultrahigh quality water (HPLC grade,

igma–Aldrich) was used for all aqueous solutions.The chemicals were obtained from the following suppliers:

mmonium formate (97%), ammonium hydroxide solution (≥25%n water) and sodium hydroxide (≥98%) from Sigma–Aldrich–FlukaSchnelldorf, Germany), 2-propanol (99.9%) from Fisher ScientificGermany). All the model compounds (either as the free acids ors their sodium salts) were obtained from Sigma–Aldrich–Fluka,

t a purity of 97% or better except lactic acid (90%). Stock solu-ions of each model compound with a concentration of 1 g/l wererepared in deionized water and stored at 4 ◦C. Solid phase extrac-ion cartridges SupelTM-Select HLB SPE 1 g/20 ml were purchased

r. A 1218 (2011) 8561– 8566

from Supelco (Bellefonte, PA, USA). Phenex polytetrafluoroethy-lene (PTFE) syringe filters with pore size of 0.2 and 0.45 �m andvarious diameters were supplied by Phenomenex (Aschaffenburg,Germany).

2.2. Sample preparation

2.2.1. Pulp bleaching effluentsThe pulp bleaching effluent sample was obtained from com-

bined effluents of a totally chlorine free (TCF) bleaching ofhardwood sulfite pulp, prior to the biological effluent treatment,from Lenzing AG, Lenzing, Austria. The sample was stored at 4 ◦Cand was warmed to room temperature before sample preparation,which started with vacuum filtration through 0.4 �m membranefilters. As the concentration of analytes was too low for thedirect analysis, effluent samples were concentrated by solid-phaseextraction (SPE) according the following procedure [24]: the SPEcartridge was conditioned with 10 ml methanol and 10 ml ofdeionized water, 200 ml of sample was loaded with 1 drop persecond. Salts are not retained on the SPE cartridge, which thusallows the separation of organic analytes from salts in the load-ing step. Carboxylic acids were eluted from the cartridge with amethanol/acetonitrile mixture (v/v = 1:1). The organic solvent wasremoved in vacuum under exclusion of oxygen, and the samplewas redissolved in deionized water. This removes long-chain car-boxylic acids in amounts according to their solubility in water[25]: acids with up to 10–12 carbon atoms are still in the sam-ple, larger ones are completely removed. This step is necessaryto protect the separation capillary from clogging. Evaporation ofthe organic solvent did not cause changes in the analyte compo-sition which was proved by GC–MS analyses of silylated samplesbefore and after the treatment (recovery above 90%). Prior toinjection, the sample was filtered through a 0.2 �m membranefilter.

2.2.2. Aged paper extractsAqueous extracts from an old book and aged papers were pre-

pared as follows: 17 g of air dry book paper were extracted in 230 mlof deionized water containing 10% methanol for 45 h. The extractwas brought to a volume of about 25 ml by freeze-drying and wasfiltered through a 0.2 �m membrane filter prior to injection.

Characteristics of the book paper: publication year 1924, brit-tle paper, double fold number: 2. The weighted average molecularweight (Mw) of the cellulose as determined by GPC in DMAc/LiCl[26] was between 100 and 120 kg/mol.

2.3. CE-ESI-MS

CE-MS analysis was performed on a G1600 Agilent capil-lary electrophoresis system (Agilent Technologies, Waldbronn,Germany) in combination with an Agilent 6320 series ion trap massspectrometer equipped with an Agilent CE-ESI-MS sprayer (Agi-lent Technologies). For separation, a fused-silica capillary (AgilentTechnologies) with a total length of 60 cm and an inner diameterof 50 �m was used. For maintaining constant performance overtime, the capillary was flushed daily with 0.1 M sodium hydroxidefor 10 min, water for 10 min and background electrolyte (BGE) for5 min. Between the actual runs, the capillary was preconditionedby flushing 5 min with water and 5 min with BGE. BGE and sampleswere filtered through 0.2 �m membranes. Hydrodynamic injec-tion was used at 50 mbar for 10 s followed by injection of buffer

at 50 mbar for 5 s. The separation voltage was 20 kV, the resultingcurrent was 20 �A. The capillary was thermostated at 25 ◦C.

The sheath liquid was delivered by an Agilent 1200 series iso-cratic pump equipped with a 1:100 splitter. System control, data

matogr. A 1218 (2011) 8561– 8566 8563

as

3

3

tao1leflt

pActwsvtc

lTi0o

2Ftothw

rhtiop

tcfae

iwdt

lBtctB

Fig. 1. Electropherograms of standard LMM carboxylic acids with optimized CE-ESI-MS parameters: total ion electropherogram (a) and extracted ion electropherogramsfor glucuronate (b), decanoate (c), 8-hydroxyoctanoate (d), xylonite (e), threonate

A. Bogolitsyna et al. / J. Chro

cquisition and data analysis were performed using Agilent Chem-tation for CE and Agilent LC/MSD Trap software for MS.

. Results and discussion

.1. Method optimization

For method optimization a model mixture was used con-aining succinic (1,4-butanedioic acid), n-valeric (n-pentanoiccid), pelargonic (n-nonanoic acid) and sebacic acid (1,8-ctanedicarboxylic acid) dissolved in water at concentrations of0 mg/l each. The acids were chosen to represent various chain

engths as well as mono-acids and di-acids. Optimized MS param-ters were dry gas temperature and flow, nebulizing gas pressure,ow rate, content of ammonium hydroxide and organic solvent inhe sheath liquid.

As organic solvents for the sheath liquid, methanol and 2-ropanol were tested. In agreement with Hagberg [5] as well ashrer and Buchberger [27], the results were successful in thease of both alcohols, but 2-propanol has an additional advan-age of improving the ionization of acids. Hence, 2-propanolas used for further concentration optimization. Four different

heath liquid concentrations were tested, having an organic sol-ent content of 30, 50, 70 and 80%. The results showed thathe peak area and intensities increased with higher 2-propanolontent.

Optimal concentration of ammonium hydroxide in the sheathiquid was tested with solutions of 0.025, 0.05, 0.075 and 0.1%.his parameter did not have a strong influence on peak areas andntensities, and there was no significant difference in the range of.025–0.075%, whereas a slight increase was observed in the casef 0.1% ammonium hydroxide.

The sheath liquid flow rate was studied in the range of–6 �l/min. A flow rate of 2 �l/min was insufficient for separation.rom 3 to 5 �l/min, a slight increase of peak areas and intensi-ies was monitored for all tested standard acids. At a flow ratef 6 �l/min a decreased intensity was observed. Resulting fromhese tests, a sheath liquid with a concentration of 0.1% ammoniumydroxide and 80% 2-propanol content at a flow rate of 4 �l/minas used.

With regard to nebulizing gas pressure, an optimal signalesponse was found between 12 and 16 psi, the range of 5–20 psiaving been tested. Values below 12 psi did not produce a satisfac-ory signal at all, while pressures above 16 psi caused lower peakntensities. A fine tuning experiment in the range of 12–16 psi gaveptimal results for 14 psi, which then was chosen as the workingressure for the nebulizing gas.

Dry gas temperatures of 200 ◦C, 250 ◦C, 300 ◦C and 350 ◦C wereested during parameter optimization. The signal intensities forarboxylic acids increased with higher temperatures up to 300 ◦C,urther temperature increase caused the intensity to decreasegain. Thus, a dry gas temperature of 300 ◦C was used for furtherxperiments.

Dry gas flow rates between 3 and 7 l/min did not significantlynfluence the signal intensities. While between 3 and 5 l/min there

as no difference in peak areas and intensities at all, a minorecrease in intensity was observed in the range of 5–7 l/min. Fur-her experiments used a dry gas flow of 5 l/min.

Ammonium formate was used as the BGE for the separation ofow molecular mass (LMM) carboxylic acids. The concentration ofGE was varied between 10 and 40 mM. While for the model mix-

ure analysis a BGE concentration of 10 mM was sufficient, moreomplicated multi-component samples demanded higher concen-rations to maintain separation quality. On the other hand, highGE concentrations of 40 mM caused instabilities in the capillary

(f), glycerate (g), azelate (h), succinate (i), malate (j). Capillary 60 cm × 50 �m I.D.;electrolyte, 20 mM ammonium formate; pH 9; applied potential, 20 kV.

current and lower peak intensities. The pH of the BGE was stud-ied in the range between 9 and 11. pH values above 10 caused CEcurrent instability and therefore cannot be applied. A satisfactoryseparation was achieved with 20 mM ammonium formate buffer ofpH 9.

3.2. Analysis of LMM carboxylic acids

After optimization of the fundamental parameters, the methodwas tested on a model mixture containing nine LMM carboxylicacids (Fig. 1; Table 1), among them two sugar acids, representativefor effluents with wood processing and degradation products. Themixture was reliably separated within 15 min, detection was byESI-MS.

Limits of the method arise for the determination of carboxylicacids with molecular mass below 70. For example, acetic acid(60 g/mol) is outside the MS-detection limit. Also short diacids can-not be MS-detected due to their high ionization potential and thedouble charge (z = 2): injection of oxalic acid did not result in a sig-nificant peak, even at comparatively high concentration. Aliphatic

acids with long chains (>C12) were not analyzed as they had beenexcluded from the sample during sample pre-treatment to avoidproblems with capillary clogging.

8564 A. Bogolitsyna et al. / J. Chromatogr. A 1218 (2011) 8561– 8566

Table 1Model mixture content and method evaluation.

Peak ID Compound Mw m/z in(−)ESIa

Migration time(min)

Linear correlationcoefficient, R

LOD (mg/l) LOQ (mg/l) Repeatability,% (n = 7)

Time Area

Aliphatic carboxylic acidsb Glucuronic acid 194 193 4.5 0.9973 0.73 2.45 0.3 3c Decanoic acid 172 171 4.4 0.9677 1.10 3.67 0.5 8d 8-Hydroxyoctanoic acid 160 159 4.5 0.9894 0.63 2.11 0.6 12e Xylonic acid 166 165 4.8 0.9992 0.58 1.94 0.7 10f Threonic acid 136 135 5.1 0.9973 0.55 1.82 0.5 4g Glyceric acid 106 105 5.8 0.8532 0.69 2.31 0.8 16h Azelaic acid 188 187 6.8 0.9925 0.91 3.03 0.3 22i Succinic acid 118 117 12.1 0.9520 0.92 3.07 0.5 17j Malic acid 133 132 12.5 0.9921 0.25 0.85 0.6 16

Lignin-derived compoundsb Acetovanillone 166 165 5.7 0.9951 0.19 0.64 0.6 15c 4-Hydroxyacetophenone 136 135 5.9 0.9959 0.16 0.53 0.7 20d Vanillin 152 151 6.2 0.9981 0.31 1.03 0.7 22e Ferulic acid 194 193 6.5 0.9953 0.30 1.01 0.8 18f 4-Hydroxybenzaldehyde 122 121 6.6 0.9946 0.09 0.31 0.8 14g Vanillic acid 168 167 7.0 0.9965 0.86 2.87 0.4 12

0.9959 0.53 1.78 0.2 20

3

acma

abwHsnsis

3

martLrifptRaos

3

eta

Fig. 2. Electropherograms of standard lignin derivatives with optimized CE-ESI-MSparameters: total ion electropherogram (a) and extracted ion electropherograms for

h 4-Hydroxybenzoic acid 138 137 7.5

a Post-run single ion.

.3. Analysis of lignin fragments

The method optimized for the separation of LMM carboxyliccids was also applied to the separation of phenolic modelompounds (Fig. 2). The lignin model mixture contained lignin frag-ents characteristic for softwood, hardwood and herbaceous lignin

s well as typical lignin degradation products (Table 1) [28,29].Injecting a model mixture of seven representative lignin-related

romatic compounds under conditions optimal for the LMM car-oxylic acids produced good results, confirming that the methodas successful as well for the analysis of lignin-derived compounds.ence, the method can be used for the simultaneous separation and

ubsequent identification of both LMM carboxylic acids and phe-olic lignoid compounds. Although the electropherogram showedome overlapping, this does not cause any serious problem for peakdentification and integration, which is done based on the well-eparated ion traces extracted from the electropherogram.

.4. Evaluation of the method

A method evaluation has been carried out to compare theethod against other options for LMM acid and lignin fragment

nalysis. Limit of detection (LOD), limit of quantification (LOQ) andepeatability (relative standard deviations, RSD%) for the migra-ion time (MT) and the peak area (PA) are given in Table 1. LOD andOQ were quantified from the threefold and tenfold signal-to-noiseatio, respectively. The repeatability was checked by repetitivenjection of the pulp bleaching effluent sample (n = 7). RSD valuesor MT and PA were less than 0.8% and 3–22%, respectively. Com-arison with literature data showed acceptable repeatability forhe ESI-MS detection. Van Pinxteren and Herrmann [7] reportedSD values of 0.2–0.5% for MT and 4–21% for PA in the analysis ofliphatic and aromatic carboxylic acids. Also in comparison withther literature accounts [15,18,27] the parameter values in thistudy were similar or better.

.5. Analysis of pulp bleaching effluents

The developed method was applied to the analysis of efflu-nts from pulp bleaching stages of the TCF (totally chlorine free)ype. The samples contained a large number of aliphatic carboxyliccids, which comprise short-chain mono- and di-acids, short-chain

acetovanillone (b), 4-hydroxyacetophenone (c), vanillin (d), ferulic acid isomers (e),4-hydroxybenzaldehyde (f), vanillic acid (g), 4-hydroxybenzoic acid (h). Capillary60 cm × 50 �m I.D.; electrolyte, 20 mM ammonium formate; pH 9; applied potential,20 kV.

A. Bogolitsyna et al. / J. Chromatogr. A 1218 (2011) 8561– 8566 8565

Fig. 3. Base peak electropherogram of combined bleaching effluent prior biologicalt

hnateptwttspcm

cps

TC

Fig. 4. Base peak electropherogram of aqueous extract from naturally aged bookpaper. For peak labeling see Table 3.

Table 3Compounds identified in the aqueous paper extract.

Peak ID Compound Mw m/z in(−)ESIa

1 Xylonic acid 166 1652 Threonic acid 136 1353 Vanillic acid 168 1674 Glyceric acid 106 1055 Lactic acid 90 896 Glycolic acid 76 757 Unidentified sugar acid 196 1958 Succinic acid 118 1179 Malic acid 134 133

10 Tartaric acid 150 149

reatment. For peak labeling see Table 2.

ydroxyacids, and medium-chain monoacids; components origi-ating from lignin detected in the sample were vanillic, syringicnd cinnamic acids (Fig. 3; Table 2). Peak assignment and quan-ification has been performed using the post-run single-ion tracesxtracted from the electropherogram, and were verified by com-arison of the migration time with standard compounds. Due tohe lack of standard compounds, not every peak was verified in thisay. Identification of compounds by ion fragmentation was rather

ime consuming due to the complexity of sample. Combination ofhe CE-MS analysis with a GC–MS approach (after derivatization,uch as silylation) provided additional information on the sam-le composition as it offers access to a much broader database ofhemical compounds which was used to support the data for CE-MSethod development.Generally, difficulties in the analysis arise from the high con-

entration of inorganic salts in the sample, which might hamper aerfect separation of some compounds in the effluent sample andhift the migration times.

able 2ompounds identified in the bleaching effluent sample.

Peak ID Compound Mw m/z in(−)ESIa

1 Syringic acid 198 1972 8-Hydroxyoctanoic acid 160 1593 Threonic acid 136 1354 Decanoic acid 172 1715 2-Hydroxyhexanoic acid 132 1316 Octanoic acid 144 1437 Tricarboxylic acid (citric or isocitric) 192 1918 Cinnamic acid 148 1479 Xylonic acid 166 165

10 Vanillic acid 168 16711 Unidentified sugar acid 196 19512 Azelaic acid 188 18713 Hexanedioic acid 146 14514 2-Ketogluconic acid 211 21015 2,4,5-Trihydroxypentanoic acid 150 14916 Malonic acid 104 10317 Unidentified sugar acid 166 16518 Succinic acid 118 11719 Malic acid 134 13320 Maleic acid 116 115

a Post-run single ion.

11 Maleic acid 116 115

a Post-run single ion.

3.6. Analysis of extract from aged paper

The extract from aged paper (Fig. 4; Table 3) mainly containedLMM mono- and di-acids, originating from far-reaching carbohy-drate degradation, as well as sugar acids. As mentioned above,carboxylic acids with very short carbon chain length (e.g. formicacid and acetic acid) were not detected due to their low molecularmass beyond the grasp of the MS detector, but they can be deter-mined by other means, e.g. conventional CE with UV-detection.As to the phenolic compounds from lignin, only vanillic acid wasdetected in the sample. The concentrations of the other lignin-derived compounds in the sample were too low to be identified.A matrix effect has been observed during the analysis of the paperextract sample as well, but it was by far not as strong as comparedto the effluent sample and did not disturb the separation. This canmainly be attributed to the missing inorganic salt freight comparedto the bleaching effluent.

4. Conclusions

The present CE-MS method offers an approach to direct analysisof most individual organic components, especially polar ones suchas acids and phenols, in some overly complex mixtures of carbo-hydrate and lignin degradation products, as occurring in industrialpulp processing or upon aging of cellulosic materials. The directanalysis refers to the fact that no derivatization is necessary as

in GC analysis (for volatility) which renders the method simpleand rapid, and the loss of components during such pretreatmentsteps is avoided. For unknown components not being identifiableby comparison to standards, the ESI-MS-hyphenation additionally

8 matog

ptaicpavo

cuoCoprpiodlcs

A

Ntp

DCgMAD

[[[[

[

[[[[[

[

[

[

[[

[

[Biomacromolecules 3 (2002) 959.

[27] W. Ahrer, W. Buchberger, J. Chromatogr. A 854 (1999) 275.

566 A. Bogolitsyna et al. / J. Chro

rovides (albeit limited) indications as to the chemical nature ofhese compounds. A comparably fast and efficient separation ofliphatic carboxylic acids and phenolic compounds from the stud-ed analytes can be achieved in one single run without the need tohange solvents. Only sample pre-concentration might be requiredrior the injection in case of very dilute solutions. The method ispplicable to any aqueous or water soluble sample and requiresery small sample amounts as the injection volume is in the rangef nanoliters.

The method presented can be complemented by other analyti-al methods as for instance GC–MS, which on one hand offers thetilization of more comprehensive compound databases, but on thether hand entails a derivatization step. Compared to GC–MS, theE-MS method shows lower sensitivity, but presents informationn the unaltered sample composition as no derivatization or sam-le preparation is involved as in the case of GC, and it offers moreapid and robust analysis. Although method improvements couldossibly further reduce the negative matrix effects of inorganic salts

n the analyte samples, the present method, being the first reportf the simultaneous determination of (hemi)cellulose and ligninegradation products by CE-ESI-MS in very complex mixtures of

ignocellulosic degradation products, is expected to find wide appli-ation in the pulp and paper industries as well as in conservationcience of historic cellulosic objects.

cknowledgements

The authors would like to thank Kyujin Ahn, MSc, University ofatural Resources and Life Sciences, Vienna, for providing data on

he book sample and Dr. Sonja Kirschnerova, for the book samplereparations.

The financial support by COST Action FP0901, the Christianoppler Research Society (CD laboratory “Advanced Cellulosehemistry and Analytics”) and Lenzing AG, Lenzing, Austria, is

ratefully acknowledged. We appreciate the helpful advice of Dr.arkus Himmelsbach and Dr. Manuela Haunschmidt, Institute of

nalytical Chemistry, Johannes-Kepler-University Linz, Austria andr. G. Götzinger, Lenzing AG, Lenzing, Austria.

[[

r. A 1218 (2011) 8561– 8566

References

[1] M. Strlic, J. Kolar, Ageing and Stabilization of Paper, National and UniversityLibrary, Ljubljana, 2005.

[2] R. Malinen, E. Sjöström, Paperi ja Puu 11 (1975) 728.[3] K. Niemelä, R. Alen, in: E. Sjöström, R. Alen (Eds.), Analytical Methods in Wood

Chemistry, Pulping, and Papermaking, Springer, Heidelberg, Germany, 1999.[4] H. Hirschlag, R. Küster, Fresen. J. Anal. Chem. 263 (1998) 274.[5] J. Hagberg, J. Chromatogr. A 988 (2003) 127.[6] E. Dabek-Zlotorzynska, R. Aranda-Rodriguez, L. Graham, J. Sep. Sci. 28 (2005)

1520.[7] D. van Pinxteren, H. Herrmann, J. Chromatogr. A 1171 (2007) 112.[8] E. Sjoholm, N.-O. Nilvebrant, A. Colmsjo, J. Wood Chem. Technol. 13 (1993)

529.[9] A.-L. Dupont, C. Egasse, A. Morin, F. Vasseur, Carbohydr. Polym. 68 (2007) 1.10] T. Javor, W. Buchberger, O. Faix, Anal. Chim. Acta 484 (2003) 181.11] C. Klampfl, W. Buchberger, Trends Anal. Chem. 16 (1997) 221.12] D. Volgger, A. Zemann, G. Bonn, J. High Res. Chromatogr. 21 (1998) 3.13] D.T. Nguyen, H. Lerch, A. Zemann, G. Bonn, Chromatographia 46 (1997)

113.14] S.M. Masselter, A.J. Zemann, G.K. Bonn, J. High Res. Chromatogr. 19 (1996)

131.15] H. Sawada, C. Nogami, Analyt. Chim. Acta 507 (2004) 191.16] B. Baena, A. Cifuentes, C. Barbas, Electrophoresis 26 (2005) 2622.17] W.-C. Yang, F.E. Regnier, J. Adamec, Electrophoresis 29 (2008) 4549.18] Y. Iinuma, H. Herrmann, J. Chromatogr. A 1018 (2003) 105.19] A. Carrasco-Pancorbo, L. Cerretani, A. Bendini, A. Segura-Carretero, G. Lercker,

A. Fernandez-Gutierrez, J. Agric. Food Chem. 55 (2007) 4771.20] J.J.B. Nevado, G.C. Penalvo, V.R. Robledo, G.V. Martinez, Talanta 79 (2009)

1238.21] A.M. Gomez-Caravaca, V. Verardo, A. Segura-Carretero, M.F. Caboni, A.

Fernandez-Gutierrez, J. Chromatogr. A 1209 (2008) 238.22] M.M. Yassine, E. Dabek-Zlotorzynska, P. Schmitt-Kopplin, Electrophoresis 30

(2009) 1756.23] A.L. Dupont, A. Seeman, B. Lavedrine, Talanta (submitted for publication).24] Supel-Select HLB SPE Instruction/Data Sheet & Troubleshooting Guide;

Data Sheet No. T708013, SUPELCO, Bellefonte, PA, 2008. http://www.sigmaaldrich.com/etc/medialib/docs/Supelco/Datasheet/t708013.Par.0001.File.tmp/t708013.pdf (accessed 7.02.11).

25] J. Dee, H.S. Stoker, Organic and Biological Chemistry, Cengage Learning,Brooks/Cole, 2009.

26] J. Röhrling, A. Potthast, T. Rosenau, T. Lange, G. Ebner, H. Sixta, P. Kosma,

28] E. Adler, Wood Sci. Technol. 11 (1977) 169.29] D. Fengel, G. Wegener, Wood: Chemistry, Ultrastructure, Reactions, Walter de

Gruyter, Berlin/New York, 1989.

81 

 

    

 Paper II   Degradation products of lignocellulosics in pulp mill effluents 

– comparison and evaluation of different gas chromatographic 

techniques for a comprehensive analysis. 

 

 Holzforschung, vol. 66, pp. 917‐925. 

 

     

Bogolitsyna, A.; Becker, M.; Borgards, A.; Liebner, F.; Rosenau, T.; Potthast, A. (2012) 

 

 

 

 

 

 

 

  

Holzforschung, Vol. xx, pp. xxx–xxx, 2012 • Copyright © by Walter de Gruyter • Berlin • Boston. DOI 10.1515/hf-2012-0008

Degradation products of lignocellulosics in pulp mill effl uents – comparison and evaluation of different gas chromatographic techniques for a comprehensive analysis

Anna Bogolitsyna 1 , Manuel Becker 1 , Andrea Borgards 2 , Falk Liebner 1 , Thomas Rosenau 1 and Antje Potthast 1, *

1 Department of Chemistry and Christian Doppler Laboratory “ Advanced cellulose chemistry and analytics ” , University of Natural Resources and Life Sciences, Vienna , Austria

2 Process Innovation , Lenzing AG, A-4860 Lenzing , Austria

* Corresponding author.Department of Chemistry and Christian Doppler Laboratory “ Advanced cellulose chemistry and analytics ” , University of Natural Resources and Life Sciences, Muthgasse 18, A-1190 Vienna, AustriaE-mail: antje.potthast@boku.ac.at

Abstract

Pulp mill effl uents contain potentially valuable compounds and will gain even greater importance than today in future biore-fi nery scenarios as possible resources. Analysis of effl uents of process streams is not straightforward because of the vast vari-ety of substances embedded in the complex inorganic matrix of high concentration. In the present comparative investiga-tion, different combinations of gas chromatography (GC) and derivatisation techniques were tested and critically evaluated aiming at a comprehensive description of all major compound classes (low-molecular weight and long-chain carboxylic acids, lignin fragments, carbohydrates and aldehydes). As derivatisa-tion techniques, trimethylsilylation, (trimethyl silyl)methylation and methylation were applied. In addition, the effl uents were analysed by pyrolysis-GC/mass spectrometry (MS), either directly or with simultaneous methylation. The study provides comprehensive information on the composition of effl uents, the suitability of the various combinations of derivatisation and GC/MS techniques. The discrimination effects of the dif-ferent approaches are compared. The comprehensive analytical approach led to a coherent balance in terms of the contribution of the major portion of organic compounds to the total organic carbon (TOC) content in pulp mill effl uents.

Keywords: derivatisation-GC/MS; lignocellulosics; pulp mill effl uents; pyrolysis-GC/MS; total organic carbon (TOC).

Introduction

Knowledge on reaction products and exact mechanisms of the degradation of (hemi)cellulosics and lignin under the

conditions of pulping followed by elemental chlorine-free (ECF) and totally chlorine-free (TCF) bleaching is still insuf-fi cient. The products contribute signifi cantly to the process liquor streams in hardwood and softwood pulp bleaching and to the resulting total organic carbon (TOC) level of the effl uent.

In today ’ s pulp and paper industries, the standard approach to evaluation of bleaching effl uents is an estimation of the fi nal TOC level, without a more detailed analysis of the effl u-ents ’ chemical composition. This is partly due to the fact that the interest in the organic compounds had been rather low and also because of the overly diffi cult comprehensive analyses of these extremely complex effl uents. Pulp bleaching effl uents contain myriad carbohydrate and lignin-derived compounds in an aqueous matrix of inorganic salts with high concentra-tion (Bogolitsyna et al. 2011 ). Fatty acids, resin acids and ste-rols, and constituents of tall oil have been reported in effl uents of kraft mills in Finland by gas chromatography (GC)-FID analysis (Holmbom 1980 ). A large number of hydroxymono-carboxylic acids from carbohydrate degradation and dicar-boxylic acids were also observed, which are arising in the course of pulping process (Niemel ä 1988 ).

Over the past few years, the drive to better understand the effl uent composition of pulp mills has increased due to envi-ronmental concerns and the resulting pressure to lower TOC levels. Moreover, the attractiveness of utilisation of the organic compounds present in the effl uents is increasing in the context of biorefi nery scenarios. Furthermore, more detailed informa-tion about the constituents of effl uents would allow better opti-misation of both bleaching sequences and effl uent treatments. The change from ECF to TCF pulp bleaching will probably enhance this interest. The analytical methodology to address effl uent compositions should have a high sensitivity and the ability to simultaneously determine degradation products of both carbohydrates and lignin (Bogolitsyna et al. 2011 ).

GC/mass spectrometry (MS) methodology fulfi ls most of these requirements. However, the aliphatic mono-, di- and hydroxy acids derived from carbohydrates usually require a pre-column derivatisation to ensure the necessary volatility of the analytes.

A common derivatisation method is trimethylsilylation (Malinen and Sj ö str ö m 1975 ; Niemel ä and Alen 1999 ), which allows deri-vatising a wide variety of compounds as the silylation occurs at alcoholic, phenolic and carboxylic groups. Silylation of carbohy-drates, on the other hand, leads to multiple peaks for each sugar isomer, and the presence of water is not tolerable. Multiple peaks can be signifi cantly diminished by oximation prior to the silyla-tion step (Laine and Sweeley 1973 ; Andrews 1989 ). This approach simplifi es signifi cantly the interpretation of the chromatograms.

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

2 A. Bogolitsyna et al.

The limited tolerance of water in trimethylsilylation can be circumvented by methylation, to which less dry conditions are acceptable. Well-known methylation reagents are boron trifl uoride/alcohol (Pietrogrande et al. 2010 ), diazomethane (DAM), trimethylsulfonium hydroxide (TMSH), N -methyl- N -(trimethylsilyl)trifl uoroacetamide (MSTFA) and methanol/sulphuric acid. Overall, the most reproducible results have been achieved with DAM (Hirschlag and K ö ster 1998 ). DAM is highly toxic, explosive, and its synthesis has to be done shortly before derivatisation. Trimethylsilyl (TMS)-DAM could substitute DAM (Seyferth et al. 1968 ; Presser and Hufner 2004 ; K ü hnel et al. 2007 ). It is not explo-sive, tolerates water, is much less toxic and commercially available.

Silylation and methylation methods provide informa-tion mainly on low-molecular weight (LMW) compounds. Products with higher molecular weight (HMW) need addi-tional techniques. Pyrolysis (Py)-GC/MS was suggested for the analysis of HMW compounds in pulp bleaching effl uents (Ristolainen et al. 1999 ). Esterifi cation with tetramethylam-monium hydroxide (TMAH) (Kukkola et al. 2006 ) was done before Py-GC/MS.

Headspace (HS)-GC/MS is a specifi c technique for analy-sis of volatile compounds. Losses or changes of the compo-sition during sample preparation are avoided. This method has been applied for the analysis of LMW carboxylic acids (Cruwys et al. 2002 ) and aldehydes (Caffaro -Filho et al. 2010 ) in wastewater.

In the present work, a comparison of different derivatisa-tion methods and different GC/MS methods has been con-ducted, aiming at a more comprehensive description of pulp bleaching effl uent samples than hitherto possible.

Materials and methods

All chemicals (Sigma-Aldrich, Schnelldorf, Germany) were of the highest purity available and were used without further purifi cation. Reagent-grade solvents were used for all extractions and workup proce-dures. Ultra-high-quality water was used for all aqueous preparations.

Pretreatment of bleaching effl uents

The sample was obtained from an effl uent stream (prior to the biological treatment) of a TCF bleaching plant from hardwood sulphite pulping. The sample was stored at 4 ° C. Filtration: at r.t. through “ blue ribbon ” paper fi lters. Samples were concentrated by lyophilisation. Solid con-tent of the lyophilised sample was about 2000 mg l -1 . Approximately 10 % of the sample corresponds to the organic substances (result of ion chromatography), and the remainder (inorganic salts) contain Cl - , SO 4

2- , Na + , Ca 2 + , etc. The TOC is about 330 mg l -1 . The subsequent analyses were performed according to the scheme presented in Figure 1 .

Ethoximation-trimethylsilylation (EtOx-TMS)

Lyophilised sample, 10 mg, was mixed with 200 μ l of pyridine, con-taining 200 μ g methyl α -d-galactopyranoside (internal standard) and 8 mg ethoxyamine (40 mg in 1 ml of pyridine). The mixture was heated in a closed vial at 70 ° C for 1 h. After EtOx, 200 μ l pyridine

containing 4-(dimethylamino)pyridine (1.5 mg in 1 ml of pyri-dine), as well as 200 μ l of N , O -bis(trimethylsilyl)trifl uoro acetamide (BSTFA) containing 10 % trimethylchlorosilane, was added. The mixture was heated at 70 ° C for 2 h, then diluted with ethyl acetate (0.6 ml) and fi ltered. An aliquot of 0.2 μ l was injected into the GC/MS system. The composition of the mixture was identifi ed fi rst by means of mass spectra libraries Wiley9 and NIST08; afterwards, the results were compared with standard compounds and quantifi ed. Calibration based on peak area of a compound-specifi c target ion (Becker et al. 2012 ) was accomplished by up to 10 different concen-tration levels ranging from 0.16 μ g ml -1 to 333 μ g ml -1 per reference standard. The linearity of the calibration curve was controlled, and regression coeffi cients were better than 0.99 for most metabolites. When the respective standard was not available, a structurally simi-lar standard was taken which had the same amount of hydroxyl and carboxyl groups.

Methylation with DAM

A fresh portion of DAM was prepared from Diazald ( N -methyl- N -nitroso- p -toluenesulfonamide), and the derivatisation was performed according to Black (1983) . An excess of the DAM solution in diethyl ether was added to 100 mg of the lyophilised sample, and the mixture was left to react for 24 h under stirring at r.t.

Excess DAM can be readily recognised by the yellow colour of the reaction mixture – it was easily destroyed by the addition of ace-tic acid, which generates methyl acetate as a volatile by-product. The solution was dried over sodium sulphate, fi ltered through a 0.45- μ m syringe fi lter and analysed by GC/MS.

Caution DAM is toxic by inhalation or by contact with the skin or eyes. It may explode in contact with sharp edges, such as ground-glass joints or even scratches in glassware. All reactions have to be performed behind a glass shield and in the hood, and suitable eye protection must be worn.

Methylation with TMS-DAM

The procedure of methylation with TMS-DAM, as fi rst published by Seyferth et al. (1972) , was modifi ed as follows. Lyophilised sample, 100 mg, was dissolved in 1 ml of a dimethylformamide:methanol mixture (v/v = 9:1), and 40 μ l of a 2.0-M solution of TMS-DAM in diethylether was added. The mixture was left to react in a loosely capped vial for 30 min at r.t. The reaction vessel must not be tightly closed as N 2 is formed. Excess reagent was destroyed by adding ap-proximately 5 μ l of acetic acid and allowing the mixture to stand for 30 min at r.t. The mixture was dried under a stream of N 2 , and

Der

ivat

izat

ion

Met

hod

Bleaching effluent

Filtration

Lyophilisation

Oximation-silylation DAM TMS-DAM

GC-MS HS-GC-MSPy-GC-MS

TMAH

Figure 1 General analysis scheme for the bleaching effl uent.

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

Comparative GC analysis of pulp mill effl uents 3

the remainder was dissolved in ethyl acetate. The solution was dried over Na 2 SO 4 , fi ltered through a 0.45- μ m syringe fi lter and analysed by GC/MS.

Methylation with TMAH

Lyophilised sample, 5 mg, was moistened with 5 ml of TMAH solution (25 % in methanol). The mixture was allowed to react for 30 min at r.t., then dried in a fl ow of N 2 at 20 ° C (Peuravuori and Pihlaja 2007 ).

HS-GC/MS analysis

HS analyses were performed in standard 20-ml glass vials. Of the sample, 3 ml was salted-out by adding 0.7 g NaHSO 4 . Vials were heated at 80 ° C in an Agilent Agitator sample shaker for 2 h. An ali-quot of 250 μ l of HS volume was injected into GC/MS at a split ratio of 5:1 at 240 ° C inlet temperature. Method optimisation showed that the salting-out step increased the HS sensitivity even for the effl uent sample, which already had a high content of inorganic salts. 20-ml mililitre vials showed a better recovery compared to 10-ml vials as they give more space to the volatile compounds.

GC/MS conditions

Instrument: GC 6890N/MSD 5973B equipped with a fused silica HP-5ms (30 m, 0.25 mm, 25 μ m). Carrier gas: He (total fl ow 27.5 ml min -1 at 46.9 kPa pressure; resulting fl ow in column 0.9 ml min -1 . Temperature program: 50 ° C (5 min), → 280 ° C (5 ° C min -1 ). Aliquots, 0.2 μ l, of the dissolved samples were injected at 260 ° C inlet temperature in splitless mode, if not stated otherwise. EI mode at 70 eV. Data were acquired and processed with the MSD Chemsta-tion E.2.01.1177 software (Agilent Technologies, USA).

Py-GC/MS conditions

Curie-point Py-GC/MS was performed with a CPP-40 pyrolyser (GSG) coupled to a GC 6890 and MSD 5973 (Agilent Technologies). The sample was pyrolysed at 600 ° C (Fecralloy TM ) for 10 s. Carrier gas: He. Inlet: 250 ° C, split 1:20. Fused silica column: HP-5ms (30 m, 0.25 mm, 25 μ m); column fl ow of 1.0 ml min -1 . Oven program: 50 ° C (5 min), → 180 ° C (5 ° C min -1 ), → 280 ° C (10 ° C min -1 ), holding time 9 min. Auxiliary temperature: 250 ° C (18 min), 10 ° C min -1 to 280 ° C (14 min). Mass spectrometer: EI mode (70 eV); ion source 230 ° C; quadrupole 150 ° C, 7.6 · 10 -6 Torr.

HS-GC/MS conditions

Instrument: GC 7890A/MSD 5975C equipped with a fused silica HP-5ms (30 m, 0.25 mm, 25 μ m) column. Carrier gas: He, 10.2 ml min -1 at 67.467 kPa pressure; column fl ow 1.2 ml min -1 . Temperature program: 50 ° C (5 min), → 70 ° C (5 ° C min -1 ), → 280 ° C (15 ° C min -1 ) holding time for 5 min. EI mode at 70 eV. Data acquisition as indica-ted above; for butanal, pentanal and hexanal in SIM mode. Heptanal, octanal, nonanal and decanal were assessed semi-quantitatively.

Acid hydrolysis

Lyophilised sample, 50 mg, was mixed with 0.25 ml of 72.2 % H 2 SO 4 and stirred for 3 – 3.5 h at r.t. Distilled water (4.25 ml) was added, and the mixture was stirred for 1.5 h at 110 ° C. After cooling to r.t., the sample was neutralised with NaHCO 3 and fi ltered through a 0.2- μ m

membrane fi lter. After lyophilisation, the solid remainder was dis-solved in 1 ml of deionised water (HPLC grade), and aliquots were injected to the LCMS system.

LC-ESI-MS conditions

HPLC instrument: PL Hi-Plex Na (300 × 7.7 mm, 10 μ m) column at 85 ° C. Eluent: deionised water (HPLC grade), 0.2 ml min -1 , injection volume 20 μ l. UV detection 200 – 300 nm (3D array). Electrospray ionisation (ESI)-MS: nebuliser gas pressure 25 psi; dry gas fl ow 10 l min -1 at 300 ° C. Scanning over a mass range of 50 – 2200 m/z in both negative and positive ESI modes.

Results and discussion

Performance of various GC/MS techniques

The compounds in bleaching effl uents were subdivided into the groups LMW substances, carboxylic acids, fatty acids, lignin fragments, carbohydrates and aldehydes. The ratio between these compound classes and the quantifi cation of the identifi ed components was dependent on the method.

The EtOx-trimethylsilylation method is particularly sen-sitive towards short-chain carboxylic acids (C 6 and below) and carbohydrates (saccharides). The silylation reagent reacts cleanly with LMW compounds. The reliability is less satis-factory with HMW compounds, such as with carboxylic acids of longer-chain length ( > C 6 ) and aromatic lignin fragments. Figure 2 a and Table 1 (third and fourth columns) show a typical chromatogram of the analyte after EtOx-TMS and the typical data, respectively.

For the determination of monosaccharides, EtOx is an imperative derivatisation step prior to trimethylsilylation, whereas for the analysis of LMW acids, EtOx has no infl u-ence, and trimethylsilylation is already suffi cient as a single derivatisation step. For comparative purposes, a combination of both EtOx and trimethylsilylation was also performed. The technique is especially useful because the major components in the complex mixture are reliably detected and accurately quantifi ed, i.e., mainly glycolic acid, 3,4-dihydroxybutanoic acids, other hydroxy acids with a chain length smaller than C 7, and carbohydrate-derived acids. Longer carboxylic acids and their hydroxy- and oxo-congeners, as well as fatty acids, are underestimated.

DAM-GC/MS

Treatment of bleaching effl uent samples with DAM in com-bination with GC/MS analysis is a good method for the deter-mination of all subgroups mentioned above, including those with HMW (Figure 2b, Table 1). The DAM reagent converts carboxylic acids into their methyl esters and phenols into their methyl ethers. A principal drawback of the method is the cancerogenity and explosiveness of DAM, which opposes its use in routine analysis and causes the necessity of its repeated synthesis each time shortly before sample analysis. Nevertheless, analysis of the bleaching effl uents according to this method provided a valuable piece of information: it gave

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

4 A. Bogolitsyna et al.

a good overview on the effl uent ’ s composition and was, in particular, well suited for identifying a large number of trace hydroxy acids and oxoacids that were not detected in such a large number by other methods. The DAM approach is the method of choice, if the focus is on the identifi cation and presence of such minor constituents. However, the insensitiv-ity towards sugar acids and LMW hydroxy acids, as well as lignin-derived phenolic acids, is a distinct drawback, as these compounds are the main constituents. Analysis of the DAM-derivatised sample is shown in Figure 2a (see also Table 1, last column).

TMS-DAM derivatisation (Seyferth et al. 1972 ), instead of DAM, improves detection of short-chain carboxylic acids, but at the expense of sensitivity. An advantage of the method is that traces of water do not interfere with the reaction (contrary to the trimethylsilylation method) and that the reagent is sig-nifi cantly less toxic and less diffi cult to handle than DAM. The disadvantage is the generally lower sensitivity compared to the EtOx-TMS approach or the DAM derivatisation. In the analy-sis of the bleaching effl uents, DAM and TMS-DAM were, to a certain extent, complementary. The advantage of the former – detection of trace longer-chain hydroxyl acids, oxoacids and fatty acids – was not maintained with TMS-DAM, which in turn was better suited for identifying LMW acids and hydroxyl acids that were less reliably detected by DAM.

Py-GC/MS yields best results with concomitant methyla-tion of hydroxyl and carboxyl groups of the effl uent compo-nents. Esterifi cation/etherifi cation in situ was done by TMAH, which is a powerful methylating agent under pyrolytic condi-tions. The chromatogram of the bleaching effl uent, when ana-lysed by Py-GC/MS with concomitant TMAH-methylation, is shown in Figure 2c; the identifi ed compounds are listed in Table 1, fi fth column. The performance was particularly successful for the analysis of fatty acids and lignin fragments. By contrast, carboxylic acids and their hydroxy and oxo vari-ants were detected rather incompletely. The same applied to carbohydrates and sugar acids. Hence, no representative image of such constituents can be obtained by Py-GC/MS; therefore, the method needs to be complemented by analy-sis according to the EtOx-TMS protocol, which is particu-larly strong at determining those compounds. To prove that the non-aromatic (not lignin-derived) compounds detected by Py-GC/MS originate from the sample and are no artifacts of the pyrolysis procedure, experiments with representative model mixture of (hydroxy-)carboxylic acids with different lengths of carbon chains were conducted, which were sub-jected to identical pyrolysis conditions. The chromatograms did not show any formation of additional compounds. Only the compounds present in the model mixture were identifi ed, except C 6 -sugar acids, which were not at all detected due to degradation.

It has been shown that the alkali metals calcium, potas-sium, magnesium and sodium infl uence Py results/analysis by alteration of volatile yield (Fahmi et al. 2007 ), interact-ing as catalyst (Raveendran et al. 1995 ), and changing com-pound degradation pathways (Liden et al. 1988 ). Therefore, quantitative analysis is hard to achieve by this method as the outcome and compound response is strongly dependent

a

b

d

Retention time (min)5

0

5.0×104

1.0×105

5×1056×105

2.0×106

6×106

8×106

0

Inte

nsity

Inte

nsity

0

1×106

4.0×1066.0×1068.0×1061.0×107

Inte

nsity

1×106

2×106

3×106

4×106

2×1073×107

0

Inte

nsity

Pen

tana

lB

utan

al

Hep

tana

l

Hex

anal

Int.

stan

dard

Non

anal

2-N

onen

alD

ecan

al

10 15 20

2010

1

2 80

1133

72

34

32

31 8

81

82

14 1221

36 3735

389375

94

969774,

769558

98 77

10099,26

39

53

55

253

7

40

34 41

4212

43 44 46

4521

37

3847 50,

265748

39

25

2

45

721312,11,14

10 20

217318 22

2374

757778

5428,29

7955

3056 3940

2426

27

7653

15 19

817

7

9

6

316

25

1

55

1617

8

14

51

76 5256

103 6364 69

65

70 71

776649 6768 101

54

18

57

101

5459

56

60 616263 64

65102

85,90,91 92,

86,18

83 8788

89

84

30 40

20 2510 15 30

20 3010 5040

c

Figure 2 GC/MS chromatograms of the bleaching effl uent after EtOx-silylation (a); after methylation with DAM (b); py-GC-MS chromatogram of the bleaching effl uent after methylation with TMAH (c); HS-GC-MS chromatogram (d), for peak labelling, see Table 1.

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

Comparative GC analysis of pulp mill effl uents 5

Table 1 Compounds identifi ed in the samples according to different GC/MS methods.

Number in chromatogram

(Figure 2c)

Compound Conc. (mg l -1 ) Retention time (min)

Common name Systematic name

Oxim-TMS-GC-MS

Oxim-TMS-

GC-MS

Py-GC-MS DAM-GC-MS

Low molecular mass carboxylic acids:

1 Lactic acid 2-Hydroxypropanoic acid 2.9 11.3 4.2 3.3 2 Caproic acid Hexanoic acid 1.0 11.6 8.6 – 3 Glycolic acid Hydroxyacetic acid 57.3 11.8 3.8 3.7 4 2-Hydroxybutyric acid 2-Hydroxybutanoic acid 0.5 13.2 – – 5 Oxalic acid Ethanedioic acid 3.4 13.4 – – 6 3-Hydroxypropionic acid 3-Hydroxypropanoic acid 19.9 13.7 – 5.4 7 Malonic acid Propanedioic acid 4.4 15.4 – 8.3 8 Succinic acid Butanedioic acid 5.5 18.4 12.6 10.7 9 Glyceric acid 2,3-Dihydroxypropanoic acid 0.8 19.1 – – 10 Maleic acid ( Z )-Butenedioic acid 2.2 19.2 – – 11 Fumaric acid (E )-Butenedioic acid 2.4 19.4 12.1 – 12 Pelargonic acid Nonanoic acid 2.5 19.5 18.5 14.0 13 Citraconic acid (2 Z )-2-Methylbut-2-enedioic acid 1.6 19.6 – – 14 Glutaric acid Pentanedioic acid 2.9 20.7 15.9 12.6 15 2,4-Dihydroxybutyric acid 2,4-Dihydroxybutanoic acid 0.2 21.2 – – 16 3,4-Dihydroxybutyric acid 3,4-Dihydroxybutanoic acid 70.4 21.7 – 13.3 17 Malic acid Hydroxybutanedioic acid 2.8 23.1 – 11.7 18 Adipic acid Hexanedioic acid 1.4 23.2 19.1 14.3 19 4-Hydroxybutyric acid 4-Hydroxybutanoic acid 0.9 23.6 – – 20 Threonic acid 2,3,4-Trihydroxybutanoic acid 1.9 25.0 – – 21 Pimelic acid Heptanedioic acid 3.3 25.6 21.9 15.7 22 Ribonic acid, lactone – 1.8 27.6 – – 23 Tartaric acid 2,3-Dihydroxybutanedioic acid 2.7 26.8 – – 24 Xylonic acid – 2.2 29.7 – – 25 Azelaic acid Nonanedioic acid 10.7 29.7 27.1 18.6 26 Sebacic acid Decanedioic acid 3.5 31.7 29.3 19.6 27 Gulonic acid – 1.1 32.8 – – 28 Glucuronic acid – 4.9 34.1 – – 29 Gluconic acid – 4.2 34.4 – – 30 Mannonic acid, lactone – 2.7 33.8 – – 31 Acrylic acid Prop-2-enoic acid – – 2.1 – 32 Propionic acid Propanoic acid – – 2.2 – 33 Enanthic acid Heptanoic acid – – 12.2 – 34 Methylsuccinic acid Methylbutanedioic acid – – 13.7 11.3 35 Capric acid Decanoic acid – – 21.3 – 36 2-Hydroxyvaleric acid 2-Hydroxypentanoic acid – – 21.6 – 37 8-Hydroxycaprylic acid 8-Hydroxyoctanoic acid – – 22.0 16.4 38 Tricarballylic acid Propane-1,2,3-tricarboxylic acid – – 23.8 16.6 39 Suberic acid Decanedioic acid – – 24.5 17.1 40 Levulinic acid 4-Oxopentanoic acid – – – 9.7 41 3-Hydroxyvaleric acid 3-Hydroxypentanoic acid – – – 12.8 42 6-Oxoenanthic acid 6-Oxoheptanoic acid – – – 13.8 43 2-Hydroxyfumaric acid 2-Hydroxybutenedioic acid – – – 14.4 44 7-Oxocaprylic acid 7-Oxooctanoic acid – – – 15.3 45 Caprylic acid Octanoic acid – – – 15.5 46 8-Oxopelargonic acid 8-Oxononanoic acid – – – 16.7 47 9-Oxopelargonic acid 9-Oxononanoic acid – – – 16.9 48 9-Oxocapric acid 9-Oxodecanoic acid – – – 18.0 49 3-Carboxyadipic acid 1,2,4-Butanetricarboxylic acid – – – 18.4 50 4-Oxosuberic acid 4-Oxooctanedioic acid – – – 18.7 51 4-Hydroxysuberic acid 4-Hydroxyoctanedioic acid – – – 18.8 52 4-Hydroxyazelaic acid 4-Hydroxynonanedioic acid – – – 20.2

Long-chain carboxylic acids: 53 Myristic acid Tetradecanoic acid 1.5 30.7 – – 54 Pentadecanoic acid Pentadecanoic acid 3.1 32.7 32.8 21.6 55 Palmitic acid Hexadecanoic acid 3.8 34.5 34.3 22.6 56 Stearic acid Octadecanoic acid 3.1 38.0 36.7 24.5

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

6 A. Bogolitsyna et al.

Number in chromatogram

(Figure 2c)

Compound Conc. (mg l -1 ) Retention time (min)

Common name Systematic name

Oxim-TMS-GC-MS

Oxim-TMS-

GC-MS

Py-GC-MS DAM-GC-MS

57 1,9-Nonanedicarboxylic acid

Undecanedioic acid – – 31.5 20.7

58 Lauric acid Dodecanoic acid – – 26.4 – 59 Palmitoleic acid 9-Hexadecenoic acid – – 33.9 – 60 Margaric acid Heptadecanoic acid – – 35.6 – 61 Oleic acid 9-Octadecenoic acid – – 36.4 – 62 Linoelaidic acid 9,12-Octadecadienoic acid – – 37.3 – 63 Nonadecyclic acid Nonadecanoic acid – – 37.7 25.4 64 Arachidic acid Eicosanoic acid – – 38.6 26.3 65 Behenic acid Docosanoic acid – – 40.3 27.9 66 11-Oxolauric acid 11-Oxododecanoic acid – – – 19.0 67 10-Oxo-1,9-nonanedi-

carboxylic acid10-Oxoundecanedioic acid – – – 19.3

68 8,9-Dihydroxybehenic acid 8,9-Dihydroxydocosanoic acid – – – 19.7 69 Heneicosylic acid Heneicosanoic acid – – – 27.1 70 Tricosylic acid Tricosanoic acid – – – 28.8 71 Lignoceric acid Tetracosanoic acid – – – 29.7

Lignin-derived compounds: 72 Benzoic acid 1.2 16.4 14.5 – 73 trans - Cinnamic acid ( E )-3-Phenylprop-2-enoic acid 2.6 24.1 – – 74 Vanillic acid 4-Hydroxy-3-methoxybenzoic acid 2.3 29.1 28.1 – 75 Terephthalic acid Benzene-1,4-dicarboxylic acid 1.9 29.7 25.8 – 76 3,4-Dihydroxybenzoic acid 1.8 30.4 28.1 18.9 77 Syringic acid 4-Hydroxy-3,5-dimethoxy-

benzoic acid 2.1 31.9 30.9 20.4

78 p- Coumaric acid (E)-3-(4-Hydroxyphenyl)-2-propenoic A

2.1 32.6 – –

79 Caffeic acid 3-(3,4-Dihydroxyphenyl 2-propenoic acid

3.1 34.3 – –

80 m -Methoxytoluene 3 -Methoxy-1-methylbenzene – – 11.9 – 81 Veratrol 1,2-Dimethoxybenzene – – 16.2 – 82 1,4-Dimethoxybenzene – – 16.7 – 83 m -Toluic acid 3-Methylbenzoic acid – – 17.9 – 84 p -Toluic acid 4-Methylbenzoic acid – – 18.2 – 85 3-Methylveratrole 3-Methyl-1,2-dimethoxybenzene – – 18.9 – 86 2,5-Dimethoxytoluene 2,5-Dimethoxy-1-methylbenzene – – 19.2 – 87 2,4-Dimethoxytoluene 2,4-Dimethoxy-1-methylbenzene – – 19.3 – 88 1,2,3-Trimethoxybenzene – – 21.0 – 89 m -Anisic acid 3-methoxybenzoic acid – – 21.7 – 90 1,2,4-Trimethoxybenzene – – 22.6 – 91 p -Anisic acid 4-methoxybenzoic acid – – 22.7 – 92 4-Methyl- m -anisic acid 4-Methyl-3-methoxybenzoic acid – – 24.1 – 93 3,4,5-Trimethoxytoluene 3,4,5-Trimethoxy-1-

methylbenzene – – 24.2 –

94 Isophthalic acid Benzene-1,3-dicarboxylic acid – – 26.2 – 95 Gentisic acid 2,5-Dihydroxybenzoic acid – – 27.2 – 96 p -Methylphthalic acid 4-Methyl-benzene-1,2-dicar-

boxylic acid – – 27.6 –

97 3,5-Dimethoxybenzoic acid – – 27.7 – 98 3,4-Dimethoxypheny l-

acetic acid3,4-Dimethoxybenzeneacetic acid

– – 28.9 –

99 2,3,6-Trimethoxybenzoic acid – – 30.6 – 100 p -Methoxyterephthalic

acid4-methoxy-benzene-1,4-dicar-boxylic A

– – 30.8 –

101 Trimellitic acid 1,2,4-Benzenetricarboxylic acid – – 33.1 21.6 102 Phthalic acid Benzene-1,4-dicarboxylic acid – – 40.5 – 103 Pyromellitic acid 1,2,4,5-benzenetetracarboxylic

acid – – – 24.6

(Table 1 continued)

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

Comparative GC analysis of pulp mill effl uents 7

on Py conditions, macroscopic sample structure and sample matrix, such as alkali metal concentration. To increase repro-ducibility, the sample amount should be low ( ∼ 0.2 mg). On the other hand, in the case of pulp bleaching effl uents with low concentrations of individual components, this might lead to impaired detectability of such compounds (Berezkin 1983 ; Wampler 2007 ). A compromise was found by pyrolysing a sample amount of 2 mg for qualitative analysis.

Application of HS-GC/MS for determination of LMW carboxylic acids was described by Cruwys et al. (2002) . In the case of the complex multi-component mixtures of analytes as studied in the present paper, this approach did not produce rea-sonable results for such LMW acids (Figure 2d). However, an important outcome of the HS-GC/MS method was the rather selective identifi cation and quantifi cation of volatile alde-hydes in the effl uent, which were not covered by the methods presented above due to the lyophilisation pretreatment of the analytes that eliminated these volatiles. HS-GC/MS allowed detection of the full range of aldehydes with a chain length of up to 12 carbon atoms in the effl uent sample.

Comparison of the GC/MS methods

The most comprehensive picture of the constituents of bleach-ing effl uents from the pulp and paper industry is a combination of EtOx-TMS-GC/MS and TMAH-Py-GC/MS. The former is the method of choice for analysis of LMW carboxylic acids, carbohydrates and sugar-derived compounds, while the latter is particularly suitable for the analyte part of fatty acids and lignin-derived products. If needed, volatiles, mainly compris-ing aliphatic aldehydes, can be covered in addition by HS-GC/MS (Figure 3 ). Trace compounds from the hydroxy acid and oxoacid compound classes can be advantageously detected by complementary DAM-GC/MS and TMS-DAM-GC/MS analyses, which are also suitable for initial screening of effl u-ent samples.

The pre-capillary derivatisation methods in combination with GC/MS show some discrimination effects. These meth-ods are highly sensitive to the members in a particular sub-stance class, but constituents of other substance classes are incompletely detected. Figure 4 is a comparative evaluation of the methods according to the number of individual com-pounds detected per compound class. Note that the label “ fatty acids ” also comprises medium-chain oxoacids and hydroxyl acids (between C 6 and C 12 ), which are not contained in the fi rst group of “ LMW acids ” (chain length C 6 and below). From this fi gure, the same recommendation can be derived as given above, i.e., a combination of EtOx-TMS-GC/MS and TMAH-Py-GC/MS is the most reliable.

The most consistent results achieved by a single method are obtained by either the EtOx-TMS or the DAM approach. These two methods permit a general survey on all the major compound groups present in the effl uent, but each fails to provide suffi ciently detailed and suffi ciently comprehensive data. Methylation with DAM is a good method for the deter-mination of LMW acids and works best for fatty acids. The EtOx-TMS method is particularly successful for the determi-nation of carboxylic acids, sugars and carbohydrate-derived

0

10

20

30

40

Ald

ehyd

es

Ext

ract

ives

Sug

ars

Lign

in-d

eriv

ativ

es

Fatty

aci

ds

Num

ber o

f com

poun

ds

Ox-silyl DAM Py TMS-DAM HS

LMW

aci

ds

Figure 4 Number of compounds in different classes according to different GC/MS and Py-GC-MS methods.

Sugars

Oxim-silyl-GC-MS DAM-GC-MS Py-GC-MS HS-GC-MS

LMMcarboxylic

acidsFatty acids Lignin-

derivatives Aldehydes

Figure 3 Major groups of compounds and corresponding suitable methods for qualitative and quantitative analysis (dashed lines cor-respond to less suitable methods).

components, but tends to discriminate long-chain acids and lignin-derived compounds. Py-GC/MS provides data mainly on the content of lignin-derived compounds and long-chain carboxylic acids, but does not give reliable information on the LMW carboxylic acid and carbohydrate part of the analyte. This methods gives semiquantitative data for comparative purposes.

Summarising the results of all described GC-MS methods applied to the bleaching effl uents analysis allows the conclu-sion that the sample contains mainly LMW carboxylic acids (carbohydrate degradation products) and long-chain carboxy-lic acids, as well as lignin fragments. Intact monosaccharides, aldehydes and amino acids are contained in trace amounts. The two main constituents are glycolic acid and 3,4-dihy-droxybutanoic acid. Aliphatic carboxylic acids are detected in the whole range up to 25 carbon atoms chain length. They are represented mainly by monoacids including several unsaturated acids, remnants of the extractive and fatty matter of the woody starting materials used in pulping and bleach-ing. Lignin fragments comprise mainly methoxybenzenes; hydroxy-, methoxy- and methyl-benzoic, benzenedicarboxy-lic, phenylpropenoic acids; and methoxybenzaldehydes.

The observation that only a few lignin derivatives are detected by GC-MS after EtOx-TMS and methylation with DAM is due to the HMW of the lignin fraction, which has to be fragmented to LMW fragments by Py-GC/MS. To check this, the sample was subjected to LCMS analysis before and

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

8 A. Bogolitsyna et al.

after an acidic hydrolysis under conditions typical for deg-radation of oligosaccharidic and polysaccharidic materials. The molecular weights of the constituents in the original sam-ple were mainly in the range of 200 – 600, with a minor part being in the 600 – 1000 range. As no signifi cant changes in the molecular weight were noticed after hydrolytic treatment, the conclusion was drawn that the HMW compounds in the sample correspond to a lignin fraction rather than a carbohy-drate matrix. This is in agreement with the observations by Py-GC/MS.

To get an overview on the actual concentrations of the main effl uent compounds, quantifi cation of LMW car-boxylic acids, fatty acids, lignin derivatives and monosac-charides has been performed (see Table 1, third column). The two main components of the effl uent, glycolic and 3,4-dihydroxy butanoic acid, make up more than two-thirds of the organic matter in the sample, their individual con-tents being 57 mg l -1 and 70 mg l -1 , respectively. It should be noted that the presence of such high amounts of 3,4-dihydroxybutanoic acid has not been reported before for such effl uents. This might be due to the fact that the com-pound is present in a dynamic equilibrium between the open chain and the lactone forms, which make it LC-silent so that it was missed by all previous LC-based analytical approaches. A mechanism for oxidative hexose degradation under conditions relevant for pulp bleaching was described by Hollingsworth (1991) . The proposed main degrada-tion products are the same as found in the present study, i.e., glycolic acid and 3,4-dihydroxybutanoic acid, as also observed by Niemel ä and Sj ö str ö m (1986) during cellulose degradation. The other predominant LMW carboxylic acid compounds in the mixture, such as 3-hydroxypropanoic, azelaic or succinic acid, originate from competitive minor degradation pathways.

Conclusion

The analysis results of bleaching effl uents were dependent on the method applied; no method was free from discrimination effects. The EtOx-TMS or the DAM approach might provide a general map, but not a detailed picture. A combination of EtOx-TMS-GC/MS and Py-GC/MS provides the most com-prehensive account, which can be further complemented by the HS-GC/MS method to cover trace volatile compounds in the effl uent. Such a combined approach allows for identifi ca-tion of a high number of components in the wastewater effl u-ents. The main compounds were 3,4-dihydroxybutyric acid and glycolic acids, along with smaller amounts of other LMW carboxylic acids. In addition, a large number of fatty acids, dimethoxybenzenes, methoxybenzoic acids and other lignin-derived compounds are present.

Acknowledgements

The fi nancial support by the Christian Doppler Research Society (CD lab “ Advanced Cellulose Chemistry and Analytics ” ) and by Lenzing AG, Lenzing, Austria, is gratefully acknowledged.

References

Andrews, M.A. (1989) Capillary gas-chromatographic analysis of monosaccharides: improvements and comparisons using trif-luoroacetylation and trimethylsilylation of sugar O-benzyl- and O-methyl-oximes. Carbohydr. Res. 194:1 – 19.

Becker, M., Liebner, F., Rosenau, T., Potthast, A. (2012) Ethoximation-silylation approach with expanded retention index for mono- and disaccharide analysis by GCMS. J. Chromatogr. A, in press.

Berezkin, V.G. Chemical Methods in Gas Chromatography (Journal of chromatography library, V.24). Elsevier Science Publishers, Amsterdam, The Netherlands, 1983.

Black, T.H. (1983) The preparation and reactions of diazomethane. Aldrichim. Acta 16:3 – 10.

Bogolitsyna, A., Becker, M., Dupont, A.-L., Borgards, A., Rosenau, T., Potthast, A. (2011) Determination of carbohydrate- and lignin-derived components in complex effl uents from cellulose process-ing by capillary electrophoresis with electrospray ionization-mass spectrometric detection. J. Chromatogr. A 1218:8561 – 8566.

Caffaro-Filho, R.A., Wagner, R., Umbuzeiro, G.A., Grossman, M.J., Durrant, L.R. (2010) Identifi cation of alpha-beta unsatu-rated aldehydes as sources of toxicity to activated sludge bio-mass in polyester manufacturing wastewater. Water Sci. Technol. 61:2317 – 2324.

Cruwys, J.A., Dinsdale, R.M., Hawkes, F.R., Hawkes, D.L. (2002) Development of a static headspace gas chromatographic proce-dure for the routine analysis of volatile fatty acids in waste waters. J. Chromatogr. A 945:195 – 209.

Fahmi, R., Bridgwater, A.V., Darvell, L.I., Jones, J.M., Yates, N., Thain, S., Donnison, I.S. (2007) The effect of alkali metals on combustion and pyrolysis of Lolium and Festuca grasses, switch-grass and willow. Fuel 86:1560 – 1569.

Hirschlag, H., K ö ster, R. (1998) Development of a solid-phase extraction and derivatization method for polar carboxylic acids from aqueous extracts of inorganic multi-component incineration residues. Fresenius J. Anal. Chem. 263:274 – 280.

Hollingsworth, R.L. (1991) Process for the preparation of 3,4-dihy-droxybutanoic acid and salts thereof. EP 91114350.1.

Holmbom, B. (1980) A Procedure for Analysis of Toxic Compounds in Pulp and Paper-Mill Waste-Waters. Pap. Puu-Pap. Tim. 62:523 – 531.

K ü hnel, E., Laffan, D.D.P., Lloyd-Jones, G.C., Mart í nez del Campo, T., Shepperson, I.R., Slaughter, J.L. (2007) Mechanism of methyl esterifi cation of carboxylic acids by thrimethylsilyldiazomethane. Angew. Chem. Int. Ed. 46:7075 – 7078.

Kukkola, J., Knuutinen, J., Paasivirta, J., Herve, S., Pessala, P., Schultz, E. (2006) Characterization of high molecular mass material in ECF and TCF bleaching liquors by Py-GC/MS with and without TMAH methylation. J. Anal. Appl. Pyrolysis 76:214 – 221.

Laine, R.A., Sweeley, C.C. (1973) O-methyl oximes of sugars. Analysis as O-trimethylsilyl derivatives by gas-liquid chroma-tography and mass spectrometry. Carbohydr. Res. 27:199 – 213.

Liden, A.G., Berruti, F., Scott, D.S. (1988) A kinetic-model for the production of liquids from the Flash Pyrolysis of Biomass. Chem. Eng. Commun. 65:207 – 221.

Malinen, R., Sj ö str ö m, E. (1975) The formation of carboxylic acids from wood polysaccharides during kraft pulping. Pap. Puu. 11:728 – 736.

Niemel ä , K. (1988) Gas-Liquid Chromatography-Mass Spectrometry Studies on Pine Kraft Black Liquors. 3. The Liberation of Carboxylic-Acids in the Initial Phase of Pulping. J. Chromatogr. 446:247 – 252.

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

Comparative GC analysis of pulp mill effl uents 9

Niemel ä , K., Alen, R. (1999) Characterization of pulping liquors. In: Analytical Methods in Wood Chemistry, Pulping, and Papermaking. Eds. Sj ö str ö m, E., Alen, R. Springer, Heidelberg, Germany. pp. 193 – 231.

Niemel ä , K., Sj ö str ö m, E. (1986) The conversion of cellulose into carboxylic-acids by a drastic alkali treatment. Biomass 11:215 – 221.

Peuravuori, J., Pihlaja, K. (2007) Advanced TMAH and TMAAc thermochemolysis – pyrolysis techniques for molecular character-ization of size-separated fractions from aquatic dissolved organic matter. Anal. Bioanal. Chem. 389:475 – 491.

Pietrogrande, M.C., Bacco, D., Mercuriali, M. (2010) GC-MS analy-sis of low-molecular-weight dicarboxylic acids in atmospheric aerosol: comparison between silylation and esterifi cation deriva-tization procedures. Anal. Bioanal. Chem. 396:877 – 885.

Presser, A., Hufner, A. (2004) Trimethylsilyldiazomethane – a mild and effi cient reagent for the methylation of carboxylic acids and alcohols in natural products. Monatsh. Chem. 135:1015 – 1022.

Raveendran, K., Ganesh, A., Khilar, K.C. (1995) Infl uence of mineral matter on biomass pyrolysis characteristics. Fuel 74:1812 – 1822.

Ristolainen, M., Alen, R., Toivanen, J. (1999) Characterization of totally chlorine-free effl uents from kraft pulp bleaching III – analytical pyrolysis of high-molecular-mass hardwood-derived material. J. Anal. Appl. Pyrol. 52:225 – 237.

Seyferth, D., Dow, A.W., Menzel, H., Flood, T.C. (1968) Trimethylsilyldiazomethane and trimethylsilylcarbene. J. Am. Chem. Soc. 90:1080 – 1082.

Seyferth, D., Menzel, H., Dow, A.W., Flood, T.C. (1972) Trime-thylsilyl-substituted diazoalkanes. I. Trimethylsilyldiazomethane. J. Organomet. Chem. 44:279 – 290.

Wampler, T.P. (2007) Applied Pyrolysis Handbook. CRC Press, Boca Raton, London, New York.

Received January 16, 2012. Accepted April 26, 2012 .

Bereitgestellt von | Universität für Bodenkultur Wien (Universität für Bodenkultur Wien)Angemeldet | 172.16.1.226

Heruntergeladen am | 22.06.12 11:28

91 

    

 Paper III   Evaluation of different derivatisation approaches for gas 

chromatographic‐mass spectrometric analysis of 

carbohydrates in complex matrices of biological and synthetic 

origin. 

 

 Journal of Chromatography A, vol. 1281, pp. 115‐26. 

 

 

 

 

 

Becker, M.; Zweckmair, T.; Forneck, A.; Rosenau, T.; Potthast, A.; Liebner, F. (2013) 

 

       

Ess

Ma

b

a

ARRAA

KGCOOTT

1

lcuebQuopecp

t

0h

Journal of Chromatography A, 1281 (2013) 115– 126

Contents lists available at SciVerse ScienceDirect

Journal of Chromatography A

jou rn al h om epage: www.elsev ier .com/ locat e/chroma

valuation of different derivatisation approaches for gas chromatographic–masspectrometric analysis of carbohydrates in complex matrices of biological andynthetic origin

. Beckera, T. Zweckmaira, A. Forneckb, T. Rosenaua, A. Potthasta, F. Liebnera,∗

University of Natural Resources and Life Sciences, Department of Chemistry, Konrad-Lorenz-Str. 24, A-3430 Tulln, AustriaUniversity of Natural Resources and Life Sciences, Department of Crop Sciences, Division of Viticulture and Pomology, Konrad Lorenz-Straße 24, A-3430 Tulln, Austria

r t i c l e i n f o

rticle history:eceived 22 June 2012eceived in revised form 11 January 2013ccepted 14 January 2013vailable online 21 January 2013

eywords:C/MSarbohydrate analysis-isopropylidenationximationrimethylsilylation

a b s t r a c t

Gas chromatographic analysis of complex carbohydrate mixtures requires highly effectiveand reliable derivatisation strategies for successful separation, identification, and quanti-tation of all constituents. Different single-step (per-trimethylsilylation, isopropylidenation)and two-step approaches (ethoximation–trimethylsilylation, ethoximation–trifluoroacetylation,benzoximation–trimethylsilylation, benzoximation–trifluoroacetylation) have been comprehensivelystudied with regard to chromatographic characteristics, informational value of mass spectra, easeof peak assignment, robustness toward matrix effects, and quantitation using a set of referencecompounds that comprise eight monosaccharides (C5–C6), glycolaldehyde, and dihydroxyacetone. Ithas been shown that isopropylidenation and the two oximation-trifluoroacetylation approaches areleast suitable for complex carbohydrate matrices. Whereas the former is limited to compounds thatcontain vicinal dihydroxy moieties in cis configuration, the latter two methods are sensitive to traces of

rifluoroacetylation trifluoroacetic acid which strongly supports decomposition of ketohexoses. It has been demonstratedfor two “real” carbohydrate-rich matrices of biological and synthetic origin, respectively, that two-stepethoximation–trimethylsilylation is superior to other approaches due to the low number of peaksobtained per carbohydrate, good peak separation performance, structural information of mass spectra,low limits of detection and quantitation, minor relative standard deviations, and low sensitivity toward

matrix effects.

. Introduction

Analysis of complex carbohydrate mixtures is still a chal-enging issue with regard to separation and identification of allonstituents and reliable quantitation of target compounds. Nat-ral, carbohydrate-rich samples are ubiquitous and include plantxtracts, body fluids, or hydrolysates of polysaccharides (e.g.,iomass processing) and glycoproteins (O- and N-glycans etc.).uantitation of the sugar composition might be required for prod-ct or process control, structure elucidation, or understandingf metabolic processes. Deviations in the natural sugar finger-rint of healthy grapevine parts (leaves, roots), for example, arexpected to provide immediate information about stress situationsaused by drought stress, enhanced UV radiation, fungal attack, or

arasites.

On the other hand, carbohydrates play a considerable role inechnical applications where they are used as nonionic surfactants

∗ Corresponding author. Tel.: +43 1 47654 6452; fax: +43 1 47654 6059.E-mail address: falk.liebner@boku.ac.at (F. Liebner).

021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.chroma.2013.01.053

© 2013 Elsevier B.V. All rights reserved.

(alkyl polyglucosides), natural adhesives, feedstock for bioplas-tics (e.g., polyurethane, polylactide, polyalkanoate), acid stabilizersin food technology, or as a source for bioethanol, organic sol-vents, or fine chemicals such as hydroxymethylfurfural and2,5-dimethylfurane [1]. The huge global sugar consumption, theincreasing number of carbohydrate-based industrial applications,and the awakening awareness of the necessity of utilizing alter-native carbohydrate sources are some of the reasons that recentlyfortified efforts to establish synthetic approaches to low-molecularcarbohydrates. The reaction of formaldehyde with calcium hydrox-ide, known as formose reaction, is regarded as a promising syntheticapproach in this respect even though the complexity of the reac-tion mixtures would render the isolation of individual compoundsa challenging task.

Regardless of their biological or synthetic origin, both of thesample matrices described above are characterized by a complexcarbohydrate composition with constituents that vary significantly

in molecular weight and polarity [2,3]. Depending on their origin,analysis of the sugar pattern is interfered by further constituentsthat add to the complexity of the matrix, and may interfere withanalysis and derivatisation [4].

1 atogr

tmrmlattdEtcti

ssctbblstyiea�chstvm

cicth

sotcdobnsf

cr

pfmdbgcot

16 M. Becker et al. / J. Chrom

Capillary gas chromatography (GC) is a powerful separationechnique that complements high performance liquid chro-

atography and capillary electrophoresis. In particular, the higheproducibility, lower sensitivity of the separation column towardatrix effects, and the availability of comprehensive mass spectra

ibraries render GC in combination with mass spectroscopy (MS) powerful technique for quantitative analysis of complex mix-ures. However, unlike HPLC, polar, hydrophilic, low-volatile, orhermally sensitive compounds like carbohydrates are not suited toirect analysis by GC/MS and require preceding derivatisation [5].ffective derivatisation strategies have been developed to increasehe signal intensity and stability of derivatives, to improve theompound-specific fragmentation pattern, and to allow for simul-aneous analysis of both lower and higher molecular compoundsn one run.

High-performance GC/MS analysis requires derivatisationtrategies that allow for simultaneous analysis of a wide range oftructurally different targeted and non-targeted compounds, suffi-ient chromatographic resolution, and high informational value ofhe obtained mass spectra. The latter are both required for unam-iguous peak assignment and quantitation, respectively, and cane controlled by choosing a suitable derivatisation reagent [6]. Sily-

ation, methylation, acetylation, and trifluoroacetylation (TFA) areingle-step derivatising techniques that are widely employed inhe analysis of polyalcohols and non-reducing sugars [7]. The anal-sis of reducing sugars is more challenging due to the variety ofsomers that co-exist in aqueous solutions with instable hemiac-tals being usually the dominating species [8]. Derivatisation of anldohexose, for example, can theoretically afford five tautomers:- and �-pyranose, �- and �-furanose, and the respective open-hain aldohexose [9]. Two- or multi-step derivatisation proceduresave been developed for GC/MS analysis of carbohydrate-richamples to reduce the number of isomer-derived peaks ando improve both chromatographic resolution and informationalalue of mass spectra, which is of particular use for complexatrices [10].Two-step derivatisation approaches commonly convert first the

arbonyl group into a specific derivative that no longer supportsnterconversion of the different tautomers described above [6]. Thisan be accomplished either by reduction, which affords the respec-ive primary alcohols (alditols), or by oximation using (substituted)ydroxylamine.

In the second step, the remaining “un-protected” primary andecondary hydroxyl groups are derivatised such as by (triflu-ro) acetylation (alditol acetates, oxime acetates), which leaveshe originally polar carbohydrates lipophilic compounds with aonsiderably higher volatility. However, even though two-steperivatisation approaches have commonly the great advantage thatnly one or two derivatives for each sugar is formed, their applica-ility suffers sometimes from certain weaknesses such as the largeumber of manual processing steps needed or the fact one and theame derivative can be formed from different sugars as reportedor alditol acetates [10].

Unlike reduction, oximation quantitatively converts reducingarbohydrates that preferably form cyclic hemiacetals into the cor-esponding open-chain aldose derivatives.

During this process, the anomeric carbon atom is converted viarochiral aldehyde groups into a stereo isomeric center that canorm the corresponding syn- and anti-isomers. The benefit of oxi-

ation from the analytical point of view is that the number oferivatisation products obtained from reducing carbohydrates cane reduced from five to two peaks [8,11], and even down to one sin-

le peak for aldoses if the latter are dehydrated and subsequentlyonverted into the respective aldonitrile acetates [10]. In general,ximation simultaneously improves the chromatographic resolu-ion; it disperses and, therefore, separates peaks originating from

. A 1281 (2013) 115– 126

reducing sugars and those derived from alditols or lactones [8] andprotects �-keto acids from decarboxylation [5].

The conversion of reducing sugars into oxime derivativeshas been recently advanced using O-alkylated or O-aralkylsubstituted hydroxylamines, such as O-methylhydroxylamine[8],O-ethylhydroxylamine [4,12] and O-benzylhydroxylamine [8],commonly applied as the respective hydrochloride, instead of theunsubstituted oximation reagent [13]. This measure has been con-firmed as affording a higher oximation performance due to bothhigher nucleophilicity of the amine and higher informational valueof the respective mass spectra [14].

Cu(II)-catalyzed formation of cyclic O-isopropylidene deriva-tives [15] under acidic conditions using acetone as derivatisingagent is another technique that has been successfully appliedto carbohydrate analysis. The low mass increment of 20 g mol−1

introduced per hydroxyl group (96 g mol−1 for trifluoroacetylation,72 g mol−1 for trimethylsilylation) is probably the major advantageof this derivatisation approach [5].

The aim of the current study was to evaluate different derivati-sation approaches with regard to their performance for GC/MSanalysis of carbohydrate-rich, complex matrices. The evaluationcriteria were derivatisation efficiency, chromatographic charac-teristics (number of isomers formed per compound, resolutionetc.), analytical sensitivity (limit of detection), informational valueof mass spectra and hence reliability of peak assignment, repro-ducibility of quantification results, and robustness toward matrixeffects. Besides common single-step derivatisation techniques,such as per-trimethylsilylation and O-isopropylidenation, two-step approaches comprising sequential oximation (with O-ethyl-or O-benzylhydroxylamine) followed by trifluoroacetylation ortrimethylsilylation were also studied. The different derivatisationapproaches were compared using a set of eight monosaccharides(C5 to C6), glycolaldehyde (GA), and dihydroxyacetone (DHA). Eventhough the latter two compounds are not members of the monosac-charide family in the strict sense due to the lacking stereo center,these di- (GA) and trioses (DHA) were included in the study as theyare involved in many metabolic processes. The different derivatisa-tion methods were also applied to “real” carbohydrate-rich samplesof biological and synthetic origin to verify the results obtainedfor the set of reference compounds described above. Furthermore,their robustness toward matrix effects was investigated based onrecovery rates, which were determined by a standard additionapproach.

2. Materials and methods

2.1. Chemicals and reagents

All reference compounds (Table 1), the internal standardmethyl �-d-galactopyranoside, calcium hydroxide, anhy-drous pyridine, 36% aqueous formaldehyde stabilized withCH3OH, anhydrous CuSO4, molecular sieve 3 A, anhydroussodium carbonate, dichloromethane (DCM), ethyl acetate,N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), N-methyl-bis(trifluoroacetamide) (MBTFA), O-ethylhydroxylaminehydrochloride, O-benzylhydroxylamine hydrochloride and4-(dimethylamino)pyridine (DMAP) were purchased fromSigma–Aldrich (Sigma–Aldrich Handels GmbH, Vienna, Austria).d-glucose-13C6, 99 atom% 13C6 (13C-glucose) were supplied byIsotec, Miamisburg, OH, USA. Methanol, sulfuric acid, hydrochloricacid, and anhydrous acetone were obtained from Carl Roth (Graz,

Austria). Acetone used for preparation of O-isopropylidene deriva-tives was further dried over freshly vacuum-dried molecular sieve3 A in an argon atmosphere for two days. Sodium hydroxide wasobtained from Merck (Vienna, Austria). All standards, chemicals

M. Becker et al. / J. Chromatogr

Tab

le

1C

har

acte

rist

ics

of

dif

fere

nt

der

ivat

ised

carb

ohyd

rate

stan

dar

ds.

Nam

e

Mol

ecu

lar

form

ula

Mw

Nu

mbe

r

ofsu

bsti

tuen

tsM

olec

ula

r

wei

ght

of

der

ivat

ive

(nom

inal

)

Nu

mbe

r

of

pea

ks

per

com

pou

nd

TMS

Oxi

me

ISP

TMS

ISP

EO–T

MS

BO

–TM

S

EO–T

FA

BO

–TFA

TMS

ISP

EO–T

MS

BO

–TM

S EO

–TFA

BO

–TFA

GA

C2H

4O

260

11

0

132

–

175

237

199

261

2

0

2

2

2

2D

HA

C3H

6O

390

2

1

1

234

188

278

339

325

387

1

1

1

1 1

1d

-(−)

-ara

bin

ose

C5H

10O

515

0

4

1

2

438

230

482

544

577

639

5

1

1

2 2

2d

-(−)

-rib

ose

C5H

10O

515

0

4

1

1/1a

438

172/

190

482

544

577

639

5

2

1

2

2

2d

-(+)

-xyl

ose

C5H

10O

515

0

4

1

2

438

230

482

544

577

639

2

1

2

2

2

2d

-all

ose

C6H

12O

618

05

12

541

260

584

646

703

765

4

1

2

2

2

2d

-(−)

-fru

ctos

e

C6H

12O

618

0

5

1

2/2a

541

260

584

646

–

–

4

2

1 2

0

0d

-(+)

-glu

cose

C6H

12O

618

0

5

1

2

541

260

584

646

703

765

2

1

2 1

2

2d

-(+)

-man

nos

e

C6H

12O

618

0

5

1

2

541

260

584

646

703

765

4

1

2

1

2

2d

-(−)

-tag

atos

e

C6H

12O

618

0

5

1

2

541

260

584

646

–

–

5

1 2

2

0

0

GA

=

glyc

olal

deh

yde,

DH

A

=

dih

ydro

xyac

eton

e.a

Two

isom

ers

avai

labl

e

per

com

pou

nd

.

. A 1281 (2013) 115– 126 117

and reagents were of GC-grade purity and used without furtherpurification.

2.2. Preparation of grapevine leaf samples

Grapevines of Vitis vinifera L. cv. Pinot noir clone Fr 1801 graftedon SO4-rootstock were grown in 3 L pots filled with loamy soil. Har-vesting was at the phenological growth stage of pea-sized berries(BBCH 75). Leaves were collected at midday. The harvested leaveswere immediately frozen in liquid nitrogen and stored at −80 ◦Cuntil needed for further processing.

The leaf samples were milled and homogenized for 4 min at30 Hz using a CryoMill (Retsch GmbH & Co KG, Haan, Germany).Afterward, the leaf material was lyophilized and stored at −20 ◦C.10 mg of the freeze-dried material was used for derivatisation.

2.3. Formose reaction

0.56 g (7.56 mmol) powdered calcium hydroxide was placed ina 100 mL Schott bottle and dissolved in 25 mL of water (solution A).In a second vessel, 6.75 mL of aqueous 36% formaldehyde (2.5 g,83.25 mmol) was diluted to a final volume of 25 mL containing5 mg of glycolaldehyde (solution B). Both of the solutions were pre-heated to the desired reaction temperature of 60 ± 2 ◦C using an oilbath. The formose reaction was started by adding solution B to solu-tion A. After 20 min the reaction was stopped by adding 5.62 mL of10% (w/v) hydrochloric acid. After cooling to room temperature, thesolution was neutralized using 5% (w/v) sodium hydroxide. A sam-ple of 50 �L of lyophilized “formose” was used for derivatisationand further analysis by GC/MS.

2.4. Derivatisation methods

Per-trimethylsilylation (TMS). A defined amount of freeze-driedgrapevine leaves (10 mg), a crude product of the formose reac-tion (50 �L) or respective reference compound (1–100 �g), wasdissolved in 200 �L pyridine which contained 75 �g mL−1 methyl�-d-galactopyranoside as internal standard (IS) and 300 �g mL−1

of DMAP as a silylation catalyst. Derivatisation was accomplishedby adding 120 �L of BSTFA and heating the mixture to 70 ◦C for 2 h.The derivatised samples were kept at −20 ◦C until further analysis.

O-isopropylidenation (ISP). 0.1 mL of a solution of the internalstandard 13C-glucose in 50% aqueous methanol (2 �g �L−1) wasadded to the sample or reference compound. Subsequently, themixture was freeze-dried and stored in vacuum over P4O10 at roomtemperature for two days. Then, a spatula tip of anhydrous copper-(II)-sulfate was added to the sample prior to dissolution in 625 �Ldry acetone. Derivatisation was started by adding 7.5 �L of conc.H2SO4. The reaction was stopped by adding 80 mg of anhydroussodium carbonate.

Ethoximation (EO). The respective sample or reference com-pound was dissolved in 200 �L pyridine. Pyridine (15 �L)containing 5 mg mL−1 methyl �-d-galactopyranoside (IS) and2 mg mL−1 DMAP was added. Subsequently, 200 �L of a solutioncontaining 40 mg mL−1 O-ethylhydroxylamine hydrochloride inpyridine was added, and the mixture was heated to 70 ◦C for 1 h.

Benzoximation (BO). Similar to O-ethoximation; 200 �L of a solu-tion containing 40 mg mL−1 O-benzylhydroxylamine hydrochlo-ride in pyridine was added for variant BO–TMS; only half theamount of O-benzylhydroxylamine hydrochloride (20 mg) wasadded for BO–TFA. The reaction mixtures were kept at 75 ◦C for40 min. A final sample volume of 1.2 mL was adjusted by adding

pyridine and ethyl acetate (O-benzyloxime-trimethylsilyl deriva-tives), respectively, prior to injection into the GC/MS.

Trimethylsilylation (TMS) of oximated samples (EO–TMS, BO–TMS).BSTFA (120 �L) was added to the oximated solution that contained

1 atogr

t(cs

5wotca

wt

2

mCs05ssf

8(0i(

2

EcrsQbsl

2

plcloorq1rfrmoi

aE

18 M. Becker et al. / J. Chrom

he sample or reference, internal standard and silylation catalystsee trimethylsilylation). The mixture was kept at 70 ◦C for 2 h. Afterooling to ambient temperature, ethyl acetate was added to theolution and stored at −20 ◦C.

Trifluoroacetylation (TFA) of oximated samples (EO–TFA, BO–TFA).0 �L (EO–TFA) and 80 �L (BO–TFA) of MBTFA, respectively,as added to the oximated solution that contained the sample

r reference, internal standard and silylation catalyst (see per-rimethylsilylation). The mixture was kept at 70 ◦C for 2 h. Afterooling to ambient temperature, pyridine was added to the solutionnd cooled down to 4 ◦C.

Standard addition. The “formose” and grapevine leaf samplesere spiked with 50 �l of a carbohydrate standard solution prior

o freeze-drying at a concentration of 0.5 mg mL−1 per compound.

.5. GC/MS analysis

GC/MS analysis was performed on an Agilent 6890N gas chro-atograph coupled with an Agilent 5973 mass selective detector.

olumn: HP-5MS (30 m × 0.25 mm × 0.25 �m; J&W Scientific, Fol-om, CA, USA); carrier gas: helium; injector: 280 ◦C; column flow:.9 mL min−1; purge flow: 32.4 mL min−1, 0.6 min; oven program:0 ◦C (2 min), 5 ◦C min−1, 280 ◦C (20 min); MS: EI mode, 70 eV,ource pressure: 1.13 · 10−7 Pa, source temperature: 230 ◦C. Thecan range was set from 50 to 950 Da, except ISP, which rangedrom 35 to 950 Da.

The derivatised samples were diluted with ethyl acetate (TMS:80 �L; EO–TMS, BO–TMS: 665 �L; ISP: 567.5 �L) or pyridineEO–TFA and BO–TFA: 665 �L) and filtered (PTFE membrane,.45 �m, 13 mm diameter) prior to injection. Aliquots of 0.2 �l were

ntroduced into the splitless injector using an autosampler 7683BAgilent Technologies, USA).

.6. Peak identification and quantification

Peak quantification was accomplished using MSD Chemstation.2.01.1177 (Agilent Technologies, USA). Peaks were assigned byomparing their retention times and mass spectra with those ofespective reference compounds. Assignment of the syn and antitereoisomers was accomplished according to literature [8,16,17].uantitative evaluation based on selected ions. Calibration wasased on peak height and peak area and accomplished using sub-ets of the reference compounds at five different concentrationevels.

.7. Method validation

The developed method was validated for 10 reference com-ounds with respect to linearity, limit of detection (LOD),

imit of quantification (LOQ), precision, and accuracy. Regressionoefficients were determined serving as acceptance criterion forinearity. Linearity was evaluated using the linear regression of thebserved signal with respect to concentration. The LOD and LOQf compounds were calculated according to the 3s and 10s crite-ion, i.e., the threefold or tenfold standard deviation of the noiseuantified via single point height calibration after DIN 32465:2008-1 [18]. Method precision was investigated by determining theelative standard deviation (RSD) of different measurements per-ormed on three independent samples on the same day. Additionalecovery tests were conducted to evaluate the accuracy of theethod. Leaf samples were spiked with a well-defined amount

f a standard mixture. After extraction, the recovery rates of the

ndividual reference compounds were determined.

The following fragment ions were excluded from quantitativenalysis due to their origin of derivatisation reagent: TMS andO–TMS (m/z 73); EO–TFA (69); BO–TMS 73 (91), except in the case

. A 1281 (2013) 115– 126

of BO–TFA, whose mass spectra show only m/z 91. The stability ofthe GC/MS performance (retention time, response, peak geometry)was confirmed by repetitive analysis of identical standards.

2.8. Statistical analysis

Statistical analysis was based on mass spectra and collected as aset of raw intensities and normalized during calculation. Discrim-inant analysis of mass ions was computed using SAS EnterpriseGuide 4.1 (SAS Institute, Cary, NC, USA) after reducing the numberof included ions; this was necessary because of the limited sam-ple size supported by the software. Pre-selection was either basedon an abundance of single molecular or fragment ions (>100) oron an abundance of certain ions accumulated for all 106 individ-uals (>500). The classification variable was based on the individualderivation strategies, while p < 0.01 was used to select significantions for further calculation. Multivariate data analysis was per-formed with Origin 8.6 software (OriginLab, Northampton, MA,USA). The hierarchical cluster analysis of the resulting mass spectrawas created using the group average cluster method and Euclideandistance type calculation.

3. Results and discussion

The reference compounds used in this study represent the greatvariety of carbohydrates found in the plant kingdom; these wereselected to cover the respective wide range of molecular weight. Onthe other hand, various sets of mono and disaccharides of identi-cal or very similar molecular weight were included to evaluate theperformance of the different derivatisation strategies with regardto chromatographic (retention time, peak height, width, and geom-etry, peak separation) and mass spectroscopic criteria (response,fragmentation behavior), and matrix effects. An overview of refer-ence compounds is given in Table S1 (supplementary part).

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.chroma.2013.01.053.

3.1. Chromatographic properties: separation of compound peaks

The number of derivatives per analyte, the quality of peak sep-aration, and the duration of the chromatographic run are the maincriteria for the efficiency of a derivatisation method.

Single-step O-isopropylidenation (ISP) using anhydrous coppersulfate as a catalyst was confirmed to be an efficient derivatisationstrategy as it usually affords only one peak for most of the carbohy-drates; hence, that method generates the lowest number of peaks.This is due to the particular stereochemical requirements for ace-tonidation, i.e., the presence of at least one vicinal diol moiety thatcarries the two neighboring hydroxyl groups in syn-configuration.Amongst the studied reference compounds only d-(-)-riboseand d-(-)-fructose afforded two different cyclic ISP derivativeseach, namely 2,3-O-isopropylidene-1,5-anhydro-d-ribofuranoseand 2,3-O-isopropylidene-d-ribofuranose for d-(-)-ribose and1,2:4,5-di-O-isopropylidene-d-fructopyranose and 2,3:4,5-di-O-isopropylidene-d-fructopyranose for d-(-)-fructose, which is inagreement with the literature [15,19,20]. Glycolaldehyde, the onlypossible aldodiose, gives no isopropylidene derivative because itcontains only one single OH group [21].

Baseline peak separation using the weakly polar stationaryphase specified above, succeeded for almost all acetonides in

the set of carbohydrates studied. Except for the two peak pairs,namely (1) 2,3:4,5-di-O-isopropylidene-d-fructopyranose and1,2:5,6-di-O-isopropylidene-d-glucofuranose and (2) 2,3:4,6-di-O-isopropylidene-d-allopyranose and 2,3:5,6-di-O

atogr

-n

ttfprto(aadd)abfmaodttwa

soltatotame5acOa((am((t(dc

apaonvNacdt

M. Becker et al. / J. Chrom

isopropylidene-d-mannofuranose, partial overlapping couldot be avoided.

Trimethylsilylation, another single-step derivatisation methodhat is frequently used in GC/MS, was confirmed to afford a dis-inctly higher number of peaks when compared to ISP. Here, theormation of up to five isomers led to significant peak overlap-ing and a higher number of coeluting compounds. Based on theetention times, the mass spectra of TMS derivatives of authen-ic compounds, and literature data [9,22,23], peak overlapping wasbserved for: (1) d-(-)-arabinose (open chain form), d-(-)-ribose�-furanose) and (�-pyranose), (2) d-(-)-fructose (�-furanose)nd d-(+)-mannose (�-pyranose), (3) d-(-)-fructose (�-furanose)nd d-(-)-tagatose (�-furanose), (4) d-allose (�-furanose) and-(-)-tagatose (�-furanose), (5) d-(-)-tagatose (�-pyranose) and-allose (�-pyranose), (6) d-(-)-tagatose (�-pyranose) and d-(--fructose (open chain form), and (7) d-(+)-glucose (�-pyranose)nd d-(-)-tagatose (open chain form). Unlike the results describedy García-Raso et al. [22], five instead of four peaks were obtainedor the aldopentoses d-(-)-ribose and d-(-)-arabinose. The MS frag-

entation pattern evidenced that both of the additional peaksre due to the formation of the TMS derivatives of the respectivepen-chain aldopentose isomers, which are sometimes neglectedue to their low relative intensity [23]. Glycolaldehyde was fur-her shown to be quantitatively converted into the correspondingrimethylsilylated dimer under TMS derivatisation conditions,hich is assumed to follow a similar mechanism as trimerisation of

cetaldehyde.Two-step oximation–trimethylsilylation derivatisation was

hown to be superior to simple per-trimethylsilylation in termsf chromatographic resolution because of the comparativelyow number of isomers formed (cf. Fig. 1). This is due tohe fact that oxime formation “blocks” the anomeric carbontom and prevents the carbohydrates from further isomeriza-ion via the open-chain configuration. Furthermore, the usef O-ethylhydroxylamine instead of O-benzylhydroxylamine forhe oximation step prior to silylation was found to result in

better peak separation performance. This is assumed to beainly due to the somewhat higher affinity of the BO moi-

ty with the used stationary phase (polydimethylsiloxane with% phenyl groups), which can lead to increased retention timesnd peak broadening. Therefore, a higher number (=10) ofoeluting (overlapping) compounds were found for the mixed-benzyloxime-TMS (BO–TMS) derivatives: (1) d-(-)-tagatose (syn)nd d-(-)-fructose (anti), (2) d-(-)-ribose (anti), d-(+)-xylosesyn) and d-(-)-arabinose (syn), (3) d-allose (syn), d-(+)-mannosesyn) and d-(+)-mannose (anti), and (4) d-(+)-glucose (syn, anti)nd d-allose (anti). For EO–TMS derivatisation, only six iso-ers coeluted in three pairs: (1) d-(+)-mannose and d-allose

anti-oxime isomers), (2) d-(+)-xylose and d-(-)-arabinose, and3) d-allose and d-(+)-mannose (syn-ethoxime isomers each). Fur-hermore, some of the syn/anti pairs obtained from both EO–TMSd-(-)-arabinose, d-(-)-ribose and d-(-)-fructose), and BO–TMSerivatisation (d-(+)-mannose and d-(+)-glucose), respectively,ould not be separated.

In contrast, ethoximation–trifluoroacetylation (EO–TFA)llowed for a satisfying separation of the syn/anti isomerairs of all studied carbohydrates, except d-(-)-fructosend d-(-)-tagatose. Unfortunately, both of the combinedximation–trifluoroacetylation approaches increased theumber of peaks per analyte. This is due to the virtually unpre-entable presence of traces of water during derivatisation with,O-bis(trimethylsilyl)trifluoroacetamide. This inevitably gener-

tes sufficient amounts of trifluoroacetic acid that catalyze theonversion of cyclic oximes, formed from ketohexoses such as-(-)-fructose and d-(-)-tagatose, into lactams [8]. However due tohe unpredictable extent of lactam formation, both peak ratio and

. A 1281 (2013) 115– 126 119

intensity of the formed products is inconsistent. This effect is lesspronounced for BO–TFA compared to EO–TFA.

Three pairs of coeluting compounds were found for theEO–TFA derivatisation approach and the applied chromato-graphic conditions: (1) d-(-)-ribose (anti) and d-(-)-arabinose(anti), (2) d-(+)-glucose (anti) and d-(+)-mannose (anti), (3)d-(+)-glucose (syn) and d-(+)-mannose (syn). For BO–TFA this num-ber increased to four: (1) d-(-)-arabinose (anti) and d-allose (anti),(2) d-(+)-xylose (syn), d-allose (syn) and d-(-)-arabinose (syn), (3)dihydroxyacetone and d-(-)-ribose (anti), and (4) d-(+)-mannose(anti) and d-(-)-ribose (syn).

The mass increment that can be gained through the introduc-tion of substituents of higher molecular weight commonly shiftsthe respective peaks toward higher retention times, which canimprove the resolution of complex spectra. However, this effectwas not observed for the aldodiose glycolaldehyde when TMS wasreplaced by TFA in the second derivatisation step after oximationwith O-ethoxyamine. Instead of eluting at higher retention time,the respective peak underwent a significant shift toward shorterretention time which is in accordance with the literature [24–26].

The number of coeluting compounds in a complex carbohydratemixture depends on many factors. Next to the concentration of indi-vidual constituents, the number and ratio of the isomers that areformed for each sugar has a significant impact on the quality ofthe chromatographic result. For complex matrices of structurallysimilar carbohydrates including their isomers, a tailored mass shiftafforded by specific derivatisation reagents can decisively improvethe resolution.

Depending on the type, number, and molecular weight of theprotective groups introduced, derivatisation commonly shiftsthe peaks of the respective analytes toward longer retentiontimes if a weak polar phase is used, as was the case in this study(cf. Table 1). For all studied reference compounds, the retentiontimes of the respective derivatives increased in the order EO–TFA(15.85 min) < ISP (21.58 min) < BO–TFA (25.76 min) < EO–TMS(28.89 min) < TMS (29.89 min) < BO–TMS (38.28 min) which con-firms that the impact of the mass gain is superimposed by theresulting overall polarity and volatility of the derivatives [27].Amongst the mixed oxime–trimethylsilyl derivatives, BO–TMSsugars were confirmed to have significantly higher retentiontimes compared to the EO–TMS derivatives; this is assumed to bedue to both higher mass increment (EO: 43, BO: 105) and �–�interactions between the benzyl groups and the phenyl groups ofthe stationary phase which are stronger than the London forcesthat are contributed by the ethyl groups of the EO–TMS derivatives.

3.2. Informational value of the mass spectra

Unlike BO–TFA, ISP derivatives form the least intensive [M-15]+

fragments by cleaving-off one methyl group. This is similar to thederivatisation methods TMS, EO–TMS, BO–TMS, which commonlyallow for deducing M+. Thus, an intensive m/z 190 fragment ion canbe observed for the ISP derivative of xylose (Fig. 2b) that is formedby eliminating one CH3 group either from the 2,2-dimethyl-1,3-dioxolane or the 2,2-dimethyl-1,3-dioxane moiety [19].

ISP-derivatives of carbohydrates are known to give com-plex mass spectra. Next to the intensive [M-15]+ fragment, acouple of typical signals are found in the lower mass range,such as those at m/z 43 and 59, and are associated withthe formation of C2H3O+ and C3H7O+, respectively. Also m/z85 can be observed in all mass spectra of ISP carbohydratederivatives, except for ISP-DHA. 1,2:4,5-di-O-isopropylidene-d-

fructopyranose (ISP-Fru1) can be distinguished from the later elu-ting 2,3:4,5-di-O-isopropylidene-d-fructopyranose (ISP-Fru2) dueto its different mass fragment pattern (ISP-Fru1:m/z100, 117, 144;ISP-Fru2: m/z171, 205, 229). Similarly, 2,3-O-isopropylidene-1,

120 M. Becker et al. / J. Chromatogr. A 1281 (2013) 115– 126

Fig. 1. Chromatograms of a set of eight monosaccharides (C5–C6), GA, and DHA after TMS (a), ISP (b), EO–TMS (c), BO–TMS (d), EO–TFA (e), and BO–TFA (f) derivatisa-tion. Peak assignment for (a) and (c–f): 1: glycolaldehyde; 2: dihydroxyacetone; 3: arabinose; 4: ribose; 5: xylose; 6: fructose; 7: mannose; 8: tagatose; 9: allose; 10:glucose; *: internal standard (methyl�-d-galactopyranoside); a: anti-isomer; b: syn-isomer; c: �-furanose; d: �-furanose; e: �-pyranose; f: �-pyranose; g: open chain form;h: unknown compound; i: dimer; j: furanose. Peak assignment for (b): 1:1,3-O-isopropylidene-2-propanone, 2a: 2,3-O-isopropylidene-1,5-anhydro-d-ribofuranose, 2b:2,3-O-isopropylidene-d-ribofuranose; 3: 1,2:3,4-di-O-isopropylidene-l-arabinopyranose; 4: 1,2:3,5-di-O-isopropylidene-d-xylofuranose; 5: 1,2;3,4-di-O-isopropylidene-�-d-tagatofuranose; 6a: 1,2:4,5-di-O-isopropylidene-d-fructopyranose; 6b: 2,3:4,5-di-O-isopropylidene-d-fructopyranose; 7: 1,2:5,6-di-O-isopropylidene-d-glucofuranose;8 annop

5idmwco

: 2,3:4,6-di-O-isopropylidene-d-allopyranose; 9: 2,3:5,6-di-O-isopropylidene-d-mropylidene acetone.

-anhydro-d-ribofuranose, which elutes earlier than 2,3-O-sopropylidene-d-ribofuranose, can be distinguished by theifferent m/z values of their [M-15]+ fragments (m/z 172 vs

/z 190). ISP-ketohexoses show an intense [M-15]+ at m/z 245,hile aldohexoses show similar fragmentation patterns with

hanging fragment intensities and characteristic mass fragmentsf m/z 101 and 187. This is not available in ketohexoses. The

furanose; *: internal standard (13C-Glucose); x1: unknown compound; x2: diiso-

ISP derivatives of the ketohexoses d-fructopyranose (1,2:4,5-di-O-isopropylidene-d-fructopyranose) and �-d-tagatofuranose(1,2;3,4-di-O-isopropylidene-�-d-tagatofuranose) afford the frag-

ment pair m/z 72 and 117. This pair which is characteristic for1,2-O-isopropylidene derivatives of ketohexoses [20] cannot befound for the 1,2:5,6-di-O-isopropylidene-derivative of the aldo-hexose d-glucofuranose.

M. Becker et al. / J. Chromatogr. A 1281 (2013) 115– 126 121

F nt der1 BO–TB

am[awDnttafwboirt3odfa

esdnf1

mtf(sa

ig. 2. Electron impact (EI) mass spectra of glucose derivatives obtained by differe,2:3,5-Di-O-isopropylidene-d-xylofuranose, (c) EO–TMS: syn-xylose, EO–TMS, (d)O–TFA.

The mass spectra of carbohydrate TMS-derivatives contain series of typical fragment ions, such as m/z 73 [(CH3)3Si]+,/z 75[(CH3)2Si OH]+, 147 [(CH3)2Si O Si(CH3)3]+, 103

(CH3)3Si O CH2]+, 205 [(CH3)3Si O CH2 CH O Si(CH3)3]+

nd 217 [(CH3)3Si O CH = CH CH O+ Si(CH3)3] (pyranoses),hich can be used for both identification and quantification [28].ifferentiation between aldoses and ketoses, furanoses and pyra-oses, and open-chain isomers can be accomplished using theseypical fragment ions. The fragment m/z 217 on the other hand isypical for cyclic isomers of aldoses and the ratio between m/z 204nd 217 can be used to distinguish furanoses from pyranoses. Whileuranoses form an intensive m/z 217 fragment but have a veryeak tendency to form m/z 204, pyranoses can be easily identified

y their m/z 204:217 ratio which is typically close to or even abovene [9]. Thus, the two isomers of per-trimethylsilylated xylose,.e., �- and �-d-xylopyranose-TMS forms, have an m/z 204:217atio of 2.25 and 2.13, respectively. Open chain TMS-isomers, onhe other hand, can be detected by the very intense fragment m/z06 and higher retention times compared to the TMS derivativesf the cyclic isomers. Ketohexoses such as d-(−)-fructose and-(−)-tagatose can be distinguished from aldohexoses, as only the

ormer affords an intensive m/z 437 fragment for both furanosesnd pyranoses [5,9].

The mass spectra of the studied carbohydrates after sequentialthoximation and trimethylsilylation were confirmed to be veryimilar. Thus, the respective EO–TMS (Fig. 2c) and BO–TMS (Fig. 2d)erivatives of xylose differ from each other mainly in their promi-ent m/z 91 benzylium fragments. Minor differences are due to

ragments formed from EO–TMS derivatives, such as those at m/z74, 236, and 338.

In contrast, significant differences were observed between theass spectra of the mixed EO–TFA and BO–TFA derivatives, respec-

ively. While the mass spectrum of the BO–TFA derivative of xylose,

or example, is dominated by one single mass fragment at m/z 91Fig. 2e), which is caused by the well-known benzylium-tropyliumtabilization ([Ph-CH2]+ → [C7H7]+) [29], the EO–TFA analog gives

complex spectrum of mass fragments (Fig. 2f). The fragmentation

ivatisation approaches: (a) TMS: pertrimethylsilylated �-d-xylopyranose, (b) ISP:MS: syn-xylose, BO–TMS, (e) EO–TFA: syn-xylose, EO–TFA, (f) BO–TFA: syn-xylose,

pattern here is minted by signals such as those at m/z 97 [COCF3]+

and [M-113]+; the latter is formed by elimination of one triflu-oroacetoxy group [OCOCF3]+. No differences in the EO–/BO–TFAfragmentation pattern were observed between the syn- and anti-isomers except a higher overall response for the BO–TFA.

A hierarchical, unsupervised cluster analysis (HCA, Fig. 3) wasperformed to compare the mass spectra of different carbohydratederivatives, aiming at the revelation of associated data and clas-sification of observations. HCA cluster formation was shown tobe affected by the different derivatisation approaches. Cluster 1(see Fig. 3), which comprises all per-trimethylsilylated, EO–TMSand most of the BO–TMS derivatives, is split into three sub-clusters (d, g, j), of which (d) can be further grouped into theper-trimethylsilylated furanoses (a), per-trimethylsilylated pyra-noses (b), and the TMS derivatives of open-chain ketohexoses,dihydroxyacetone plus the EO–TMS derivatives of dihydroxyace-tone and glycerin aldehyde (c). Amongst the TMS-pyranoses, thosewith an m/z 204:217 ratio ≥1 are located on the left and those closeto 1 on the right-hand side of the sub-cluster. Low variability wasobserved between the BO– and EO–TMS derivatives. They formedmixed groups of aldohexoses (sub-cluster (g)), and aldopentosesplus ketohexoses (sub-cluster (j)). Within these sub-clusters theEO–TMS derivatives (branches (f) and (i)) can be distinguished fromthe BO–TMS derivatives (branches (e) and (h)). Cluster 2 comprisesthe BO–TMS derivatives of dihydroxyacetone, glycolaldehyde, andthe BO–TFA derivatives (branch (k)). For the BO–TFA derivatisationapproach, the lowest distance between the respective derivativeswas observed in the HCA, which is due to the high level of similarityin their mass spectra. Cluster 3 contains the EO–TFA derivatives.Here, considerable differences were observed for the mass spec-tra of both EO–TFA derivatives of aldohexoses (l) and EO–TFAderivatives of aldopentoses (m). Furthermore, the great distancebetween the two branches reveals the pronounced dissimilarities

between the EO–TFA derivatives of aldohexoses and aldopentoseswith regard to their EI fragmentation behavior. Of all of the studiedderivatisation approaches, the mass spectra of the ISP deriva-tives, which are described by cluster 4, were shown to have the

122 M. Becker et al. / J. Chromatogr. A 1281 (2013) 115– 126

F set of eight carbohydrates (C5–C6), glycolaldehyde and dihydroxyacetone after derivati-s

hIftg

afoifa

pnnsaEatda

3

odeT

ig. 3. Hierarchical cluster analysis of the electron impact mass spectra (70 eV) of aation using six different approaches.

ighest distance between individual compounds. The EO–TFA andSP derivatives of dihydroxyacetone were further confirmed toollow fragmentation mechanisms that distinguish them fromhose of the other studied carbohydrates, as it is evident from theirreat distance to other derivatives within the respective clusters.

The value of the mass spectroscopic data with regard to peakssignment can be summarized as follows: the pronounced dif-erences between the mass spectra of individual ISP derivativesf carbohydrates mentioned above provide most of the structuralnformation among the studied derivatisation approaches; there-ore, they allow for an unerring distinction of aldohexoses such asllose, glucose, or mannose.

Per-trimethylsilylation supports differentiation of furanoses,yranoses, and open chain isomers but is inferior to isopropylide-ation, as it neither allows �- and �-isomers to be distinguished,or different members of a particular group of carbohydrates,uch as aldohexoses or ketohexoses. The latter also applies forll approaches that include an oximation step (EO–TMS, BO–TMS,O–TFA and BO–TFA). In addition, differentiation between syn- andnti-isomers is not possible, too. BO–TFA has been shown to providehe least suitable derivatisation approach for complex carbohy-rate mixtures, as the respective mass spectra provide the leastmount of structural information.

.3. Quantitative analysis

Five-point-calibration curves were calculated for all derivatives

f each of the carbohydrates (C5–C6), glycolaldehyde (GA), andihydroxyacetone (DHA) that belonged to the studied set of ref-rence compounds (cf. Fig. 4, calibration curves for �-d-xylose).he calibration data obtained confirmed that all investigated

Fig. 4. Calibration curves for the different derivatives of �-d-xylose.

derivatisation approaches are basically suitable for quantitativeanalysis of monosaccharides using GC/MS. The data further demon-strates that the peak areas of all identified compounds (cf. TableS1) increase proportionally with concentration. The correlationcoefficients of the respective calibration curves for the derivatisedisomers, which covered a concentration range of three orders ofmagnitude, were shown to range from 0.9831 to 0.9999 (cf. TableS1, supplementary part). Comparative calibrations based on either

peak area or peak height of the respective syn-, anti-, furanose, orpyranose isomers have confirmed that quantification of the stud-ied set of carbohydrates (C5–C6), GA and DHA can be accomplishedusing one of these peaks only, i.e., a simultaneous consideration

atogr

ofoipdfomiauIwT1v(B5bsbercv

a�icEcttor

3

pponMoarafmardSpi“

chapa

M. Becker et al. / J. Chrom

f all peaks originating from one compound is not a prerequisiteor quantitation. For sensitivity reasons, quantification of mixedxime-TMS or oxime-TFA derivatives should be based on the synsomers, as anti isomers commonly produce lower peak areas andeak heights [12]. Median values were used to compare the limit ofetection (LOD) and the relative standard deviation (RSD) of the dif-erent derivatisation strategies because of their robustness towardutliers. TMS (mean 5.28, median 0.63) and BO–TMS (mean 1.72,edian 0.76) derivatisation were shown to have the lowest LOD

n column values, followed by EO–TMS (mean 5.98, median 1.27)nd BO–TFA (mean 7.42, median 2.95). Higher LOD in column val-es were observed for EO–TFA (mean 15.20, median 11.61) and

SP (mean 19.22, median 12.18). The relative standard deviationas calculated based on peak area of the standard compounds.

MS (mean 1.99, median 1.00) and BO–TMS (mean 1.02, median.03) derivatisation were confirmed to produce the lowest RSDalues, followed by the EO–TMS (mean 1.93, median 1.24) and ISPmean 2.62, median 1.96) variants. Higher values were observed forO–TFA (mean 5.94, median 4.74) and EO–TFA (mean 5.80, median.25), while the other derivatisation approaches had a median valueelow two percent. Mixed derivatisation involving an oximationtep afforded less isomers than single TMS or TFA derivatisationy “setting” the configuration at the anomeric carbon atom; how-ver, our assumption that this goes along with higher analyticalesponses and lower detection limits, did not prove true. On theontrary, per-trimethylsilylation afforded the lowest LOD and LOQalues due to their large m/z 204 and 217 fragments.

Based on the calibration curves of all studied derivatisationpproaches and reference compounds (exemplarely shown for-d-xylose, Fig. 4), it can be summarized that sequential benzox-

mation and trifluoroacetylation afford the highest slope of thealibration curves, followed by the EO–TFA, BO–TMS, ISP, TMS, andO–TMS derivatisation approaches. The calibration curves werealculated by excluding those mass fragments that originated fromhe derivatisation reagent except for the BO–TFA approach wherehis method cannot be applied as the respective derivatives affordnly one intense peak at m/z 91, derived from the oximationeagent.

.4. Analysis of matrix effects

The term “matrix effect” is used to describe sample-specifichenomena in analytical chromatography that can impede botheak assignment and quantitation of target compounds. Bothccurrence and extent of matrix effects largely depend on theumber, type and amount of compounds present in a sample.ass spectrometric detectors are considered to be the detectors

f choice for gas chromatographic analysis of complex matricess they provide structure-specific information additionally to theetention characteristics which greatly supports peak assignmentnd quantitation in particular for coeluting compounds. However,or complex matrices such as multiple carbohydrate, isomer-rich

ixtures that afford very similar MS fragmentation pattern, thevailability of structure-related information does not imperativelyesolve all problems related to peak assignment and, therefore,oes not necessarily improve the preciseness of quantitation [30].eparation of target and non-target compounds (“matrix”) by high-erformance separation techniques (e.g., solid-phase extraction)

s an approach to overcome such problems but bears the risk ofloosing” a certain fraction of the target compounds.

Direct derivatization of complex samples such as apricot [31],itrus [32], cherry, and apple [33] has been shown to allow for

ighly reproducible quantification of different target compoundsnd does not necessarily require preceding removal of the sam-le matrix. The broad range of compounds (sugars, sugar alcohols,cids, amino acids, e.g.) that can be analyzed in one run, the

. A 1281 (2013) 115– 126 123

low expenditures of labor and time, and the applicability forhigh-throughput analysis [31,34] are further advantages of thisapproach.

We propose that the effect of a given sample matrix on thereliability of quantitation results can be described by a sample-specific average matrix impact MI value which is calculated as themean of the absolute matrix impact values of single (target, refer-ence) compounds MISC (Eq. (1)). These MISC values can be calculatedby comparing the recovery rates of the reference compoundsusing two calibration methods that feature a different sensitiv-ity toward matrix effects. These two calibration approaches arebased on standard addition which is known to allow for compensat-ing matrix effects [35]. However, depending on how quantitationis performed, the recovery rates of single compounds as deter-mined by standard addition have a different sensitivity towardmatrix effects. While calibration based on peak area commonlyshows a stronger response to matrix effects, calibration based onpeak height is known to be less affected by the matrix of com-plex samples and less erroneous for constituents present in lowconcentration [26] provided that the peak maxima of different com-pounds do not lay upon each other [36,37]. However, care must betaken with regard to column overloading as calibrations based onpeak height are more sensitive leading to non-linear correlationsbetween peak height and concentration [38].

MISC = 1n

n∑SC=1

√(MISC )2 (1)

MISC = (RSC,height − RSC,area) × 100RSC,height

(2)

RSC,height =(

mSC(S+spiked) − mSC(S)

mSC(spiked)

)× 100% (3a)

RSC,area =(

mSC(S+spiked) − mSC(S)

mSC(spiked)

)× 100% (3b)

mSC(S+spiked): amount of target compound in spiked sample; mSC(S):amount of target compound in non-spiked sample; mSC(spiked):amount of target compound added (spiked).

The MISC values calculated according to Eq. (2) thereforereport the deviation (%) of the matrix-affected “peak areacalibration”-based recovery rates (RSC,area, Eq. (3b)) from the non-matrix-affected “peak height calibration”-based recovery rates(RSC,height, Eq. (3a)) of single (target) compounds for a particularderivatisation approach.

The above proposed method has been complimentary applied to(1) verify the results obtained with the above-described standardcalibration method and (2) to quantify the matrix effects for thedifferent derivatisation approaches using two “real” carbohydrate-rich samples of biological (grapevine leaves) and synthetic origin(crude product obtained from the “formose” reaction betweenformaldehyde and aqueous calcium hydroxide).

The chromatograms of the formose reaction mixture (A) and ofthe leaves of Vitis vinifera (B) after EO–TMS derivatisation are shownin Fig. 5. While the formose reaction mixture afforded a rathercomplex chromatogram with a considerable number of coelutingcarbohydrates, sugar alcohols, and sugar acids, the grapevine leafsample is less complex but is characterized by high concentrationsof individual compounds.

The average matrix impact values MI shown in Table 2 revealthat sequential ethoximation–trimethylsilylation is a derivati-

sation approach, whose chromatographic separation and massspectrometric fragmentation behavior is comparably robusttoward matrix effects. This is evident from the low MI values forboth of the studied matrices (formose 9.1, leaves 10.4) which results

124 M. Becker et al. / J. Chromatogr. A 1281 (2013) 115– 126

Fig. 5. Chromatograms of “real” carbohydrate-rich matrices of biological and synthetic origin after sequential ethoximation–trimethylsilylation: (A) crude product of thereaction of formaldehyde with calcium hydroxide (formose-reaction); (B) grapevine leaves (Vitis vinifera), peak description: 1: malic acid; 2: tartaric acid; 3: ribose; 4: internalstandard (methyl-galactose); 5: fructose; 6: syn-glucose; 7: anti-glucose; 8: gluconic acid; 9: myo-inositol; 10: sucrose.

Table 2Matrix impact (MISC) values of individual reference compounds (cf. Eq. (2)) and average matrix impact values MI (cf. Eq. (1)) calculated for the different derivatisationapproaches that have been applied to study the composition of carbohydrate-rich complex matrices of different origin.

TMS EO–TMS BO–TMS EO–TFA BO–TFA ISPd

Form. Leaf Form. Leaf Form. Leaf Form. Leaf Form. Leaf Form. Leaf

d-allose 2.0 5.0 6.5 −2.8 b b −38.6 −25.9 6.5 −51.6 −31.8 −16.0d-(−)-arabinose 2.3 −3.5 −3.1 −0.1 −8.3 −8.3 −19.6 14.4 −14.5 −63.0 −32.9 43.2dihydroxyacetone n.d. n.d. −5.3 12.1 3.2 −5.3 13.9 9.8 b b −49.3 −4.6d-(−)-fructose −51.6 −11.9 −34.8 −67.5 −15.3 −72.1 a a a a −0.4 −25.8d-(+)-glucose 4.3 3.4 5.6 4.8 7.4 −159.3 6.1 −41.0 23.3 c −36.4 −104.0glycolaldehyde n.d. n.d. 5.1 −8.8 16.9 6.6 b b −6.6 −0.1 a a

d-(+)-mannose 16.2 7.0 −16.9 0.3 b b −3.3 b c c −106.9 −28.8d-(−)-ribose −28.1 −10.5 −1.8 −2.9 −10.8 −13.6 12.5 24.0 −20.9 c −62.0 n.d.d-(−)-tagatose 10.2 7.2 2.3 1.9 0.4 −9.5 a a a a −30.9 n.d.d-(+)-xylose 10.9 −2.6 9.3 2.5 1.2 −14.9 −9.1 7.8 2.3 8.7 −41.8 −4.1

MI value 15.7 6.4 9.1 10.4 8.0 36.2 14.7 20.5 12.4 30.8 43.6 32.4

Form, formose reaction; Leaf, leaves of Vitis vinifera; n.d., not detected.a No derivative formed (ISP) or partial decomposition of derivative due to the presence of TFA.b Deconvolution of coeluating standard and/or target compounds not possible.c Deconvolution of coeluating standard or target with non-target compounds of matrix not possible.d All values divided by 10.

Table 3Evaluation of the studied derivatisation approaches for GC/MS-based carbohydrate analysis of complex sample matrices.

Derivatisationapproach

Chromatographic properties MS informationalvalue

Quantitativecharacteristics

Matrix effect Method labo-riousness

Ranking

Peak Complete detection ofreference compounds

LOD RSD

Count per compound Separation

EO–TMS o o + o + + + o 1TMS − − + o + + + + 2BO–TMS o o + o + o o o 3EO–TFA o + − + − + o o 4BO–TFA o + − − o o o o 5ISP + + − + − + − − 6

Rating of characteristics: “+” = good, “o” = moderate, “−” = negative.

atogr

f(SMstaelw

fc

b((tc

4

dccaptic

amdrldfdctnbc

pawtocprdcrrdap

stw

[

[[[[[[[[

[[[[

[

[[[

[

M. Becker et al. / J. Chrom

rom both the small differences between the two recovery ratesRSC,area and RSC,height) and their high absolute values (cf. Table S2,upplementary part). Per-trimethylsilylation affords similar lowI values however a larger variance was observed for the two

tudied carbohydrate-rich matrices (formose 15.7, leaves 6.4). Dis-inctly higher MI values were obtained for the other derivatisationpproaches which indicate a stronger sensitivity toward matrixffects. The unsatisfying high MI values for ISP derivatisation areikely due to the high sensitivity of this method toward traces of

ater [39].Supplementary data associated with this article can be

ound, in the online version, at http://dx.doi.org/10.1016/j.hroma.2013.01.053.

Some of the BO–TMS, EO–TFA and BO–TFA derivatives could note included in the matrix impact considerations for several reasons:a) they could not be obtained due to structural limitations (ISP),b) they decomposed due to the presence of traces of TFA, or (c)hey coeluted either with spiked standards or target and non-targetompounds from the sample matrix.

. Conclusion

The present study investigated different single- and two-steperivatisation approaches for GC/MS analysis of carbohydrate-richomplex matrices using a test kit of eight monosaccharides, gly-olaldehyde, and dihydroxyacetone. Based on the revealed prosnd cons of the studied methods which comprise chromatographicroperties, informational value of mass spectra and robustnessoward matrix effects, the following conclusions can be made lead-ng to the proposed derivatisation performance ranking shown inlosing Table 3.

ISP has been demonstrated to be an useful derivatisationpproach for GC/MS analysis of carbohydrates in simple matrices asass spectroscopic information, signal intensity (formation of 2,2-

imethyl-1,3-dioxolane moieties; [5]), and the chromatographicesolution is fairly good for a certain number of compounds. Theatter is due to the fact that only one peak is formed per carbohy-rate except for ribose and fructose. ISP, however, is of limited valueor analysis of complex sugar mixtures because carbohydrates thato not have a vicinal cis-diol moiety in their structure cannot beonverted into the respective ISP derivative. The high sensitivity ofhis particular derivatisation approach toward water, and the longeutralization times required, further prevent this method fromeing used in high throughput analysis of “real” carbohydrate-richomplex matrices.

Per-trimethylsilylation affords less complex mass spectra com-ared to ISP or TFA derivatisation and allows for an unambiguousssignment of furanose and pyranose isomers. TMS derivatisationas also shown to be least sensitive to matrix effects and had

he lowest LOD values. However, the performance and throughputf this method suffers somewhat from (1) the enhanced count ofoeluting peaks, which complicates data analysis, and (2) the com-aratively long equilibration time required to obtain reproducibleatios of the isomeric derivatives [40]. Oximation of carbohy-rates is a well-known means to reduce the complexity of gashromatograms, as the conversion of the carbonyl group into theespective oxime derivative diminishes the number of isomers. Theetention times of structurally similar carbohydrates become moreispersed and the informational value of the mass spectra increasess a larger variation is achieved with regard to the fragmentationattern [12].

Sequential oximation–trimethylsilylation has been demon-trated to be an efficient derivatisation approach regardless theype of oximation reagent used. Both EO–TMS and BO–TMS affordell-resolved chromatograms and mass spectra of similarly high

[[

[[

. A 1281 (2013) 115– 126 125

informational value. The use of O-benzylhydroxylamine insteadof O-ethylhydroxylamine increases the retention times of therespective derivatives considerably, which is due to the highermass increment of the benzyl group, and might be advantageousin some cases. Compared to EO/BO–TMS, sequential oximation-trifluoroacetylation has some disadvantages, the decomposition ofketohexoses by traces of trifluoroacetic acid is probably the mostserious one. Furthermore, the mass spectra of carbohydrate BO–TFAderivatives are of very low informational value. Therefore, thismethod is applicable for non-complex matrices where target andmatrix compounds do not interfere.

Based on the conducted study, it can be concluded that sequen-tial ethoximation and trimethylsilylation (EO–TMS) features thehighest GC/MS performance due to the low number of isomersformed, satisfying chromatographic resolution, high informationalvalue of the mass spectra, low LOD, LOQ and RSD values, and highrobustness toward matrix effects. Although somewhat more labo-rious than single step methods, the sum of the features describedabove render EO–TMS a derivatisation approach that can be recom-mended for GC/MS-based carbohydrate analysis of complex samplematrices.

Acknowledgement

The authors gratefully acknowledge the financial support ofthe Christian Doppler Research Society through the CD-laboratoryfor “Advanced Cellulose Chemistry and Analytics”, of the AustrianFederal Ministry of Agriculture, Forestry, Environment and WaterManagement (FP 100196), and of the Austrian Science Fund FWF(P21203-B16).

References

[1] E. Rudnik, Compostable Polymer Materials, Elsevier, Oxford, Boston, 2008.[2] P. Decker, H. Schweer, R. Pohlmann, J. Chromatogr. A 244 (1982) 281.[3] M.L. Sanz, I. Martinez-Castro, J. Chromatogr. A 1153 (2007) 74.[4] M.M. Koek, B. Muilwijk, M.J. van der Werf, T. Hankemeier, Anal. Chem. 78 (2006)

1272.[5] K. Blau, J.M. Halket, Handbook of Derivatives for Chromatography, Wiley, New

York, 1993.[6] V. Zaikin, J. Halket, A Handbook of Derivatives for Mass Spectrometry, IM Pub-

lications, Chichester, 2009.[7] D.R. Knapp, Handbook of Analytical Derivatization Reactions, Wiley, New York,

1979.[8] M.A. Andrews, Carbohydr. Res. 194 (1989) 1.[9] A. García-Raso, M. Fernández-Díaz, M.I. Páez, J. Sanz, I. Martínez-Castro, J. Chro-

matogr. A 471 (1989) 205.10] A.I. Ruiz-Matute, O. Hernandez-Hernandez, S. Rodriguez-Sanchez, M.L. Sanz, I.

Martinez-Castro, J. Chromatogr. B: Analyt. Technol. Biomed. Life Sci. 879 (2011)1226.

11] G.E. Black, A. Fox, J. Chromatogr. A 720 (1996) 51.12] M. Becker, F., Liebner, T., Rosenau, A. Potthast, submitted to Talanta.13] B.S. Mason, H.T. Slover, J. Agr. Food Chem. 6 (1958) 551.14] O. Fiehn, J. Kopka, R.N. Trethewey, L. Willmitzer, Anal. Chem. 72 (2000) 3573.15] S. Morgenlie, Carbohydr. Res. 41 (1975) 285.16] W. Funcke, C. von Sonntag, Carbohydr. Res. 69 (1979) 247.17] J.R. Snyder, Carbohydr. Res. 198 (1990) 1.18] Chemische Analytik, Nachweis-, Erfassungs- und Bestimmungsgrenze, Norme-

nausschuss Materialprüfung, DIN 32645:2008-11.19] D.C. De Jongh, K. Biemann, J. Am. Chem. Soc. 86 (1964) 67.20] S. Morgenlie, Carbohydr. Res. 80 (1980) 215.21] V.G. Zaikin, J.M. Halket, Eur. J. Mass Spectrom. 10 (2004) 555.22] A. García-Raso, I. Martínez-Castro, M.I. Páez, J. Sanz, J. García-Raso, F. Saura-

Calixto, J. Chromatogr. A 398 (1987) 9.23] M. Paez, I. Martínez-Castro, J. Sanz, A. Olano, A. Garcia-Raso, F. Saura-Calixto,

Chromatographia 23 (1987) 43.24] P. Englmaier, Carbohydr. Res. 144 (1985) 177.25] W.A. König, H. Bauer, W. Voelter, E. Bayer, Chem. Ber. 106 (1973) 1905.26] J.P. Zanetta, W.C. Breckenridge, G. Vincendon, J. Chromatogr. A 69 (1972)

291.27] K. Lakszner, L. Szepesy, L. Vida, Chromatographia 19 (1984) 304.

28] S. Moldoveanu, V. David, J. Chromatogr. Libr. (2002) 930.29] P.S. Kalsi, Spectroscopy of Organic Compounds, New Age International, New

Delhi, 2004.30] J. Hajslova, J. Zrostlikova, J. Chromatogr. A 1000 (2003) 181.31] Z.F. Katona, P. Sass, I. Molnar-Perl, J. Chromatogr. A 847 (1999) 91.

1 atogr

[[

[[[

[

26 M. Becker et al. / J. Chrom

32] Z. Fuzfai, I. Molnarperl, J. Chromatogr. A 1149 (2007) 88.

33] Z. Fuzfai, Z.F. Katona, E. Kovacs, I. Molnar-Perl, J. Agr. Food Chem. 52 (2004)

7444.34] I. Molnar-Perl, J. Chromatogr. A 845 (1999) 181.35] K. Mastovska, S.J. Lehotay, M. Anastassiades, Anal. Chem. 77 (2005) 8129.36] R.W. Dixon, G. Baltzell, J. Chromatogr. A 1109 (2006) 214.

[

[[

. A 1281 (2013) 115– 126

37] L.R. Snyder, J.J. Kirkland, J.W. Dolan, Introduction to Modern Liquid Chromatog-

raphy, Wiley, Hoboken, NJ, 2010.

38] R.L. Grob, E.F. Barry, Modern Practice of Gas Chromatography, Wiley-Interscience, Hoboken, NJ, 2004.

39] R.F. Brady Jr., Adv. Carbohydr. Chem. Biochem. 26 (1971) 197.40] J. Lehrfeld, J. Chromatogr. A 120 (1976) 141.

104 

    

 Paper IV  

 Ethoximation‐silylation approach for mono‐ and disaccharide 

analysis and characterization of their identification 

parameters by GC/MS. 

 

 Talanta, vol. 115, pp. 642‐651. 

 

   

 

 

 

 

Becker, M.; Liebner, F.; Rosenau, T.; Potthast, A. (2013) 

 

 

 

 

 

 

 

Talanta 115 (2013) 642–651

Contents lists available at SciVerse ScienceDirect

Talanta

0039-91http://d

n CorrE-m

journal homepage: www.elsevier.com/locate/talanta

Ethoximation-silylation approach for mono- and disaccharide analysisand characterization of their identification parameters by GC/MS

M. Becker n, F. Liebner, T. Rosenau, A. PotthastUniversity of Natural Resources and Life Sciences, Department of Chemistry, Division of Chemistry of Renewables and Christian-Doppler Laboratory“Advanced Cellulose Chemistry and Analytics”, Muthgasse 18, A-1190 Vienna, Austria

a r t i c l e i n f o

Article history:Received 1 February 2013Received in revised form17 May 2013Accepted 21 May 2013Available online 28 May 2013

Keywords:CarbohydratesDerivatizationGC/MSEthoximationRetention indexTrimethylsilylation

40/$ - see front matter & 2013 Elsevier B.V. Ax.doi.org/10.1016/j.talanta.2013.05.052

esponding author. Tel.: +43 1 47654 6412; faxail address: manuel.becker@boku.ac.at (M. Be

a b s t r a c t

The qualitative and quantitative analysis of complex carbohydrate mixtures is a challenging problem.When tackled by GC/MS, close retention times and largely similar mass spectra with no specific featurescomplicate unambiguous identification, especially of monosaccharides. An optimized pre-capillaryethoximation-silylation GC/MS method for determination of monosaccharides and disaccharides wasapplied to a wide range of analytes (46 compounds). The two-step derivatization resulted in a pair of synand anti peaks with specific retention and intensity ratio. The resulting dataset of mass spectra wassubjected to a PCA-based pattern recognition. An oxime peak identifier (OPI) of the carbohydrateanalytes, based on the combination of an internal standard and the corresponding syn/anti peak ratios,increased the reliability of the identification of reducing carbohydrates. Finally, the introduced EtOx-TMSderivatization method was applied to four different carbohydrate matrices (agave sirup, maple sirup,palm sugar, and honey).

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Carbohydrate analysis is diverse, multifaceted, and encounteredin a large number of applications. One important area is the identi-fication and quantification of low-molecular weight carbohydrates,especially monosaccharides and small oligosaccharides, in differentmatrices. Examples where this type of analysis is needed are theevaluation of stress responses for plant breeding [1], analysis of com-plex mixtures from different biorefinery scenarios for production ofbiofuels and biomaterials [2], and food quality monitoring [3].

Analysis of mono- and disaccharide constituents in complexmixtures is commonly performed by MS-hyphenated gas chromato-graphy (GC/MS) following an appropriate pre-capillary derivatization[4]. Gas chromatography today is an affordable and widespreadtechnique and is superior to capillary electrophoresis and high perfor-mance liquid chromatography due to its relatively high resolution andsensitivity [5]. Its main limitation arises from the similar molecularweights of the carbohydrate analytes, and determination beyond trisa-ccharides is usually not feasible. All carbohydrates, before analysis byGC/MS, require a suitable derivatization to convert them into volatilederivatives since they naturally exhibit high polarity, pronouncedhydrophilicity with a strong tendency to hydrogen bonding, and near-zero volatility [6]. The derivatization strategies aim at enhancing

ll rights reserved.

: +43 1 47654 6059.cker).

signal intensity and compound stability, increasing the informationcontent of the mass spectra, and improving quantification [7].

A wide range of derivatization strategies is available for GC/MScarbohydrate analysis, and the type of pre-capillary derivatization isthe main factor that distinguishes the different approaches. Silylationand trifluoroacetylation reactions are single-step derivatization meth-ods that are widely employed in analysis of polyalcohols and non-reducing sugars [8,9]. Nearly all functional groups are present in therelevant analytes and most of them are problematic in gas chromato-graphic analysis in one way or another due to their polarity andhydrogen bonding capacity. These groups, which include hydroxyl,amine, amide, phosphate and thiol groups can be converted into theirtrialkylsilyl derivates by displacement of the active proton [10,11]. Thereagent N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA), introducedby Stalling et al. [12], has the ability to react with all common proticsites and has become a quasi-standard derivatization reagent. Thehigh volatility of the derivatization by-products, namely trimethylsilyltrifluoroacetamide and trifluoroacetamide, is an additional advantageover alternative silylation reagents [11]. Trimethylsilyl chloride (TMS-Cl) has been added to BSTFA as a catalyst to increase the silyl donorstrength [6]. When used on its own as reagent, TMS-Cl is oftencombined with pyridine, which acts as a basic auxiliary and HCl trap,as it does with other trialkylsilyl halides [13]. The advantages ofpyridine are its catalytic capability [14] and the increased stability ofthe silylated products in its presence [15]. A mixture of 4-(dimethyl-amino)pyridine (DMAP) and pyridine, when used at ambient tem-peratures, has been shown to minimize side reactions during

M. Becker et al. / Talanta 115 (2013) 642–651 643

silylation and acylation, and thus reduce the complexity of thechromatograms [16].

The analysis of mixtures of reducing monosaccharides in solutionis generally hindered by their structural similarity and the existence ofup to five isomers (α- and β-pyranosides, α- and β-furanosides, andthe open-chain form) per monosaccharide [17,18]. The eight hexoses,for instance, may form up to 40 of these configurational isomers, eachof which might appear as a separate peak in the chromatograms.After silylation, these isomer peaks increase the risk of peak over-lapping, especially in complex mixtures, although a well-separatedpeak from one of the isomers can be used for quantification andmight also improve the possibility of compound identification [19,20].

Oximation of carbonyl functionalities eliminates the occurrenceof furanosidic and pyranosidic isomers, and thus significantlyreduces chromatogram complexity. The sensitivity of the sugaranalysis is boosted because of the increasing signal intensity withdecreasing amounts of isomers and peaks [21,22]. At the sametime, the formed oximes also result in only two peaks correspond-ing to the syn (E) and anti (Z) forms of the oxime. Non-reducingcarbohydrates (e.g., sucrose, trehalose, raffinose) show only onepeak, since they cannot form the corresponding oxime due to themissing aldehyde (hemiacetal) function [23].

Combining silylation (for rendering the compound volatile) andoximation (for reducing the numbers of isomers, and thus, peaks)consequently requires a two-step derivatization strategy: (1) con-version into oximes with hydroxylamine or alkoxyamines, prior to(2) trialkylsilylation with an appropriate silylating agent. Oximationis always carried out as the first derivatization step. An additionaladvantage of an oximation step is the protection of α-keto acidsagainst decarboxylation [24]. In previous work, reducing carbohy-drates were analyzed as their oxime [25], O-methyloxime [18],O-ethyloxime [8], and O-benzyloxime trimethylsilyl derivatives [18].

The distinction of monosaccharides by MS-hyphenated chroma-tographic techniques, based on retention times and mass spectra, israther difficult and attempts have revealed the general difficulties inthis type of carbohydrate analysis [26]. The derivatized compoundsof similar molecular weight show only small differences in retentiontime, which results in partial peak overlapping or even complete co-elution. The mass spectra of these structural isomers are quite similarand differences are mainly related to fragment intensity. This similarfragmentation pattern of carbohydrate isomers largely hampers theuse of databases for identification purposes.

In a previous study of carbohydrate derivatization with oximationreagents followed by trimethylsilylation, we applied ethoximationfollowed by trimethylsilylation derivatization to complex matrices ofbiological (grapevine leaf) and synthetic (formose mixture) origins.This method showed advantages compared to other derivatizationapproaches due to low limits of detection and quantitation, minorrelative standard deviations, and low sensitivity toward matrixeffects [26]. Peak assignment was supported by NMR and providedinformation on the peak ratios: 2-deoxyhexoses and ketohexosesformed almost equal ratios of the syn and anti peak, while aldopen-toses and aldohexoses were solely reported as the syn form [27,28].However, complex matrices may easily contain more sugars than areavailable as reference standards and syn/anti peak ratios might beinfluenced by matrix effects.

All these difficulties point to the great need for a standardizedand robust analytical method that overcomes at least some of thecurrent limitations in qualitative and quantitative mono- anddisaccharide analysis, especially regarding the problems of similarmass spectra and close retention times. Obligatory methodologicalrequirements are high derivatization efficiency, reproducibility,and stability of products, low number of by-products, and aconstant ratio of syn/anti forms of the oximes. Other important,but less pressing, demands are easy handling, acceptable analysistimes, and low overall costs. Today's requirements also include

compatibility with automated derivatization robots and the pos-sibility for simultaneously analysis of a wide range of other organic(non-carbohydrate) compounds (e.g., in metabolome analysis).

In this study, we communicate a two-step derivatization methodthat begins with an initial formation of O-ethyloximes, followedby trimethylsilylation, for optimized GC/MS analysis of mono- anddisaccharides in complex matrices. Derivatization conditions andderivatization efficiency were optimized. The chromatographic andmass spectra characteristics for 46 carbohydrates ranging fromcarbohydrate-related C2- and C3-bodies to monosaccharides(tetroses to heptoses) and up to disaccharides and one trisaccharide.Of particular interest were the separation and identification char-acteristics of derivatized carbohydrates with similar structures andidentical masses. An oxime peak identifier (OPI) to improve carbo-hydrate identification is presented, which was elaborated based onthe combination of an internal standard and the retention of thecorresponding syn/anti peaks of the reducing carbohydrates.

2. Material and methods

2.1. Chemicals and reagents

The reference compounds (Supplemental Table S1) included 1,3-dihydroxyacetone dimer, glycolaldehyde dimer, methylglyoxal solu-tion (ca. 40% in H2O), D-(+)-glyceraldehyde, D-(−)-threose, L-(+)-ery-thrulose, D-(−)-erythrose, D-(+)-xylose, D-(−)-lyxose, D-ribulose,D-psicose, D-(−)-tagatose, D-allose, D-(+)-mannose, L-(+)-gulose,D-(+)-galactose, D-(+)-talose, D-altrose, D-(+)-glucose, D-apiose solution,D-(−)-arabinose, D-(−)-fructose, D-(−)-ribose, L-sorbose, 2-deoxy-D-ribose, 2-deoxy-D-glucose, D-(+)-fucose, L-(+)-rhamnose, D-(+)-digitox-ose, D-glucoheptose, D-(+)-trehalose (glucose-α,α′-(1-1)-glucose),sucrose (glucose-α-(1-2)-fructose), D-(+)-turanose (glucose-α(1-3)-fructose), D-(+)-maltose monohydrate (glucose-α-(1-4)-glucose), D-(+)-cellobiose (glucose-β-(1-4)-glucose), D-lactose monohydrate(galactose-β-(1-4)-glucose), lactulose (galactose-β-(1-4)-fructose),xylobiose (xylose-β(1-4)-xylose), maltulose monohydrate (glucose-α(1-4)-fructose), leucrose (fructose-(1-5)-glucose), β-gentiobiose(glucose-β-(1-6)-glucose), D-(+)-melibiose (galactose-(1-6)-glu-cose), palatinose hydrate (glucose-(1-6)-fructose), D-(+)-raffinose,the internal standard methyl α-D-galactopyranoside, C7–C40 saturatedalkane mixture, anhydrous pyridine, ethyl acetate, N,O-bis(trimethyl-silyl)trifluoroacetamide (BSTFA), O-ethylhydroxylamine hydrochloride,4-(dimethylamino)pyridine (DMAP), dimethyl formamide (DMF),dimethyl sulfoxide (DMSO) and trimethylsilyl chloride (TMS-Cl); allwere purchased from Sigma-Aldrich-Fluka (Sigma-Aldrich, Schnell-dorf, Germany). All standards, chemicals, and reagents were of GCgrade and used without further purification. L-glycero-D-Mannohep-tose and D-sedoheptulose were kindly provided by P. Kosma, Depart-ment of Chemistry, BOKU, Vienna, Austria.

2.2. Carbohydrate samples

Different carbohydrate-containing samples of biological originwere analyzed with the EtOx-TMS method, focusing on fructose,glucose, and sucrose content. Orange honey from Spain (Allos GmbH,Drebber, Germany), buckwheat honey (Rainbauer, St. Magdalena,Austria), honeydew honey from EU and non-EU countries (Honig-mayr, Tenneck, Austria), agave sirup from Mexico (Allos GmbH,Drebber, Germany), maple sirup (grade C) from Canada (DennreeGmbH, Töpen, Germany), and palm sugar (Gula Java Brut/Amanprana,coconut blossom sugar) from Indonesia (Noble House, Brasschaat,Belgium) were purchased from local supermarkets.

M. Becker et al. / Talanta 115 (2013) 642–651644

2.3. Optimized method

2.3.1. Derivatization: O-ethyloximation (EtOx)/trimethylsilylation(TMS)

Different volumes (0.5 ml–1000 ml) of carbohydrate model solu-tions (0.4 mg/ml in MeOH/H2O, 80:20, v/v) and the honey samplewere lyophilized. The lyophilized model mixtures (containing 0.2 mg–400 mg per carbohydrate) and the honey sample (2.33 mg) weredissolved in 200 ml of anhydrous pyridine containing 40 mg/ml O-ethylhydroxylamine hydrochloride and 1 mg/ml methyl α-D-galacto-pyranoside (internal standard, IS) and were heated at 70 1C for 1 h.Subsequently, 200 ml of a solution of 1.5 mg/ml DMAP in pyridinewas added to the mixture, followed by 200 ml of BSTFA (containing10% TMCS). The mixture was heated at 70 1C for 2 h. The derivatizedsamples were kept at −20 1C until analysis.

2.3.2. GC/MS analysisGC/MS analysis was performed on an Agilent 7890A gas

chromatograph coupled with an Agilent 5975C mass selectivedetector. Column: HP-5 MS (30 m�0.25 mm�0.25 mm; J&WScientific, Folsom, CA, USA); carrier gas: helium, injector: 280 1C;column flow: 0.9 ml/min; purge flow: 32.4 ml/min, 0.6 min; ovenprogram: 50 1C (2 min), 5 1C/min, 280 1C (20 min); MS: EI mode,70 eV, source pressure: 1.13�10–7 Pa, source temperature: 230 1C.Scan range was set from 50 to 950 Da. The derivatized sampleswere diluted with ethyl acetate (600 ml) and filtered prior toinjection. Aliquots of 0.2 ml were introduced into the splitlessinjector with an Agilent GC Sampler 120.

2.4. Peak identification and quantification

Peak assignment and quantification was accomplished withMSD Chemstation E.2.01.1177 (Agilent Technologies, USA). Peakswere assigned by comparing their retention times and massspectra with those of respective reference compounds. Calibrationcurves were based on peak areas obtained for up to ten concen-tration levels.

2.5. Method validation

The optimized method was validated for all reference com-pounds with respect to linearity, limit of detection (LOD), limitof quantification (LOQ), and precision. Regression coefficients weredetermined as acceptance criterion for linearity. The linearityof the calibration curve, based on peak area, was evaluatedwith respect to concentration, corrected by an internal standard.The LOD and LOQ were calculated according to the 3s- and10s-criteria, i.e., the three- and tenfold standard deviation of thenoise quantified by single point calibration (8.33 mg/ml), accordingto DIN 32465:2008–11 [29]. The stability of the GC/MS perfor-mance (retention time, peak response and peak geometry) wasconfirmed by repetitive analysis of identical standards.

2.6. Statistical analysis

Multivariate data analysis was performed with Origin 8.6 soft-ware (OriginLab, Northampton, MA, USA). Principal componentanalysis (PCA), an unsupervised clustering method, was used tocompare GC/MS originated mass spectra of different carbohydratesto uncover data structures that account for a large percentage ofthe total variance and to create new hypothetical constructs thatmay be employed to predict or classify observations into groups.Each mass spectrum was collected as a set of raw intensities andnormalized during calculation. PCA was applied using the elevenmass fragments with the largest fragment intensity. The calculated

eigenvalues greater than one were used as criteria to determinethe number of components.

3. Results and discussion

3.1. Method optimization

The chromatographic and mass spectrometric characteristics ofthe ethoxime-TMS derivatives of 46 carbohydrates and carbohydrate-related compounds were analyzed. The substances were selected in away that comprised all compounds relevant in biorefinery analytics,plant hydrolysates, and plant metabolom analysis.

The derivatization procedure and GC/MS conditions were opti-mized using a solution that contained mono-, di- and trisaccharides inwater/methanol solution (80:20, v/v) at a concentration of 100 mg/mleach. The effects of different solvents (pyridine, N,N-dimethylform-amide, dimethyl sulfoxide) on the efficiency of ethoximation andsilylation were studied. The success of the ethoximation reaction wasinvestigated with regards to the concentration of O-ethylhydro-xylamine hydrochloride (20–60 mg/ml solvent) and the time of thederivatization (30–120 min). Further parameters studied for optimi-zation included the volume ratio between solvent and silylationreagent (1:1–2:3, v/v), silylation temperature (room temperature,50 1C, 70 1C), and silylation time (30–240 min). The effect of catalystson the derivatization efficiency was examined at various concentra-tions of DMAP (0–1.5 mg/ml) and TMS-Cl (0–10%). Derivatizedsamples were either diluted with ethyl acetate or dried with nitrogengas and re-dissolved in ethyl acetate prior to GC/MS analysis.Optimization of the GC/MS program mainly involved the heatingrate (4–12 1C min−1) and the column flow (0.6–1.3 ml min−1).

We verified that an oximation time period of 1 h at 70 1C wassufficient to complete the conversion of saccharidic carbonyl groupsinto oximes, as no non-ethoximated TMS-derivatives were observed.The optimization of the silylation procedure showed no significantpeak area increase between 1–2 h of derivatization, so that 1 happeared to be a sufficiently long reaction time. However, to ensurecomplete derivatization of compounds in complex matrices, silyla-tion time was set to 2 h at 70 1C. The high volatility of derivatizationby-products and the possibility for injection of the reaction mixturedirectly into GC without harming the GC columns led to a preferencefor use of BSTFA over other silylation agents [11].

The overall two-step derivatization strategy is straightforwardand facile. The combination of oximation and trimethylsilylation canbe carried out in the same vial and allows injection of the reactionmixture into GC/MS without any additional sample manipulation.

Different techniques were applied to enhance method sensitivity,such as stripping the solvent in a stream of nitrogen and removingexcess BSTFA by rotavaporation. However, these attempts to reducethe volume of the reaction mixture led to precipitation of sugarphosphates, so they were not subsequently performed so that therange of application was not limited. Pyridine, DMF, and DMSO weretested as solvents to optimize the efficiency of ethoximation andsilylation. Eventually, pyridine was chosen as the most promisingsolvent for the combination of both ethoximation and silylation. Theadvantage of pyridine is its catalytic effect, in addition to its action asa solvent for derivatized compounds. The amount of the derivatiza-tion mixture used must be sufficient to cover the complete sampleduring the ethoximation process; see experimental section. Thevolumes applied should be as small as possible to maintain goodmethod sensitivity, but at the same time enough solvent must bepresent to avoid precipitation of excess O-ethylhydroxylaminehydrochloride.

The derivatization catalysts TMCS and DMAP were used tofurther increase the silylation efficiency. The combination of DMAPand TMCS with BSTFA was used for the first time in the present

M. Becker et al. / Talanta 115 (2013) 642–651 645

study. Both reagents, on their own, bring about an increase insilylation efficiency, but their combination had a distinct synergis-tic effect. Since the two catalysts induced formation of by-productsduring the oximation reaction, they were added afterward, in thesilylation step, together with BSTFA as the actual reagent. Theaddition of the two derivatization catalysts to the silylationmixture had no negative influence on the previously formedoximation products.

Optimum chromatographic peak resolution was obtained witha heating rate of 5 1C min−1 and a gas flow of 0.9 ml min−1. Even incases of structurally closely related carbohydrates (e.g., membersof the aldohexose series), a qualitative differentiation was easilypossible. No complete overlapping of the syn and anti peak of acarbohydrate with the peaks of other carbohydrates was observedunder the conditions investigated.

3.2. Method validation

In most cases, biological samples (and their hydrolysates) con-tain larger amounts of glucose, fructose, and sucrose, while othercarbohydrates are present in smaller amounts. Therefore, a broadconcentration range was covered for calibration curves, in order toallow quantification of carbohydrates with largely different concen-trations in one run.

Aliquots (0.2 ml) of the derivatization mixture (1.2 ml), containingcarbohydrate concentrations from 0.16 mg/ml mg to 333 mg/ml, wereinjected in splitless mode. Linearity of the calibration plots wassatisfactory for most metabolites, with regression coefficients betterthan 0.99. The high linearity of peak area-based calibration curvesconfirmed that quantitative analysis was possible by means of thesingle ethoxime-TMS isomer peaks (either syn or anti). Quantificationlimits were below 1 pg and up to 70 pg on-column; only methyl-glyoxal, glucoheptose, and raffinose show higher values (see Supple-mental Table S1). No MS detector saturation or non-linearity of thecalibration curves was observed, not even at the highest concentra-tions tested. Calibration equations and response factors are given inSupplemental Table S1 and examples are shown in Supplemental Fig.S2. As the syn/anti ratio increased, so did the sensitivity for therespective carbohydrate, so that carbohydrates with only one elutingpeak, as well as those with very prominent syn peaks, showed LODswell below 1 pg on-column (Supplemental Table S1).

Fig. 1. Gas chromatogram with MS detection of a mixture of 46 carbohydrates and caderivatives according to the optimized analytical procedure (carbohydrate concentrationhigher resolution of EtOx-TMS derivatized pentoses, hexoses and disaccharides is givSupplemental Table S1 (supplementary part).

3.3. GC-separation characteristics

3.3.1. Peak separationThe challenge of carbohydrate analysis is the good baseline

separation of isomers of high structural similarity. The chromato-graphic behavior of 46 carbohydrates was examined in the presentstudy. The carbohydrates were grouped according to their numberof carbon atoms (e.g., tetroses, pentoses, hexoses, etc.), the avail-ability and position of a carbonyl group (e.g., non-reducing sugars,ketoses, and aldoses) and stereoisomers within one class (e.g.,aldohexoses: glucose, mannose, gulose, etc.), see Fig. 1 and Supple-mental Fig. S1a and b. A temperature increase of 5 1C/min in the GCoven program resulted in a good chromatographic resolution of allcarbohydrates investigated. High molecular weight derivatives,such as the trisaccharide raffinose, show high elution times(62.11 min) and resulted in a relatively low detector response. Forless complex mixtures, a higher temperature increase of the ovenprogram is feasible to increase the sample throughput. Correspond-ing to theory, only one peak can be observed for the non-reducingcarbohydrates, such as methyl-galactopyranoside, sucrose, treha-lose, and raffinose, as no oximation occurs and thus no syn/antioximes are formed. In addition, the reducing carbohydrate apiose,the aldopentoses arabinose and ribose, the ketohexoses sorbose andfructose, and the disaccharide lactulose only generated one peak,due to insufficient separation of the syn and anti oxime peaks. Eventhough both isomers occurred, the deconvolution of syn and antipeak was not possible due to the equal fragmentation patterns.Dihydroxyacetone does not produce syn and anti forms uponoximation, and hence only one peak was observed.

Apart from these exceptions, all other mono- and disaccharidesshowed the typical occurrence of two peaks (syn/anti) ofethoxime-TMS derivatives (Fig. 1; Supplemental Fig. S1a and b).Several co-eluting anti peaks were observed, such as those ofgalactose and glucose, as well as mannose and gulose, while theirsyn peaks could neatly be separated (Supplemental Fig. S1a). Theoptimized conditions provided at least one well-separated peakfor each carbohydrate. However, with increasing number andconcentration of carbohydrates in the mixtures, especially in thecase of hexoses and disaccharides, overlapping of peak tails wasevident (e.g., in the mixture of all 46 saccharides). Even though thereplacement of hydroxylamine [30,31] by O-ethyl hydroxylaminedid not afford a significant improvement of chromatographic

rbohydrate-derived compounds converted to ethoxime-trimethylsilyl (EtOx-TMS)42.5 mg/ml; internal standard concentration 166.6 mg/ml; injected volume 0.2 ml). Aen in Supplemental Fig. S1a and b. Description of peak assignment is given in

Fig. 3. Chromatogram of an EtOx-TMS derivatized honey sample (Orange/Spain).For peak description, see Table 1.

M. Becker et al. / Talanta 115 (2013) 642–651646

resolution, the ethyl substituent is considered to increase thenucleophilicity of the derivatisation reagent, impede follow-upreactions that could occur in complex matrices, and simplify massspectrometric fragmentation. For highly complex sugar mixtures,we recommend applying a calibration based on peak heightinstead of peak area, to avoid integration errors caused by peaktailing and peak overlapping.

3.3.2. Structural conformation and chromatographic retentionObvious retention time differences were evident between the

different groups of carbohydrates, ranging from tetroses to dis-accharides (Figs. 1 and 2). However, the individual isomers withinone carbohydrate family, with their equal molecular weights andsimilar structures, also showed small differences in their retentiontimes. The relationship between molecular weight and retentiontime (Supplemental Fig. S3) of the 46 carbohydrate analytesshowed a very good correlation (R2¼0.989).

Besides the molecular weight, the conformation of the oximederivatives is an important factor influencing the chromatographicretention of the isomers [32]. Snyder [28] concluded from NMRspectroscopy that the acyclic oximes of aldopentoses adopt differ-ent geometries: the derivatives of arabinose and lyxose have aplanar “zigzag” arrangement, whereas those of ribose and xyloseshow a “bent” or “sickle” conformation. We can assume that thesetypes of differences are also responsible for differences in theretention time within the groups of pentoses, hexoses, or heptoses.Close retention times of stereoisomers (Fig. 1; SupplementalFig. S1a and b) complicate the identification process, and yet theybecame the crucial factor and often the only means of distinguish-ing different carbohydrates [23].

Different from the Kovats retention index for isothermal elu-tion conditions that is based on non-linear increments added tothe retention times of consecutive peaks [33], a constant incre-ment is added to the retention times of respective precedent peaksfor the linear retention index (LRI) or programmed-temperatureretention index (PTRI) which is commonly used in temperature-

Fig. 2. Course of retention time (left y-axis) and corresponding molecular weight (right yderivatives.

programmed gas chromatography. The calculation of LRI can beaccomplished according to the non-logarithmic equation proposedby Van Den Dool and Kratz [34,35].

The retention data, obtained for compounds found in an EtOx-TMSderivatized Spanish orange honey sample (Fig. 3) and the correspond-ing compounds in the carbohydrate standard solution, were used tocalculate the values of relative retention (rG; Eq. (1)) and the linearretention index (LRI). Although complex biological matrices are knownto cause retention time drifts, which can negatively affect the rGprecision [36], the rG values in this study showed the most stablevalues between the standard solution and the honey sample (Table 1).The LRI showed stable values with low variability. The advantage ofboth indices is that the calculation is based on one peak of thecompound and can be applied to non-reducing carbohydrates orwhen the corresponding oxime peak coelutes.

-axis) of 46 carbohydrate analytes converted to ethoxime-trimethylsilyl (EtOx-TMS)

Table 1Overview of calculated retention values between a carbohydrate standard compound solution and a honey sample (Orange/Spain).

No Compound Standard compound solution Orange honey

R.T. [min] rG LRI R.T. [min] rG LRI

1 Methyl-galactopyranoside 30.770 1.000 1858.67 30.781 1.000 1859.212 Fructose 32.706 1.063 1956.63 32.734 1.063 1958.083 Glucose (syn) 33.306 1.082 1987.61 33.348 1.083 1989.784 Glucose (anti) 33.762 1.097 2011.67 33.799 1.098 2013.675 Sucrose 45.197 1.469 2711.14 45.210 1.469 2712.076 Cellobiose (syn) 46.759 1.520 2822.78 46.758 1.519 2822.707 Maltulose (anti) 47.076 1.530 2846.07 47.074 1.529 2845.928 Maltulose (syn) 47.107 1.531 2848.35 47.114 1.531 2848.869 Maltose (syn) 47.263 1.536 2859.81 47.268 1.536 2860.1810 Turanose (anti) 47.484 1.543 2876.05 47.498 1.543 2877.0811 Turanose (syn) 47.546 1.545 2880.60 47.548 1.545 2880.7512 Maltose (anti) 47.757 1.552 2896.11 47.770 1.552 2897.0613 Palatinose (anti) 48.561 1.578 2953.26 48.559 1.578 2953.1214 Palatinose (syn) 48.726 1.584 2964.96 48.718 1.583 2964.40

R.T.¼Retention time.rG¼Relative retention¼R.T. compound/R.T. internal standard (α-methyl-galactopyranoside).LRI¼Linear retention index according to Van Den Dool and Kratz [34].

M. Becker et al. / Talanta 115 (2013) 642–651 647

3.3.3. Retention time shift of the syn/anti oxime peak pairsThe elution order of the syn and anti oxime peaks is directly

associated with the structural properties of the carbohydrate, suchas the position of carbonyl group (keto/aldo), molecular weight,and type of carbohydrate subunits in disaccharides. The presentstudy showed an elution order of the syn peak appearing beforethe corresponding anti peak for aldohexoses and aldoheptoses.The aldotrioses, aldotetroses, and aldopentoses showed the oppo-site chromatographic behavior, with the syn peak following theanti peak (Fig. 4).

The greatest time shift between syn and anti peaks was observedfor aldoheptoses. Fructose-containing disaccharides (maltulose,turanose, leucrose, and palatinose) showed a relatively small timeshift and the anti peak eluted before the syn peak, similar to the twooxime peaks of ketohexoses. All other investigated disaccharidesshowed a syn peak that eluted before the anti peak, as in thealdohexose group (Fig. 4).

The data in Fig. 4 show a coherent effect of the position of theglycosidic linkage on the elution order. We observed a syn peakelution order of 1,4-linked disaccharides (xylobiose, lactose, cello-biose, maltulose, and maltose), before 1,3- (turanose), 1,5- (leucrose)and 1,6-linked disaccharides (gentiobiose, palatinose, and melibiose).Garcia-Raso et al. [37] proposed that the glycosidic bond affects theretention time via its effect on the overall molecular shape of thedisaccharide, suggesting that the higher retention of the 1,6-di-sacharides is due to their greater conformational flexibility. Their dataon monosaccharides also indicated that more planar forms translateinto higher retention times [37]. These results fully agreed with ourfindings on the retention characteristics of the ethoxime derivatives.

Close retention times of the syn peaks and a similar syn/antipeak time shift were observed for the following carbohydrates:allose/gulose, galactose/talose, and turanose/leucrose, which com-plicated the identification of these sugars. In any case, the elutionorder of syn and anti peaks and their time shifts were character-istic parameters of the individual carbohydrates that were used fortheir unambiguous identification.

3.3.4. Syn/anti peak ratiosThe different carbohydrate groups showed characteristic syn/anti

peak ratios after ethoximation (Fig. 5). The silylation proceduremaintained the ratio of the two oxime peaks [27]. The mass spectraof syn and anti forms of the oximes were generally quite similar andreliable structural information could not be easily derived. The peaks

were identified in previous studies based on 1H NMR data in pyridine-d5 [27] and D2O [28]. The major oxime peak formed by aldosescorresponded to the syn form, the smaller peak to the anti peak. Dueto both the lack in reference data for peak assignment of the syn- andanti-isomers of oximated ketoses and 2-deoxysugars and the fact thatthe oxime isomer pairs were found to have similar peak areas, thepeak caused by the first eluting isomer is now consistently termed as“peak 1” and that of the later eluting respective isomer as “peak 2”. Wecalculated the syn/anti peak ratios of ethoxime-trimethylsilyl (EtOx-TMS) derivatized saccharides (cf. Supplemental Table S1) at a con-centration level of 83.3 mg/ml for each carbohydrate (Fig. 5). In general,aldoses (C3–C6) showed a high syn/anti peak ratio from 3.93 (galac-tose) to 9.34 (mannose). Ketoses (C4–C6) had low oxime peak ratios,ranging from 0.54 (tagatose) to 1.38 (ribulose), if the peak area of thefirst eluting isomer is divided by the peak area of the second elutingisomer. Using the same calculation procedure, small oxime peak ratioswere also observed for 2-deoxysugars, whereas the 6-deoxysugars(fucose and rhamnose) showed values similar to aldoses. Despiteserious attempts, separation of the syn/anti peak pair of the twoethoxime-TMS isomers of fructose was unfortunately not possible, sothe peak ratio for that monosaccharide is unity. Disaccharides contain-ing galactose, glucose, or xylose (lactose 4.76, xylobiose 4.93, cello-biose 5.75, gentiobiose 7.35, maltose 7.39, and melibiose 7.69) yieldedhigh syn/anti peak ratios compared to fructose-containing disacchar-ides, which gave values in the range of ketoses (palatinose 1.32,turanose 1.38, and leucrose 1.71).

The observed syn/anti peak ratios of monosaccharides were inagreement with the values of the smaller range of O-methyloxime-TMS derivatized carbohydrates studied by Snyder [28] and Funckeand Sonntag [27]. We observed similar syn/anti peak ratios formonosaccharides and disaccharides that contain the same mono-saccharide as the reducing sugar. Hence, we can assume that thereducing end of disaccharides plays an important role in deter-mining the syn/anti peak ratio of the EtOx-TMS-disaccharides, andthereby, their chromatographic behavior.

3.4. Characteristics of mass spectra

The mass spectra of 83 ethoxime-TMS and TMS (sucrose, treha-lose, and raffinose) derivatives were investigated with regard to theirfragmentation patterns and structure-related mass ions. The spectrashowed a nearly equal distribution of mass fragments between thecorresponding syn and anti peaks as well as similarities betweendifferent groups of carbohydrates. The only difference between two

Fig. 4. Retention time of syn peaks (■) and retention time shift (○) between the syn peak and corresponding anti peak or peak 1 and peak 2 of different ethoxime-trimethylsilyl (EtOx-TMS) derivatized carbohydrates. A positive retention time shift corresponds to the appearance of the anti peak after the corresponding syn peak. The firsteluting isomer of oximated ketoses and 2-deoxysugars has been termed as “peak 1” and that of the later eluting respective isomer as “peak 2”.

Fig. 5. Isomer ratios (based on peak area) between the syn peaks and corresponding anti peaks or peak 1 and peak 2 of different ethoxime-trimethylsilyl (EtOx-TMS)derivatized carbohydrates. The first eluting isomer of oximated ketoses and 2-deoxysugars has been termed as “peak 1” and that of the later eluting respective isomer as“peak 2”.

M. Becker et al. / Talanta 115 (2013) 642–651648

corresponding oxime peaks was recognized as a lower overallintensity of fragments of the anti isomer compared to syn isomerspectra, especially evident for aldohexoses and aldopentoses.MacLeod et al. [38] also observed these differences in methoxime-TMS derivatized keto sugars, where both isomers showed very lowintensities of the molecular ion [M] or the [M - CH3] ion.

The ethoxime-carrying moiety in the mass spectra of the EtOx-TMS derivatized aldohexose D-glucose, selected as example (Sup-plemental Fig. S4a), appeared as the m/z 174 ion (C2H5ON=C-CH-OTMS). This fragment contained the former reducing end ofglucose. The spectrum of non-oximated TMS-glucose is differentfrom that of EtOx-TMS-glucose; although it contains some

Fig. 6. PCA score plot of the eleven fragment ions (m/z 73, 103, 129, 147, 174, 204,205, 217, 307, 319 and 361) with the largest intensity from the mass spectra of 46EtOx-TMS derivatized carbohydrates. Each number in the score plot represents onespecific ethoxime-TMS derivative as listed in Supplemental Table S1.

M. Becker et al. / Talanta 115 (2013) 642–651 649

common fragments to both derivatives (m/z 73, 103, 147, and 217)it also shows typical fragments as m/z 191, 204 (205 as isotopiccontribution) and 435, whereas m/z 319 is absent or very low. Themass spectrum of EtOx-TMS D-fructose (Supplemental Fig. S4b), a2-ketohexose, did not yield the m/z 174 ion; instead, two ions m/z378 and m/z 277 occurred, which represented a C4- and C3-partwith ethoxime moiety, respectively. Once more, all other fragmentions were independent of the oxime functionality and are similarto the non-ethoximated TMS-derivatives. Laine and Sweeley [39]concluded that the increased stability of bonds directly adjacent tothe oxime carbon of ketohexoses caused this difference whencompared to aldohexoses. They support their proposal further bythe missing m/z 319 ion in spectra of the derivatized ketohexoses.Ethoximation gave rise to key fragment peaks in the spectra,which differed between different carbohydrate types (cf. the twoexamples in Supplemental Fig. S4a and b), and thus introduce agood tool for compound distinction.

Although the mass spectra of EtOx-TMS derivatized stereoiso-meric carbohydrates are congruent with regard to the fragment ionpattern, distinguishing between carbohydrate groups may be possi-ble based on fragment intensities (e.g., different intensities of massions or characteristic mass ion ratios). Generally, aldoses and ketosesof the respective carbohydrate groups showed opposite intensitybehavior of certain mass ions. Aldotetroses showed highm/z 117, 174,204 and low m/z 103, while ketotetroses fragmented into high m/z103 and low m/z 117, 174, 204. Aldopentoses and ketohexoses werecharacterized by high intensities of m/z 103, 217, 307 and lowintensity of m/z 204, and the opposite behavior was observed forketopentoses (highm/z 204; lowm/z 103) and aldohexoses (highm/z174, 204, 319; low m/z 103, 217). 6-Deoxyhexoses (rhamnose:6-deoxy-L-mannose; fucose: 6-deoxy-L-galactose) showed m/z 117as a major ion. This ion was also observed in the mass spectra of thedideoxyhexose digitoxose, but not in 2-deoxy-D-glucose, a decreaseof m/z 204 and m/z 319 was observed here.

Previous studies have shown that the mass spectra of oxime-TMS derivatized disaccharides contain fragments of a cyclicmonosaccharide-TMS part that includes the glycosidic oxygenand fragments resulting from an open-chain TMS-monosaccharidewith the oxime group [31]. In the present study, we observed aneffect of monomer composition of the disaccharide on the frag-ment intensity patterns. Fructose-containing disaccharides (mal-tulose, palatinose, turanose, leucrose, lactulose, and sucrose)produced a very low m/z 174 compared to disaccharides withoutfructose. No specific mass fragmentation was observed accordingto the position of the glycosidic linkage, just intensity differences.This is consistent with previous studies [31,40], suggesting arelationship between mass fragmentation and the structuralcharacteristics of 1,2- or 1,6-linked disaccharides, estimated fromthe relative intensity of fragment ions, but without giving moredetailed information.

Based on the sample mass fragmentation and ion intensity data inthe current study, we attempted a pattern recognition of carbohydratemass fragments using principal component analysis (PCA), an unsu-pervised exploratory data analysis technique. Where no prior knowl-edge about the data was introduced during statistical analysis, furthersample information was used for interpreting PCA analysis [41]. Theprincipal components (PCs) were used to discover and interpret thedependencies that exist among the variables and to examine relation-ships that may exist among the spectra of individual carbohydrates[42]. Discrimination analysis techniques were not carried out becausethey require an assignment of the spectra into groups before analysis.This introduces user bias into the data analysis through emphasizingthe differences between assigned groups instead of the differencesbetween mass fragment intensities [41].

PCA was performed on the relative intensities of the elevenhighest fragment ions of each mass spectrum of the EtOx-TMS

derivatives of 32 monosaccharides, 13 disaccharides, and onetrisaccharide and afforded one overall score plot for a total of 83saccharides (Fig. 6). The score dataset was clustered into differentgroups associated with different types and sizes of carbohydrates.The first two components combined account for 52.84% of thetotal variance. The 2D PCA score plot was used for objectivecomparison of the carbohydrate mass spectra profiles.

Principal component 1 (PC1) separated the clusters of mono-saccharides, disaccharides, the trisaccharide raffinose, and methyl-galactopyranoside (internal standard), capturing 28.09% of the totalvariance. Principal component 2 (PC2) reflected 24.75% of the totalvariance and separated the clusters of deoxysugars, trioses, aldo-hexoses, and ketopentoses from the cluster of aldopentoses andketohexoses, implying high similarity between the members of eachgroup, which was also described above. The score plot of PC1 andPC2 demonstrates an easy differentiation between three groups ofdisaccharides, containing either (1) xylose–xylose; (2) glucose–glu-cose and glucose–galactose or (3) glucose–fructose and galactose–fructose. The PCA method revealed that the relative abundances ofeleven fragment ions only (m/z 73, 103, 129, 147, 174, 204, 205, 217,307, 319 and 361) were sufficient for a reliable classification.

3.5. Oxime peak identifier (OPI)

Based on our findings, the existence of two corresponding oximepeaks and their characteristic time shift between the syn and anti peak(Fig. 4) of the EtOx-TMS derivatized carbohydrates can be used asdirect and reliable parameters to distinguish between carbohydrateswith similar mass spectra. Therefore, we have introduced an oximepeak identifier (OPI). The oxime peak identifier (OPI) represents arelative retention, based on one of the two EtOx-TMS peaks (syn/anti),while the other peak is used as dynamic standard in combinationwitha static internal standard. The OPI value is calculated according toEq. (2). This novel approach offers significantly more positive matchidentifications. The equation of the OPI is similar to the calculation of arelative retention [43]. The principle of the positive identification of aparticular carbohydrate is the existence of a second peak at a definiteposition, while the static internal standard acts as anchor point to setthe unambiguous relation to other carbohydrates.

rG¼ RTtarget=RTIStd ð1Þ

OPI¼ 100 n ðRTtarget−RTcorresponding oxime peakÞ=ðRThigh−RTlowÞ ð2Þ

M. Becker et al. / Talanta 115 (2013) 642–651650

rG: relative retention, OPI: oxime peak identifier, RTtarget:retention time of an oxime isomer peak, RTIStd: retention time ofthe internal standard, RTcorresponding oxime peak: retention time of thecorresponding oxime peak isomer, RThigh: retention time of thecompound with the latest elution, either the internal standard orthe corresponding oxime peak isomer, RTlow: retention time of thecompound with the earliest elution, either the internal standard orthe corresponding oxime peak isomer.

The OPI of the carbohydrate derivatives (Fig. 7, right) shows anincreasing value for the syn peak, while the value of the anti peakdecreases. For carbohydrates eluting before glycolaldehyde (OPI-syno1, OPIanti41), the syn peak elutes before the anti peak. Incontrast, for carbohydrates from maltulose on (OPIsyn41, OPIan-tio1), the anti peak elutes before the syn peak. The conventionalrG-plot (Fig. 7, left) also starts with glycolaldehyde at an rG valueof 0.29576 and the rG value increases with increasingelution order of the carbohydrate. rG values smaller than unity(up to fucose) indicate a retention below the internal standard(methyl α-D-galactopyranoside), while rG values greater than 1(starting from 2-deoxy-D-glucose) come from analytes elutingafter the internal standard.

The graph of conventional rGs (Fig. 7, left) shows several plateausin the group of tetroses, pentoses and hexoses, which indicated“regions” where the distinction between individual carbohydrates isvery difficult, if not impossible, due to missing retention timedifferences between stereoisomers and their similar mass spectra.Compared to the rG, the graph of the novel OPIs (Fig. 7, right) alsoshows plateaus, but most of the carbohydrates within one plateau arecoming from different carbohydrate groups and can be distinguishedfrom the mass spectra and the syn/anti peak characteristics. Comparedto rG, the OPI shows a better differentiation of individual carbohydrateas well as a higher resolution (Supplemental Table S2). In contrast tothe rG, which only considers the elution order, the OPI combines theelution order and the characteristic time shift between syn and anti

Fig. 7. Comparison of the relative retention (rG, based on one static internal standard)corresponding syn or anti peak/peak 1 or peak 2 of the carbohydrate) for ethoxime-trimeketoses and 2-deoxysugars has been termed as “peak 1” and that of the later eluting re

peaks, which ensures positive match identification of all testedcarbohydrates. The OPI value accounts for the positive assignment oftwo peaks, and the calculation of the corresponding peak values canvalidate the positive identification.

3.6. Application to carbohydrate matrices

Four different carbohydrate-containing matrices were lyophi-lized to remove water residues and subsequently analyzed withthe introduced EtOx-TMS derivatization method. The samplematrices were compared for their fructose, glucose, and sucrosecontents on a dry mass basis (Fig. 8).

The honey group, consisting of buckwheat, orange, and honey-dew honeys, showed similar fructose/glucose/sucrose-ratios. Thefructose content within the group ranged from 52.21% to 54.31%,followed by glucose, ranging from 44.99% to 47.19%. Slightly higheramounts of glucose were observed in the honeydew honeycompared to the other two honeys, while orange honey showedthe highest sucrose amount within the honey group (2.1%). Maplesirup and palm sugar were characterized by high sucrose amounts(91.97% and 85.93% respectively) and low fructose and glucosecontents. In contrast, agave sirup showed the highest amount offructose (93.2%) among all analyzed samples.

4. Conclusion

Gas chromatography, hyphenated to mass spectrometry, is anappropriate method for analyzing carbohydrates; however, theexistence of ubiquitous mass ions and very similar mass spectralimit the use of mass spectroscopy for carbohydrate identification.Differentiations between ketoses and aldoses, as well as molecularweight differences, are useful but insufficient for positive com-pound identification by mass spectrometry. A distinction between

and the oxime peak identifier (OPI, based on one static internal standard and thethylsilyl (EtOx-TMS) derivatized carbohydrates. The first eluting isomer of oximatedspective isomer as “peak 2”.

Fig. 8. Analysis of fructose/glucose/sucrose-ratio (%) in dry mass of different sugarmatrices: agave sirup, maple sirup, palm sugar, and honeys (buckwheat, orange,and honeydew).

M. Becker et al. / Talanta 115 (2013) 642–651 651

isomeric carbohydrates is only possible using retention times andchromatographic behavior as discrimination parameters, but asthis is not generally successful, different derivatizations are usuallyperformed to improve their chromatographic behavior.

The combination of oximation followed by silylation for carbohy-drate analysis is a particularly powerful approach for carbohydrateanalysis, its full potential has not yet been exploited. We have aimed atadvancing the ethoxime-trimethylsilyl derivatization method withregard to both practical aspects and data interpretation. This methodis probably the best one amongst all related procedures that werehitherto assayed by the authors. Together with the proposed oximepeak identifier (OPI), this method overcomes several pressing pro-blems in GC/MS carbohydrate analysis and allows for a reliable andstraightforward identification of a large range of analytes.

The chromatographic and mass spectra characteristics of 46different carbohydrates were used to evaluate possible unambig-uous identification criteria. Specific parameters for the individualcarbohydrates were the area ratio of syn and anti oxime peaks, theirelution order, and the retention time shift between the two peaks.Two of these parameters were incorporated into the oxime peakidentifier (OPI), which allows reliable identification of the carbohy-drates. In addition, in contrast to conventional retention indices, theOPI is based on the second peak of the syn/anti pair as dynamicstandard. In the case of disaccharides, the anti peak order and syn/anti peak area were influenced by the monosaccharide moieties,while the syn peak elution was effected by the position of theglycosidic linkage. We are confident that the present EtOx-TMSderivatization with GC/MS analysis and the developed oxime peakidentifier evaluation can help to enhance positive carbohydrateidentification, even in highly complex biological samples. Our hopeis that this approach will become widely accepted in different areasof carbohydrate analysis and quantification.

Acknowledgments

The financial support by the Christian Doppler Research Society(CD lab “Advanced Cellulose Chemistry and Analytics”) and LenzingAG, Austria are gratefully acknowledged. We thank P. Kosma,

Department of Chemistry, BOKU, Vienna, Austria, for donating thetwo sugars L-glycero-D-manno-heptose and D-sedoheptulose.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.talanta.2013.05.052.

References

[1] B. Vinocur, A. Altman, Curr. Opin. Biotechnol. 16 (2005) 123–132.[2] S.N. Naik, V.V. Goud, P.K. Rout, A.K. Dalai, Renewable Sustainable Energy Rev.

14 (2010) 578–597.[3] A. Cifuentes, Electrophoresis 27 (2006) 283–303.[4] C. Birkemeyer, J. Chromatogr. A 993 (2003) 89–102.[5] M.L. Sanz, J. Sanz, I. Martinez-Castro, J. Chromatogr. A 1059 (2004) 143–148.[6] D.R. Knapp, Handbook of Analytical Derivatization Reactions, Wiley, New York,

1979.[7] V. Zaikin, J. Halket, A Handbook of Derivatives for Mass Spectrometry, IM

Publications, Chichester, 2009.[8] M.M. Koek, B. Muilwijk, M.J. van der Werf, T. Hankemeier, Anal. Chem. 78

(2006) 1272–1281.[9] P.M. Medeiros, B.R. Simoneit, J. Chromatogr. A 1141 (2007) 271–278.[10] C.F. Poole, W.F. Sye, S. Singhawangcha, F. Hsu, A. Zlatkis, A. Arfwidsson,

J. Vessman, J. Chromatogr. A 199 (1980) 123–142.[11] R.P. Evershed, Advances in silylation, in: K. Blau, J.M. Halket (Eds.), Handbook

of Derivatives for Chromatography, second ed.,Wiley, New York, 1993,pp. 51–108.

[12] D.L. Stalling, C.W. Gehrke, R.W. Zumwalt, Biochem. Biophys. Res. Commun. 31(1968) 616–622.

[13] M. Makita, W.W. Wells, Anal. Biochem. 5 (1963) 523–530.[14] P.S. Mason, E.D. Smith, J. Gas Chromatogr. 4 (1966) 398.[15] Z. Zhang, A. Hibberd, J. Zhou, Anal. Chim. Acta 577 (2006) 52–61.[16] K.R. Anumula, P.B. Taylor, Anal. Biochem. 203 (1992) 101–108.[17] A. García-Raso, M. Fernández-Díaz, M.I. Páez, J. Sanz, I. Martínez-Castro,

J. Chromatogr. A 471 (1989) 205–216.[18] M.A. Andrews, Carbohydr. Res. 194 (1989) 1–19.[19] W. Amelung, M.V. Cheshire, G. Guggenberger, Soil Biol. Biochem. 28 (1996)

1631–1639.[20] C. Laine, T. Tamminen, A. Vikkula, T. Vuorinen, Holzforschung 56 (2002)

607–614.[21] C. Rumpel, M.F. Dignac, Soil Biol. Biochem. 38 (2006) 1478–1481.[22] O. Fiehn, J. Kopka, R.N. Trethewey, L. Willmitzer, Anal. Chem. 72 (2000)

3573–3580.[23] A.I. Ruiz-Matute, O. Hernandez-Hernandez, S. Rodriguez-Sanchez, M.L. Sanz,

I. Martinez-Castro, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 879 (2011)1226–1240.

[24] K. Blau, J.M. Halket, Handbook of Derivatives for Chromatography, Wiley, NewYork, 1993.

[25] B.S. Mason, H.T. Slover, J. Agric. Food Chem. 6 (1958) 551–554.[26] M. Becker, T. Zweckmair, A. Forneck, T. Rosenau, A. Potthast, F. Liebner,

J. Chromatogr. A 1281 (2013) 115–126.[27] W. Funcke, C. von Sonntag, Carbohydr. Res. 69 (1979) 247–251.[28] J.R. Snyder, Carbohydr. Res. 198 (1990) 1–13.[29] DIN 32645:2008-11 Chemische Analytik; Nachweis-, Erfassungs- und Bestim-

mungsgrenze unter Wiederholbedingungen; Begriffe, Verfahren, Auswertung.[30] Z. Fuzfai, Z.F. Katona, E. Kovacs, I. Molnar-Perl, J. Agric. Food Chem. 52 (2004)

7444–7452.[31] M.L. Sanz, J. Sanz, I. Martinez-Castro, Chromatographia 56 (2002) 617–622.[32] A. García-Raso, I. Martínez-Castro, M.I. Páez, J. Sanz, J. García-Raso, F. Saura-

Calixto, J. Chromatogr. A 398 (1987) 9–20.[33] E. Kováts, Helv. Chim. Acta 41 (1958) 1915–1932.[34] H. Van Den Dool, P.D. Kratz, J. Chromatogr. A 11 (1963) 463–471.[35] B.d.A. Zellner, C. Bicchi, P. Dugo, P. Rubiolo, G. Dugo, L. Mondello, Flavour Fragr.

J. 23 (2008) 297–314.[36] N. Strehmel, J. Hummel, A. Erban, K. Strassburg, J. Kopka, J. Chromatogr. B:

Anal. Technol. Biomed. Life Sci. 871 (2008) 182–190.[37] A. Garcia-Raso, M.I. Paez, I. Martinez-Castro, J. Sanz, M.M. Calvo, J. Chromatogr.

A 607 (1992) 221–225.[38] J.K. MacLeod, I.L. Flanigan, J.F. Williams, J.G. Collins, J. Mass Spectrom. 36

(2001) 500–508.[39] R.A. Laine, C.C. Sweeley, Carbohydr. Res. 27 (1973) 199–213.[40] I. MolnarPerl, K. Horvath, Chromatographia 45 (1997) 321–327.[41] M.S. Wagner, B.J. Tyler, D.G. Castner, Anal. Chem. 74 (2002) 1824–1835.[42] N.H. Timm, Applied Multivariate Analysis, Springer-Verlag, New York, 2002.[43] L. Ettre, Pure Appl. Chem. 65 (1993) 819.

115 

  

  

 Paper V  

 A novel method to analyze the degree of acetylation in 

biopolymers. 

 

 Journal of Chromatography A, vol. 1372, pp. 212‐220. 

 

   

 

 

 

 

Zweckmair, T. 1, Becker, M. 1, Ahn, K., Hettegger, H., Kosma, P., Rosenau, T., Potthast, A. 

(2014) 

1 These authors contributed equally to this work. 

       

A

TAa

Ub

a

ARRAA

KCDGMPSZ

1

gcaioihsblaaccs

h0

Journal of Chromatography A, 1372 (2014) 212–220

Contents lists available at ScienceDirect

Journal of Chromatography A

jo ur nal ho me pag e: www.elsev ier .com/ locate /chroma

novel method to analyze the degree of acetylation in biopolymers

. Zweckmaira,1, M. Beckera,1, K. Ahna, H. Hetteggera, P. Kosmab, T. Rosenaua,. Potthasta,∗

Department of Chemistry, Division of Chemistry of Renewable Resources and Christian-Doppler Laboratory “Advanced Cellulose Chemistry and Analytics”,niversity of Natural Resources and Life Sciences-Vienna, Muthgasse 18, A-1190 Vienna, AustriaDepartment of Chemistry, Division of Organic Chemistry, University of Natural Resources and Life Sciences-Vienna, Muthgasse 18, A-1190 Vienna, Austria

r t i c l e i n f o

rticle history:eceived 16 July 2014eceived in revised form 20 October 2014ccepted 25 October 2014vailable online 31 October 2014

eywords:arbohydratesegree of acetylationC–MS

a b s t r a c t

A novel approach to measure the degree of acetylation in biopolymers applying a combination ofZemplén-deacetylation by sodium methanolate and GC–MS methodology is introduced. The devel-opment focuses on very low limits of detection to cover also samples with extremely low degreesof acetylation which hitherto eluded accurate determination. Free acetic acid or inorganic acetates,often present in biopolymer samples, do not disturb the quantification. Two techniques to measure theZemplén-released methyl acetate were comparatively assessed, direct injection of the liquid phase anda SPME-based approach, the former being more straightforward, but being inferior to the latter in sen-sitivity. By applying isotopically labeled methyl acetate released from 4-O-(13C2-acetyl)-vanillin as theinternal standard, influences, such as varying moisture contents, are corrected, improving the overall

ethyl acetateolysaccharidesPMEemplén transesterification

method reliability to a large extent. The combination of Zemplén-release of acetyl groups in biopolymersas methyl acetate, in connection with its accurate quantification by SPME–GC–MS, was found to be themethod of choice for routine, yet very accurate analysis of a wide range of acetylation degrees of biopoly-mers, showing satisfying analytical parameters along with easy handling and widest applicability. Limitof detection for acetylated cellulose samples is 0.09 nmol/mg, for hemicellulose samples 0.48 nmol/mg.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

Acetylation is a chemical reaction that replaces an acidic hydro-en in a chemical compound by an acetyl (CH3CO ) group,ommonly abbreviated as “Ac” [1–3]. Ester formation with aceticcid is a nearly ubiquitous feature of plant polysaccharides, mostmportantly hemicelluloses and pectins [4,5]. Also lignin, the sec-nd main component of woody matter besides polysaccharides,s acetylated to some low extent [6]. The partial acetylation ofemicelluloses causes some solubility in water and certain organicolvents. While a hydroxyl group can be both a potent hydrogenond donor as well as a strong hydrogen bond acceptor, acety-

ation of the hydroxyl group cancels its H-bond donor capabilitynd switches the H-bond acceptor activity from a hydroxyl- to

carbonyl-oxygen group. Non-acetylated or deacetylated hemi-

elluloses – with a higher content of free hydroxyl groups –onsequently show much lower solubility in organic solventsuch as acetone or chloroform than their natural, acetylated

∗ Corresponding author. Tel.: +43 1 47654 6454; fax: +43 1 47654 6059.E-mail address: antje.potthast@boku.ac.at (A. Potthast).

1 These authors contributed equally to this work.

ttp://dx.doi.org/10.1016/j.chroma.2014.10.082021-9673/© 2014 Elsevier B.V. All rights reserved.

counterparts. Natural cellulose is free of acetylation, contrary tohemicelluloses. It displays a complex network of strong intra- andintermolecular hydrogen bonds that accounts for its insolubilityin water and most organic solvents. When artificially acetylated –cellulose acetates of different degree of substitution are importantindustrial products [7] – the hydrogen bond network is weakenedand solubility is improved. Solubility of cellulose acetates stronglydepends on the degree of substitution. Cellulose 2.5-acetate andcellulose 3-acetate are fully soluble in organic solvents such asacetone or chloroform.

In woody and annual plant hemicelluloses, most of the natu-rally present acetyl groups are lost during pulping. Those groupsare esters of acetic acid with the polysaccharide as co-reactingpolyalcohol, and they show the typical reactivity of esters: cleavagecan be effected both under acidic and alkaline conditions. Whileacidic ester cleavage produces alcohol and free acid and stopsat an equilibrium determined by the law of mass action, alka-line saponification yields alcohol and the salt of the acid and isthus quantitative and irreversible. According to the pulping pro-

cess used, the acetates in hemicelluloses are cleaved in acidic oralkaline medium, and either free acetic acid or the respective salt(inorganic acetate) is liberated. The cleavage of acetates must notbe disregarded as minor and unimportant side reaction: on the one

matog

hitaaitt

pieolap

lltcfnaata

pttnfcatabbtaasaalal

ttattsa

qmnhictmc

T. Zweckmair et al. / J. Chro

and, released acetic acid might have a noticeable effect on the pHn liquid process streams, e.g. in prehydrolysis processing steps. Onhe other hand, the reaction might even be used to retrieve aceticcid as a valuable side product. Pulp and fiber producers in Austriand elsewhere, for instance, have tuned their pulp processing linesn such a way that acetic acid becomes an important byproducthat is obtained in food-grade quality and contributes significantlyo overall revenue.

In pulp, fibers and paper, partial acetylation causes changedroperties, even if the degree of substitution is rather low. This

s especially true for acetylation of surface hydroxyl groups whichffects changes in reactivity, hydrophobicity, water interaction andther surface properties. A low degree of acetylation in cellu-osics may arise from acetyl groups that survived the alkaline orcidic conditions during pulping, but also re-acetylation in laterrocessing stages might occur.

The acetyl linkage, a typical ester bond, is chemically ratherabile if the pH deviates from neutral for more than 2–3 units, alka-ine conditions usually being more detrimental to ester integrityhan acidic ones. As a consequence, all polysaccharidic materialsarrying acetyl groups do generate acetic acid, mostly in volatileorm, or contain traces of acetic acid to some extent, if water isot rigorously excluded. The adsorptively bound acetic acid orcetate is generally hard to distinguish from covalently boundcetate, especially since there are no general methods to track backhe origin of analytically determined acetate to either free aceticcid/inorganic acetate or esterified (organic) acetate.

Most of the methods for determination of acetyl groups inolysaccharides (and also many other organic materials) rely onhe cleavage of the ester and the subsequent quantification ofhe released acetic acid [8–11]. For this quantification many tech-iques are placed at the analytical chemists’ disposal, reaching

rom titrations over enzymatic methods to spectroscopic andhromatographic techniques. As manifold as the quantificationpproaches are, there is one general drawback: they cannot dis-inguish between bound and free acetate (bound and free aceticcid). In other words, acetate released from the ester groups cannote differentiated from free acid/acetate that was already presentesides the covalently bound acetyl groups. The available methodshus just “see” the overall acetate present, being the sum of boundcetate (esters, acetyl groups) and free acetate (adsorbed aceticcid, including acetic acid salts). This is particularly problematic forystems involving acetylating agents, such as procedures used tocetylate polysaccharides, temporarily protect hydroxyl groups ascetates or surface-hydrophobize polysaccharides by acetylation:arge amounts of acetylating agents, and their hydrolysis productcetic acid as well, might be present in adsorbed form, and mightargely falsify analytical results for covalently bound acetate.

When sodium methanolate in methanol is used for determina-ion of the degree of acetylation, a simple gravimetric analysis ofhe deacetylated compounds [12] or a classical titrimetric methodfter distillation and cleavage of methyl acetate [13,14] is applied,hat requires at least a few hundred milligrams of sample. Whenheir concentration is sufficiently high, acetate functionalities canelectively be monitored by NMR techniques and quantified withn error below 3–5% [15–17].

Having dealt with the chemistry of polysaccharidic acetates foruite some time, in particular with the side reactions of the acetateoiety in cellulose acetates [18], we felt the increasingly pressing

eed to have both a sensitive and selective analytical method atand which would allow to reliably quantify bound acetyl groups

n polysaccharides and related carbohydrate-based materials, espe-

ially in the presence of free/adsorbed acetic acid and acetate. Inhis paper, we communicate the related attempts and present a

ethod for the accurate determination of bound acetate in polysac-harides, based on a transesterification/gas chromatographic (GC)

r. A 1372 (2014) 212–220 213

determination protocol. Due to the analysis principle, which is dif-ferent from previous approaches, free acetate/acetic acid do notinterfere with the method nor influence the results negatively, noteven if present in large excess.

2. Materials and methods

2.1. Chemicals

The following chemicals were obtained from Sigma–Aldrich andFluka (Schnelldorf, Germany): acetic acid, anhydrous methanol,methyl acetate, sodium methanolate (0.5 M in methanol), vanillin,acetyl chloride and acetyl chloride-13C2 (99 atom% 13C). Diethylether, dichloromethane, anhydrous sodium sulfate, calcium chlo-ride and pyridine were obtained from VWR (Vienna, Austria). THF,3 A molecular sieve and glacial acetic acid were purchased fromCarl Roth (Graz, Austria). THF and pyridine were stored over freshlydried 3 A molecular sieve before use. All other chemicals were of thehighest purity available and were used without further purification.

2.1.1. Samples2.1.1.1. Acetylated model compounds. Xylan (Birchwood), gumarabic (acacia tree), eugenol acetate, ethyl acetate, linalylacetate, bornyl acetate, �-d-glucopyranose pentaacetate, cel-lobiose ocataacetate and cellulose acetate were purchased fromSigma–Aldrich (Schnelldorf, Germany). Pectin (apple, classicAU202) was obtained from Herbstreith & Fox (Neuenbürg,Germany).

2.1.1.2. Hemicelluloses and cellooligomers. The hemicellulosic sam-ples Spruce 1, Spruce 2, Birch 1 and Birch 2, showing a broad rangeof degree of acetylation, were obtained from Abo Akademi, Turku,Finland. In addition, a sample showing a very low degree of acety-lation was obtained from Lenzing AG, Lenzing, Austria.

Acetolysis of cellulose [19] and subsequent preparative chro-matography were used to obtain acetylated cellotriose andcellohexaose.

2.1.1.3. Pulp and paper samples. Whatman filter paper no.1, cottonlinters, and rag paper were subjected to artificial aging by hangingapproximately three grams of each paper in a 100 mL closed vialwith 5 mL of glacial acetic acid at r.t. for up to 350 days. A booksample from 1968 was purchased from a second hand store andmeasured without any pretreatment.

2.2. Hardware

2.2.1. GC–MS conditions for the analysis of the liquid phaseGC–MS analysis was performed on an Agilent 7890A gas

chromatograph equipped with a Gerstel PTV inlet, coupled withan Agilent 5975C mass selective detector. Column: HP-5MS (30m × 0.25 mm i.d. × 0.25 �m film thickness; J&W Scientific, Fol-som, CA, USA); glass liners packed with quartz wool for thePTV-inlet were obtained from Gerstel (Mülheim an der Ruhr,Germany); carrier gas: helium; column flow: 1.5 mL/min; purgeflow: 50.0 mL/min, 0 min; PTV inlet temperature program: 40 ◦C(2 min), 20 ◦C/min to 100 ◦C (0 min), then 25 ◦C/min to 270 ◦C(6 min); oven temperature program: 30 ◦C (2 min), 10 ◦C/min to50 ◦C (0 min), then 25 ◦C/min to 270 ◦C (2 min); MS: EI mode, 70 eV,source pressure: 1.13*10−7 Pa at 230 ◦C. Quadrupole temperature:150 ◦C, transfer line: 200 ◦C; MS data acquisition in scan and sin-gle ion-mode (SIM): scan range: 45–400 amu; SIM: m/z 60 (acetic

acid), m/z 74 (methyl acetate) and m/z 65 (toluene) at 100 ms dwelltime each.

Aliquots of 0.3 �L were injected into the PTV inlet with an Agi-lent GC Sampler 120. The GC–MS as well as the autosampler were

2 matog

cg

2

mdlawatto

abtehtcda

tnhi

tpptataad

wsw

2v(itaGEw

2

22m

2smsm

14 T. Zweckmair et al. / J. Chro

ontrolled by Agilent MSD Chemstation E.02.01 (Agilent Technolo-ies, Santa Clara, CA, USA).

.2.2. GC–MS conditions for the analysis by SPMEGC–MS analysis was carried out on an Agilent 7890A gas chro-

atograph coupled with an Agilent 5975C triple axis mass selectiveetector (MSD). The GC was equipped with two inlets, an Agi-

ent multimode as well as an Agilent split-/splitless inlet. Bleednd temperature optimized septa (Agilent part no. 5183-4757)ere used for covering both inlets. For the analysis of methyl

cetate, a VF-WaxMS column (30 m × 0.25 mm i.d. × 0.25 �m filmhickness; J&W Scientific, Folsom, CA, USA) was connected tohe multimode inlet on one end and to the MSD on thether.

The SPME-parameters optimized should briefly be explained:gitator temperature: temperature of the agitator used to equili-rate the sample vial; incubation time: time used for equilibratinghe sample at a specific agitator temperature without enrichment;nrichment time: time used to expose the fiber into the sampleeadspace for analyte enrichment; desorption time: time usedo expose the fiber into the inlet to facilitate analyte desorption;onditioning time: time used to expose the fiber at a specific con-itioning temperature for preparing the fiber for the subsequentnalyte enrichment.

The split-/splitless inlet was used for fiber conditioning prioro enrichment. A 5 m × 0.25 mm i.d. fused silica capillary was con-ected on one end to the split-/splitless inlet to ensure a moderateelium carrier gas pressure build-up. The other end just protruded

nto the GC-column oven.The multimode-inlet was operated under the following condi-

ions: constant column flow: 0.9 mL/min using helium carrier gas,urge flow: 60.0 mL/min (2.54 min); injector: 90 ◦C constant. Tem-erature gradient profile: 30 ◦C (1 min), 8 ◦C/min to 50 ◦C (0 min),hen 10 ◦C/min to 200 ◦C (2 min). The MSD was operated in EI-modet 70 eV ionization energy and 1.13 × 10−7 Pa. Ion source tempera-ure: 230 ◦C, quadrupole: 150 ◦C, transfer line: 200 ◦C. The data wascquired in SIM mode selecting 74 m/z for detection of the methylcetate analyte as well as 76 m/z for detection of its carbon-labelederivate at 50 ms dwell time each.

The fully automatized operation of all SPME steps was realizedith a CTC-PALxt autosampler which was controlled by Chronos

oftware v.3.5 (Axel Semrau, Spockhövel, Germany). SPME fibersere obtained from Supelco (Bellefonte, PA, USA).

.2.2.1. Characterization of acetyl vanillin and 4-O-(13C2-acetyl)-anillin. An Agilent XCT 6320 Ion Trap-MS coupled to an ESI-sourceboth Agilent Technologies, Santa Clara, CA, USA) was used. Theon trap was operated by LC/MSD Trap software 5.3 (Bruker Dal-onik GmbH, Bremen, Germany). NMR spectra were recorded on

400 MHz BRUKER Avance II NMR spectrometer (Rheinstetten,ermany) using TopSpin 2.1 (Bruker) for data processing. A Perkinlmer Frontier IR Single-Range spectrometer (Waltham, MA, USA)as used in ATR mode for FTIR measurements.

.3. Methods

.3.1. Method 1: Direct liquid-phase analysis

.3.1.1. Internal standard. Toluene was added to the sodiumethanolate solution as an internal standard.

.3.1.2. Sample preparation. Paper or pulp samples were cut into

mall pieces (approx. 1–2 cm2). 50 mg of the carbohydrate poly-er sample were transferred into a 1.5 mL vial. The vials were

ealed with 1.3 mm silicon/PTFE septum crimp caps. 700 �L sodiumethanolate solution containing 1 �L/mL toluene was added

r. A 1372 (2014) 212–220

through the septum. After 5 min reaction time the sample wasready for analysis.

2.3.2. Method 2: Analysis via gas phase2.3.2.1. Internal standard. 4-O-(13C2-acetyl) vanillin was selectedto generate isotopically labeled methyl 13C2-acetate in situ, actingas an internal standard. The synthesis was carried out according to[20]. In brief: a two-necked round-bottomed flask was equippedwith a stirring bar, septum and a drying tube containing CaCl2 asa drying agent. Vanillin (395 mg, 2.56 mmol) was dissolved in THF(1.75 mL) and the solution was cooled to 0 ◦C. A solution of acetylchloride-13C2 (250 mg, 3.08 mmol) in THF (1 mL) was added drop-wise to the ice cooled solution of vanillin. After completion of theaddition, a solution of pyridine (250 �L, 3.10 mmol) in THF (1 mL)was carefully added dropwise to the reaction mixture. Immediatelya white precipitate appeared. The reaction mixture was stirred for1 h at 0 ◦C and for 6 h at room temperature. The progress of the reac-tion was monitored with TLC (n-heptane/ethyl acetate, v/v = 1:2).The white precipitate was filtered off and washed with diethyl ether(2 mL). The turbid filtrate was evaporated and the residue was dis-solved in hot THF (2 mL). After filtration the filter cake was washedwith hot THF (1 mL) and the filtrate was evaporated. The residuewas dissolved in CH2Cl2 (3 mL) and washed with water and brine(each 1 mL). The organic layer was carefully dried with anhydrousNa2SO4 and filtered. The solvent was evaporated yielding a color-less oil. The liquid product was allowed to solidify under ambientconditions and then dried under high vacuum at room temper-ature overnight. Yield: 404 mg of a white solid (1.91 mmol, yield75%); TLC (n-heptane/ethyl acetate, v/v = 1:2): vanillin Rf = 0.47; 4-O-acetyl vanillin Rf = 0.55; FTIR [cm−1]: 1688, 1598, 1506, 1471,1426, 1394, 1277, 1180, 1123. 1H NMR (400.13 MHz, CDCl3) ı[ppm] = 9.97 (s, 1H), 7.52 (s, 1H), 7.50 (d, 1H, J = 9.0 Hz), 7.24 (d,1H, J = 7.8 Hz), 3.93 (s, 3H), 2.37 (dd, 3H, J = 130.3, 7.0 Hz), 1H–13Cheteronuclear coupling clearly visible through the duplet of dupletsat 2.37 ppm; 13C NMR (100.62 MHz, CDCl3) ı [ppm] = 191.0 (CHO),168.3 (d, 13C O, J = 60.5 Hz), 152.0 (Cq), 145.0 (Cq), 135.2 (Cq),124.7 (CarH), 123.4 (CarH), 110.9 (CarH), 56.1 (CH3), 20.6 (d, 13CH3,J = 60.5 Hz), 13C–13C homonuclear coupling clearly visible by thetwo duplets at 168.3 and 20.6 ppm. ESI-MS (5 ppm in THF, positiveionization mode) [m/z]: 153.0 (100%, [M−13CH3

13CO+H]+, C8H9O3),218.9 (12%, [M+Na]+, 13C2

12C8H10O4Na).

2.3.2.2. Sample preparation. 10 mg of 13C2-labeled internalstandard were dissolved in 4 mL of anhydrous methanol (stock A).Approximately 2 mg of cellulose or paper sample were transferredinto a 1.5 mL vial for analysis. 50 �L of anhydrous methanol and20 �L of stock A were added to each sample. 300 �L of sodiummethanolate was transferred to every sample before the vials weresealed with 1.3 mm silicon/PTFE septa crimp caps. The sampleswere immediately ready for analysis by GC–MS.

Hemicellulosic samples were prepared in analogy topaper/cellulose samples except for using 150 �L of anhydrousmethanol, 40 �L of stock A and 180 �L of sodium methanolatesolution.

2.3.3. Statistical data evaluationPairs of datasets were compared using the fitCmpData x-

function of Origin 9.0 (OriginLab, Northhampton, MA, USA). The testfunction compares two datasets, which are fitted with the same fit-ting function, by applying an F-test to determine whether the twodata sets are significantly different from each other.

T. Zweckmair et al. / J. Chromatogr. A 1372 (2014) 212–220 215

acety

3

3

gstasmtwafbcIwataafbatsom

haspmtcmafeb

3c

3

tsc(tZhc

Fig. 1. Zemplén de

. Results and discussion

.1. Reaction principle – the Zemplén reaction

Zemplén et al. [12] described the reaction between acetylroups and sodium methanolate (sodium methanolate in methanololution), which results in methyl acetate formation. The transes-erification usually starts immediately upon methanolate additionnd proceeds rapidly. The overall reaction time depends on theample type and interpenetration of the sample. The lower theolar mass or the lower the sample concentration, the faster is

he full completion of the reaction [12]. The Zemplén reaction [21],hich is regularly applied in carbohydrate synthesis to remove

cetyl and other acyl protecting groups, is used in the present caseor accurate distinction between free acetic acid and covalentlyound acetyl groups. Only the latter ones are transesterified uponatalytic action of anhydrous sodium methanolate in methanol.n this process, an equimolar amount of methanol is consumed

hile generating the corresponding amount of methyl acetateccording to Fig. 1 [13,21]. The Zemplén reaction proceeds quanti-atively, a prerequisite to the envisaged analytical application. Freecetic acid or the corresponding salt are not converted to methylcetate by transesterification, which renders the method specificor ester functionalities [22]. Free acids in a sample, however, wille neutralized and thus require sufficient amounts (beyond cat-lytic quantities) of sodium methanolate. Transesterification ofriglycerides and carboxylic acid esters to generate their corre-ponding methyl esters has also been reported upon the actionf an excess of methyl acetate in the solvent system sodiumethanolate/methanol [22,27,28].Because of its low boiling point (57 ◦C) and the resulting

igh volatility, gas-chromatography is the method of choice fornalyzing the methyl acetate formed, allowing for a fast andensitive detection. The transesterification can conveniently beerformed directly in the GC injection vial after adding sodiumethanolate/methanol to the analyte. Several sample introduc-

ion techniques are available to transfer the analyte to theolumn: liquid sampling, headspace and headspace solid-phase-icroextraction (SPME). Since we aimed to analyze also trace

mounts of acetyl groups and heating was considered detrimentalor the sample material, the headspace technique was not consid-red any further. Hence, liquid sample injection and SPME haveeen evaluated and will be addressed in more detail.

.2. Determination of methyl acetate by liquid injection – proof ofoncept and limitations

.2.1. Zemplén conditionsZemplén et al. [21] conducted all experiments in less concen-

rated sodium methanolate and at gram scale compared to ourtudy. Different reaction times were reported by Zemplén andoworkers. In case of 30 g octacetyl cellobiose 120 mL of methanolabs.) and 30 mL sodium methanolate (0.1 M) were added, the reac-

ion was performed at room temperature for 3 h. According toemplén, the transesterification reaction at non-reducing carbo-ydrates, sugar alcohols and glycosides proceeded fast and wasompleted e.g. for hexaacetyl-mannitol within 3 min.

lation mechanism.

In the current study, 0.5 M sodium methanolate was usedfor transesterification. No difference in yield for reaction timesbetween 5 min and 48 h were observed for a powdered glucosepentaacetate. It was concluded that deacetylation proceeded quan-titatively within short reaction time, which is an excellent featurefor a rapid pre-column treatment.

3.2.2. Interference of acetic acidThe transesterification was first performed straightforward in

the GC injection vial adding sodium methanolate to the analyte. Thesolution was directly injected after 5 min. Free acetic acid, whichhas to be considered as an integral trace substance in biologicalsamples, is transformed into the corresponding salt in the highlyalkaline sodium methanolate solution. Hence it does not interferewith the reaction. Still, trace amounts of ubiquitous acetic acid andacetate (sodium salt) do enter the liner upon injection, and we havetested whether this causes any interference with the determina-tion. If acetic acid is injected directly with methanol present, anesterification of acetic acid to methyl acetate at GC inlet tempera-tures higher than 70 ◦C was observed, however, this is not the caseif methanolate is present. In order to obtain a reasonable methylacetate peak shape, decrease of the PTV inlet temperature to 40 ◦Cduring injection was necessary.

The capability of the sodium methanolate treatment to detectsolely acetyl groups but no free acetic acid has been evaluated byacetic acid spiking experiments. Even the presence of a high excessof acetic acid did not lead to an increased formation of methylacetate.

3.2.3. Peak shape optimizationIn order to optimize the peak shape of methyl acetate differently

packed liners (silanized glass wool, quartz wool, Tenax TATM or Car-botrap BTM) and different inlet temperatures (30–120 ◦C) have beenstudied. The best peak geometry of methyl acetate was achievedwith a quartz wool liner packing material at 40 ◦C inlet temperatureand an initial oven temperature of 30 ◦C.

3.2.4. Internal standardToluene was selected as an internal standard to be added to each

sample mixture at a concentration of 1 �L/mL. It was chosen, since itis miscible with methanol, inert under the Zemplén conditions, andcan be detected quite easily at a different retention time relative tomethyl acetate (Fig. 2b and c). The sodium methanolate itself didnot show any byproduct which could interfere with the compoundsof interest (Fig. 2d).

The compounds methyl acetate, acetic acid as well as toluene(internal standard) were detected without coelution (Fig. 2a). Inthe presence of methanolate no free acetic acid was detected.The comparison between the methyl acetate reference compoundand methyl acetate formed from glucose pentaacetate in sodiummethanolate showed no difference with regard to peak retentionand peak area response (Fig. 2b and c).

3.2.5. CalibrationIn order to assess the effect of the actual deacetylation step

on generating methyl acetate, calibration was carried out with

216 T. Zweckmair et al. / J. Chromatogr. A 1372 (2014) 212–220

Fig. 2. GC/SIM-MS chromatograms of (a) methyl acetate (authentic sample) and acetic ain 0.5 M sodium methanolate; (c) glucose pentaacetate (0.23 mg/mL) in 0.5 M sodium mblank. Every sample contained toluene (1 �L/mL) as an internal standard.

Fig. 3. Calibration curves of methyl acetate as reference compound in methanol(mc

gt

a(tpeedMtTa

3

ml

By injecting the complete reaction mixture, sodiummethanolate is accumulated in the GC liner at the height ofthe syringe injection, while methanol solvent passes through.The sodium methanolate could partly be transferred onto the

Table 1Liquid phase validation samples analyzed by GC–MS.

�) and methyl acetate formed from �-d-glucose pentaacetate in sodiumethanolate (©); response ratio: Aanalyte/Ainternal standard; concentration ratio:

analyte/cinternal standard.

lucose pentaacetate. Direct calibration with methyl acetate is notrivial because of its high volatility (b.p. 57 ◦C).

The calibration with methyl acetate directly and with methylcetate formed from glucose pentaacetate and sodium methanolateafter 5 min reaction time) gave similar slopes in the linear calibra-ion curve (methyl acetate: k = 1.2582; methyl acetate by glucoseentaacetate: k = 1.2471; Fig. 3). The statistical comparison ofxperimental data and their mathematical function is crucial tovaluate the statistical similarity of different data sets. The twoatasets were compared by Origin 9.0 (OriginLab, Northhampton,A, USA). The analysis revealed that the two datasets are not sta-

istically different at a 0.05 significance level (f(2, 14) = 0.87644).hus, it was concluded that the quantification was not affected byny other parameters.

.2.6. Analytical figures of meritThe evaluation of the analytical figures of merit revealed that

ethyl acetate can be detected well at the desired sensitivity: theimit of detection of 6.95 pmol/mg of sample was determined based

cid in methanol (MeOH), 1 �L/mL each; (b) methyl acetate standard (0.28 mg/mL)ethanolate producing 0.219 mg/mL methyl acetate; (d) 0.5 M sodium methanolate

on the three sigma criterion of the noise quantified via a single-point calibration [23]. Accuracy of 100.89% was determined by theaddition of a known amount of methyl acetate to a hemicellulosicsample. The linear dynamic range was valid within four orders ofmagnitude.

3.2.7. Application to polysaccharide samplesIn the next step, hemicellulose and pectin samples were ana-

lyzed to show the applicability of the method with real-worldcarbohydrate biopolymer samples, using the glucose pentaacetate-based calibration. Three selected polysaccharides were successfullyanalyzed with an acceptable precision revealing meaningful valuesbased on literature data [4,24] (see Table 1). No direct referencevalues on the degree of acetylation of the respective samples areavailable from the manufacturers.

The stability of the GC/MS performance (retention time, peakresponse and peak geometry) of methyl acetate released fromsodium methanolate treatment of glucose pentaacetate was con-firmed by repetitive analysis of methyl acetate as a referencestandard. In order to obtain reasonable analyte signal, 50 mg of therespective sample were analyzed. Lower sample amounts are alsopossible depending on the sample structure. Reducing the sam-ple amount is particularly important for analysis of samples frombiological or historic (paper, textile) origin, where sample amountmatters.

3.2.8. Limits of the method and cautionary considerations

Polysaccharides Methyl acetate (nmol/mg) (n = 3)

Xylan (birch wood) 684.32 ± 40.54Gum arabic (acacia) 767.43 ± 17.30Pectin from apple (classic AU202) 3561.35 ± 13.92

matogr. A 1372 (2014) 212–220 217

ciaalils

mtitpasitHi

lTratm

pdmam

3c

dipdao

3

SDsmtrsefit

3

dm9

3a

T. Zweckmair et al. / J. Chro

olumn unless appropriate trapping materials such as quartz wooln combination with a Gerstel PTV-inlet system are used. When

larger number of samples is analyzed by the liquid technique,ccumulation of sodium methanolate in the liner becomes a prob-em, as the trapping capacity of the absorber material in the liners limited to approximately 40 samples. Selection of appropriateiner packing materials would improve the situation, but not fullyolve this problem.

Toluene as an internal standard fulfills the common require-ents, such as varying injection volume. However, toluene is not

aking part in the actual deacetylation process. Therefore, possiblenfluences affecting the final methyl acetate yield, such as mois-ure content or adsorptive interactions of the analyte in the gashase with the materials present in the vial (i.e. cellulose), as wells discrimination effects cannot be corrected for by this internaltandardization approach. A more sophisticated internal standard-zation strategy is needed. Isotope Dilution Mass Spectrometry ishe preferred method to eliminate all discriminations mentioned.owever, an isotopically labeled standard must be appropriately

ntroduced and needs to be optimized for signal yield.The clear benefit of the liquid sampling GC/MS method is the

ess laborious protocol for sample preparation and easy handling.he high reactivity of the methanolate allows for fast and completeeactions at room temperature, which is particularly suitable fornalysis of heat-sensitive substrates. The transesterification reac-ion takes place independent of whether the samples are soluble in

ethanolate/methanol or are only suspended or dispersed.To overcome the above mentioned weaknesses of the liquid-

hase sampling GC/MS method, the vapor phase was analyzedirectly by SPME. This approach would overcome the aboveentioned difficulties, in particular the contamination of inlet

nd column head by traces of the rather aggressive sodiumethanolate.

.3. Determination of methyl acetate by SPME: optimization andalibration

It should be noted that all the parameters associated with SPMEo show a strong interdependency. In order to ensure traceability

n method development, every parameter (fiber type, inlet tem-erature, agitator temperature, incubation time, enrichment time,esorption time, conditioning time) was optimized individually in

step-by-step sequence. While varying a specific parameter, all thethers were kept constant.

.3.1. Fiber selectionAs a first step, screening was conducted among common

PME fiber types, such as polyacrylate (white), PEG (purple),VB/CAR/PDMS (gray) and CAR/PDMS (black). All tested fibers

howed a rather comparable performance towards enrichment ofethyl acetate. However, the black fiber (CAR/PDMS) enabled up

o 120 injections with a single fiber rendering this fiber type supe-ior compared to the other fibers tested. Hence, the black fiber waselected for further investigations due to its higher stability. Fornhancing SPME-fiber stability after conditioning, an additionalber cooling step lasting 5 min was implemented before startinghe next analyte enrichment cycle.

.3.2. Parameter optimizationThe shape of the methyl acetate peak was optimized by applying

ifferent inlet temperatures. The sharpest and most symmetricalethyl acetate peak could be obtained at an inlet temperature of

0 ◦C.As a next step, the agitator temperature was varied between

0 ◦C and 40 ◦C in 5 ◦C steps. Keeping in mind that a highergitator temperature might cause secondary reactions of the

Fig. 4. Methyl acetate signal vs. (a) agitator temperature, (b) desorption and (c)incubation time; normalization was carried out for each individual parameter.

sample in sodium methanolate and methanol, emphasis was puton maximizing the methyl acetate peak area at moderate agita-tor temperatures. 35 ◦C agitator temperature gave the highest peakareas for methyl acetate (Fig. 4a). Thus, this setting was selected forsubsequent studies.

Incubation time influences the total peak area, and this parame-ter was varied between 10 and 25 min in 5 min steps. The optimumwas reached at 20 min incubation time yielding the highest peakareas (Fig. 4c).

Also enrichment time had to be optimized. Preliminary investi-gations using the CAR/PDMS (black) fiber type have shown thatfiber overloading occurs easily, which does affect fiber stabilityon the long run. Thus, analyte concentration and enrichment timewere considered important parameters to be investigated simulta-neously. For covering a certain dynamic range, varying amounts ofthe highest and lowest concentrated samples and standards wereanalyzed. At the same time, enrichment time was systematicallydecreased in order to prevent the fiber from being overloaded.

Desorption time was investigated within a range of 3–12 min in3 min steps. Clear evidence was found that increasing desorptiontime enhances methyl acetate peak height (Fig. 4b). The rather timeconsuming desorption process due to (a) moderate inlet temper-ature (90 ◦C) and (b) initially rather low oven temperature (30 ◦C)might be a reasonable explanation for methyl acetate peak heightenhancement [25]. 12 min desorption time was selected for all fur-ther investigations.

The fiber conditioning time adds considerably to total analysistime and hence was varied between 10 and 50 min in 10 min steps.It was found that already 10 min of fiber conditioning are sufficientfor preparing the fiber to enrich methyl acetate on the fiber surfacewithout observing any analyte carry over from the previous sample.

3.3.3. Degree of acetylation (degree of substitution, DS) in theanalyzed samples

Acetylated samples to be analyzed by this method have verydifferent properties. Some are soluble in dry organic solvents (i.e.peracetylated mono- and oligosaccharides, peracetylated or nearlyperacetylated polysaccharides) while others have to be measured

in suspended or dispersed form, such as hemicelluloses or partiallyacetylated celluloses, as they are insoluble in most organic sol-vents. In addition, the degree of substitution (DS) of acetyl groupscan vary from fully acetylated to minute traces. For cellulose, the

218 T. Zweckmair et al. / J. Chromatogr. A 1372 (2014) 212–220

Fig. 5. Universal calibration curves (at the same time preparation-specific cal-ibration curves) of methyl acetate covering all sample types; response ratio:A0

mhahdp

aamoo

aspara

1ab

aoc2To

ha

Fig. 6. Methyl acetate produced during consecutive injection of independently

TA

analyte/Ainternal standard; concentration ratio: canalyte/cinternal standard. y = a + bx; slope:.9377 ± 0.0015; intercept: 0.0829 ± 0.0073; r2 = 0.9997.

aximum degree of acetylation is three, in this case all threeydroxyl groups per anhydroglucose unit are esterified. Tracecetylations may cover only one hydroxyl group among severalundred anhydroglucose units. The protocol has to be adoptedepending on whether only trace amounts of acetyl groups areresent or if a higher degree of acetylation has to be analyzed.

A rough threshold can be drawn around 3.5 �mol of methylcetate released per GC-vial according to the above protocol. For

dynamic range between 0.605 nmol and 3.50 �mol of releasedethyl acetate, two minutes of enrichment time will cover a degree

f substitution (DS) range up to 0.283 (Fig. 5a) in typical celluloser aged paper samples.

When decreasing the sample amount continuously, a virtu-lly infinite DS can be covered: for analyzing acetylated oligomeramples (e.g. peracetylated cellooligomers), a considerable sam-le amount e.g. 5 mg is dissolved in acetone. After transferring anppropriate aliquot corresponding to 20 �g, the solvent was evapo-ated prior to the Zemplén reaction. This setup covers samples with

very high DS of acetyl groups.For the analysis of e.g. liquid samples soluble in methanol,

0 �L of sample were dissolved in 10 mL of methanol. Subsequently 20 �L aliquot was analyzed without carrying out evaporationefore Zemplén reaction.

Some specimen might generate fiber overloading due to a highmount of methyl acetate generated from 2 mg of sample materialr because of lack of solubility in acetone or methanol (e.g. hemi-ellulose samples). Such samples are analyzed in a modified way:0 sec of enrichment time are applied for a DS-range up to 0.726.his setup covers a dynamic range between 3.19 nmol and 11 �molf released methyl acetate (Fig. 5b).

Unknown samples are consequently first analyzed using theemicellulose route as a first approximation. Only if the methylcetate signal intensity is not sufficient, appropriate aliquots

able 2nalytical figures of merit in SPME–GC–MS.

Limit of detection (nmol/mg)a Accuracy (%) RSD (n = 3

Low end o

Cellulose (paper) sample 0.09 101.45 1.01

Hemicellulose sample 0.48 101.26 0.97

a Considering 2 mg of solid sample to be analyzed.

prepared samples (from the same sample material) and samples stored at roomtemperature for a specified period of time.

dissolved in acetone or methanol should be analyzed. If the methylacetate signal intensity is still considered to be too low, the samplematerial can be analyzed directly similar to paper samples.

In classical GC–MS quantification the response ratio is plot-ted against the concentration ratio for calibration [26]. Since a13C-labeled internal standard is used (see below), practically no dis-crimination between the analyte and the internal standard occurs.Thus, the calibration itself is virtually independent of enrichmenttime which allows generating a universal calibration curve inde-pendent of the sample setup applied. For clarity, the two samplepreparation setups are given individually in Fig. 5. For routinequantification, two separate calibration curves are used to providehighest quality of data fit. Fig. 5 shows the universal calibration ofmethyl acetate covering the whole range of different sample typeswithin a single dynamic range.

3.3.4. Internal standardWe have selected 4-O-(13C2-acetyl)-vanillin as a standard,

which was prepared from vanillin and 13C2-acetyl chloride (seeexperimental). This standard comprises both the Zemplén reactionand the adsorption of methyl acetate at the SPME fiber, and addsconsiderably to the robustness of the method.

After calibration (Fig. 5) with 4-O-(13C2-acetyl)-vanillin to gen-erate methyl 13C2-acetate in situ, method stability was criticallyassessed over a longer period of time. Independent sampleswere prepared from the same sample material and differenttime intervals were chosen before GC analysis. Fig. 6 shows thatSPME–GC–MS yields valid results independent of sample storageat ambient conditions. Thus, it can be concluded that the inter-nal standardization approach is indeed compensating for (a) anyfurther parameter affecting the yield of methyl acetate releasedaccording to Zemplén deacetylation (e.g. moisture affecting sodium

methanolate stability) and (b) the practically unavoidable time-dependent loss of small amounts of analyte, even though thesample vials are properly sealed and handled.

) (%)

f calibration curve Center of calibration curve High end of calibration curve

0.86 0.860.65 1.59

T. Zweckmair et al. / J. Chromatogr. A 1372 (2014) 212–220 219

Table 3Analysis of acetylated model compounds in SPME–GC–MS (n = 3).

n(OAc)/n(compound)

Determined Target

Eugenol acetate 1.05 ± 0.09 1Ethyl acetate 1.07 ± 0.02 1Linalyl acetate 1.01 ± 0.03 1Bornyl acetate 1.07 ± 0.01 1Glucose pentaacetate 5.30 ± 0.09 5Cellobiose octaacetate 8.38 ± 0.20 8Cellotriose, peracetylated 10.98 ± 0.40 11

3

maaosm

isdpti

3

dasw

cesalei

aoads

Fig. 7. Acetylation of paper samples, dependency on exposure time to acetic acidvapor. The insert table gives the acetyl group content for the starting materials before

TA

Cellohexaose, peracetylated 19.01 ± 0.08 20Cellulose acetate (Mw 100 kDa) 41.92 ± 0.46 wt% acetyl 39.3 wt% acetyl

.3.5. Analytical figures of meritTypical analytical figures of merit were determined as sum-

arized in Table 2, using the examples of a cellulose (paper)nd a hemicellulose specimen. Limit of detection was determinedccording to the 3-sigma-criterion of the noise quantified by ane-point-height calibration [23]. Accuracy was determined bytandard addition of vanillyl acetate to the two specified sampleatrices.Relative standard deviation was evaluated by the analysis of

ndependent standards. At the lowest end of the calibration curvetill an RSD-value of around one percent was obtained. In the mid-le range of the calibration RSD-values significantly below oneercent were achieved, which exceeds the expected RSD-valuesypically known in GC–MS involving derivatization [26] by approx-mately one order of magnitude.

.3.6. Validation and application to real samplesSeveral acetylated model compounds were analyzed for vali-

ation purposes as summarized in Table 3. In summary, a goodgreement between the measured O-acetyl content and the corre-ponding target values were obtained (Table 3). A very broad bandidth can be analyzed accurately by SPME–GC–MS.

As a next step, the SPME–GC–MS method was used to analyzeellulose (paper) samples. The paper samples are an example ofxtremely low degrees of acetylation. These analytes are of interestince historic papers – especially if stored largely sealed – generatecetic acid due to aging processes, which in turn might cause acety-ation. Although of low-degree, this acetylation might have severeffects on mechanical and surface properties as well as on cellulosentegrity.

Different types of cellulose (paper) were exposed to acetic acidnd hence contained adsorbed acetic acid as well as trace amounts

f covalently bound acetyl groups. In order to get variable contentsn exposure kinetic was recorded. A clear trend in acetylation forifferent paper types (Whatman, rag, book paper and cotton lintersample) was observed applying the SPME method developed, cf.

able 4nalysis of various hemicellulosic samples, data in (nmol/mg) (n = 3).

Hemicellulose origin Quantified

(nmol/mg) DS (n(OAc)/n(anhyd

Spruce 1 1048.20 ± 28.62 0.178 ± 0.010b

Spruce 2 260.77 ± 43.39 0.043 ± 0.010b

Birch 1 2174.47 ± 166.23 0.317 ± 0.011c

Birch 2 1479.05 ± 85.37 0.208 ± 0.011c

Beech 0.98 ± 0.03a 0.0001 ± 0.0100c

a Between LOD and LOQ.b n(OAc)/n(anhydrohexose).c n(OAc)/n(anhydroxylose).d Average standard deviation: 1.05 mg acetate/g of sample.e Very heterogeneous sample.f Determined by cleaving the acetate-groups by sodium hydroxide and quantifying the

exposure to acetic acid vapor (values indicated by an arrow).

Fig. 7. As expected, cotton linters and Whatman paper showed anextremely low degree of acetylation (between LOD and LOQ) asthey are virtually free of any acetate-containing hemicelluloses:the commercial cleaning removes remaining acetyl traces by analkaline purification step. Both book paper and rag paper have anotably higher DS of acetyl groups, mainly due to a higher naturalcontent originating from hemicelluloses.

Finally, several hemicellulose samples, revealing a rather broaddistribution in degree of acetylation, were analyzed as summarizedin Table 4.

As shown in Table 4, in all cases HPLC generates (sometimesexceedingly) higher values of acetate contents. This is due to the factthat the HPLC method is not able to accurately distinguish betweenfree acetic acid and covalently bound acetate and thus determinesacetates as sum parameter while the GC–MS method is selectivefor covalently bound acetate. A rather low degree of coverage wasobtained in case of sample Spruce 2. In case of the Beech sam-ple a much lower amount (approx. factor 100) of covalently boundacetate was found. This result demonstrates the disadvantages ofthe HPLC-based approach where particularly small concentrationsof covalently bound acetate might be largely falsified by the pres-ence of physically adsorbed acetic acid/acetate.

Standard deviations of the values for comparison were calcu-lated based on a mean RSD-value. In case of the values obtainedfrom SPME, the actual standard deviation values are shown. Thus,

the analytical precision of the SPME-method might be equal or evenbetter compared to the method used for comparison.

HPLC results for comparisonf

ro-monosaccharide)) (nmol/mg)d Coverage (%)

1180.48 ± 28.83 88.791297.90 ± 31.69 20.09e

2198.81 ± 53.69 98.892092.51 ± 51.10 70.68

69.41 ± 3.37 1.40

acetic acid generated by LC–UV after neutralization.

2 matog

4

ocqGopdmuawsipaftifiwTco

A

“aFlt

R

[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

20 T. Zweckmair et al. / J. Chro

. Conclusion

In the present work, methods for the analysis of the degreef acetylation in carbohydrate biopolymers according to a proto-ol combining a Zemplén-deacetylation reaction and subsequentuantification of quantitatively generated methyl acetate byC–MS were developed and comparatively assessed. The analysisf the liquid phase using toluene as an internal standard is a sim-le and fast method with a wide linear dynamic range. However,ue to shortcomings induced by the presence of excess sodiumethanolate, which can cause problems at the equipment (col-

mn head), an alternative route was found in an SPME–GC–MSpproach. Using 4-O-(13C2-acetyl)-vanillin as the internal standard,hich generates methyl 13C2-acetate in situ, additional influences

uch as inconstant moisture affecting sodium methanolate stabil-ty and adsorptive interactions of the analyte with the materialsresent in the sample vessel are completely compensated for. Inddition, the sample amount required was significantly reducedrom approximately 50 to 2 mg and even below. The robustness ofhe SPME–GC–MS method is somewhat limited by the fiber stabil-ty as well as by a rather narrow dynamic range, but it allows for therst time to analyze samples with very low degrees of acetylationith high accuracy and precision using only a few mg of material.

he Zemplén-SPME–GC–MS approach proved to be the method ofhoice for the analysis of various acetylated polysaccharides andther biopolymers.

cknowledgements

The financial support of the Christian Doppler LaboratoryAdvanced Cellulose Chemistry and Analytics” is gratefullycknowledged. We thank Dr. Risto Korpinen (Abo Akademi,inland) and Lenzing AG (Lenzing, Austria) for providing hemicellu-osic samples and reference data for acetyl groups. We are thankfulo Dr. Markus Bacher for recording the NMR spectra.

eferences

[1] C. Choudhary, C. Kumar, F. Gnad, M.L. Nielsen, M. Rehman, T.C. Walther, J.V.Olsen, M. Mann, Lysine acetylation targets protein complexes and co-regulatesmajor cellular functions, Science 325 (2009) 834–840.

[2] K.S. Fritz, J.J. Galligan, M.D. Hirschey, E. Verdin, D.R. Petersen, Mitochondrialacetylome analysis in a mouse model of alcohol-induced liver injury utilizingSIRT3 knockout mice, J. Proteome Res. 11 (2012) 1633–1643.

[3] T. Brock, Protein acetylation: much more than histone acetylation, Cayman(2011) 1–3, Article 2152.

[4] A. Teleman, M. Nordström, M. Tenkanen, A. Jacobs, O. Dahlman, Isolation and

characterization of O-acetylated glucomannans from aspen and birch wood,Carbohydr. Res. 338 (2003) 525–534.

[5] Z. Kostalova, Z. Hromádková, A. Ebringerová, Structural diversity of pectinsisolated from the Styrian oil-pumpkin (Cucurbita pepo var. styriaca) fruit, Car-bohydr. Polym. 93 (2013) 163–171.

[

r. A 1372 (2014) 212–220

[6] G. Henriksson, in: M. Ek, G. Gellerstedt, G. Henriksson (Eds.), Vol. 1 WoodChemistry and Wood Biotechnology, 6. Lignin, Pulp and Paper Chemistry andTechnology, de Gruyter, Berlin, 2009, pp. 121–145.

[7] N. Davis, S. Hon, in: O. Olabisi (Ed.), Handbook of Thermoplastics, 14. CellulosePlastics, Marcel Dekker, New York, 1997, pp. 331–348.

[8] D. Klemm, B. Philipp, T. Heinze, U. Heinze, W. Wagenknecht, ComprehensiveCellulose Chemistry; Volume 1: Fundamentals and Analytical Methods, WILEY-VCH, Weinheim, 1998.

[9] O.B. Wurzburg, in: Methods in Carbohydrate Chemistry; vol. IV, Acetylation,Academic Press, New York, 1964, pp. 286–288.

10] S. Levigne, M. Thomas, M.C. Ralet, B. Quemener, J.F. Thibault, Determination ofthe degrees of methylation and acetylation of pectins using a C18 column andinternal standards, Food Hydrocoll. 16 (2002) 547–550.

11] M.M.H. Huisman, A. Oosterveld, H.A. Schols, Fast determination of the degreeof methyl esterification of pectins by head-space GC, Food Hydrocoll. 18 (2004)665–668.

12] G. Zemplén, A. Gerecs, I. Hadacsy, The saponification of acetylated carbohy-drates, Ber. Dtsch. Chem. Ges. 69 (1936) 1827–1829.

13] R.L. Whistler, A. Jeanes, Quantitative estimation of acetyl in carbohydrateacetates, Ind. Eng. Chem. 15 (1943) 317–318.

14] L. Zhu, J.P. O’Dwyer, V.S. Chang, C.B. Granda, M.T. Holtzapple, Structural fea-tures affecting biomass enzymatic digestibility, Bioresour. Technol. 99 (2008)3817–3828.

15] V.W. Goodlett, J.T. Dougherty, H.W. Patton, Characterization of celluloseacetates by nuclear magnetic resonance, J. Polym. Sci. A-Polym. Chem. 9 (1971)155–161.

16] D.W. Lowman, ACS Symposium Series 688, 1998, pp. 131–162.17] C.M. Buchanan, J.A. Hyatt, D.W. Lowman, Two-dimensional NMR of polysaccha-

rides: spectral assignments of cellulose triesters, Macromolecules 20 (1987)2750–2754.

18] T. Rosenau, A. Potthast, A. Hofinger, P. Kosma, Isolation and identification ofresidual chromophores in cellulosic materials, Macromol. Symp. 223 (2005)239–252.

19] K. Hess, K. Dziengel, Über Cellotriose und ihre Derivate, Ber. Dtsch. Chem. Ges.68 (1935) 1594–1605.

20] D.R. Kelly, S.C. Baker, D.S. King, D.S. de Silva, G. Lord, J.P. Taylor, Studies of nitrileoxide cycloadditions, and the phenolic oxidative coupling of vanillin aldoximeby Geobacillus sp. DDS012 from Italian rye grass silage, Org. Biomol. Chem. 6(2008) 787–796.

21] G. Zemplén, Géza Zemplén Neuere Ergebnisse der Kohlenhydratforschung, Ber.Dtsch. Chem. Ges. (A and B Series) 74 (1941) A75–A92.

22] S.J. Jankowski, Determination of carboxylic acids present as esters in plasticiz-ers and polymers by transesterification and gas chromatography, Anal. Chem.37 (1965) 1709–1711.

23] Chemische Analytik, Nachweis-, Erfassungs- und Bestimmungsgrenze, Norme-nausschuss Materialprüfung, DIN 32645:2008-11.

24] M. Fischer, E. Arrigoni, R. Amadò, Changes in the pectic substances of applesduring development and postharvest ripening. Part 2: Analysis of the pecticfractions, Carbohydr. Polym. 25 (1994) 167–175.

25] H. Prosen, L. Zupancic-Kralj, Solid-phase microextraction, TrAC – Trends Anal.Chem. 18 (1999) 272–282.

26] M. Becker, T. Zweckmair, A. Forneck, T. Rosenau, A. Potthast, F. Liebner,Evaluation of different derivatisation approaches for gas chromatographic-mass spectrometric analysis of carbohydrates in complex matricesof biological and synthetic origin, J. Chromatogr. A 1281 (2013)115–126.

27] J. Blasko, R. Kubinec, B. Husová, P. Prikryl, V. Pacákova, K. Stulík, J. Hradilová,

Gas chromatography/mass spectrometry of oils and oil binders in paintings, J.Sep. Sci. 31 (2008) 1067–1073.

28] A.J. Dijkstra, Revisiting the mechanisms of low-temperature, base-catalysedester interchange reactions, OCL – Ol. Corps Gras Lipides 15 (2008)208–212.

125 

  

  

 Paper VI  

 The museum in a test tube – Adding a third dimension to the 

evaluation of the impact of volatile organic acids on paper. 

 

 Polymer Degradation and Stability, vol. 130, pp. 109‐117. 

 

   

 

 

 

 

Becker, M.; Meyer, F.; Jeong, M.‐J.; Ahn, K.; Henniges, U.; Potthast, A. (2016)  

          

lable at ScienceDirect

Polymer Degradation and Stability 130 (2016) 109e117

Contents lists avai

Polymer Degradation and Stability

journal homepage: www.elsevier .com/locate /polydegstab

The museum in a test tube e Adding a third dimension to theevaluation of the impact of volatile organic acids on paper

M. Becker a, F. Meyer b, M.-J. Jeong a, c, K. Ahn a, U. Henniges a, A. Potthast a, *

a University of Natural Resources and Life Sciences, Department of Chemistry, Christian-Doppler-Laboratory “Advanced Cellulose Chemistry and Analytics”,Department of Chemistry, Konrad-Lorenz-Str. 24, A-3430, Tulln, Austriab Staatliche Museen zu Berlin (National Museums in Berlin), Kupferstichkabinett (Museum of Prints and Drawings), Matth€aikirchplatz 4, D-10785, Berlin,Germanyc Department of Biological and Environmental Science, Dongguk University-Seoul, Seoul, 100-715, Republic of Korea

a r t i c l e i n f o

Article history:Received 25 January 2016Received in revised form19 May 2016Accepted 29 May 2016Available online 31 May 2016

Keywords:Acetic acidFormic acidPaper degradationStatic headspace gas chromatographyVolatile organic compounds (VOC)Cellulose analysis

* Corresponding author.E-mail address: antje.potthast@boku.ac.at (A. Pott

http://dx.doi.org/10.1016/j.polymdegradstab.2016.05.00141-3910/© 2016 Elsevier Ltd. All rights reserved.

a b s t r a c t

Collections of art on paper, libraries and archives in general, are striving for optimal storage conditionsfor cultural heritage in their deposits and exhibition facilities. Therefore, properly estimating the risk ofvolatile organic compounds emitted from historic and recent storage materials on paper-based collec-tions is of utmost importance. Applying an optimized static headspace gas chromatography and massspectrometry approach combined with cellulose-specific analysis provides insight into the potentialdegradation effects and mechanisms acting on two different paper samples (Whatman No.1 and historicrag paper) used as degradation indicators. Acetic and formic acid, two powerful volatile organic acids,were quantified with differing amounts, depending on the type of historic storage material tested. Molarmass and carbonyl group content were used to monitor cellulose degradation of the indicator papers.Combining these results in 3D plots helps to visualize synergies that evolve from mutual emission ofacetic and formic acids. In addition, choosing between the two paper-based degradation indicators helpsto evaluate different phenomena: Whatman No.1 reacts toward acid hydrolysis and is more sensitive,while rag paper helps to evaluate buffering phenomena as occurring in original materials.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

In the last 10 years, indoor air quality has gained increasingresearch interest. Volatile organic compounds (VOCs) have beenreported to be present at high levels in museum environments.They are associated with damages of cultural heritage objects.Therefore, cultural heritage objects kept in museum collectionsemerged as the focus of research because theymay find themselvesexposed to high levels of VOCs as well. Possible emission sourcesare construction materials, furnishings, materials used for storageor presentation of objects, or the objects themselves [1]. Thenegative effect of slow but constant exposure to detrimental off-gassing is often discovered only after noticeable damage hasoccurred on pieces of historic or artistic value. VOCs are known tocause distinct deterioration phenomena on inorganic materials,such as corrosion on metals or efflorescence on calcareous

hast).

26

materials and glass, which appears shortly after the contamination[2]. However, the decomposition of organic materials, such as pa-per, in the presence of VOCs manifests itself in accelerated hydro-lytic depolymerization, which is only recognizable when it hasalready led to a significant loss of physical stability or a change inthe optical properties, thus in an advanced stage of deterioration.Other factors besides VOCs, such as detrimental or unfavorableclimate and lighting conditions and intrinsic contamination withreactive substances, can contribute to the accelerated depolymer-ization of organic materials. Even in ideal storage conditions,organic materials are subjected to a continuous, albeit slow, processof natural deterioration. For these reasons, it is difficult to evaluatethe effect of VOCs (especially those that are acidic) on organicmaterials, such as paper, and to assess the risk they pose.

Among VOCs, acetic acid has been the focus of research for quitesome time. Dupont and Tetreault [3] found an increased degrada-tion of paper substrates with increasing concentration of aceticacid. The lowest concentration of 3 mg per m3 acetic acid did notlead to a noticeable degradation. An increase from 3 mg per m3 to20 mg per m3 noticeably accelerated the deterioration of the paper,

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117110

and an increase to 200 mg per m3 caused a further, yet decelerated,increase of degradation. In later publications by the same authors,the results are put into a different perspective. Significant cellulosedegradation (decreased DP and loss in mechanical strength) couldnot be shown in the presence of 50 ppm acetic acid, formaldehyde,and furfural. Only formic acid vapor led to measurable cellulosedegradation [4]. According to Pedersoli et al., the impact of volatileacetic acid on the aging process of papers is probably negligible, atleast at levels of 0.7 mg per m3: for neutral paper without anyalkaline and acid components, the absorption of acetic acid woulddecrease the pH of their aqueous extracts from 7 to about 5.5 [5].Zou et al. showed, for paper with a pH above 4.0, that the contri-bution of acid catalysis to the rate of cellulose depolymerization byhydrolysis is minor [6]. In acidic papers, which contain stronger orhigher concentrations of acids, or both, resulting in a pH below 4,the absorption of additional acetic acid is not expected to impactsignificantly on the decay rate, but a final proof is needed. Likewise,papers with a filler material, such as calcium carbonate, would notbe negatively impacted because the alkaline components wouldneutralize organic acids absorbed from the environment, which areno longer available for chemical reactions with the cellulose.

To assess the compatibility of materials for the storage anddisplay of cultural assets, the Oddy-Test [7] is often suggested. TheOddy-Test is an accelerated corrosion test inwhich samples of lead,copper, and silver are exposed to the test material at increasedlevels of humidity and temperature in a tightly closed vessel. If thetest materials emit VOCs, they will corrode the respective metalpieces. This observation leads to a negative evaluation of the testmaterial as potentially harmful to metal-based museum artefacts,for which the test was initially developed. Subsequently, the Oddy-Test has also been used in a more general context to assess thecompatibility of materials with non-metallic, even organic, objects.In addition to its simple and inexpensive nature, one key advantageof the Oddy-Test lies in its low specificity, which enables thedetection of a broad spectrum of volatile compounds and theestimation of the general emission potential of a material. However,a qualitative or quantitative determination of VOCs is not possible.Furthermore, despite the use of the Oddy-Test for 30 years, it hasnot been adequately standardized, and a broad variation of pro-cedures is described in the literature leading to hardly comparableresults [8e11]. Overall, the Oddy-Test has proven valuable as anexclusion test: a material that did not cause any corrosion duringthe test is expected to be safe in combinationwith museum objectsof both inorganic and organic composition [11]. However, whetherthe VOC-induced corrosion of test metals caused by a materialunder extreme conditions can be a meaningful indicator of itsgeneral deteriorating effect on cultural assets, especially organicmaterials under moderate climate conditions, remainsquestionable.

Strli�c et al. [12] proposed a new rapid test method (we refer to asStrli�c -Test) for assessing the compatibility of materials for housingand display of paper-based heritage objects. In this test, a piece ofthe material to be tested is enclosed in a vial with a reference paper(Whatman No. 1 or ground wood paper) and, subsequently, isartificially aged. The impact of the VOCs emitted by the test ma-terials on the reference papers is evaluated by determining thedecrease of the degree of polymerization of the reference paper.

To judge the effect of formic and acetic acid emitted fromdifferent materials used in archival collections on paper, and to seehow strong the effect of the acids really is connected to paperdegradation, we aimed at combining the quantitative emissionanalysis of volatile organic acids with the modified Oddy-Testdeveloped by Strli�c et al. As an example, the collection of thedrawings and prints of Karl Friedrich Schinkel (1781e1841) storedin the Kupferstichkabinett (Museum of Prints and Drawings) of the

Staatliche Museen zu Berlin (National Museums in Berlin) in cabi-nets made of laminated chipboard panels were chosen to demon-strate our concept. To get an initial idea about the VOCs beingpresent at the current storage situation and to assess their con-centrations, qualitative and quantitative determination of the VOCsinside the cabinet were performed using active sampling in com-binationwith gas chromatography andmass spectrometry (GC-MS)and high-performance liquid chromatography (HPLC) respectively.

To estimate the risk of the detected VOCs on the paper-basedartworks stored within the cabinets, and to take measures ac-cording to the determined risk, it seemed to be reasonable to getmore information about the emission and the damaging potentialof the single materials being present at the current housing situa-tion. Acetic acid and formic acid, two VOCs with high destructionpotential, require more detailed research. The emissions of formicand acetic acid from different storagematerials used in the Schinkelcollection were examined using headspace GC-MS. To extend thescope of the test further, the degree of polymerization and theentire molar mass distribution and carbonyl group profiles wereanalyzed. This approach yields additional data on oxidativeprocesses.

2. Material and methods

2.1. Chemicals

Acetic acid, formic acid, N,N-dimethylacetamide, lithium chlo-ride, toluene-d8 and 1,2-dimethoxybenzol (common name “vera-trol”), 2-(2-chloro-6-fluorophenyl)-acetoaceton (“acetoaceton”),methanol-d3, aceton-d6, 4-pentan-1-ol, 3-pentanone, and hexanalwere purchased from Sigma-Aldrich (Sigma-Aldrich Han-delsGmbH, Vienna, Austria). All standards and chemicals were ofHPLC- or GC-grade purity and were used without furtherpurification.

2.2. Sample material

The selection of sample materials reflects the storage situationfound in theMuseum of Prints and Drawings in Berlin and is furthermotivated and supported by previous studies [13]. A historic ragpaper similar to the one Schinkel used, representative materialsfrom the paperboard mounts that were in direct contact with thedrawings and prints, materials that constitute the portfolio inwhich they are stored, and cabinet materials were used for theexaminations. The dates of origin vary from 1800 to 1990 (Table 1).

Whatman filter paper No. 1 and historic rag paper (material no.1; Table 1) were additionally used for the modified Strli�c -Test.

2.3. Analysis of formic acid and acetic acid emission potential bystatic headspace GC-MS with selected-ion monitoring (SHS-GC/SIM-MS)

Sample materials were cut into small pieces (approx. 1e2 mm2)and, depending on the calibration range of the respective VOC,between 100 mg and 250 mg of the samples were transferred in a10 mL crimp-cap vial. The vials were sealed with a crimp capcontaining UltraClean silicon/PTFE septa (45� shore A, 3 mm). Staticheadspace GC-MS analysis of formic acid and acetic acid was per-formed on a 7890A gas chromatograph coupled with a 5975C massselective detector (Agilent Technologies). Samples were introducedby an auto sampler (Agilent GC Sampler 120). Agitator settings:incubation time 1800 s, incubation temperature 120 �C, and syringetemperature 130 �C. Injection: 250 mL, split ratio 5:1, 240 �C. Sep-aration: Fused silica HP-5MS column (30 m � 0.25 mm � 0.25 mm;J&W Scientific, Folsom, CA, USA) using helium as carrier gas.

Table 1Composition and date of origin of analyzed sample materials.

No. Material Composition Date of origin

Schinkel object 1 Historic rag paper similar to Schinkel’s papersstacked up on an open shelf

Low amount of ligneous fibers,surface sizing with gelatin containing alum

ca. 1800

2 Historic paperboard mounts (boards to whichthe drawings are mounted)

Colored rag papers (different coloring materials),high content of ligneous fibers,surface sizing with gelatin-containing alum,starch paste

ca. 1845

Portfolio 3 Cardboard of the historic portfolios Several layers of paper,high content of ligneous fibers

ca. 1845

4 Fabrics (flaps of the historic portfolios) Linen (contains tissues with different colors) ca. 18454a Fabric Inner tissue of portfolio, color: grey4b Fabric Inner tissue of portfolio from boundary area,

color: brown4c Fabric Cover tissue of portfolio, color: red-brown

5 Leather (spine of the historic portfolios) Uncolored sheep skin ca. 1845Cabinet 7 Cabinet shelves Plywood, made of beech ca. 1990

8 Fabric II (cover of the shelves) Corduroy fabric ca. 19909 Cabinet wall Particle boards,

lamination with melamineca. 1990

Controls 4d Fabric-new (flaps of more recent portfolios) Cover tissue of the recent portfolio, color: grey-white ca. 1960/706 Leather-new (spine of more recent portfolios) Uncolored leather, unknown origin ca. 1970

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117 111

Column flow: 1.2 mL min�1. Oven Program: 30 �C for 5 min, then5 �C per min to 70 �C, and 15 �C per min to 280 �C, then hold for5 min. The mass spectrometer (MS) was operated in the electronimpact mode at 70 eV (230 �C, 1.5 � 10�5 Torr). Selected ionmonitoring (SIM)modewas applied for formic acid atm/z 46, aceticacid at m/z 60, and toluene-d8 at m/z 98 to detect peak area. Theaddition of 2.5 mL of a toluene-d8/veratrol mixture (1 mL mL�1) wasused as the internal standard for headspace measurements.

2.4. Peak identification and quantification

Peak assignment and quantification were accomplished usingMSD Chemstation E.2.01.1177 (Agilent Technologies, USA). Peakswere assigned by comparing their retention times andmass spectrawith those of respective reference compounds. Calibration wasbased on peak area and was accomplished using sub-sets of thereference compounds at four different concentration levels.

2.5. Artificial aging

To evaluate the effect of degradation, each material was artifi-cially aged with Whatman No. 1 composed of a minimum a-cel-lulose content of 98% or historic paper (Table 1), according to thetest proposed by Strli�c [12]. In short, a defined amount of indicatorpaper, in our case Whatman No. 1 and historic rag paper, ismounted in the volume of tightly closed 100 mL Schott bottle that,at the bottom, contains a defined mass of the suspicious materialsthat needs to be checked for its volatile emissions. Deviating fromStrli�c et al. [12], the aging was carried out in small (10 mL) crimpcap vials sealed with UltraClean™ silicon/PTFE septa (45� shore A,3 mm) because of the small amount of the available original sam-ples and the fact that the 100 mL Schott flasks showed measurablegas leaking at an elevated temperature. Sample weights of(91.7 ± 0.3) mgWhatman No.1 or historic rag paper were hinged inthe air volume and (183.3 ± 0.3) mg of original samples to be testedwere placed at the bottom of the same vial. Samples of the storagematerials to house the collection and papers were preconditionedin a desiccator at room temperature (25 �C, relative humidity:52.9%) prior to aging for 24 h using saturated magnesium nitratesolution. Samples were aged at 100 �C for 5 days. After aging, thepaper samples were cooled at room temperature for 1 h beforeopening the vial.

2.6. Molar mass and carbonyl content measurement

The hydrolysis and oxidation characterization of paper bydifferent materials were carried out by measuring molar mass andcarbonyl group content. CCOA (carbazole-9-carbonyl-oxy-amine)labeling for measuring the carbonyl group content was performedaccording to earlier reports [14,15]. After labeling, the paper sam-ples were dissolved in N,N-dimethylacetamide containing 9% oflithium chloride (w/v) after solvent exchange. The measurementwas performed on the GPC system consisting of a fluorescencedetector (TSP FL2000), a multiple-angle laser light scattering de-tector (Wyatt Dawn DSP with argon ion laser, l0 ¼ 488 nm), and arefractive index detector (Shodex RI-71). The separation was car-ried out on a set of four PLgel mixed-A LS columns (20 mm,7.5 � 300 mm, Varian/Agilent). N,N-dimethylacetamide containing0.9% lithium chloride (w/v) was used for the mobile phase. Thesystem was operated at a flow rate of 1.0 mL min�1 with an injec-tion volume of 100 mL. The fluorescence of the CCOA label wasdetected at an excitation wavelength set at 290 nm, and theemission wavelength was set at 340 nm. Data evaluation was per-formed with standard Chromeleon 4, Astra 4.73, and GRAMS/32software packages.

3. Results and discussion

It is well known that the chromatographic separation of formicacid and acetic acid in the presence of air is difficult to achieve and,therefore, method optimization with respect to the selection anddilution of a suitable internal standard was needed before actuallyquantifying the two VOCs. Different GC-MS conditions were eval-uated to ensure an optimal chromatographic resolution andsensitivity towards the target compounds.

The optimized analysis for the two organic acids is furtherapplied to quantify their amounts in historic materials that wereused in the Schinkel collection. The impact of these historic mate-rials on cellulose is further elucidated by a modified Strli�c-Test thatis evaluated by molar mass and carbonyl group determinations. Inthe end, the four sets of analytical datae the amount of two organicacids, carbonyl groups, and molar mass e are plotted in 3D graphsto visualize their mutual interaction.

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117112

3.1. Optimization of SHS-GC/SIM-MS analysis

Gas flow conditions from 0.7 to 1.2 mL min�1 were studied withrespect to the chromatographic resolution of formic acid and aceticacid. A gas flow of 1.2 mLmin�1 shows the most suitable separationunder applied conditions. With respect to the temperature condi-tions, the vials were heated to 120 �C to achieve the evaporation ofacetic acid. The syringe that transfers the sample volume into theGC injector was heated to 130 �C. This temperature gradient be-tween sample and syringe was necessary to avoid compoundcondensation within the syringe.

Different compounds were evaluated as internal standard.Methanol-d3 and aceton-d6 were excluded because they eluted tooclose to the target compounds; 4-pentan-1-ol was discardedbecause of the inconsistent behavior of the peak area at differenttemperatures. Acetoaceton was excluded because of inappropriatepeak geometry (tailing). The most appropriate compounds as in-ternal standard were 3-pentanone, hexanal, and toluene-d8.Toluene-d8 was selected from these compounds because it is notgenerated during paper production or in the course of aging pro-cesses of thematerials tested, therefore allowing for the applicationof the method in a wide range of samples.

Veratrol was used as an appropriate solvent for the internalstandard. It eluted after 16.90 min under the conditions chosen.Therefore, the MS detector was turned off for 65 s at this point toavoid overloading with veratrol. An internal standard solution of2.5 mL was added to each sample containing 0.94 mg toluene-d8 permL veratrol. The internal standard had to be diluted because evenwhen the compound was added via a mL-scaled Hamilton syringethrough the septum, the internal standard was too concentrated.The addition of the standard by small-scaled pipettes could not beapplied through the septum because a loss of internal standardsubstance could not be eliminated because of evaporation duringvial heating process.

The mass spectrometer was operated in selected ion monitoring(SIM) mode to enhance the detection limit of the target compounds(Table 2). SIM mode was conducted for formic acid atm/z 46, aceticacid at m/z 60, and toluene-d8 at m/z 98 to detect the peak area(Fig. 1). The selection of ions was carried out according to their m/zintensity and target specificity. This procedure helped to avoidinterference with other compounds and ensured the specificity ofthe selected ion at the current time for the individual targetcompound.

The quantification of formic acid and acetic acid was achieved bydissolving them in veratrol to provide different concentrations. Thedetector response of the acids is normalized to the response oftoluene-d8 used as internal standard. The calibration ranged from0.31 mg to 61 mg per mL veratrol for formic acid and from 0.26 mgto 20.98 mg per mL veratrol for acetic acid and is expressed as theratio between analyte and internal standard (Fig. 2).

3.2. SHS-GC/SIM-MS analysis of acetic acid and formicaciddresults of storage materials

Table 3 combines the emission data of the different materials

Table 2Retention, quantification, andmass spectra characteristics of analyzed standard compounhighest ions of the target compound.

Compound R.T. [min] SIM ions [m/z] LODa

Formic acid 1.70 46 0.10Acetic acid 2.57 60 0.06Toluene-d8 5.72 98 e

a pmol/on column.

obtained by SHS-GC/SIM-MS analysis. Large differences of aceticacid concentrations were observed among the samples. The leathersample that was taken from the cover of the spine of historicportfolios released the highest amount of acetic acid during anal-ysis, followed by the cabinet shelves and the cabinet walls. Thelowest emission of acetic acid was observed at the historic ragpaper. Leather can emit high acetic acid content due to variouscauses. Acetic acid is used in leather production during acidunhairing and in the pickling process, in which the acid binds toanimal protein [16]. Residual amounts of acetic acid could releasefrom the leather over time or the leather may act as absorbers foracetic acid in an appropriate environment. Because of those factors,leather is not suitable for storage or presentation materials. Theother Schinkel exhibits e rag papers similar to the Schinkel objects,the paperboard mount as well as the compounds of the portfolio e

released onlyminor amounts of acetic acid. The cabinet shelves andwalls were identified as the major source of emitting acetic acid[13]. There are actually only two materials that emit detectableamounts of formic acid: fabric 4a and 4b. Formic acid is frequentlyapplied in textile production as an efficient medium for neutrali-zation of cellulosic fibers after alkaline process steps or as a fixingagent for dyes in printing [17]. However, modern textile processingcould include more washing steps and, therefore, new fabrics nolonger contain formic acid.

Acetic acid is frequently emitted from wood from which hemi-celluloses carrying acetylated side groups are the main source.Whereas wood and wooden products are well-known sources ofacetic acid [18e20], pulping generally leads to deacetylation and,therefore, hemicelluloses in paper carry fewer side groups andshould not emit that much acetic acid any longer. However, inliterature, acetic acid was frequently identified as a degradationproduct of paper [4,5].

3.3. Impact of VOCs emitted from historic sample materials onWhatman No. 1 and rag papers

3.3.1. Molar mass and degradation rateTwo different paper types have been chosen for the test of

emissions from different materials. Therefore, Whatman No.1 andhistoric rag paper were used as “aging indicators.” Whatman No. 1paper allows for studying the impact of volatiles on pure cellulose,whereas the other paper, historic rag paper, also includes hemi-celluloses and small amounts of fillers and sizing agents, whichusually lead to more stable behavior in aging tests and resemblesthe actual situation of the papers present in the collection.

With respect to their impact on the selected paper probes, theSchinkel exhibit materials can be classified into two groups (Fig. 3).One group consists of the historic rag paper (1), the cabinet shelves(7), and the cabinet wall consisting of melamine-coated particleboard (9). In this group, the emitted VOCs led to a relatively smalldegradation of cellulose in both the Whatman No. 1 paper and thehistoric rag paper. In contrast, the second group caused a higherdegradation of cellulose in both paper materials. Here, a significantdecrease in molar mass was observed in the presence of thefollowing samples: cardboard of the historic portfolios of 1845 (3),

ds in selected ionmonitoring (SIM) mode. Mass spectra characteristics report the five

LOQa Mass spectra [scan mode] [m/z (% intensity of total spectra)]

0.34 29 (34.0); 46 (31.5); 45 (24.2); 44 (6.1); 32 (2.7)0.21 43 (32.5); 45 (30.1); 60 (21.3); 42 (5.5); 29 (4.0)e 98 (45.4); 100 (29.1); 42 (4.6); 70 (4.3); 99 (3.5)

1 2 3 4 5 6

Rel

ativ

e in

tens

ity

Rel

ativ

e in

tens

ity

Scan mode (TIC)

SIM: m/z 46 (formic acid)

SIM: m/z 60 (acetic acid)

SIM: m/z 98 (toluene-d8)

Retention time [min]

Fig. 1. SHS-GC-MS chromatograms (waterfall orientation) of a standard solution containing formic acid (2.44 mg), acetic acid (2.098 mg), and toluol-d8 (2.35 mg) as internal standardper mL air volume. Analysis of compounds was carried out in selected ion monitoring (SIM) modedformic acid (m/z 46), acetic acid (m/z 60), and toluene-d8 (m/z 98).

Fig. 2. Calibration curves of formic acid (A) and acetic acid (B). Response ratio and concentration ratio were calculated on peak areas between target compound and the internalstandard toluene-d8.

Table 3Amounts of emitted formic acid and acetic acid from Schinkel library samples after heating for 4 h at 120 �C, analyzed by SHS-GC/SIM-MS.

Origin No. Material mg formic acid/g sample mg acetic acid/g sample mg total acids/g sample

Schinkel objects 1 Historic rag paper 0 0.45 0.452 Historic paperboard mounts 0 24.30 24.30

Portfolio 3 Cardboard 0 22.70 22.704a Fabric 27.73 21.53 49.264b Fabric 32.95 19.93 52.884c Fabric 0 19.95 19.955 Leather 0 175.50 175.50

Cabinet 7 Cabinet shelves 0 95.20 95.208 Fabric II 0 10.70 10.709 Cabinet wall 0 48.60 48.60

Controla 4d Fabric-new 0 15.31 15.31

a The control samples represent a modern reference specimen.

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117 113

fabric (4b), and both leather samples (5 and 6).In two cases, historic paperboard mounts (2) and the corduroy

fabric (fabric II; 8), the reactions differed among Whatman No. 1and historic rag papers.

Overall, the trends in Whatman No. 1 and rag paper are com-parable, but the differences in the historic rag paper are less

pronounced because the Mw of the historic rag material is initiallylower than the one of Whatman No. 1, therefore leaving lessaccessible cellulose chains for further chain scission. Also, rag paperis a more complex system with hemicelluloses, sizing agents, andearth alkaline fillers, such as calcium carbonate, that interact withthe VOCs. The acids might be partly neutralized by the fillers and

Un-treated

Aged 5 days

Whatman*2

Historic rag paper (1)

Historic paperboard mounts (2)

Cardboard (3)

Fabric (4b)

Leather (5)

Leather-new (6)

Cabinet shelves (7)

Fabric II (8)

Cabinet wall (9)

0

50

100

150

200

250

300

350

400

Wha

tman

No.

1 [M

w]

Un-treated

Aged 5 days

Whatman*2

Historic rag paper (1)

Historic paperboard mounts (2)

Cardboard (3)

Fabric (4b)

Leather (5)

Leather-new (6)

Cabinet shelves (7)

Fabric II (8)

Cabinet wall (9)

0

50

100

150

200

250

His

toric

rag

pape

r [M

w]

Fig. 3. The impact of historic storage materials on the weight-averaged molar mass of Whatman No.1 (left) and historic rag paper (right). Figures in brackets refer to the samplesdescription in Table 1.

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117114

may be immobilized to a certain extent and kept from interactingwith the polysaccharides by the surface sizing, and if theywere ableto interact with the polysaccharides, hemicellulose (because of itsless-ordered structure) is more prone to degradation than cellulose.A similar observationwas described recently by Kida et al. [21], whofound that hemicellulose-containing Japanese paper reacted lesswith iron ions and acids than Whatman No. 1 filter paper.

For further clarification of this assumption, the rate constants ofthe DP loss in the two cellulose probes were calculated andcompared once exposed to the historic storage materials selectedfrom the Schinkel collection (Fig. 4).

The calculation is based on Equation (1).

k’t ¼ 1DPnt

� 1DPnt0

(1)

where DPnt refers to the DP after aging in the closed vial environ-ment with an emitting source and DPnt0 refers to the DP withouthistoric sample material present. The DP was calculated from theMw determined by molar mass measurement.

The number of chain scissions after 5 days of exposure todifferent materials shows the relative effect of the respective his-toric materials on rag and Whatman No. 1 paper. The degradation

Aged 5 days

Whatman*2

Historic ra

g paper (1)

Historic paperboard mounts (2)

Cardboard (3)

Fabric (4b)

Leather (5)

Leather-new (6)

Cabinet shelves (7)

Fabric II (

8)

Cabinet wall (9

)02468

101214161820

elpmas

o/wt'k/

elpmast'k

Whatman No.1Historic rag paper

Fig. 4. Rate constants k0t of DP loss in Whatman No. 1 and rag paper. Figures inbrackets refer to the samples description in Table 1.

rates are not very different, except at the fabric sample (4b).Interestingly, rag paper tends to exhibit a slightly higher rate formost sample materials (i.e., rag paper degrades faster compared toWhatman No. 1 paper in the comparative test applied). However,care has to be taken with the interpretation because the moisturecontent of historic rag papers has not been evaluated. Therefore,there is a possibility that the difference of moisture may somewhatinfluence cellulose degradation.

There are two significant exceptions from this observation inwhich Whatman No. 1 filter paper degrades faster: historic paperboard mounts (2) and fabric (4b). For the fabric and the emittedformic acid, the Whatman No. 1 paper has little protection to offerand, therefore, is degraded much faster than rag paper.

3.3.2. Carbonyl group content and prevailing degradationmechanism

Fig. 5 shows the total carbonyl group content of both indicatorpapers. This value is determined by selective fluorescence labelingand includes reducing end groups at the cellulose polymers and allother oxidized groups along the cellulose backbone. With regard toan increase of carbonyl groups, the question arises whether thesegroups actually originate from oxidation along the cellulose back-bone or if the increase in carbonyl groupsmostly reflects the freshlygenerated reducing end groups (REG) that form upon chain scis-sion. The differentiation between aldehydes e mainly present asREG e and keto groups would therefore be very attractive. By aneasy mathematical operation, the content of the REG of the cellu-lose is derived from dividing the number of average molar mass Mnby the mass of a single anhydroglucose unit [22]. This calculatedvalue reveals the degree of chain cleavage of the cellulose polymersas a result of acid hydrolysis degradation. By subtracting the totalcontent of all carbonyl groups from the calculated amount of REG,the amount of keto groups (pure keto) is obtained. This value in-dicates the proportion of functional groups modified by oxidativeprocesses and enables an insight into the prevailing degradationprocess, hydrolysis, or oxidation.

Negative values for carbonyl groups (Fig. 5, left) result if the totalamount of carbonyl groups measured is smaller than the theoret-ical value of reducing end groups calculated from Mn. The estima-tion of pure keto groups is only possible if reducing ends arepresent as such. In cotton pulp, such as Whatman No. 1, this oftenseems not to be the case because pre-processing leads to oxidationof REG to the corresponding gluconic acids; therefore, a negativevalue is obtained. Still the pure keto group number yields valuableinformation on the state of oxidation of the cellulose when

Untreate

d Whatm

an

Whatman

(pure)

Historic

rag pap

er (1)

+ Whatm

an*2

Historic

rag pap

er (1)

+ Whatm

an*3

Historic

paperb

oard m

ounts (2)

-1

Historic

paperb

oard m

ounts (2)

- 2

Cardboard

(3)

Fabric

(4b)

Leather

(5)

Leather-

new (6

)

Cabinet

shelv

es (7

)

Fabric

II (8)

Cabinet

wall (9

)-5

0

5

10

15

20

C=O

[μm

ol*g

-1]

CO total REG calc pure keto calc

Whatman paper No. 1

Un-trea

ted hist

oric ra

g paper

(1)

historic

rag pap

er (1)

(pure)

Historic

rag pap

er (1)

+ Whatm

an*2

Historic

rag pap

er (1)

+ Hist

oric ra

g paper

(1)*3

Historic

paperb

oard m

ounts (2)

- 1

Historic

paperb

oard m

ounts (2)

- 2

Cardboard

(3)

Fabric

(4b)

Leather

(5)

Leather-

new (6

)

Cabinet

shelv

es (7

)

Fabric

II (8)

Cabinet

wall (9

)0

5

10

15

20

25

30

35

40

C=O

[μm

ol*g

-1]

C=O total REG calc. pure keto calc.

Rag paper

Fig. 5. Carbonyl group contents in Whatman paper No. 1 (left) and historic rag paper (right). The amount of total carbonyl groups (C]O) determined by fluorescence labeling wasfurther divided into a calculated amount of reducing end groups (REG) and carbonyl groups introduced along the cellulose molecule by oxidation (pure keto). The dotted lines showthe original values calculated for reducing ends and pure keto groups for faster comparison.

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117 115

different treatments are compared. Overall, the carbonyl groupcontent in Whatman No. 1 filter paper is lower than in historic ragpaper. This is because Whatman No. 1 is (a) a new product and (b)does not contain any hemicelluloses that contribute to the amountof carbonyl groups. Whatman No. 1 paper reacts rather sensitivelytowards hydrolytic influences, which can be seen in Fig. 5. All his-toric materials exert an oxidative impact on the indicator papersbecause they do not emit acids only; however, the total degree ofoxidation remains low.

Another interesting general observation is the increment incarbonyl groups. The carbonyl group content in Whatman No. 1ranges from about 6 to 21 mmol g�1, while the carbonyls in ragpaper range from about 21 to 36 mmol g�1. Therefore, the range ofchanges indicated is broader in Whatman No. 1 than in historic ragpapers (i.e., an increment factor of 3.5 in Whatman No. 1 comparedto only 1.7 in historic rag paper). This simply indicates that ragpapers have a higher buffering capacity towards oxidation andhydrolysis, which is very likely accomplished by the precence ofhemicelluloses, which are more easily oxidized and hydrolyzed.Hemicellulose can, to some extent, serve as sacrificial materialprotecting the cellulosic polymer. Interestingly, the presence offormic acid in the fabric sample leads not only to a strong hydro-lysis, as expected, but also to a loss in keto groups, mainly visible inthe rag paper sample, probably because of its reductive nature.Although the plot indicates that the keto groups are lost, we canonly conclude that there is an overall loss on carbonyls (aldehydes/keto) as the number of reducing ends exceeds the total number ofmeasured carbonyls. In any case, we can see a clear correlationbetween emitted volatile organic acids and degradation caused inthe chosen indicator papers.

The benefit of adding the oxidative potential of VOCs emittedfrom historic materials to themodified test is obvious and can serveas an extra parameter in future material analyses. While the testedmaterials from the Schinkel collection did not contain very largeamounts of oxidative VOCs, the potential was visible as oxidationmainly occurred in the presence of historic samples and not for thefresh materials. As we are still dealing with a sum parameter foroxidative and reductive effects influenced by the VOC compositionand also by the paper substrate, we refrain from discussing smallerindividual changes.

3.4. Correlation between VOCs and cellulose analysis

When overviewing the emission potential of all sample mate-rials, the data show emissions of acetic acid from all samplesanalyzed and emission of formic acid from some of them. Similarobservations by NMR analysis of the same samples were reportedby Meyer et al. [13] by a different approach. Based on the quanti-fication of formic and acetic acids in the historic sample materials,the deteriorating impact of fabrics (4) and leather (5) are explain-able and expected. They belong to the group of materials with hightotal acid emission potential and they have been assessed asparticularly critical for storing purposes. However, based on thetotal amount of VOCs emitted, cabinet shelves and their wallsshould have more impact on the deterioration of the indicatorpapers. The co-existence of emitted formic and acetic acids fromthe fabrics caused the strongest degradation effects. This assump-tionwas further studied by plotting the impact of acetic and formicacid mutually in 3D graphs (Fig. 6).

Obviously, the negative effect of the emitted formic acid ontodegradation of the indicator papers is very strong. The higher pKaof formic acid (pKa 3.75) compared to acetic acid (pKa 4.75) has theexpected stronger effect on cellulose hydrolysis compared to aceticacid.

4. Conclusion

The control of storage environments for cultural heritage is ofparamount importance for its preservation. Next to temperature,relative humidity, and light, VOCs have gained increased researchinterest. However, the evaluation of storage materials and of theenvironment in which paper-based objects are kept still needsimprovement of existing methods.

The Strli�c-Test was extended from the previously publishedanalysis of viscosity (i.e., DPmeasurement to a full analysis of molarmass together with profiling of oxidized functionalities in additionto the quantification of acetic and formic acid). To comparedifferent papers, the number of chain scissions is used as a measureof degradation. The carbonyl group profiling allows one to addresswhether oxidation takes place in addition to hydrolytic chaincleavage and to what degree this happens, depending on differentmaterials to which the papers are exposed. By further combining

Fig. 6. Plot of 3D correlation between acetic and formic acid in relation to cellulose degradation expressed as K0t (left) and cellulose oxidation expressed in amount of carbonylgroups (C]O) in mmol g�1 (right) at Whatman No. 1 paper (above) and historic rag paper (bottom).

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117116

the emission of acetic and formic acid, and the modified Strli�c-Test,a more-dimensional relationship of factors leading to cellulosedegradation was obtained.

In Whatman No. 1, paper is used as indicator paper, a moresensitive reaction towards hydrolytic degradation can be expected,and Whatman No. 1 can serve as a more universal indicator, simplyfor deciding whether a material is suitable for storing organicmaterials and not only paper. If the collection is paper based, as inthe present case of the Schinkel collection, a rag paper will betterreflect reality. The rag paper with its different additional matrixcompounds (hemicellulose, sizing agents, alkaline buffers, etc.) willgenerally respond in a way similar to Whatman No. 1 but will alsoshow its buffering capacity toward some VOC cocktails.

The similar degradation characteristic of historic rag paper andWhatman No. 1 paper in the presence of the respective historicsample materials validates the experimental setup and confirmsthe interpretation of the obtained data.

The mutual evaluation of two prominent VOCs e formic acidand acetic acid e and their impact on the two most descriptiveparameters of cellulose integrity e carbonyl group content and

molar mass e allows for an in-depth study of degradation mech-anisms and is able to explain increased deterioration despite VOCemissions that are seemingly low at first sight. We postulate syn-ergistic effects that influence the storage quality of paper-basedobjects induced by the co-emission of different acids. Anincreased understanding of this effect will greatly contribute todesigning safe environments for our paper-based cultural heritage.

Acknowledgement

The financial support by the Federal Ministry of Education andResearch (BMBF), Germany, Grant 01UB910 and by the ChristianDoppler Research Society through the CD-laboratory for “AdvancedCellulose Chemistry and Analytics” is gratefully acknowledged.

References

[1] P.B. Hatchfield, Pollutants in the museum environment: practical strategies forproblem solving in design, in: Exhibition and Storage, Archetype Publications,London, 2002.

[2] C.M. Grzywacz, Monitoring for Gaseous Pollutants in Museum Environments

M. Becker et al. / Polymer Degradation and Stability 130 (2016) 109e117 117

(Tools for Conservation), Getty Conservation Institute, Los Angeles, 2006.[3] A.L. Dupont, J. T�etreault, Cellulose degradation in an acetic acid environment,

Stud. Conservation 45 (2000) 201e210.[4] J. T�etreault, A.L. Dupont, P. B�egin, S. Paris, The impact of volatile compounds

released by paper on cellulose degradation in ambient hygrothermal condi-tions, Polym. Degrad. Stab. 98 (2013) 1827e1837.

[5] J.L. Pedersoli, F.J. Ligterink, M. van Bommel, Non-destructive determination ofacetic acid and furfural in books by solid-phase micro-extraction (SPME) andgas chromatography-mass spectrometry (GC/MS), Restaurator 32 (2011)110e134.

[6] X. Zou, T. Uesaka, N. Gurnagul, Prediction of paper permanence by acceleratedaging II. Comparison of the predictions with natural aging results, Cellulose 3(1996) 269e279.

[7] W.A. Oddy, An unsuspected danger in display, Museums J. 73 (1973) 27e28.[8] L. Robinet, D. Thickett, A new methodology for accelerated corrosion testing,

Stud. Conservation 48 (2004) 263e268.[9] J.A. Bamberger, E.G. Howe, G. Wheeler, A variant Oddy test procedure for

evaluating materials used in storage and display cases, Stud. Conservation 44(1999) 86e90.

[10] L.R. Lee, D. Thickett, Selection of materials for the storage or display ofmuseum objects, Br. Mus. Occas. Pap. 111 (1996) 13e15. The British MuseumCompany Ltd, 46 Bloomsbury Street, London WC1B.

[11] L.R. Green, D. Thickett, Testing materials for use in the storage and display ofantiquities - a revised methodology, Stud. Conservation 40 (1995) 145e152.

[12] M. Strli�c, I.K. Cigi�c, A. Mo�zir, D. Thickett, G. De Bruin, J. Kolar, et al., Test forcompatibility with organic heritage materials - a proposed procedure, e-Preservation Sci. 7 (2010) 78e86.

[13] F. Meyer, D. Hansen, V. Knjasev, G. Volland, The “Schinkel’s legacy” project at

the Kupferstichkabinett Berlin, Restaurator 35 (2014) 81.[14] J. R€ohrling, A. Potthast, T. Rosenau, T. Lange, A. Borgards, H. Sixta, et al.,

A novel method for the determination of carbonyl groups in cellulosics byfluorescence labeling. 2. Validation and applications, Biomacromolecules 3(2002) 969e975.

[15] J. R€ohrling, A. Potthast, T. Rosenau, T. Lange, G. Ebner, H. Sixta, et al., A novelmethod for the determination of carbonyl groups in cellulosics by fluores-cence labeling. 1. Method development, Biomacromolecules 3 (2002)959e968.

[16] A.D. Covington, T. Covington, Tanning Chemistry: the Science of Leather, RoyalSociety of Chemistry, 2009.

[17] W. Bernard, Druck von Textilien: mechanische und chemische Technologie,Springer-Verlag, 2013.

[18] S. Colak, G. Colakoglu, Volatile acetic acid and formaldehyde emission fromplywood treated with boron compound, Build. Environ. 39 (2004) 533e536.

[19] L.T. Gibson, B.G. Cooksey, D. Littlejohn, N.H. Tennent, A diffusion tube samplerfor the determination of acetic acid and formic acid vapours in museumcabinets, Anal. Chim. Acta 341 (1997) 11e19.

[20] M. Risholm-Sundman, M. Lundgren, E. Vestin, P. Herder, Emissions of aceticacid and other volatile organic compounds from different species of solidwood, Holz als Roh - Werkst. 56 (1998) 125e129.

[21] K. Kida, A. Potthast, M. Inaba, N. Hayakawa, The effect of iron ions fromprussian blue pigment on the deterioration of Japanese paper, Restaurator 36(2015) 251.

[22] A. Potthast, U. Henniges, G. Banik, Iron gall ink-induced corrosion of cellulose:aging, degradation and stabilization. Part 1: model paper studies, Cellulose 15(2008) 849e859.

135 

  

  

 Paper VII  

 Comparative hydrolysis analysis of pulps and papers: 

carbohydrate compositions and uncovering supramolecular 

structure of cellulose. 

 

 To be submitted. 

 

   

 

 

 

 

Manuel Becker, Kyujin Ahn, Markus Bacher, Chunlin Xu, Anna Sundberg, Stefan Willför, 

Thomas Rosenau, Antje Potthast 

        

Comparative hydrolysis analysis of cellulose samples: carbohydrate compositions and uncovering 1 

the supramolecular structure of cellulose 2 

 3 

 4 

Manuel Beckera, Kyujin Ahna, Markus Bachera, Chunlin Xub, Anna Sundbergb, Stefan Willförb, Thomas 5 

Rosenaua, Antje Potthasta* 6 

a Department of Chemistry, Division of Chemistry of Renewables and Christian‐Doppler Laboratory 7 

“Advanced  Cellulose  Chemistry  and  Analytics”,  University  of  Natural  Resources  and  Life 8 

Sciences, Muthgasse 18, A‐1190 Vienna, Austria 9 

b Process Chemistry Centre, C/o Laboratory of Wood and Paper Chemistry, Åbo Akademi University, 10 

Porthaninkatu 3, 20500 Turku, Finland 11 

 12 

* Corresponding author: 13 

Phone: +43 1 47654 6454, Fax: +43 1 47654 6059, E‐mail address: antje.potthast@boku.ac.at 14 

 15 

 16 

 17 

 18 

 19 

 20 

 21 

 22 

 23 

 24 

 25 

 26 

 27 

 28 

 29 

 30 

 31 

 32 

 33 

 34 

 35 

Abstract  36 

Knowledge  about  carbohydrate  composition  of  cellulose  samples  is  essential  for  their 37 

characterization  and  further  processing.  Therefore,  a  novel  strategy  for  the  determination  of 38 

inaccessible  cellulose  fractions  is  described.  The  proposed  method  based  on  the  sulfuric  acid 39 

hydrolysis and acid methanolysis,  followed by GC‐MS analysis of  the  corresponding products. The 40 

study  involved the analysis of 42 polysaccharides and cellulose samples. A statistical comparison of 41 

polysaccharides’ monomer patterns with those of cellulosic sample had been carried out to estimate 42 

their polysaccharide distribution. The comparison of hydrolysis data was used to calculate the index 43 

of  recalcitrance  cellulose  fraction  (CRCI)  to  elucidate  the  structure  of  cellulosic  samples  further. 44 

Results of the proposed approach had been compared to NMR data and weight‐average molecular 45 

mass data. The application of  the proposed analysis method  to e‐beam  treated or artificially aged 46 

samples uncovered structural rearrangements and alteration at the monomer level. 47 

 48 

Keywords 49 

Methanolysis, cellulose, crystallinity, electron beam irradiation, hemicellulose, sulfuric acid hydrolysis 50 

 51 

Abbreviations 52 

CG2H = Comparison of glucose data obtained by two hydrolysis methods 53 

CRCI = Index of recalcitrance cellulose fraction  54 

E‐beam = Electron beam irradiation 55 

ECF = Elemental chlorine free 56 

GC‐FID = Gas chromatography ‐ flame ionization detection 57 

GC‐MS = Gas chromatography ‐ mass spectrometry 58 

GPC = Gel permeation chromatography 59 

TMP = Thermo mechanical pulping 60 

 61 

Highlights  62 

Monomer analysis of cellulose and hemicellulose‐containing samples 63 

Comparison of polysaccharides’ subunit pattern with cellulose sample pattern by PCA 64 

Supramolecular  structure  elucidation  of  cellulose  by  sulfuric  acid  hydrolysis  and  acid 65 

methanolysis 66 

Announcing the index of recalcitrance cellulose fraction (CRCI) 67 

Comparison of CRCI (based on proposed CG2H‐method) and CrI (based on NMR analysis) 68 

Application of the proposed analysis method to e‐beam treated or artificially aged samples 69 

 70 

1 Introduction 71 

The  analysis  of  supramolecular  structure  and  carbohydrate  composition  of  cellulose  samples  is 72 

essential  for characterization of  their chemical and physical properties and  further processing. The 73 

sample properties of the monomer composition, crystallinity and amorphous fraction were affected 74 

by  aging  processes  and  electron  beam  (e‐beam)  irradiation  treatments,  inducing  structural 75 

rearrangements.1‐4  The  crystallinity  index  (CrI)  has  been  used  to  describe  the  relative  amount  of 76 

crystalline material  in cellulose.5 The  index based on the assumption, that cellulose contains a two‐77 

phase system with crystalline  (highly ordered) and non‐crystalline (highly disordered) regions.6 This 78 

simple concept of cellulose structure provides an approximation  to  the real structure of cellulose.5 79 

The non‐crystalline portion of cellulose, frequently called amorphous has the meaning of complete 80 

lack  of  order.  Different  hypotheses  were  made,  such  as  the  possibility  of  the  existence  of  a 81 

continuous range of degree of order  in cellulose, since  long chain cellulose molecules are known to 82 

pass both  crystalline  and  amorphous  regions.5 A  less‐ordered or paracrystalline  'in‐core'  structure 83 

with  somewhat  larger  mobility  than  that  of  the  crystalline  cellulose  was  also  announced  in 84 

literature.7  The  quantification  of  the  crystalline  and  amorphous  fraction  of  a  sample  provides 85 

information about their physical and mechanical properties, aging processes and structural integrity.8 86 

The analysis of these fraction ratio was analyzed by measurement methods such as x‐ray diffraction 87 

(XRD)3, 9‐11, Fourier transform infrared spectroscopy (FTIR)12‐13, near infrared (NIR) spectroscopy13 und 88 

solid‐state  13C CP‐MAS NMR7, 10, 14‐15. Recent studies have been  found,  that crystallinity value varies 89 

significantly,  depending  on  the  choice  of  these  measurement  methods,  applied  calculations,  or 90 

both.10‐11, 13  Inconsistent crystallinity values were also observed at XRD and  13C CP‐MAS NMR, only 91 

the  order  of  crystallinity  for  analyzed  cellulosic  samples  was  relatively  consistent  within  each 92 

measurement  technique.10‐11  Another  study  used  FTIR  and  NIR  spectroscopy  to  analyses  the 93 

crystallinity changes during paper aging. These authors observed  significantly different crystallinity 94 

values  of  the  cellulosic  material.  The  errors  in  the  measurement  were  too  large  to  allow  a 95 

quantitative interpretation, so it was not possible to identify any apparent tendency during extended 96 

aging.13 The question, which method provides the most accurate evaluation of cellulose crystallinity 97 

could not be answered. 98 

The  carbohydrate  composition  of  cellulosic  materials  shows  glucose  predominantly  from  the 99 

cellulose fraction and smaller amounts of pentoses, hexoses, and deoxy sugar from the hemicellulose 100 

fraction. Hydrolysis is a necessary tool to cleave the glycosidic bond of polysaccharides. Enzymatically 101 

and acid hydrolysis is known to form single/subunits of a polysaccharide. Acid hydrolysis is a standard 102 

method and  involves  sulfuric acid,  trifluoroacetic acid, and hydrochloric acid under  strong or mild 103 

conditions.16  Monosaccharides  liberated  from  polysaccharides  were  subsequently  analyzed  with 104 

HPLC17, GC18 or NMR19‐20. 105 

Unfortunately,  there are differences of  the hydrolysis rate of monosaccharides and  their glycosidic 106 

linkages. Hydrolysis efficiency is affected by hydrolysis conditions like the type of acid, acid strength, 107 

duration, and temperature.21 The efficiency also depends on parameters associated with the origin of 108 

the sample  like  the  type(s) of glycosidic bond, position of bondage, sugar structure  (α‐/β‐anomer), 109 

chain order degree of cross‐linking and number of hydrogen bonds between sugars.21 110 

Strong  hydrolysis  with  sulfuric  acid  leads  to  a  release  of  reducing monosaccharides. While  acid 111 

methanolysis represents a hydrolysis method under mild conditions, the free monosaccharides were 112 

converted  into  their corresponding methyl glycosides, and  the  carboxyl groups of uronic acids are 113 

esterified with methyl  groups.22  These methyl  glycosides  formed  by  acid methanolysis  lose  their 114 

anomeric configuration and equilibrate to α‐ and ß‐anomers.23 The occurrence of up to four isomers 115 

after  acid methanolysis  reduces  the  risk of  complete peak overlapping,  and  the  constant  ratio of 116 

isomers enables the possibility of compound identification and quantification on one single peak.23‐24 117 

On  the  other  hand,  the  formation  of  more  isomers  per  compound  leads  to  an  increase  in 118 

chromatogram complexity and may decrease the sensitivity of sugar analysis, because of decreasing 119 

signal intensity with increasing amount of isomers.25 The advantages of acid methanolysis compared 120 

to sulfuric acid hydrolysis are much  less degradation of fragile hemicelluloses and the possibility to 121 

quantify uronic  acids.26 However, mild  acid hydrolysis was  reported  to  cleave  crystalline  cellulose 122 

structure  only  slightly,  here  strong  acid  hydrolysis  methods  are  recommended.18  Therefore,  a 123 

combined analysis of total glucose, pentose, and uronic acids  is not available by a single hydrolysis 124 

method. This  leads  to a preference  for  the methodology according  to  the  target of analysis, either 125 

cellulose or hemicellulose content.  126 

In this study, the two hydrolysis methods sulfuric acid hydrolysis and acid methanolysis were applied 127 

to characterize monomer composition of reference polysaccharides and cellulose samples, covering a 128 

wide  range  of monosaccharides  and  sugar  acid  compounds.  The  current  approach  analyzes  the 129 

fraction  of  hemicelluloses,  pectin  and  amorphous  cellulose  by  acid  methanolysis,  followed  by 130 

silylation and gas chromatography  (GC) according  to Sundberg et al.18 The determination of whole 131 

cellulose/total glucose content  is carried out using sulfuric acid hydrolysis according to Bose et al.17 132 

Thus, the subtraction of liberated glucose units by acid methanolysis from glucose units determined 133 

by acid hydrolysis has been considered as  the glucose content only  in  the cellulose  fraction of  the 134 

respective sample.27‐29 This method has been accepted to measure the cellulose fraction of pulp and 135 

paper samples.27‐28, 30  136 

The  glucose  content  determined  by  the  two  hydrolysis methods  sulfuric  acid  hydrolysis  and  acid 137 

methanolysis was  implemented  to  quantify  the  index  of  recalcitrance  cellulose  fraction  (CRCI)  to 138 

elucidate  the  supramolecular  structure  of  cellulose  of  different  cellulose  samples.  The  proposed 139 

analysis method was conducted at experimental series to answer the question, how the application 140 

of electron beam irradiation (e‐beam) treatments and aging processes affect cellulose samples about 141 

structural  rearrangements  via  CRCI  and  monomer  composition.  A  simultaneous  solid  state  NMR 142 

analysis of samples was performed to compare CRCI results, calculated by comparison of glucose data 143 

obtained by  two hydrolysis methods  (CG2H). The obtained data  from  the different methods were 144 

discussed.  The  proposed method  had  been  applied  to  samples,  either  treated with  an  electronic 145 

beam  (e‐beam) or artificially aged and compared with similar weight‐average molecular mass  (Mw) 146 

results by GPC analysis.  Lastly, we  could  successfully propose versatility of  comparative hydrolysis 147 

methods towards insight into crystallinity of cellulose. 148 

 149 

 150 

 151 

2  Material and methods 152 

2.1 Chemicals and reagents 153 

The  reference  compounds  D‐(‐)‐arabinose,  D‐(+)‐galactose,  D‐(+)‐glucose,  D‐(+)‐mannose,  L‐(+)‐154 

rhamnose  (6‐deoxy‐mannose),  D‐(+)‐xylose,  D‐(+)‐galacturonic  acid  monohydrate  (GalA),  D‐155 

glucuronic acid (GlcA), the internal standard sorbitol , anhydrous pyridine, acetic acid, ethyl acetate, 156 

sodium carbonate  (Na2CO3), N,O‐bis(trimethylsilyl)trifluoroacetamide  (BSTFA),  trimethylchlorosilane 157 

(TMCS)  and  4‐(dimethylamino)pyridine  (DMAP) were  purchased  from  Sigma‐Aldrich/Fluka  (Sigma‐158 

Aldrich  Schnelldorf, Germany). All  standards,  chemicals,  and  reagents were of GC‐grade  and used 159 

without further purification. 160 

 161 

 162 

2.2 Material 163 

All polysaccharides (Tab. 1) and cellulose samples (Tab. 2) were freeze‐dried prior analysis. For acid 164 

methanolysis, sample amounts of 1‐2 mg of polysaccharides and 10 mg (±2 mg) of cellulose samples 165 

were used for analysis. The sulfuric acid hydrolysis was conducted with sample amounts of 40 mg (±1 166 

mg) per polysaccharide and cellulose sample for analysis.  167 

 168 

Table 1: Sample list of polysaccharide standards 169 

No  Polysaccharides  Code  Origin  Producer 

P1  Arabinan  AS  Sugar Beet  Megazyme 

P2  Arabinan – Debranched  DA  Sugar Beet  Megazyme 

P3  Arabinan ‐ Linear 1,5‐α‐L‐  LA  Sugar Beet  Megazyme 

P4  Galactan  GG  Potato  Megazyme 

P5  Pectic Galactan  GL  Lupin  Megazyme 

P6  Pectic Galactan  GP  Potato  Megazyme 

P7  Galactomannan   GC  Carob  Megazyme 

P8  Galactomannan  GB  Locust bean  Sigma 

P9  Glucogalactomannan  GS  Spruce  Åbo Akademi University 

P10  Glucomannan  GK  Konjac  Megazyme 

P11  Gum arabic  GA  Acacia tree  Sigma‐Aldrich 

P12  Inulin  IN  Dahlia tubers  Sigma‐Aldrich 

P13  Pectin Classic AU202  AU  Apple  Herbstreith & Fox KG 

P14  Pectin Classic CM201  CM  Citrus  Herbstreith & Fox KG 

P15  Pectin, esterified  PC  Citrus   Sigma 

P16  Polygalacturonic acid  PG  Orange  Sigma 

P17  Rhamnogalacturonan  RG  Soy Bean  Megazyme 

P18  Stachyose  ST  Stachys tuberifera Sigma‐Aldrich 

P19  Xylan  LG  Beech  Lenzing AG 

P20  Xylan  XB  Birch  Sigma 

P21  Xyloglucan  XG  Tamarind  Megazyme 

 170 

 171 

Table 2: Sample list of cellulose samples. 172 

No.  Sample  Code  Origin  Producer 

F01  Cotton Linters  CL  Cotton  Buckeye 

F02  Wheat bran  BR  Wheat  unknown 

Hardwood pulp 

F03  Hardwood ‐ Kraft pulp  HK  Birch  unknown 

F04  Hardwood ‐ Sulfite pulp  HS  Beech  Lenzing AG 

Softwood pulp 

F05  Softwood ‐ Kraft pulp  SK Spruce (70%), pine (30%) 

Södra 

F06  Softwood ‐ Sulfite pulp  SS  Spruce  Domsjö 

F07  Softwood ‐ TMP  TM  Spruce Åbo Akademi University 

F08  Eucalyptus paper pulp ‐ Kraft pulp  EC  Eucalyptus  ENCE 

F09 Eucalyptus paper pulp ‐ Kraft pulp – e‐beam treated*** 

EC‐E  Eucalyptus  ENCE 

F10  Hemp paper pulp ECF  HC  Hemp  Celesa 

F11  Hemp paper pulp ECF ‐ e‐beam treated***  HC‐E  Hemp  Celesa 

Paper 

F12  Book 1 (1951)  B1 

 

 

F13  Book 2 (1912)  B2 

F14  Book 3 (1892)  B3 

F15  Book 4 (1860)  B4 

F16  Mulberry tree paper  MB  Mulberry 

F17  Paper sample (historical)  PH 

 

F18  Rag paper (historical)  RH 

F19  Rag paper (modern)  RM 

F20  Rag paper (modern) ‐ artificially aged*  RM‐A 

F21  Rag paper (modern) ‐ e‐beam treated**  RM‐E 

*  =  Accelerated aging conditions were 80 °C and 65% RH for two weeks. **  =  Beta‐radiation with 60 kGy was applied. ***  =  Beta‐radiation with 120 kGy was applied. 

 173 

 174 

2.3 Sulfuric acid hydrolysis 175 

The sulfuric acid hydrolysis was conducted using a two‐step acid concentration treatment according 176 

to Bose et al. (2009) modified for GC‐MS analysis.17 During the primary hydrolysis step, 1.5 ml of 72% 177 

H2SO4 was added to the sample at room temperature for two h. For the subsequent hydrolysis step, 178 

2 ml H2O was added to obtain a 40% H2SO4 and heated in an oven at 80°C for one h. The hydrolysis 179 

solution was cooled down  in an  ice bath and stored at 4°C overnight  to polymerize  lignin  fraction. 180 

Then,  7 ml  of  an  internal  standard  solution  (150 mg  Sorbitol  /  100 ml  H2O) were  added  to  the 181 

hydrolysis solution. An aliquot of 1.5 ml was neutralized using Na2CO3 until no more bubbles appear 182 

(approx. 290 mg). The liquid was filtered (0.45 µm, 13 mm diameter) into a new GC vial and the pH‐183 

value was checked with pH‐paper and adjusted to pH 7 with 1‐2 drop(s) of acetic acid. 184 

 185 

 186 

2.4 Acid methanolysis 187 

The dried sample materials were depolymerized through acid methanolysis by addition of 2 mL 2 M 188 

HCl  in  anhydrous methanol  according  to  Sundberg  et  al.  (1996).18  A  calibration  solution  (1  ml) 189 

containing 0.1 mg/ml of sugar monomers and uronic acids, except 4‐O‐methyl glucuronic acid (4‐O‐190 

MeGlcA), was also subjected to acid methanolysis under same conditions. The samples were kept at 191 

100°C for three h. When reaction mixtures reached ambient temperatures, samples were neutralized 192 

with 100 µl pyridine. Afterward, 1 ml of an internal standard solution (0.1 mg sorbitol/ml methanol) 193 

was added to the samples. Acid methanolysis samples and calibration mixtures were evaporated in a 194 

water bath  (50°C) under nitrogen until dryness and  further dried  in a vacuum desiccator at  room 195 

temperature for 30 min. 196 

 197 

 198 

2.5 Per‐trimethylsilylation (TMS) of monosaccharides obtained from sulfuric acid hydrolysis or acid 199 

methanolysis reaction 200 

The  lyophilized hydrolysates, calibration mixtures, and reference compounds were dissolved  in 200 201 

µL pyridine and  incubated at  room  temperature  for 30 min  to ensure proper  isomer equilibration. 202 

Subsequently, 200 µL of a solution of 1.5 mg×mL‐1 DMAP as a silylation catalyst in pyridine was added 203 

to the mixture. Derivatization was accomplished by adding 200 µL BSTFA (containing 10% TMCS) and 204 

heating  the mixture  to 70°C  for 2 hours according  to Becker et al.31 The derivatized samples were 205 

kept at ‐20°C until analysis. 206 

 207 

 208 

 209 

2.6 GC‐FID and GC‐MS analysis of TMS‐derivatized hydrolysis or acid methanolysis products 210 

The  derivatized  samples  were  diluted  with  ethyl  acetate  (600  µl)  and  filtered  before  injection. 211 

Aliquots of 0.2 µl derivatized sample were  introduced  into the splitless  injector using an Agilent GC 212 

Sampler 120. GC‐MS Analysis was performed on an Agilent 7890A gas chromatograph coupled with 213 

an Agilent 5975C mass selective detector. 214 

GC‐FID  analysis  was  performed  on  a  Perkin  Elmer  Autosystem  XL  gas  chromatograph.  Analysis 215 

parameters based on Sundberg et al. 18. Column: HP‐1 (25 m × 0.20 mm x 0.11 µm; J&W Scientific, 216 

Folsom,  CA,  USA);  carrier  gas:  hydrogen,  injector  temperature:  250°C;  column  flow:  0.8 ml/min, 217 

pressure 14 psi; oven program: 100°C (1 min), 4°C/min to 170°C, 12°C/min, 300°C (7 min); detector 218 

temperature: 310°C. Aliquots of 1 µl derivatized samples were  injected  into GC  in split mode  (split 219 

ratio 1:25) by the autosampler. 220 

General GC‐MS  analysis  conditions:  Column: HP‐5MS  (30 m  ×  0.25 mm  x  25  µm;  J&W  Scientific, 221 

Folsom, CA, USA); carrier gas: helium, MS: EI mode, 70 eV, source pressure: 1.13∙10‐7 Pa, purge flow: 222 

36.3 ml/min, 0.6 min; source temperature: 230°C. Scan range was set from 43 to 950 Da.  223 

Parameters  for  analysis  of  acid methanolysis  products:  injector  temperature:  140°C  (30°C/min  to 224 

260°C); column  flow: 0.9 ml/min; oven program: 140°C  (1 min), 4°C/min  to 210°C,  then 30°C/min, 225 

260°C (5 min); inlet pressure 78.361 kPa.  226 

Parameters for analysis of sulfuric acid hydrolysis products: injector temperature: 150°C (30°C/min to 227 

260°C); column  flow: 0.9 ml/min; oven program: 120°C  (2 min), 5°C/min  to 230°C,  then 20°C/min, 228 

260°C (10 min); inlet pressure 78.361 kPa.  229 

 230 

 231 

2.7 Peak identification and quantification 232 

Peak assignment, data acquisition, and quantification of sulfuric acid hydrolysis or acid methanolysis 233 

products were accomplished using MSD Chemstation E.2.01.1177 (Agilent Technologies, USA). Peaks 234 

were  assigned by  comparing  their  retention  times  and mass  spectra with  those of  corresponding 235 

reference compounds (Fig. 1). Calibration factors were determined from the carbohydrate standard 236 

solution  after  sulfuric  acid  hydrolysis  or  acid methanolysis  by  the  ratio  of  the  total  area  of  the 237 

different  sugar unit peaks  to  the area of  the  sorbitol peak. The  calibration  factor  for 4‐O‐MeGlcA, 238 

measured during acid methanolysis, which is not commercially available in pure form, was assumed 239 

to  be  the  same  as  for  GlcA.  Up  to  four  replications  per  samples were  analyzed,  and  values  are 240 

deviating  from  the  average  of  15%  or more were  discarded. All  the  results were  calculated  on  a 241 

freeze‐dried basis. 242 

 243 

 244 Figure  1:  Left:  acid  methanolysis  of  a  carbohydrate  standard  solution  containing  arabinose, 245 rhamnose, xylose, galactose, glucose, mannose, galacturonic acid (GalA), and glucuronic acid (GlcA); 246 right:  sulfuric acid hydrolysis of a carbohydrate  standard  solution containing arabinose,  rhamnose, 247 xylose, galactose, glucose, and mannose. 248  249 

 250 

2.8 Solid‐state NMR 251 

All  solid‐state  NMR  experiments  were  performed  on  a  Bruker  Avance  III  HD  400  spectrometer 252 

(resonance  frequency of  1H of 400.13 MHz, and  13C of 100.61 MHz,  respectively), equipped with a 253 

4 mm dual broadband CP/MAS probe. The samples were swollen in deionized water overnight before 254 

measurement. 13C spectra were acquired by using the TOSS (total sideband suppression) sequence at 255 

ambient temperature with a spinning rate of 5 kHz, a cross‐polarization (CP) a contact time of 2 ms, a 256 

recycle delay of 2 s, SPINAL 64 1H decoupling and an acquisition time of 43 ms. Chemical shifts were 257 

referenced externally against the carbonyl signal of glycine with δ = 176.03 ppm. The acquired FIDs 258 

were apodized with an exponential  function  (lb = 1 Hz) before Fourier  transformation. Peak  fitting 259 

was performed with the Dmfit programme. The deconvolution of cellulose fractions was performed 260 

by spectral fitting according to the model and method of Larsson et al. (1997). Deconvolution fitting 261 

and assignment of fractions were conducted according to Larsson et al. (1997).7 262 

 263 

 264 

2.9 GPC analysis of cellulose samples 265 

The characterization of samples was carried out by measuring weight‐average molecular mass (Mw). 266 

The paper samples were dissolved in N,N‐dimethylacetamide containing 9% of lithium chloride (w/v). 267 

The  measurement  was  performed  on  the  GPC  system  consisted  of  fluorescence  detector  (TSP 268 

FL2000), multiple‐angle laser light scattering detector (Wyatt Dawn DSP with argon ion laser (λ0 = 488 269 

nm)], and a refractive index detector (Shodex RI‐71). The separation was carried out on a set of four 270 

PLgel mixed‐ALS columns (20 µm, 7.5×300 mm, Varian/Agilent)). N,N‐dimethylacetamide containing 271 

0.9% lithium chloride (w/v) was used for mobile phase. The system was operated at a flow rate of 1.0 272 

ml/min  with  an  injection  volume  of  100  µl.  Data  evaluation  was  performed  with  standard 273 

Chromeleon 4, Astra 4.73, and GRAMS/32 software packages. 274 

 275 

 276 

 277 

3 Results and discussion 278 

3.1 Acid methanolysis  279 

The  polysaccharides  showed  a  recovery  of  54.96%  –  102.19%  of  released  sugar  units  after  acid 280 

methanolysis  (Fig.  2),  except  Inulin  with  4.29%.  Inulin  showed  a  low  amount  of  released 281 

carbohydrates during analysis, because the significant polysaccharide fructose, which is also available 282 

in stachyose, decomposed during acid methanolysis, while the glucose from the terminal end of the 283 

inulin polymer remained.  284 

The primary monosaccharides  identified with acid methanolysis  in cellulose samples were glucose, 285 

xylose,  and mannose,  followed by  arabinose  and  galactose  (Fig.  3). Recovery  rates below  22.71% 286 

were observed after depolymerization, (except wheat bran with 37.1%). 287 

 288 

Arabin

an (S

ugar

Bee

t)

Arabin

an -

Debra

nche

d (S

ugar

Bee

t)

Arabin

an -

Linea

r 1,5

-a-L

- (Sug

ar B

eet)

Galacta

n (P

otat

o)

Pectic

Gala

ctan

(Lup

in)

Pectic

Gala

ctan

(Pot

ato)

Galacto

man

nan

(Car

ob)

Galact

oman

nan

(Loc

ust b

ean)

Galactu

gluco

man

nan

(Spr

uce)

Glucom

anna

n (K

onjac

)

Gum a

rabic

(Aca

cia tr

ee)

Inull

in (D

ahlia

tube

rs)

Pectin

Clas

sic A

U202

(App

le)

Pectin

Clas

sic C

M20

1 (C

itrus

)

Pectin

, este

rified

(Citr

us)

Polyga

lactu

ronic

acid

(Ora

nge)

Rham

noga

lactu

rona

n (S

oy B

ean)

Stach

yose

(Sta

chys

tube

rifer

a)

Xylan

(Bee

ch)

Xylan

(Birc

h)

Xyloglu

can

(Tam

arind

)0

10

20

30

40

50

60

70

80

90

100

Mol

ar

ratio

of c

arb

ohyd

rate

s [%

]

Arabinose Galactose Glucose Mannose Rhamnose Xylose Galacturonic acid Glucuronic acid 4-O-MeGlcA

0

10

20

30

40

50

60

70

80

90

100105

% Recovery rate

Re

cove

ry r

ate

(ca

rboh

ydra

tes

fro

m fi

ber

, )

[%]

 289 Figure 2: Carbohydrate composition of different polysaccharides, expressed as molar ratio in % of the 290 total  sugar  amount  (columns)  and  recovery  rate  of  released  carbohydrates  in  %  ()  by  acid 291 methanolysis and GC‐FID or GC‐MS analysis. Further information about samples is shown in Table 1 292 and 2. 293  294 

 295 

Cotto

n Lin

ters

Whe

at b

ran

Hardw

ood

- Kra

ft pu

lp

Hardw

ood

- Sulf

ite p

ulp

Softw

ood

- Kra

ft pu

lp

Softw

ood

- Sulf

ite p

ulp

Softw

ood

- TM

P

Eucalyp

tus p

aper

pulp

Eucalyp

tus p

aper

pul

p - e

-bea

m

Hemp

pape

r pulp

Hemp

pape

r pulp

- e-

beam

Book 1

(195

1)

Book 2

(191

2)

Book 3

(189

2)

Book 4

(186

0)

Mulb

erry

tree

pap

er

Paper

sam

ple

(hist

orica

l)

Rag p

aper

(hist

orica

l)

Rag p

aper

(mod

ern)

Rag p

aper

(mod

ern)

- ag

ed

Rag p

aper

(mod

ern)

- e-

beam

0

20

40

60

80

100M

ola

r ra

tio o

f ca

rboh

ydra

tes

[%]

Arabinose Galactose Glucose Mannose Rhamnose Xylose Galacturonic acid Glucuronic acid 4-O-MeGlcA

0

10

20

30

40

% Recovery rate

Rec

ove

ry r

ate

(car

bohy

drat

es f

rom

fib

er,

) [%

]

 296 Figure 3: Carbohydrate composition of cellulose samples, expressed as molar ratio  in % of the total 297 sugar amount (columns) and recovery rate of released carbohydrates in % () by acid methanolysis 298 and GC‐FID or GC‐MS analysis. Further information about the samples is shown in Table 1 and 2. 299  300 

 301 

3.1.1 Polysaccharides vs. cellulose samples of different origin 302 

While  the  kind  of  carbohydrate  subunits  of  polysaccharides  is  already  known,  the  relationship 303 

between subunits within a polysaccharide as well as between  individual polysaccharides enhanced 304 

sample characterization for further analysis and was used to compare with celluloses from different 305 

sources. The molar  ratio data of  sugar units of depolymerized polysaccharides obtained  from acid 306 

methanolysis were compared with corresponding data of cellulose samples. The aim was to classify 307 

observations to uncover chemical structures of different cellulose samples by resolving  information 308 

about  their  polysaccharide  composition.  A  principal  component  analysis  (PCA) was  chosen  as  an 309 

unsupervised clustering method to conduct the comparative analysis of acid methanolysis data. The 310 

results  of  samples  are  shown  in  Fig.  4  as  biplot  based  on  the molar  ratio  pattern  of  analyzed 311 

monosaccharides.  The  circles  represent  the  scores  of  polysaccharide,  the  squares  the  cellulose 312 

samples on the principal components. Vectors represent the coefficients of the molar ratio of sugar 313 

units on the principal components. 314 

The  relationship  of  sugar  units within  a  polysaccharide  can  be  observed  by  the  position  of  score 315 

points to the vector lines. The closer a point to a vector, the higher the influence of the molar ratio of 316 

the associated sugar unit in the polysaccharide, for example, observed at xylan from beech (LG) and 317 

birch  (XB), which are close to the xylose vector. The opposite direction minors the  influence, while 318 

points  projecting  in  the middle  of  two  vectors  show  an  average  influence.  All  cellulose  samples 319 

(scores) are displayed on  the negative  side  (left) of  the principal  component  (PC) 1 axis. They are 320 

located between the mannose and xylose vector and close to the glucose vector (loadings), notifying 321 

that glucose, xylose, mannose, and 4‐O‐MeGlcA are  the most  influencing carbohydrates within  the 322 

cellulose  samples, while  the  sugar units arabinose,  rhamnose, galactose as well as glucuronic acid 323 

and galacturonic acid  indicating  less  importance, because  they are orientated on  the positive  side 324 

(right) of PC1 axis. According to Fig. 4, cellulose samples are divided  into three different groups: (1) 325 

between the xylose and glucose vector, closer to the xylose vector; (2) between xylose and glucose 326 

vector, closer to the glucose vector; and (3) between glucose and mannose vector. Group 1 consists 327 

of  eucalyptus paper pulp  samples,  the hardwood  kraft  pulp, mulberry paper  and book  1  sample. 328 

Group 2 consists of rag paper and hemp paper pulp samples, the hardwood sulfite pulp and historical 329 

cellulose sample. Group 3 consists of softwood pulp samples (incl. TMP), cotton linters, and historical 330 

rag  paper  and  book  2  to  4  samples.  The  aged  and  e‐beam  treated  samples  are  shifting  in  the 331 

direction  of  the  glucose  vector,  indicating  that  the  importance  of  xylose  decrease  and  increase 332 

towards  glucose.  The  polysaccharides  form  a  large  scatter  within  the  PCA.  They  are  distributed 333 

according to the most representative monosaccharides in the polymer. The extensive distribution of 334 

the polysaccharide samples enables a functional characterization of each polymer and results of both 335 

the  diversity  of  the  studied  polysaccharides  and  because  uronic  acids  can  be  analyzed  by  acid 336 

methanolysis. 337 

A similar molar ratio pattern of polysaccharides and cellulose samples can be a result either (1) due 338 

to containing the corresponding polymer or (2) the combination of different polymers in the sample 339 

corresponds  after  depolymerization  approximately  to  the  same molar  ratio  of  a  polysaccharide. 340 

Groups,  representing  cellulose  samples  together  with  their  corresponding  polysaccharides  are 341 

“Glucogalactomannan (GS) – Softwood TMP” and “Xyloglucan (XG) – Wheat bran”. While the group 342 

of “Inulin  (IN) – Eucalyptus paper pulp, Mulberry  tree paper, etc.” consists of samples with similar 343 

molar ratio pattern, due to a combination of different polymers.  344 

A difference of  the chemical composition between historical and modern rag paper was observed. 345 

The historic rag paper revealed at acid methanolysis a very  low xylose yield (1.83 mg/g) and higher 346 

mannose content  (4.46 mg/g) compared  to modern  rag paper  (xylose: 16.43 mg/g; mannose: 0.95 347 

mg/g). This  indicates a different origin of  fiber material and has as a consequence that historic rag 348 

paper excluded from further sample comparisons.  349 

 350 

 351 

Figure 4: PCA biplot on scores (polysaccharides: ●, cellulose samples: ) and  loadings (grey) based 352 on  their molar  ratio of  carbohydrate  composition, obtained by acid methanolysis and  followed by 353 GC‐FID  or  GC‐MS  analysis.  See  Table  1  and  2  for  abbreviations  and  further  information  about 354 samples. 355  356 

 357 

3.2 Sulfuric acid hydrolysis 358 

The sulfuric acid hydrolysis data of polysaccharides revealed a specific carbohydrate pattern for each 359 

polymer  according  to  their  corresponding monosaccharide  distribution  (Fig.  5).  Galactomannans, 360 

glucomannans, xylans, galactans, and arabinans showed  their sugar units as significant compounds 361 

after analysis.  In comparison  to acid methanolysis results of polysaccharides, most of  the analyzed 362 

samples  showed different  recovery values, and none of  the  four uronic acids were observed after 363 

sulfuric  acid  hydrolysis  of  polysaccharides  (Fig.  5).  The  recovery  values  of  polysaccharides  after 364 

sulfuric acid hydrolysis ranged from 0.22% to 68.96%. The highest recovery values of polysaccharides 365 

at sulfuric acid hydrolysis showed the galactomannan, glucomannan, and xylan, which correspond to 366 

the acid methanolysis results. The different recovery values were  influenced by several aspects. On 367 

the one hand, very  low amounts of  individual carbohydrate could be destroyed during strong acid 368 

hydrolysis  treatment  of  a  sample.  In  this  case,  the  destroyed  monosaccharides  cannot  be 369 

compensated  by  calibration,  and  no  information  about  the  presence  of  the  carbohydrate  in  the 370 

sample would be available. Regarding  the data analysis of arabinans,  influences of different  chain 371 

structures and polysaccharide  linkage on monosaccharide recovery values can be observed. On the 372 

other  hand,  the  sugar  acid  content  of  up  to  56.8%  (at  polygalacturonic  acid) measured  by  acid 373 

methanolysis  was  entirely  absence  under  current  sulfuric  acid  hydrolysis  conditions,  due  to 374 

-2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 5,5-2,5

-2,0

-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5

4,5

5,0-0,4 -0,2 0,0 0,2 0,4 0,6 0,8

-0,6

-0,4

-0,2

0,0

0,2

0,4

0,6

AS

DA

LA

GG

GL GP

GCGB

GS

GK

GAIN

AU

CMPC

PG

RG

ST

LG

XB

XG

CL

BR

HK

HS

SK

SSTM

EC

EC-E

HCHC-E

B1

B2

B3

B4

MB

PH

RH

RMRM-ARM-E

Arabinose

GalactoseGlucose

Mannose

Rhamnose

Xylose

Galacturonic acid

Glucuronic acid

4-O-MeGlcA

Prin

cipa

l Com

pone

nt 2

(sc

ores

)

Principal Component 1 (scores)

= Pulp and paper samples= Polysaccharides

Prin

cipa

l Com

pone

nt 2

(lo

adin

gs)

Principal Component 1 (loadings)

decarboxylation  and  destruction  of  the  sugar  acid molecule.  This  explained  the  shallow  recovery 375 

values  of  the  citrus  and  apple  pectin,  and  the  polygalacturonic  acid,  among  the  analyzed 376 

polysaccharides. While  the  carbohydrate  pattern  and  recovery  yield  of  inulin were  equal  to  acid 377 

methanolysis data, due to the destruction of fructose during acid treatment.  378 

 379 

Arabin

an (S

ugar

Bee

t)

Arabin

an -

Debra

nche

d (S

ugar

Bee

t)

Arabin

an -

Linea

r 1,5

-a-L

- (Sug

ar B

eet)

Galacta

n (P

otat

o)

Pectic

Gala

ctan

(Lup

in)

Pectic

Gala

ctan

(Pot

ato)

Galacto

man

nan

(Car

ob)

Galacto

man

nan

(Loc

ust b

ean)

Galactu

gluco

man

nan

(Spr

uce)

Glucom

anna

n (K

onjac

)

Gum a

rabic

(Aca

cia tr

ee)

Inull

in (D

ahlia

tube

rs)

Pectin

Clas

sic A

U202

(App

le)

Pectin

Clas

sic C

M20

1 (C

itrus

)

Pectin

, este

rified

(Citr

us)

Polyga

lactu

ronic

acid

(Ora

nge)

Rham

noga

lactu

rona

n (S

oy B

ean)

Stach

yose

(Sta

chys

tube

rifer

a)

Xylan

(Bee

ch)

Xylan

(Birc

h)

Xyloglu

can

(Tam

arind

)0

10

20

30

40

50

60

70

80

90

100M

ola

r ra

tio o

f car

bohy

drat

es [%

] Arabinose Galactose Glucose Mannose Rhamnose Xylose

10

20

30

40

50

60

70

80

90

100

% Recovery rate

Rec

over

y ra

te (

carb

ohy

drat

es fr

om

fib

er,

) [%

]

 380 Figure 5: Carbohydrate composition of different polysaccharides, expressed as molar ratio in % of the 381 total sugar amount (columns) and recovery rate of released carbohydrates in % () by sulfuric acid 382 hydrolysis and GC‐MS analysis. Further information about the samples is shown in Table 1 and 2. 383  384 

 385 

The  sulfuric  acid  hydrolysis  data  revealed molar  ratios  of  glucose  ranging  from  39.6%  to  100%, 386 

followed by xylose (up to 40.3%), and mannose (up to 18.9%) within the analyzed cellulose samples. 387 

The analysis  showed molar  ratio values below one percent at  rhamnose and galactose. Moderate 388 

amounts  of  mannose  were  observed  in  softwood  TMP,  originating  from  the  high  content  of 389 

glucogalactomannan and glucomannan polysaccharides. Wheat bran showed moderate amounts of 390 

arabinose,  from  high  amounts  of  the  arabinoxylan  polysaccharide.  The  sulfuric  acid  hydrolysis  of 391 

different  cellulose  samples  showed  higher  recovery  rates  compared  to  acid  methanolysis  data 392 

because  the  sulfuric  acid  hydrolysis  affected  besides  the  amorphous  fraction  also  the  crystalline 393 

fraction of cellulosic samples and takes this fraction into account of the analysis. So, recovery rates in 394 

most cellulose samples ranged  from 51.87%  to 72.51%, except softwood TMP  (34.45%) and wheat 395 

bran (40.32%; Fig. 6). 396 

Cotto

n Lin

ters

Whe

at b

ran

Hardw

ood

- Kra

ft pu

lp

Hardw

ood

- Sulf

ite p

ulp

Softw

ood

- Kra

ft pu

lp

Softw

ood

- Sulf

ite p

ulp

Softw

ood

- TM

P

Eucaly

ptus

pap

er p

ulp

Eucaly

ptus

pap

er p

ulp -

e-be

am

Hemp

pape

r pulp

Hemp

pape

r pulp

- e-

beam

Book 1

(195

1)

Book 2

(191

2)

Book 3

(189

2)

Book 4

(186

0)

Mulb

erry

tree

pap

er

Paper

sam

ple (h

istor

ical)

Rag p

aper

(hist

orica

l)

Rag p

aper

(mod

ern)

Rag p

aper

(mod

ern)

- ag

ed

Rag p

aper

(mod

ern)

- e-

beam

0

20

40

60

80

100M

ola

r ra

tio o

f ca

rbo

hydr

ates

[%

] Arabinose Galactose Glucose Mannose Rhamnose

Xylose

0

10

20

30

40

50

60

70

80

90

100

% Recovery rate

Re

cove

ry r

ate

(ca

rbo

hyd

rate

s fr

om f

ibe

r,

) [%

]

 397 Figure 6: Carbohydrate composition of cellulose samples, expressed as molar ratio  in % of the total 398 sugar  amount  (columns)  and  recovery  rate  of  released  carbohydrates  in  %  ()  by  sulfuric  acid 399 hydrolysis and GC‐MS analysis. Further information about the samples is shown in Table 1 and 2. 400  401 

 402 

There  are  several  reasons  sulfuric  acid hydrolysis  could not be  achieved  full  recovery  rate  in  this 403 

study.  The  sulfuric  acid  hydrolysis  caused  decarboxylation  of  uronic  acids,  and  the  possible 404 

destruction  of  low‐content  carbohydrates  leads  to  an  information  loss  compared  to  acid 405 

methanolysis.  The  existence  of  oligomers was  excluded  because  no  signals  of  disaccharides  and 406 

trisaccharides were present in GC, but non‐GC‐analyzable oligomers might be present. The fractions 407 

of  fats,  proteins,  lignin,  and  minerals  were  not  considered  during  this  study,  which  is  mainly 408 

noticeable  in wheat  bran.  This  explains  the  high  recovery  rates  of  cellulose  samples  and  the  low 409 

recovery rates of specific polysaccharides at once. 410 

The acid  concentrations,  reaction  temperatures and processing  times of  the  two‐step  sulfuric acid 411 

hydrolysis method  applied  in  this  study was  optimized  by  Bose  et  al.17  and  adapted  for  GC/MS 412 

analysis by  the authors  to obtain a high  sugar  yield of  samples, which  contain both  cellulose and 413 

hemicellulose. The sugar yield  is defined as the number of monosaccharides  theoretically obtained 414 

from cellulose and hemicellulose conversion of present samples after treatment such as hydrolysis. 415 

While the recovery rate refers to the total sample weight and includes ash, lignin, and other residues. 416 

In many scientific trails, the sugar yield is used as target parameters for hydrolysis performance. The 417 

amount of the non‐analyzed fraction is of less importance if any. Therefore, little is known about the 418 

molecular structure of this inaccessible fraction, which is more difficult to hydrolyze. Crystallinity, the 419 

defined term of hornification and lignin composition within samples’ structure was mentioned, that 420 

they  interfere  with  solvent  diffusivity  into  cellulose  structure32.  The  presence  of  lignin  and 421 

hemicellulose  on  the  cellulose  surface  can  negatively  affect  the  cellulose  hydrolysis  rate.33‐35  The 422 

conditions for the two‐step sulfuric acid hydrolysis supposed by Bose et al. are  less aggressive than 423 

other  tested  hydrolysis  conditions  within  their  study.17  A  higher  sulfuric  acid  concentration  and 424 

longer  reaction  time  could  enhance  the  sugar  yield,  but  raise  the  decomposition  products  from 425 

obtained monosaccharides at  the  same  time.36‐37 The PCA of  the  sulfuric acid hydrolysis data  (see 426 

Supplemental  Figure  S1)  is  not  discussed,  due  to  the  low  information  content. Only  one  or  two 427 

different carbohydrate accounts for the most significant proportion of the samples. As mentioned by 428 

other  authors18, 26,  sulfuric  acid  hydrolysis  seems  to  be  inadequate  to  analyze  samples with  low 429 

cellulosic content or highly amorphous materials, due  to  the destruction of  information about  the 430 

respective  monomer  composition  and  sample  structure  by  carbohydrate  decomposition. 431 

Nevertheless, this hydrolysis method is required to uncover existing crystalline fractions in samples. 432 

 433 

 434 

Supplemental Figure S1: PCA biplot on scores (polysaccharides: ●, cellulose samples: ) and loadings 435 (grey) based on their molar ratio of carbohydrate composition, obtained by sulfuric acid hydrolysis 436 and  followed  by  GC‐FID  or  GC‐MS  analysis.  See  Table  1  and  2  for  abbreviations  and  further 437 information about samples. 438  439 

 440 

3.3 The index of recalcitrance cellulose fraction (CRCI)  441 

The  index of  recalcitrance  cellulose  fraction  (CRCI) of different  cellulose  samples was  calculated by 442 

subtracting the amount of glucose determined with acid methanolysis from the amount of glucose 443 

determined with sulfuric acid hydrolysis (CG2H). Three groups of different CRCI ranges (Tab. 3) were 444 

observed during analysis:  (1) a high CRCI  ‐level  group  (CRCI > 60%), which  consists of  cotton based 445 

material and hemp paper pulp; (2) a moderate CRCI‐level group (CRCI < 60%; > 40%), which consists of 446 

-2 -1 0 1 2 3 4

-1

0

1

2

3-0,5 0,0 0,5 1,0 1,5

-0,5

0,0

0,5

1,0

ASDA

LA GG

GLGP

GC

GB

GSGK

GA

IN

AU

PC

PG

RG

ST

LG

XB

XG BR

HK

SK

TM

EC

EC-E

B1

B2B3

MBPH

ArabinoseGalactose

Glucose

Mannose

Rhamnose

Xylose

Prin

cipa

l Com

pone

nt 2

(sc

ores

)

Principal Component 1 (scores)

B4, CL, CM, HC, HC-E, HS, RM-E, SS, RM-A, RH, RM,

Prin

cipa

l Co

mpo

nent

2 (

load

ings

)Principal Component 1 (loadings)

two  softwood  samples  (Sulfite,  Kraft),  Kraft  pulped  eucalyptus,  Sulfite  pulped  hardwood,  and 447 

mulberry three paper; and (3) a low level group (CRCI < 40%), containing birch sulfite pulp, a historical 448 

paper sample, a thermo‐mechanical pulped spruce sample (Softwood  ‐ TMP), and wheat bran with 449 

the lowest value. 450 

 451 

Table 3: The  index of recalcitrance cellulose fraction (CRCI) of different cellulose samples, calculated 452 as  the  difference  between  glucose  amount  from  sulfuric  acid  hydrolysis  and  glucose  amount 453 obtained by acid methanolysis (CG2H). 454 

CRCI 

level 

Sample 

mg glucose / g sample 

 CRCI [%] Sulfuric acid hydrolysis 

Acid methanolysis 

> 60%  Rag paper (modern)  712.28  36.19  67.61 

Cotton Linters  712.69  54.86  65.78 

Rag paper (historical)  703.89  59.12  64.48 

Hemp paper pulp  695.28  80.35  61.49 

>40%; <60%  Softwood – Sulfite  644.52  79.81  56.47 

Softwood – Kraft  616.98  77.31  53.97 

Eucalyptus paper pulp  544.14  52.54  49.16 

Hardwood ‐ Sulfite  587.48  101.22  48.63 

Mulberry tree paper  482.77  39.47  44.33 

< 40%  Birch ‐ Sulfite  455.2  71.99  38.32 

Paper sample (historical)  460.82  132.17  32.87 

Softwood ‐ TMP   245.53  36.91  20.86 

Wheat bran  180.06  138.36  4.17  455 

 456 

3.3.1 Comparison of CRCI (based on CG2H) and CrI (based on NMR analysis) 457 

A comparison of recalcitrance cellulose fraction  index (CRCI) values obtained from cellulose samples 458 

with crystallinity index (CrI) values by NMR analysis of corresponding samples have been conducted. 459 

The structural fractions obtained by the different analysis methods were compared.  460 

The total amount of carbohydrates in a sample equates to the total area of the C‐4 region of the 13C 461 

CP‐MAS NMR  spectra. However,  sulfuric  acid hydrolysis  is not  capable of  recovering  sample  yield 462 

completely.  A  gap  exists  between  received  and  theoretical  possible  sample  yield  by  sulfuric  acid 463 

hydrolysis.  Although  the  hydrolysis  methods  were  optimized  to  obtain  credit  the  highest 464 

monosaccharide  yield  of  cellulose  as well  as  hemicelluloses,  in  between  20  ‐  30%  of  the  sample 465 

composition  is missing.  The NMR  analysis  names  this  fraction  as  inaccessible  fibril  surface  (IAFS), 466 

which  corresponds  to  the  non‐analyzed  fraction  of  the  CG2H‐method.  The  carbohydrate  analysis 467 

conducted  by  gas  chromatogram  showed  no  di‐  or  trisaccharides  corresponding  signals  nor 468 

information  about  the  availability  of  other  existing  oligomers.  Signals  of  polymers with  unknown 469 

composition  were  observed  by  TLC  analysis  of  sulfuric  acid  reaction mixture,  but  could  not  be 470 

quantified. 471 

It follows that the amount of carbohydrates released by sulfuric acid hydrolysis equates to the total 472 

area of the NMR C‐4 region minus the IAFS fraction. According to previous findings27‐28, 30, we  imply 473 

the  assumption  that  acid methanolysis  analyzes  the  amorphous  structure  of  a  sample, while  the 474 

sulfuric  acid  hydrolysis  analyzed  both  the  amorphous  and  crystalline  structure.  Accordingly,  the 475 

amount  of  carbohydrates  released  by  acid  methanolysis  corresponds  to  the  amorphous 476 

hemicellulose (hemi) signal and the partial amorphous cellulose fraction, the accessible fibril surface 477 

(AFS) signal of the deconvoluted NMR spectra. Subtract the glucose data of methanolysis from the 478 

glucose data of sulfuric acid hydrolysis results in the CRCI. The correlation of CRCI results from cellulose 479 

samples  provided  by  the  CG2H‐method with  crystallinity  index  results  obtained  by  NMR  agreed 480 

moderately.  (R²  =  0.7737;  Fig.  7  left).  CRCI  values  obtained  from  pulp  samples  showed  a  higher 481 

correlation to NMR data than including paper samples. So, the CrI calculation according to Nocanda 482 

et  al.38  correlated  to  CRCI  values  shows  a  lower  correlation  compared  to  the  calculation  of 483 

Zuckerstätter  et  al.39  (R²  =  0.5692,  Fig.  7  right).  Zuckerstätter  et  al.  include  the  paracrystalline 484 

cellulose  fraction  to  CrI  calculation,  which  increased  the  correlation  to  CRCI  (R²  =  0.8574,  Fig.  7 485 

middle)39. This indicates a similarity of CRCI to the sum of the crystalline cellulose signals Iα, Iβ, Iα+β and 486 

the paracrystalline cellulose (Para) signal area in the deconvoluted NMR spectra. Nevertheless, CRCI is 487 

only similar to CrI, not equal and therefore named “recalcitrance”. To reduce the distances between 488 

the  indices CRCI and CrI, the hydrolysis methods should release a higher glucose yield per cellulosic 489 

samples.  However,  samples with  larger  crystallite  size  are more  resistant  to  acid  hydrolysis,  and 490 

harsh conditions (high temperatures, high acid concentrations) are required to liberate glucose from 491 

these  tightly  associated  chains.37  However,  in  this  case,  the method  could  not  be  used  for  the 492 

analysis of hemicelluloses, and thus a high information content would be lost through carbohydrate 493 

destruction, in contrast to the applied method. The most sensitive step of the proposed methodology 494 

is the dissolution of sample tissue  in the acid solution, crucial for sulfuric acid hydrolysis as for acid 495 

methanolysis. The sample crushing during preparation without affecting the crystallinity on the one 496 

hand and  suitable particle  size  for dissolution  is  the  central  key of  the method. Pulp  samples are 497 

easier to handle compared to paper samples with more solid texture.  498 

35 40 45 50 55 60 65 70 750

10

20

30

40

50

60

70

80

CrI

acc

ord

ing

to Z

uck

erst

ädte

r e

t al.

(200

9)

CRC

I

Pulp samples Paper samples

35 40 45 50 55 60 65 70 750

10

20

30

40

50

60

70

80

R2 = 0.7408

R2 = 0.49745

CrI

acc

ordi

ng t

o N

ocan

da

et a

l. (2

007)

CrI

acc

ord

ing

to Z

uck

erst

ädte

r e

t al.

(200

9)

CRC

I

Pulp samples

R2 = 0.83377

35 40 45 50 55 60 65 70 750

10

20

30

40

50

60

70

80

CRC

I

Pulp samples

 499 

Figure 7: Correlation of  index of  recalcitrance  cellulose  fraction  (CRCI) of pulp  and paper  samples, 500 obtained  by  comparison  of  glucose  data  obtained  by  two  hydrolysis  methods  (CG2H‐method) 501 following GC analysis and crystallinity  index (CrI) obtained by NMR analysis, calculated according to 502 Zuckerstädter et al. (2009) or Nocanda et al. (2007). 503  504 

 505 

3.3.2 Effect of aging and electronic beam irradiation treatment on sample integrity 506 

Electronic  beam  (e‐beam)  irradiation  treatment  was  described  to  change  cellulose  crystallinity, 507 

mostly  used  to modify  chemical  and  physical  properties  of  cellulose‐containing materials.3‐4  The 508 

procedure  is  considered  as  pre‐treatment  of  renewable  lignocellulosic  resources  to  improve 509 

monosaccharide  yield  for  bioethanol  production.40  E‐beam  irradiation  was  reported  to  induce  a 510 

reduction  of  sample's  cellulose  crystallinity,  which  enhanced  the  accessibility  of  the  surface  of 511 

cellulosic material to enzymatic hydrolysis and chemical reagents, resulting in an increased hydrolysis 512 

efficiency  to  produce monosaccharides.9, 40  The  reduction of  crystallinity by  e‐beam  radiation had 513 

also been shown most evident at doses above 100 kGy.3, 9 Eucalyptus and hemp paper pulps, and rag 514 

paper was exposed to e‐beam irradiation at 60 kGy or 120 kGy (see Tab. 2) to determine the effect of 515 

structural  changes  at  monosaccharide  level,  between  amorphous/crystalline  fractions  and  the 516 

molecular mass. The sulfuric acid hydrolysis of e‐beam treated eucalyptus paper pulp indicates a loss 517 

of  carbohydrate  yield  by  6.04%, while  the  acid methanolysis measured  an  increase  of  the  total 518 

carbohydrate  fraction  by  51.3%  (Fig.  8).  The  glucose  content, obtained  by  sulfuric  acid hydrolysis 519 

decreased (‐12.55%) in contrast to xylose, which slightly increased (7.2%) after e‐beam treatment. In 520 

contrary to the sulfuric acid hydrolysis, the glucose content obtained by acid methanolysis increased 521 

(64.89%), while e‐beam also caused a higher release of xylose yield by 42.15% after treatment, which 522 

changes  the molar glucose/xylose  ratio  from 5.3/1  to 2.8/1. The  total carbohydrate yield of hemp 523 

paper pulp analyzed by sulfuric acid hydrolysis decreased by 0.94% (Fig. 8A) after e‐beam treatment. 524 

The total carbohydrate yield obtained by acid methanolysis,  increased by 16.92%  (Fig. 8B). Sulfuric 525 

acid hydrolysis uncovered, that glucose content slightly decreased, while the xylose content slightly 526 

increased.  The  carbohydrate  composition,  analyzed  by  acid methanolysis,  show  that  glucose  and 527 

xylose increased by 11.29% and 26.92%, respectively. The total carbohydrate yield of rag paper after 528 

e‐beam  treatment,  analyzed  by  sulfuric  acid  hydrolysis,  decreased  by  ‐5.38%,  while  total 529 

carbohydrate  yield  obtained  by  acid methanolysis  data  showed  an  increase  by  66.05%  (Fig.  8). 530 

Simultaneously, the glucose content analyzed by sulfuric acid hydrolysis decreased by ‐6.04%, while 531 

glucose content obtained by acid methanolysis increased by 77.9%. The CRCI value of cellulose slightly 532 

decreased  after  e‐beam  irradiation  in  hemp  paper  pulp  (‐1.77%)  and  rag  paper  (‐4.21%),  while 533 

eucalyptus paper pulp (‐10.24%) showed a severe break down (Fig. 9). The spectral fitting results of 534 

the NMR C4‐region spectra obtained  from eucalyptus and hemp paper pulps and rag paper before 535 

and  after  e‐beam  treatment  are  shown  in  Supplemental  Figure  S2.  After  e‐beam  treatment, 536 

eucalyptus  paper  pulp  showed  slight  changes  within  the  ordered  C4‐region  (86–92  ppm).  The 537 

cellulose Iα signal (89.50 ppm) decreased from 3.26% to 2.15% of the integrated area, while the 538 

cellulose  Iβ signal  (87.86 ppm)  increased  from 2.68%  to 3.98%. Eucalyptus paper pulp showed a 539 

decrease  of  hemicellulose  signal  (81.94  ppm)  from  9.61%  to  5.10%  integrated  area  in  the  less 540 

ordered C4‐region  (80–86 ppm) after e‐beam  treatment, while  the  inaccessible  fibril surface signal 541 

(83.17 ppm)  increased  from 42.93%  to 49.91% of  the  integrated area. Hemp paper pulp showed a 542 

decrease of  cellulose  Iβ  signal  (87.97 ppm)  from 21.78%  to 14.07% of  the  integrated area  and  an 543 

increase of paracrystalline cellulose signal (88.48 ppm) from 14.01% to 19.49% of the integrated area 544 

after e‐beam treatment. The inaccessible fibril surface signal (82.93 ppm) increased from 33.15% to 545 

36.84% of  the  integrated area. The NMR analysis of  rag paper before and after e‐beam  treatment 546 

showed only minor changes of parameters within the  integrated area. The cellulose Iβ signal (87.99 547 

ppm) slightly decreased from 13.67% of the integrated area to 12.60%. The hemicellulose area (82.1 548 

ppm) and the  inaccessible fibril surface area (82.79 ppm)  increased from 3.04% to 4.28%, and from 549 

35.44% to 36.71%, respectively. Other parameters varied only by 0.5%. The effect of chain scission in 550 

cellulose,  proven  by  Saeman  et  al.41,  measured  before  and  after  e‐beam  treatment  showed  a 551 

significant  collapse  of  the  weight‐average  molecular  mass  in  every  e‐beam  treated  sample 552 

(eucalyptus paper pulp: ‐88.1%; hemp paper pulp: ‐87.4%; rag paper: ‐83.3%), shown in Fig. 9.  553 

The  ageing  process  was  also  reported  to  disturb  crystalline  structures  of  cellulose  and  induce 554 

structural  changes within  the  cellulosic  samples.  The  impact of  environmental parameters on  the 555 

ageing  of  paper  samples,  such  as  temperature,  pressure,  air moisture  content  and  presence  of 556 

oxygen, was well  studied  since degradation  kinetics  for  artificial  ageing has  been  ascertained.42‐43 557 

Thus,  artificially  accelerated  ageing was  found  to  be  suitable  to  simulate  natural  ageing.43  These 558 

parameters  in combination with paper dependent factors, such as acidity, water content, alum and 559 

calcium carbonate content of papers  influence  the speed of shortening  the cellulose chains, which 560 

implies the degradation of paper.8, 42, 44 There has been a discussion over the direction of crystallinity 561 

change  in  thermally‐aged  celluloses,  due  to  several  reports  of  both  crystallinity  increases  and 562 

decreases.1‐2 When  looking  at monosaccharide  level,  similarities  can  be  detected  between  the  e‐563 

beam and aged rag paper sample. The total carbohydrate yield of rag paper after ageing, analyzed by 564 

sulfuric  acid  hydrolysis,  also  decreased  (‐3.71%), while  total  carbohydrate  yield  obtained  by  acid 565 

methanolysis  increased  (33.21%),  shown  in  Fig.  8.  The  glucose  content  analyzed  by  sulfuric  acid 566 

hydrolysis decreased by  ‐3.85%, while glucose content obtained by acid methanolysis  increased by 567 

40.61%. A  lower CRCI  value of artificially aged  rag paper  (‐7.13%) was observed  in  contrast  to  the 568 

untreated sample, which suggests an alteration of the crystalline fraction (Fig. 9). The artificial ageing 569 

affected the weight‐average molecular mass (Mw) of rag paper (‐32.6%) but has not reached the level 570 

of the e‐beam treated sample (Fig. 9). The results are  in contrary to the study of Sandy et al.2 They 571 

observed  an  increase  of  crystalline  cellulose  fraction  after  ageing  experiments, which  involved  a 572 

hydrochloric acid pre‐treatment of their paper samples prior ageing process.2 The  introduced effect 573 

(increasing crystallinity during paper ageing) has been equated by other authors with  the effect of 574 

aging  and  used  as  a  general  notion.45  In  our  opinion,  the  acid  pre‐treatment  generates  similar 575 

conditions as an acid methanolysis.  In  combination with accelerated  thermal ageing under  cycling 576 

relative  humidity,  destruction  of  carbohydrates within  the  sample  has  been  caused.  The  ageing‐577 

related  reduction of  crystalline  areas  observed  in  our  study  seems  to be defaced  in  the  study of 578 

Sandy et al.2, due to acid‐induced destruction of the amorphous carbohydrates fraction. 579 

The  applied  treatments  (aging  or  e‐beam)  decreased  the  total monosaccharide  yield  obtained  by 580 

sulfuric  acid  hydrolysis  and  CRCI  values  at  the  same  time  of  all  samples. On  the  other  hand,  the 581 

monosaccharide  yield  obtained  by  acid  methanolysis  increased  in  all  samples  after  the  two 582 

treatment.  They  showed  also  decreasing  total  amounts  of  glucose,  measured  by  sulfuric  acid 583 

hydrolysis  and  simultaneous  equal  or  higher  amounts  of  glucose  released  by  acid methanolysis 584 

compared  to  the  untreated  samples.  Assuming  that  the  sulfuric  acid  hydrolysis  analyzed  the 585 

crystalline  and  amorphous  fraction, while  the  acid methanolysis  analyzed  the  amorphous  fraction 586 

leads to several results. The yield loss after e‐beam or ageing treatment indicates treatment‐caused 587 

carbohydrate  destruction.  The  decreased  of  CRCI,  and  the  simultaneous  rise  of  the  amorphous 588 

carbohydrate fraction in the samples indicates a reduction of samples’ crystalline fraction and imply 589 

a  structural  rearrangement of  crystalline  into  amorphous  cellulose  fraction.  The high  loss of  total 590 

glucose,  analyzed  by  sulfuric  acid  hydrolysis,  seems  to  face  a  slightly  elevated  level  of  glucose, 591 

measured  by  acid  methanolysis.  The  fact  that  no  decrease  of  amorphous  glucose  content  was 592 

observed after each  treatment, despite  reduced  total glucose content may support  the hypothesis 593 

that crystalline cellulose converts  faster  to allomorphic cellulose as allomorphic cellulose degraded 594 

under applied conditions. We exclude  the possibility  that only  the crystalline  fraction  is destroyed 595 

because cellulose degradation  is assumed to occur mainly  in the amorphous fraction rather than  in 596 

the crystalline  fraction.3 Nevertheless,  the  immediate destruction of crystalline cellulose cannot be 597 

excluded as well as  the assumption,  that carbohydrate degradation  takes place entirely within  the 598 

amorphous regions, which cannot be quantified.46 So, two associated degradation processes can be 599 

concluded:  (1)  the  conversion  from  crystalline  fraction  to  an  amorphous  fraction  and  (2)  the 600 

destruction  of  the  amorphous  and  crystalline  fraction within  a  sample,  but  their  ratio  cannot  be 601 

quantified. 602 

The weight‐average molecular mass (Mw) of e‐beam treated rag paper was observed as 4‐fold lower 603 

than of the corresponding aged sample, due to the different physical actions on chain scission of the 604 

two treatments. The eucalyptus paper pulp structure seems to be particularly vulnerable to e‐beam 605 

irradiation, since the yield of acid methanolysis released glucose and xylose,  increased dramatically 606 

after treatment compared to hemp paper pulp and rag paper, which based on other fiber origins. 607 

 608 

 609 

 610  611  612  613  614  615  616  617  618  619  620  621  622  623  624  625  626  627  628  629  630  631  632  633  634  635  636  637  638  639 Supplemental Figure S2: Signals of spectral  fitting of  the cellulose C4‐region obtained  from  13C CP‐640 MAS NMR  spectra  of  eucalyptus  and  hemp  paper  pulps  and  rag  paper  (from  above  to  bottom), 641 before  (left)  and  after  electronic  beam  (e‐beam)  treatment  (right).  The  red  line  shows  the 642 experimental spectra. The colorized solid lines represent the deconvoluted signals/fractions: Iα and Iβ 643 – Crystalline cellulose Iα and Iβ; Para – Paracrystalline cellulose; AFS – Accessible fibril surfaces; IAFS – 644 Inaccessible fibril surfaces; Hemi – Hemicellulose.  645  646 

 647 

(ppm)8081828384858687888990919293

p p y j p p

Hempcell

IAFSAFS HemiPara

Iα+β

Iβ

Iα

Hemp paper pulp

(ppm)(ppm)

8081828384858687888990919293

IAFSAFS Hemi

Para

Iα+β

Iβ

Iα

Hemp paper pulp – e-beam

(ppm)

(ppm)7980818283848586878889909192

IAFS AFSHemi

ParaIα+β

Iβ

Iα

Rag paper

(ppm) (ppm)7980818283848586878889909192

IAFS AFS Hemi

ParaIα+β

Iβ

Iα

Rag paper – e-beam

(ppm)

(ppm)7980818283848586878889909192

IAFS

AFS Hemi

Para

Iα+β

IβIα

Eucalyptus paper pulp – e-beam

(ppm)(ppm)798081828384858687888990919293

IAFS

AFS

Hemi

Para

Iα+β

IβIα

(ppm)

Eucalyptus paper pulp

no tr

eatm

ent

e-be

am

no tr

eatm

ent

e-be

am

no tr

eatm

ent

e-be

am

aged

0

100

200

300

400

500

600

700

800

900

1000

0

100

200

300

400

500

600

700

800

900

1000A)

Rel

ease

d ca

rboh

ydra

tes

[mg/

g sa

mp

le]

Sulfuric acid hydrolysis Total carbohydrates Glucose Xylose

Eucalyptuspaper pulp

Rag paper(modern)

Hemppaper pulp  

no tr

eatm

ent

e-be

am

no tr

eatm

ent

e-be

am

no tr

eatm

ent

e-be

am

aged

0

50

100

150

200

250

300

0

50

100

150

200

250

300B)

Rel

ease

d ca

rboh

ydra

tes

[mg/

g sa

mpl

e] Acid methanolysis Total carbohydrates Glucose Xylose

Eucalyptuspaper pulp

Rag paper(modern)

Hemppaper pulp  648 

Figure 8: Released carbohydrates (total amount, glucose and xylose) of untreated, electron beam (e‐649 beam) treatment (eucalyptus and hemp paper pulps: 120 kGy, and rag paper: 60 kGy) and artificially 650 aged cellulose samples. A) represents sulfuric acid hydrolysis data; B) represents acid methanolysis 651 data. 652  653 

 654 

321

,3

38,

3

438,

7

55,

1

354,

5

59,2

238

,9

no tr

eatm

ent

e-be

am

no tr

eatm

ent

e-be

am

no tr

eatm

ent

e-be

am

aged

0

10

20

30

40

50

60

70

Wei

ght-

aver

age

mol

ecul

ar m

ass

(Mw

, )

CRC

I

Mw

Inde

x of

rec

alci

tran

ce c

ellu

lose

frac

tion

(CR

CI)

0

100

200

300

400

500

600

700

Eucalyptus paper pulp

Rag paper(modern)

Hemppaper pulp

 655 Figure 9: The  index of  recalcitrance  cellulose  fraction  (CRCI,  columns) of untreated, electron beam 656 irradiated  (e‐beam) and artificially aged  cellulose  samples  (eucalyptus and hemp paper pulps: 120 657 kGy,  and  rag  paper:  60  kGy)  and  corresponding weight‐average molecular mass  (Mw, ).  Index 658 values were calculated by subtracting the amount of glucose determined by acid methanolysis from 659 the amount of glucose determined by sulfuric acid hydrolysis.  660  661 

 662 

 663 

4 Conclusion 664 

The  current  study  showed  that  the  two  applied  hydrolysis  methods,  alone  or  in  combination, 665 

followed  by  GC‐MS  analysis,  represents  a  useful  analytical  tool  for  determination  of  structure 666 

elucidation  and  carbohydrate  characterization  of  a  wide  range  of  polysaccharide  and  cellulose 667 

samples. The similarity of polysaccharides’ carbohydrate pattern compared to cellulose samples can 668 

be a result due to containing the corresponding polymer, or the combination of different polymers in 669 

the  sample  corresponds  approximately  to  the  same  molar  ratio  of  a  polysaccharide  after 670 

depolymerization. 671 

The proposed approach  involved  the parallel hydrolysis of  cellulose‐containing  samples by  sulfuric 672 

acid  and  acid  methanolysis,  and  the  subsequently  gas  chromatographic  analysis  of  obtained 673 

hydrolysis products. The comparative analysis subtracts the amount of glucose determined with acid 674 

methanolysis from the amount of glucose determined with sulfuric acid hydrolysis (CG2H). The result 675 

quantifies the recalcitrance cellulose fraction within cellulosic samples and is referred as the index of 676 

recalcitrance cellulose fraction  (CRCI). The CG2H‐method derived the carbohydrate composition and 677 

the  quantity  of  individual  monosaccharide,  while  the  13C  CP‐MAS  NMR  method  distinguishes 678 

between structural alterations based on  the different cellulose  fractions  (Iα,  Iβ and paracrystalline). 679 

Both  methods  can  detect  and  uncover  structural  rearrangements  from  the  crystalline  to  the 680 

amorphous  fraction. The NMR method  is  fast, but  requires expensive equipment, while  the CG2H‐681 

method comes along with less expensive equipment, but is more time‐consuming and requires more 682 

labor and analytical  resources. The method  is preferred  to discover background  information about 683 

treatment‐caused  structural  rearrangements  and  resolve more detailed  information  about  sample 684 

characteristic besides methods, which measure only the crystallinity index of paper samples. 685 

The  results  led  to a  classification of  three groups  among  the analyzed  samples with different CRCI 686 

ranges: high (CRCI > 60%), moderate (CRCI < 60%; > 40%), and low (CRCI < 40%). The analysis of artificial 687 

aging  or  electron  beam  irradiation  treated  cellulose  samples  uncovered  similar  tendencies  in  the 688 

structural rearrangements from crystalline to amorphous cellulose fraction and a general loss of total 689 

carbohydrates within all samples. Differences in glucose accessibility at an amorphous level between 690 

e‐beam  treated  pulp  samples  were  detected  depending  on  fiber  source  or  pulp  pre‐treatment. 691 

Despite  a  similar  behavior  of  the  treatments  at  samples’  carbohydrate  level,  e‐beam  treatment 692 

caused a significantly higher effect of chain scission at rag paper in contrast to the ageing treatment. 693 

The crystallinity‐reducing properties of the e‐beam treatment make it attractive as an application for 694 

renewable  resources. The  irradiation can  improve  carbohydrate yield by enhancing polysaccharide 695 

accessibility  towards  enzyme  or  hydrolysis  methods.  The  e‐beam  treatment  makes  cellulose 696 

containing materials more interesting as raw material for biomass to fuel conversion. 697 

The information, obtained by the proposed CG2H approach, can be used to improve the efficiency of 698 

current  technologies  to  increase  carbohydrates  accessibility  of  cellulose‐containing  samples. 699 

Nevertheless, further research is needed to extend the technique to samples of another origin. 700 

 701 

 702 

 703 

5 Acknowledgment 704 

The  financial  support  by  the  Christian  Doppler  Research  Society  (CD  lab  “Advanced  Cellulose 705 

Chemistry and Analytics”) and an STSM grant  from  the European COST B21 Action FP0901  (COST‐706 

STSM‐FP0901‐9940) is gratefully acknowledged. 707 

 708 

 709 

 710 

6 References 711 

1.  Henniges, U.;  Bjerregaard,  L.;  Ludwig,  B.;  Potthast,  A.,  Controversial  influence  of  aqueous 712 treatments on historic textiles. Polym Degrad Stabil 2011, 96 (4), 588‐594. 713 

2.  Sandy, M.; Manning, A.; Bollet, F., Changes  in  the Crystallinity of Cellulose  in Response  to 714 Changes in Relative Humidity and Acid Treatment. In Restaurator, 2010; Vol. 31, p 1. 715 

3.  Driscoll, M.; Stipanovic, A.; Winter, W.; Cheng, K.; Manning, M.; Spiese,  J.; Galloway, R. A.; 716 Cleland, M. R., Electron beam irradiation of cellulose. Radiation Physics and Chemistry 2009, 717 78 (7‐8), 539‐542. 718 

4.  Henniges,  U.;  Okubayashi,  S.;  Rosenau,  T.;  Potthast,  A.,  Irradiation  of  cellulosic  pulps: 719 Understanding  its  impact  on  cellulose  oxidation.  Biomacromolecules  2012,  13  (12),  4171‐720 4178. 721 

5.  Mann, J., Modern methods of determining crystallinity in cellulose. Pure Appl. Chem. 1962, 5 722 (1‐2), 91‐106. 723 

6.  Black,  G.  E.;  Fox,  A.,  Recent  progress  in  the  analysis  of  sugar monomers  from  complex 724 matrices  using  chromatography  in  conjunction  with  mass  spectrometry  or  stand‐alone 725 tandem mass spectrometry. Journal of Chromatography A 1996, 720 (1‐2), 51‐60. 726 

7.  Larsson,  P.  T.;  Wickholm,  K.;  Iversen,  T.,  A  CP/MAS13C  NMR  investigation  of  molecular 727 ordering in celluloses. Carbohydrate Research 1997, 302 (1–2), 19‐25. 728 

8.  Kato,  K.  L.;  Cameron,  R.  E.,  A  Review  of  the  Relationship  Between  Thermally‐Accelerated 729 Ageing of Paper and Hornification. Cellulose 1999, 6 (1), 23‐40. 730 

9.  Chung, B. Y.; Lee, J. T.; Lee, S. S.; Kim, U. J.; Wi, S. G.; Bai, H. W.; Cho, J. Y., A comparison of 731 the efficiency of electron beam  irradiation on enzymatic hydrolysis between 4 doses of 25 732 kGy and a single dose of 100 kGy for bioethanol production. Journal of the Korean Society for 733 Applied Biological Chemistry 2012, 55 (3), 385‐389. 734 

10.  Park, S.; Baker, J. O.; Himmel, M. E.; Parilla, P. A.; Johnson, D. K., Cellulose crystallinity index: 735 measurement  techniques  and  their  impact  on  interpreting  cellulase  performance. 736 Biotechnology for biofuels 2010, 3, 10. 737 

11.  Thygesen, A.; Oddershede, J.; Lilholt, H.; Thomsen, A. B.; Ståhl, K., On the determination of 738 crystallinity and cellulose content in plant fibres. Cellulose 2005, 12 (6), 563‐576. 739 

12.  Agarwal, U. P.; Reiner, R. R.; Ralph, S. A. In Cellulose crystallinity of woods, wood pulps, and 740 agricultural  fibers by FT‐Raman spectroscopy, Proceedings 16th  international symposium on 741 wood, fiber and pulping chemistry. 2011 June 8‐10. Tianjin, China. Beijing, China: China Light 742 Industry Press: 69‐74, 2011. 743 

13.  Ali, M.; Emsley, A. M.; Herman, H.; Heywood, R.  J., Spectroscopic  studies of  the ageing of 744 cellulosic paper. Polymer 2001, 42 (7), 2893‐2900. 745 

14.  Lennholm, H.; Larsson, T.;  Iversen, T., Determination of cellulose  Iα and  Iβ  in  lignocellulosic 746 materials. Carbohydrate Research 1994, 261 (1), 119‐131. 747 

15.  Maunu, S.; Liitiä, T.; Kauliomäki, S.; Hortling, B.; Sundquist, J., 13C CPMAS NMR investigations 748 of cellulose polymorphs in different pulps. Cellulose 2000, 7 (2), 147‐159. 749 

16.  Bertaud, F.; Sundberg, A.; Holmbom, B., Evaluation of acid methanolysis for analysis of wood 750 hemicelluloses and pectins. Carbohydrate Polymers 2002, 48 (3), 319‐324. 751 

17.  Bose,  S.  K.;  Barber,  V.  A.;  Alves,  E.  F.;  Kiemle,  D.  J.;  Stipanovic,  A.  J.;  Francis,  R.  C.,  An 752 improved method for the hydrolysis of hardwood carbohydrates to monomers. Carbohydrate 753 Polymers 2009, 78 (3), 396‐401. 754 

18.  Sundberg, A.; Sundberg, K.; Lillandt, C.; Holmbom, B., Determination of hemicelluloses and 755 pectins  in wood and pulp fibres by acid methanolysis and gas chromatography. Nordic Pulp 756 and Paper Research Journal 1996, 11 (4), 216‐219+226. 757 

19.  Marques, G.; Gutiérrez, A.; del Río, J. C.; Evtuguin, D. V., Acetylated heteroxylan from Agave 758 sisalana and  its behavior  in alkaline pulping and TCF/ECF bleaching. Carbohydrate Polymers 759 2010, 81 (3), 517‐523. 760 

20.  Duquesnoy,  E.;  Castola,  V.;  Casanova,  J.,  Identification  and  quantitative  determination  of 761 carbohydrates  in  ethanolic  extracts  of  two  conifers  using  13C  NMR  spectroscopy. 762 Carbohydrate Research 2008, 343 (5), 893‐902. 763 

21.  Panagiotopoulos,  C.;  Sempere,  R.,  Analytical methods  for  the  determination  of  sugars  in 764 marine samples: A historical perspective and future directions. Limnol Oceanogr‐Meth 2005, 765 3, 419‐454. 766 

22.  Huang,  Y.;  Indrarti,  L.; Azuma,  J.; Okamura, K.,  Simultaneous Determination of Xylose  and 767 Uronic‐Acid in Beech Xylan by Methanolysis. Mokuzai Gakkaishi 1992, 38 (12), 1168‐1171. 768 

23.  Laine, C.; Tamminen, T.; Vikkula, A.; Vuorinen, T., Methylation analysis as a tool for structural 769 analysis of wood polysaccharides. Holzforschung 2002, 56 (6), 607‐614. 770 

24.  Amelung, W.; Cheshire, M. V.; Guggenberger, G., Determination of neutral and acidic sugars 771 in  soil by  capillary  gas‐liquid  chromatography  after  trifluoroacetic  acid hydrolysis.  Soil Biol 772 Biochem 1996, 28 (12), 1631‐1639. 773 

25.  Rumpel,  C.;  Dignac, M.  F.,  Chromatographic  analysis  of monosaccharides  in  a  forest  soil 774 profile: Analysis by gas chromatography after  trifluoroacetic acid hydrolysis and  reduction‐775 acetylation. Soil Biol Biochem 2006, 38 (6), 1478‐1481. 776 

26.  Chambers, R. E.; Clamp, J. R., Assessment of Methanolysis and Other Factors Used in Analysis 777 of Carbohydrate‐Containing Materials. Biochem J 1971, 125 (4), 1009‐&. 778 

27.  Kilic, A.; Hafizoglu, H.;  Tümen,  I.; Dönmez,  I.  E.;  Sivrikaya, H.;  Sundberg, A.; Holmbom, B., 779 Polysaccharides  in cones of eleven coniferous species growing  in Turkey. Wood Science and 780 Technology 2010, 44 (3), 523‐529. 781 

28.  Willför,  S.;  Sundberg,  A.; Hemming,  J.; Holmbom,  B.,  Polysaccharides  in  some  industrially 782 important softwood species. Wood Science and Technology 2005, 39 (4), 245‐257. 783 

29.  Willför,  S.;  Sundberg, A.;  Pranovich, A.; Holmbom, B., Polysaccharides  in  some  industrially 784 important hardwood species. Wood Science and Technology 2005, 39 (8), 601‐617. 785 

30.  Sundberg, A.; Pranovich, A. V.; Holmbom, B., Chemical characterization of various  types of 786 mechanical pulp fines. Anglais 2003, 29 (5), 6. 787 

31.  Becker, M.; Zweckmair, T.; Forneck, A.; Rosenau, T.; Potthast, A.; Liebner, F., Evaluation of 788 different derivatisation approaches  for gas chromatographic‐mass spectrometric analysis of 789 carbohydrates  in  complex  matrices  of  biological  and  synthetic  origin.  Journal  of 790 Chromatography A 2013, 1281, 115‐26. 791 

32.  Wijaya, Y. P.; Putra, R. D. D.; Widyaya, V. T.; Ha, J.‐M.; Suh, D. J.; Kim, C. S., Comparative study 792 on  two‐step  concentrated  acid  hydrolysis  for  the  extraction  of  sugars  from  lignocellulosic 793 biomass. Bioresource Technology 2014, 164 (0), 221‐231. 794 

33.  Bhandari, N.; Macdonald, D. G.; Bakhshi, N. N., Kinetic studies of corn stover saccharification 795 using sulphuric acid. Biotechnol Bioeng 1984, 26 (4), 320‐327. 796 

34.  Singh, A.; Das, K.; Sharma, D. K., Production of reducing sugars from bagasse and rice husk by 797 acid hydrolysis. Agricultural Wastes 1984, 9 (2), 131‐145. 798 

35.  Zhang, Y.‐H. P.; Ding, S.‐Y.; Mielenz, J. R.; Cui, J.‐B.; Elander, R. T.; Laser, M.; Himmel, M. E.; 799 McMillan,  J.  R.;  Lynd,  L.  R.,  Fractionating  recalcitrant  lignocellulose  at  modest  reaction 800 conditions. Biotechnol Bioeng 2007, 97 (2), 214‐223. 801 

36.  Girisuta, B.; Janssen, L. P. B. M.; Heeres, H. J., Kinetic Study on the Acid‐Catalyzed Hydrolysis 802 of Cellulose to Levulinic Acid. Ind Eng Chem Res 2007, 46 (6), 1696‐1708. 803 

37.  Morales‐delaRosa,  S.;  Campos‐Martin,  J.;  Fierro,  J.  G.,  Optimization  of  the  process  of 804 chemical hydrolysis of cellulose to glucose. Cellulose 2014, 21 (4), 2397‐2407. 805 

38.  Nocanda,  X.;  Larsson  Per,  T.;  Spark,  A.;  Bush,  T.; Olsson,  A.; Madikane, M.;  Bissessur,  A.; 806 Iversen,  T.,  Cross  polarisation/magic  angle  spinning  13C‐NMR  spectroscopic  studies  of 807 cellulose  structural  changes  in  hardwood  dissolving  pulp  process.  In Holzforschung,  2007; 808 Vol. 61, p 675. 809 

39.  Zuckerstätter, G.; Schild, G.; Wollboldt, P.; Röder, T.; Weber, H. K.; Sixta, H., The elucidation 810 of cellulose supramolecular structure by 13C CP‐MAS NMR. Lenzinger Berichte 2009, 87, 38‐811 46. 812 

40.  Sundar, S.; Begey, N. S.; Salamanca‐Cardona, L.; Stipanovic, A.; Driscoll, M., Electron beam 813 pretreatment of switchgrass to enhance enzymatic hydrolysis to produce sugars for biofuels. 814 Carbohydrate Polymers 2013. 815 

41.  Saeman, J. F.; Millett, M. A.; Lawton, E. J., Effect of High‐Energy Cathode Rays on Cellulose. 816 Industrial & Engineering Chemistry 1952, 44 (12), 2848‐2852. 817 

42.  Barański, A., Ageing kinetics of cellulose and paper. Restaurator 2002, 23 (2), 77‐88. 818 43.  Zou,  X.; Uesaka,  T.; Gurnagul, N.,  Prediction of paper permanence  by  accelerated  aging  I. 819 

Kinetic analysis of the aging process. Cellulose 1996, 3 (1), 243‐267. 820 44.  Strlic, M.; Cigic, I. K.; Kolar, J.; de Bruin, G.; Pihlar, B., Non‐destructive evaluation of historical 821 

paper based on pH estimation from VOC emissions. Sensors 2007, 7 (12), 3136‐3145. 822 45.  Menart, E.; De Bruin, G.; Strlič, M., Dose–response functions for historic paper. Polym Degrad 823 

Stabil 2011, 96 (12), 2029‐2039. 824 46.  Emsley, A. M.; Stevens, G. C., Kinetics and mechanisms of the low‐temperature degradation 825 

of cellulose. Cellulose 1994, 1 (1), 26‐56. 826 

 827 

163 

Curriculum vitae ‐ M.Sc. agr. Manuel BECKER  

PERSONAL  DETAILS 

    

 Name:    Becker 

 Forenames:  Manuel Stefan 

 Date of Birth:   May 31, 1981 

 Place of Birth:  Heidelberg, Germany 

 Nationality:  German     

ACADEMIC  CREDENTIALS 

  

10/2001 – 07/2004   Agricultural Sciences / Bachelor 

      University of Hohenheim, Germany 

      Major: Plant sciences   

10/2004 – 11/2006   Agricultural Sciences / Master 

      University of Hohenheim, Germany 

      Major: Plant sciences 

      Main focus: Yield physiology and management (V. vinifera L.) 

    Master thesis: „Genetisches Potential von Pinot meunier untersucht

      anhand eines Wildtyp/Mutanten‐Vergleichs (hairy/non hairy 

      phenotype)“  

Since 04/2007   Doctorate at the Institute of Horticulture, Fruit‐Growing and  

    Viticulture, Department of Applied Plant Sciences and Plant 

    Biotechnology, University of Natural Resources and Applied Life

    Sciences, Vienna: „Analyse of complex carbohydrate mixtures“  

WORKING EXPERIENCE 

 

 

 

04/2007 – 04/2010   Research associate at the Institute of Horticulture, Fruit‐Growing 

    and Viticulture, Department of Applied Plant Sciences and 

    Plant Biotechnology, University of Natural Resources and  

    Applied Life Sciences, Vienna 

04/2010 – 09/2012  Research associate at the Christian‐Doppler‐Laboratory at the  

    Department of Chemistry, University of Natural Resources and

    Applied Life Science, Vienna, Austria 

01/2012 – 02/2012  Research visit at Åbo Akademi University, Turku, Finland  

    Project: Determination of carbohydrate composition in Pulps 

    and Papers 

10/2012 – 04/2014  Internship at the ministry of rural development and consumer  

    protection (MLR), Stuttgart, Baden‐Württemberg, Germany  

Since 06/2014  Head, Viticulture and plant protection Section, State research  

    institute for viticulture and pomiculture, Weinsberg, Germany 

  

164 

Erklärung 

 

Hiermit  versichere  ich,  dass  ich  die  vorliegende  Arbeit  ohne  unzulässige  Hilfe  Dritter  und  ohne 

Benutzung anderer als der angegebenen Hilfsmittel angefertigt habe; die aus fremden Quellen direkt 

oder  indirekt übernommenen Gedanken sind als diese kenntlich gemacht worden. Bei der Auswahl 

und Auswertung des Materials sowie bei der Herstellung des Manuskriptes habe ich Unterstützungs‐

leistungen von folgenden Personen erhalten: 

 

 

 

Weitere  Personen  waren  an  der  geistigen  Herstellung  der  vorliegenden  Arbeit  nicht  beteiligt. 

Insbesondere  habe  ich  nicht  die  Hilfe  eines  oder  mehrerer  Promotionsberater(s)  in  Anspruch 

genommen.  Dritte  haben  von  mir  weder  unmittelbar  noch  mittelbar  geldwerte  Leistungen  für 

Arbeiten erhalten, die im Zusammenhang mit dem Inhalt der vorgelegten Dissertation stehen. 

 

Die Arbeit wurde  bisher weder  im  Inland  noch  im Ausland  in  gleicher  oder  ähnlicher  Form  einer 

anderen Prüfungsbehörde zum Zwecke der Promotion vorgelegt. 

 

 

 

 

 

 

.......................................................... 

Ort, Datum 

 

 

 

.......................................................... 

Unterschrift (Manuel Becker)