Pp

Ma

b

c

a

ARRAA

KSHNEGPR

1

lgmnottpsei[hhpp[v

((

0d

Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265

Contents lists available at SciVerse ScienceDirect

Colloids and Surfaces A: Physicochemical andEngineering Aspects

journa l h omepa g e: www.elsev ier .com/ locate /co lsur fa

rogress towards synthetic modelling of humic acid: Peering into thehysicochemical polymerization mechanism

arios Drososa, Maria Jerzykiewiczc, Maria Louloudib, Yiannis Deligiannakisa,∗

Laboratory of Physical Chemistry, Department of Environmental and Natural Resources Management, University of Ioannina, Seferi 2, 30100 Agrinio, GreeceDepartment of Chemistry, University of Ioannina, Panepstimioupoli, 45110 Ioannina, GreeceFaculty of Chemistry, Wroclaw University, 14 F. Joliot-Curie St., 50-383 Wroclaw, Poland

r t i c l e i n f o

rticle history:eceived 20 April 2011eceived in revised form 14 August 2011ccepted 17 August 2011vailable online 25 August 2011

eywords:

a b s t r a c t

Oxidative copolymerization of gallic acid (GA) and protocatechuic acid (PA) at 1:1 ratio provides a watersoluble humic-acid-like polycondensate (HALP) which mimics fundamental physicochemical and spec-troscopic properties of natural humic acid (HA). The redox potential (Eh) of polymerization plays adeterminative role on the physicochemical, spectroscopic and H-binding properties of the HALP as wellas on the mass yield. Trends have been systematically mapped and analyzed for two Eh values, e.g. 0 mV(HALP 0) and 100 mV (HALP 100). HALP 100 has physicochemical properties, prevailing aliphatic struc-

ynthetic humic acid-bindingMRPRallic acid

ture, which resemble those of fulvic acids (FAs) or soil-type HAs. HALP 0 has a prevailing aromatic/phenolstructure which resembles lignite-like HAs. Ionic strength had a significant impact on the charge and H-binding properties of the HALP 100. Donnan volume (VD) estimates show that HALP 100 has a moreexpanded structure. A molecular model is suggested for the polymerization reactions in connection withthe observed macromolecular, spectroscopic and H-binding characteristics of the HALPs and natural HAs.

olymerizationadical

. Introduction

Up to now, analytical and spectroscopic approaches on iso-ated humic substances have provided information on the chemicalroups involved, and/or possible molecular fractions of the humicaterial. Presently, the oxidative polymerization of polyphe-

ols in soils is thought to be among the major processesf formation of natural humic substances [1]. In this context,he transformation of organic precursors takes place via oxida-ive polymerization, forming polymeric structures. However, thehysicochemical events which lead from the molecular precur-ors to the key-properties of the HA at the macromolecular level,.g. hydrophobicity/hydrophylicity, assembly dynamics, effect ofonic strength, polyelectrolyte nature, remain largely unexplored1–7]. In this context, to solve the humic puzzle, two approachesave been followed so far (i) to isolate and characterize naturalumic material, aiming to distinguish the possible humificationathways (ii) constructing humic-like model materials from sim-

le well characterized precursors under controlled conditions5,7–14]. In principle, the synthetic-model approach could pro-ide a proper model system, while careful study of the role of

∗ Corresponding author. Fax: +30 2 641039576.E-mail addresses: drosos.marios@gmail.com (M. Drosos), mariaj@wchuwr.pl

M. Jerzykiewicz), mlouloud@uoi.gr (M. Louloudi), ideligia@cc.uoi.grY. Deligiannakis).

927-7757/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.colsurfa.2011.08.016

© 2011 Elsevier B.V. All rights reserved.

the synthesis conditions could serve as guidance for understand-ing the natural humification process. Moreover the productionof a HA-like-polymer may have potential applications, parallelto the properties of natural HA, i.e. very effective ion binding,hydrophobic sequestration of organics, electron transfer, catalysis,etc. [1,2,15,16].

In the past, approaches used to produce synthetic humic mate-rials that were based in the presence of a – biotic or abiotic –catalyst. A detailed tabulation of the used – biotic or abiotic –catalysts has been compiled in reference [17]. Recently, we havedemonstrated that a HALP can be obtained without the need ofa catalyst, by using a combination of appropriate precursors, e.g.by copolymerization of GA and PA [17]. HA-like compounds canbe further obtained by oxidation of phenolic compounds in thepresence of amino acids or by condensation of reducing sugarsin the absence or presence off amino compounds [see Table 1 inRef. [17] for details, and 18]. However, so far, proper mimickingof the H-binding properties of natural HAs has been documentedonly for a gallic–protocatechuic–humic acid-like polycondensate(GA–PA–HALP). By using EPR spectroscopy it was shown thatGA–PA–HALP material can be produced via oxidative coupling reac-tions of radicals generated by O2 at alkaline pH [17]. As shown in[17] GA–PA–HALP mimics (i) the distribution of the pKa values of

carboxyl and phenolic groups, (ii) UV–vis and FT-IR properties ofnatural HA and (iii) the radical properties of the natural HA.

The differences that were reported for HALP vs. natural HAwere that (i) the 13C NMR for HALP showed significantly lower

Physic

asribfu

tsIrp

2

2

fMfcpIf

2

dpbcs

oafapaspssswsrcssi

tt2cH

[twt

tion of Hartmann–Hahn conditions [26,27]. At least 18,000 singlescans were collected for each sample. These NMR conditions arethose typically used for HA [27]. For data analysis, the spectra

M. Drosos et al. / Colloids and Surfaces A:

liphatic-C compared to natural HA, and (ii) H-binding propertieshowed a low-sensitivity to ionic strength. According to the cur-ently adopted models, e.g. Donnan theory [19–21], the effect ofonic strength is intimately linked to the macromolecular assem-lage profile of the humics. These observations apparently originaterom differences between HALP and natural HA at the macromolec-lar level.

Herein we show that by modifying the polycondensation reac-ion via the redox potential we can obtain a HALP which mimicsuccessfully the ionic strength effect on H-binding of natural HA.n combination with spectroscopic data 13C NMR, FT-IR, EPR, fluo-escence, we propose a working model for HALP explaining thesehenomena, which occur in natural HA.

. Experimental

.1. Reagents

3,4,5-Trihydroxy benzoic acid (GA) (Lot. G7384), was purchasedrom Aldrich. 3,4-Dihydroxy-benzoic acid (PA) was obtained from

erck (Lot. 54443433612). K2Fe(CN)6 (Lot. 31460) was purchasedrom Riedel-de Haan and K3Fe(CN)6 (Lot. 1.04973.0100) was pur-hased from Merck. Hydrogen peroxide (30% H2O2, Perhydrol) wasurchased from Merck (Lot. 1.07210.1000). A fresh bottle was used.

HSS Standard Leonardite HA (Lot. 1S104H) was used as referenceor comparison.

.2. Polymerization procedure

The method used here is a modification of the method originallyeveloped in our previous work [17]. Herein we repeat the updatedrotocol. The polymerization was carried out in aqueous solutionubbled with air, at controlled pH = 10.5. The redox potential wasontinuously monitored by the redox electrode immersed in theolution.

Gallic acid:protocatechuic acid: For a 1:1 molar solution, 5.245 gf GA and 4.755 g of PA, were suspended in 0.5 L of Milli-Q waterdjusted at pH 1 using HCL. The mixture was continuously stirredor 12 h at 25 ◦C and pH 1 in a closed glass jar, under controlledir bubbling, for preventing the polymerization procedure to startrior to the fully solubilization of the reagents. Then the pH wasdjusted at pH 10.5 with NaOH. Two suspensions of 0.5 L of theame material were prepared. In the first suspension, the solutionotential was adjusted to Eh = +100 mV adding 1 ml of a fresh H2O2olution. Herein this will be called HALP 100. The second suspen-ion was adjusted to Eh = +200 mV adding 12 ml of a fresh H2O2olution. Herein, this will be called HALP 200. Then the mixtureas further bubbled with a continuous flow of natural air under

tirring for up to 3 days. During that period, the initially transparenteaction mixture turned to a green and then to a dark-brown/blackolor, for the first suspension (100 mV) and brown/yellow for theecond one (200 mV). A GA–PA–HALP, synthesized under O2 atmo-phere and in the absence of H2O2 (Eh ∼ 0 mV) as described in [17],t will be referred to as HALP 0.

The redox potential was not re-adjusted during the polymeriza-ion, but the pH was always adjusted at pH 10.5. In the beginning,he redox potential was 0 mV for HALP 0, 100 mV for HALP 100 and00 mV for HALP 200, while at the end of the polymerization pro-edure, these values have changed to 60 mV for HALP 0, 140 mV forALP 100 and 210 mV for HALP 200.

Finally, the I.H.S.S. procedure for HA isolation was applied

21,22]. In brief, the pH was adjusted to ∼1 with HCl and the solu-ion was allowed to precipitate for 72 h. We found that this timeas optimum for obtaining maximum precipitation. Here we have

o mention that when the pH was between 7 and 4, gaseous bubbles

ochem. Eng. Aspects 389 (2011) 254– 265 255

were released out of the suspension. Then, the precipitate obtainedafter centrifugation at 4000 rpm/15 min, was washed with Milli-Qwater to remove monomer residues and Cl−, freeze-dried at −60 ◦Cwith a CHRIST-ALPHA 1-2LD freeze drier, and stocked until furtheruse.

2.3. Redox, O2

Redox potential of the solution Eh was measured witha Metrohm platinum redox electrode (type 6.0401.100). Theelectrode was calibrated using a reference solution of 10 mMK3Fe(CN)6:10 mM K2Fe(CN)6 having Eh = +228 mV vs. Ag/AgCl, sat-urated with 3 N KCl. O2 concentrations in solution were measuredby a O2-meter (OXI-340-B, Germany) with a O2-selective electrodetype OX-325.

2.4. UV–vis spectroscopy

UV–vis absorption spectra were recorded with a PerkinElmerLambda 35 spectrophotometer. The measurements were carriedout in Milli-Q water at HALP and Leonardite HA concentration of100 mg L−1 in the presence of 50 mM NaHCO3 solution, which wasalso used as the blank [23,24]. According to standard protocols thepH of the solutions was set to 8 [23,24]. The E4/E6 ratios werecalculated as the absorbance ratio at 465 nm and 665 nm [23].

2.5. Fluorescence spectroscopy

Fluorescence spectra of the samples were recorded with aShimadzu RF-1501 fluorescence spectrophotometer, from 465 to600 nm, with excitation of 350 nm. The HALP solutions (50 mg L−1)were adjusted to pH 9 using HNO3/NaOH.

2.6. FT-IR spectroscopy

Fourier transformed-infra red spectra were collected inabsorbance mode with a Spectrum GX PerkinElmer FT-IR spec-trophotometer using potassium bromide pellets (300 mg KBr)containing 0.3–0.6 mg of freeze-dried Leonardite HA/HALP.

2.7. Electron paramagnetic resonance (EPR) spectroscopy

EPR spectra were recorded at liquid nitrogen, with a BrukerER200D spectrometer, equipped with an Agilent 5310A frequencycounter. g-Values were calibrated vs. DPPH, g = 2.0036, which wasalso used as spin standard for radical concentration as describedearlier [25].

2.8. 13C-CP-MAS NMR spectroscopy

Solid-state 13C NMR spectra of powder samples were recordedon a Bruker NMR spectrometer at a resonance frequency of400 MHz, using a Ramped-Cross Polarization MAS with a spin-ning speed of 6.8 kHz [26,27]. A contact time of (1 ms) and a pulsedelay of (400 ms) were used. A ramped 1H-pulse decreasing thepower from 100% to 50% was used to circumvent spin modula-

were divided into chemical shift regions assigned to the chem-ical group classes alkyl-C (0–45 ppm), O–alkyl-C (45–110 ppm),aromatic C (110–160 ppm), phenolic C (140–160 ppm), carboxyl C(160–190 ppm) and carbonyl C (190–220 ppm), respectively [1,27].

256 M. Drosos et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265

200150100500

0

2

4

6

8

10

12

Yie

ld (%

)

Eh (mV)

HAL P

F(t

2

HstHttsatststa

3

3

ttwpphotp

108640

-2

-4

-6

-8

0

-2

-4

-6

-8

0.1M

0.0001M

HALP_0 E = 0mV [O ]h 2

char

ge (e

q kg

-1)

pH

0.1M

0.0001M

HALP_100 E =100mV [H O ]h 2 2

Fig. 2. Charge vs. pH data for a HALP at two ionic strengths (©) 0.1 M KNO3, (�)0.0001 M KNO3. Solid lines are NICA-Donnan fit to the data by using the parameters

TY

E

ig. 1. Mass yield of HALP formation at various polymerization redox potentials Eh

V) for a mixture of 1:1 GA:PA. The base of the percentage yields is the total mass ofhe starting reagents GA and PA.

.9. H-binding

Acid–base potentiometric titration was used to measure the-binding properties of HALP and Leonardite HA. A HALP stock

olution 8 g L−1 was prepared, incubated at pH 12 for 12 h, andhen pH was adjusted to 3.5 using HNO3. Then 12.5 ml of 2 g L−1

ALP was prepared in Milli-Q water. The suspension was allowedo equilibrate for 30 min at room temperature (25 ◦C) under con-inuous stirring and purging with 99.999% N2 prior to titration. Theuspension was then titrated using a 0.05 M NaOH solution with anutomatic Metrohm 794 Basic Titrino. The initial pH was 4.0 andhe final pH was 10.5. The duration of each titration was 7–8 h. Theame procedure was applied for I.H.S.S. Leonardite HA. The titra-ions were repeated 3 times. The titrations were carried out at ionictrengths of 0.1 M and 0.0001 M, using KNO3. The acid–base titra-ion data were analyzed using the NICA-Donnan model, see [19]nd references therein.

. Results and discussion

.1. Reaction yield

The mass yield of the HALP 100 polymer extracted, at reactionime of 3 days was 2% and there was zero yield extraction out ofhe HALP 200. The yield under O2, e.g. Eh ∼ 0 mV, giving HALP 0as 12% as already found before [17] (see Fig. 1). The base of theercentage yields It is underlined that the increase of the solutionotential resulted in an acceleration of the polymerization process,

owever this has a rather counterintuitive impact on the mass yieldf the end product, e.g. the humic acid like isolate. The polymeriza-ion process was monitored by the changes of the pH and Eh. Duringolymerization, pH, e.g. which was slowly drifting bellow 10.5 as

able 1ields and NICA-Donnan parameters for HALPs and natural HAs.

Sample Yield (%) b Q1

HALP 0 12 0.67 4.05

HALP 100 2 0.71 5.30

HALP 200 0Leonardite IHSS standard [ref. [19]] 0.68 4.47

Generic HA parameters by Milne et al. [19] 0.20–0.70 3.17

Generic FA parameters by Milne et al. [19] 0.30–0.90 5.66

Lignite HA parameters by Drosos et al. [21] 0.64–0.69 3.60

Soil HA parameters by Drosos et al. [21] 0.59–0.89 3.80

rrors: b ± 0.005, log K1, m1, m2 ± 0.05, log K2 ± 0.2, and Q1, Q2 ± 0.05.

listed in Table 1. [Top panel] HALP 100 formed in the presence of H2O2. [Lowerpanel] HALP 0 formed under O2.

detailed in [17], was re-adjusted to 10.5 using NaOH. The reactionwas completed in 7 days for HALP 0, for 3 days for HALP 100 and1 day for HALP 200. Higher Eh resulted in accelerated reaction rates.However, as we show herein higher Eh gave lower HA-like fractions,e.g. higher fractions that did no precipitate at pH 1 [88% for HALP 0(yield 12%), 98% for HALP 100 (yield 2%) and 100% for HALP 200(yield 0%)]. The non-precipitating fractions did not match the spec-troscopic properties of FA, therefore no further investigation wascarried out on them.

During polymerization, we have observed that CO2 bubbleswere released when pH was lowered to acidic values. This infor-mation, together with the finding of diminished reaction yield inHALP 100 will be further analyzed in the context of a physicochem-ical reaction model that we discuss in the following.

3.2. H-binding

The pH-dependence of the charge, determined from the H-binding data at two ionic strengths, is presented comparatively forHALP 0 vs. HALP 100 in Fig. 2. In both cases, the titration curvesshow a smeared-out titration profile, strongly reminiscent of thedistribution of the pKa values, which is a well-known property of

natural humic acids [1,19–21]. At pH = 3.9 [I = 10−4 M] the charge,for both HALP 0 and HALP 100, was −2.3 equiv. kg−1, as shown inFig. 2.

Q2 Q1 (%) Q2 (%) m1 m2 log K1 log K2

5.39 42.9 57.1 0.47 0.18 4.00 7.603.75 58.6 41.4 0.35 0.30 3.90 9.90

3.30 57.6 42.4 0.52 0.52 3.22 8.002.66 55.6 44.4 0.55 0.43 3.09 7.982.57 70.1 29.9 0.41 0.57 2.65 8.602.78 56.4 43.6 0.40 0.19 3.03 8.201.28 74.8 25.2 0.38 0.35 3.50 8.48

M. Drosos et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265 257

Table 2Donnan volume VD-values [in L kg−1] calculated for the HALP and IHSS LeonarditeHA for I = 0.0001 M and I = 0.1 M.

Material b-Value VD [L kg−1]I = 0.0001 M

VD [L kg−1]I = 0.1 M

HALP 100 0.71 ± 0.005 355 ± 10 2.6 ± 0.1

afFd

3

btTomb

shopa[iphcfaHd

cwaf

Q

wptaattQaQrt[tTmaco

800700600500400300

0.0

0.2

0.4

0.6

0.8

1.0

1.2

HALP_1 00

Abs

orba

nce

wavelengt h (nm)

E4/E6 r atio IHSS Leon ardite HA= 5.08 HALP_0 =6.79 HALP_100 =7.15

IHSS Leonard ite HA

HALP_0

H-binding data show that HALP is rich in water-accessible charge-able groups with a charge variation �Q/�pH ∼ 4.9 [equiv. kg−1]which is higher than all the �Q/�pH values reported [19] for nat-ural humic acids.

5001000150020002500300035004000

HALP_0

HALP_10 0(iv)

(iii )

(ii )

Abs

orba

nce

GA+PA(1:1 )

(i)

IHSS Leon ardite HA

HALP 0 0.67 ± 0.005 224 ± 10 2.3 ± 0.1IHSS Leonardite HA 0.68 ± 0.005 251 ± 10 2.4 ± 0.1

For HALP 0, the difference of H-binding sites between pH 10.5nd pH 4 (�Q/�pH[pH 4–10.5]) is 5.2 equiv. kg−1 [I = 10−4 M] whileor HALP 100, �Q/�pH[pH 4–10.5] = 4.9 equiv. kg−1 [I = 10−4 M] (seeig. 2). This indicates that the higher Eh of HALP 100, results in aecrease on the number of H-binding i.e. 5.2 vs. 4.9 equiv. kg−1.

.2.1. Ionic strength effectIonic strength had a significant impact on the charge and H-

inding properties of the HALP 100, see Fig. 2 upper panel. Inhe presence of 0.1 M KNO3 the initial charge is −4.1 equiv. kg−1.his shift vs. ionic strength is in accordance with what is typicallybserved for natural HAs [1,19–21], and can be described mathe-atically by a change of the activity coefficient [1,19,20], as it will

e shown in the following.The curves in Fig. 2 fall in the region of [charge vs. pH] curves

ummarized by Milne et al. (see Fig. 3 in reference [19]) for naturalumic acids. This pattern shows that the initial charge [at pH 4]f humic substances can vary from −1 to −4 equiv. kg−1, while atH 10.5 the charge can vary between −3.8 and −8.2 equiv. kg−1 as

result of the well-known polyelectrolyte pK distribution pattern19]. According to the current concept, in natural HA, increasedonic strength allows more groups to be deprotonated at a givenH thus resulting in an increase of the apparent total charge of theumic acid [1,19,20]. This results in an upshift of the charge vs. pHurve, i.e. similar to that observed for the HALP 100 in Fig. 2. Thusor HALP 0, �Q/�pH[pH 4–10.5] is 5.2 equiv. kg−1 for 10−4 M KNO3,nd 4.5 equiv. kg−1 for 0.1 M KNO3. A striking observation is thatALP 0 appears to be practically insensitive to ionic strength (I),espite the fact that it bears more H-binding groups than HALP 100.

Theoretical modelling: This salt-effect can be analyzed theoreti-ally by a pKa shift introduced via an electrostatic Boltzmann termhich incorporates both the charge of the humic macromolecule

s well ionic strength effects [20]. This concept, mathematicallyormulated by the so-called NICA-Donnan model [20]:

H = Qo + Q1 × (K1[HS])m1

1 + (K1[HS])m1+ Q2 × (K2[HS])m2

1 + (K2[HS])m2(1)

hich resulted in the fits shown by solid lines in Fig. 2. Q1 is the totalroton site density of carboxyl groups [in equiv. kg−1] and Q2 is theotal proton site density of phenol groups. K1 and K2 are the medianffinity constants for carboxyl and phenol groups, and m1 and m2re heterogeneity parameters of the carboxyl and phenol groupshat reflect the combined effect of the intrinsic-heterogeneity andhe ion-specific-heterogeneity (0 < m < 1). The used parameters Q1,2, m1, m2, pK1, pK2, are listed in Table 1. In applying relation (1), wessumed that the charge at pH 4 corresponds to the initial charge,0 [19,20] of the HALP. In natural-HA the validity of this assumption

equires further justification, and would be true only if at this pH allhe protons are derived exclusively from deprotonation of the HA19,20]. Ash content of natural HA is a major error factor in estima-ion of Q0 [19,20]. The data for HALP are free of such interferences.hus, HALP can serve as reference error-free material for the esti-

ation of charge, in H-binding experiments. Noticeably, HALP 100

ppears to have a high initial charge of Q0 = −2.3 equiv. kg−1, e.g.ompared with the literature Q0 values of −1 to −2 equiv. kg−1

bserved for natural HAs [19].

Fig. 3. UV–vis spectra for IHSS Leonardite HA (dashed line), HALP 100 (solid line)and HALP 0 (dotted line) and the E4/E6 ratio of each material.

In this model, ionic strength effects are considered to affect theDonnan volume VD (in L kg−1) which for humic and fulvic acids istaken by the empirical equation [19,20]:

log VD = −1 + b(1 − log I) (2)

where b is an adjustable parameter with positive value dependingon the humic substance and I is the ionic strength. Donnan volumeis balancing the total charge q of the humic ‘particle’ according toq

VD+

∑i

zi(ai,D − ai) = 0 (3a)

where q is in mol unit charge kg−1, and

ai,D ≡ ai exp(−ziF�D

RT

)(3b)

is the activity component-i in the Donnan volume and � D is theDonnan potential [19,20].The key-effect of relation (1) is that atincreased I, the Donnan volume will shrink (lower VD). Overall the

Wavenumber (cm-1)

Fig. 4. FT-IR signals (in KBr) of (iv) IHSS Leonardite HA, (iii) HALP 100, (ii) HALP 0and (i) gallic:protocatechuic acid [1:1].

258 M. Drosos et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265

2,01502,01002,00502,00001,99501,9900

pH 5dx"/

dH (a

rb. u

n.)

pH 12

g=2.0045

tfns

3

tped

fvflHAwfuOccvfurtpieb

Fig. 6. Solid-state 13C NMR spectra for IHSS Leonardite HA, HALP 100 and HALP 0.

700650600550500450400350

0

50

100

150

200

250

HALP_0

Fluo

resc

ence

Inte

nsity

Emmision (nm)

λExcitation=350nm

IHSS_Leonardite_HA

HALP_100

TCC

g-value

Fig. 5. EPR signal of HALP 100 in aqueous solution at pH 5 and 12.

The charge properties of HALP 0 were not influenced much byhe ionic strength (from −2.3 to −2.6 equiv. kg−1 for ionic strengthrom 0.0001 M to 0.1 M) while HALP 100 seems to be affected sig-ificantly by ionic strength (from −2.3 to −4.1 equiv. kg−1 for ionictrength from 0.0001 M to 0.1 M).

.2.2. NICA-Donnan parametersThe solid lines in Fig. 2 are NICA-Donnan fits, using the parame-

ers listed in Table 1. For comparison, we also list the NICA-Donnanarameters for I.H.S.S. Leonardite [19] as well as the generic param-ters suggested by Milne et al. [19], e.g. based on average valueserived for an extended data set for natural humic and fulvic acids.

In Table 1, Q2, i.e. phenolic groups charge is higher for HALP thanor natural HAs. Q1 values, i.e. carboxylates’ charge, fall between thealues for Leonardite HA and the higher reference values observedor natural HAs. In HALP the high Q2 originates from the pheno-ic groups of GA and PA. HALP 0 has less carboxyl charge thanALP 100, while HALP 100 has less phenolic charge than HALP 0.ccording to the NICA-Donnan model [20] m1 and m2 reflect theidth of the distribution of K1 and K2. Accordingly, the m2 value

or HALP 0 which is clearly lower than the values reported for nat-ral HAs, can be attributed to the more homogeneous nature of theH functionalities, i.e. since they are derived from only two pre-ursors, GA and PA in HALP. However, HALP 100 gives m1 valueslose to m2. An interesting observation is that the log K2 (log K1)alues for HALP 0 although comparable with the log K2 (log K1)or natural HAs, see Table 1, they are well bellow the pKa val-es, i.e. 8.4, 10.5, 11.5, for the phenolic groups of GA and PA [25],espectively. Apparently, incorporation and/or transformation ofhe GA–PA functionalities into the HALP are responsible for this

Ka-shift. This is a nice demonstration of the pKa-shift phenomenon

nduced due to incorporation of a protonable moiety into a poly-lectrolyte polymer [1]. The log K2 value of HALP 100 is found toe around 10. This value falls to the upper limit of log K2 values

able 3arbon concentration in mol kg−1 and carbon concentration as a percent mass fraction (%P-MAS 13C NMR spectroscopy.a

Material Aliphatic (0–110 ppm) Aromatic (110–140 ppm) Pheno

HALP 100 12.9 (31.5) 16.0 (29.2) 4.0 (9.HALP 0 8.4 (25.8) 13.0 (23.0) 5.5 (16Leonardite HA 15.6 (35.9) 16.0 (36.9) 3.1 (7.

a Error: ±0.2.

Fig. 7. Fluorescence spectra for IHSS Leonardite HA (dotted line), HALP 100 (solidline) and HALP 0 (dashed line).

reported for natural HAs [19], however uncertainties in its determi-nation, e.g. as discussed in [1,17,25], might result in an overestimateof this value.

The b factor in HALP 0 is found to be 0.67, while the b factor forHALP 100 is 0.71. This means that when the ionic strength is 0.1 M,the Donnan volume for HALP 0 is 2.19 L kg−1 and for HALP 100 is2.63 L kg−1, but for ionic strength of 0.0001 M, the Donnan volumefor HALP 0 is 224 L kg−1 and for HALP 100 is 355 L kg−1.

The Donnan volume should be viewed as a hydrodynamic vol-

ume occupied by the organic polymer and the hydrating watermolecules associated with its structure [1,19,20]. In this contextthe data in Table 2, indicate that HALP 100 has more expandable,

), in brackets, for Leonardite HA, HALP 0 and HALP 100 determined by solid-state

lic (140–160 ppm) Carboxyl (160–190 ppm) Carbonyl (190–220 ppm)

8) 7.5 (18.3) 4.6 (11.2).9) 8.5 (26.1) 2.7 (8.2)

1) 7.8 (18.0) 0.9 (2.1)

Physic

eH

3

3

HFpswlrll

3

tFam(ac(1(op8iHu

3

dEonrTi(rria4g

srcr[apnLupwe

M. Drosos et al. / Colloids and Surfaces A:

.g. more flexible, conformation than HALP 0, as seen also by the-binding data in Fig. 2.

.3. Spectroscopy

.3.1. UV–visLeonardite HA has smaller E4/E6 ratio (5.08) than HALP, while

ALP 0 has smaller E4/E6 ratio (6.79) than HALP 100 (7.15) (seeig. 3). The absolute meaning of the E4/E6 ratio is a rather com-lex issue [1,16] however trends based on HA and FA from variousources show that E4/E6 tends to be larger when the moleculareight becomes smaller and/or when the condensation degree is

ower, e.g. a hydrolyzed humic substance has much smaller E4/E6atio than an untreated one [1,16,17]. FA structures tend to havearger E4/E6 ratio than HA, which might imply a more Fulvic-likeess condensed structure for HALP 100 vs. HALP 0.

.3.2. FT-IRFT-IR spectra for HALP 0 and HALP 100 (Fig. 4) exhibited the

ypical FT-IR bands of humic substances [1]. For comparison, inig. 4 we display an FT-IR spectrum for IHSS Leonardite HA andn FT-IR spectrum for a 1:1 molar GA:PA mixture—with no poly-erization. The FT-IR peaks are as follows: regions of 3370 cm−1

H-bonded OH groups), 3195 cm−1 (OH stretching of carboxyl OHnd phenolic OH), 2608 cm−1 (hydrogen-bonded OH stretching ofarboxyl and phenolic OH), 1716 cm−1 (phenolic esters), 1614 cm−1

aromatic C C, H-bonded C O, and/or dissociated COOH groups),385 cm−1 (aromatic ring frequency), 1285 cm−1 and 1195 cm−1

C–O stretching and OH deformation of COOH, C–O stretchingf aryl esters), 1115 cm−1 (C–O stretching of polysaccharide orolysaccharide-like substances), and the small peaks at 895 cm−1,35 cm−1, 772 cm−1, 740 cm−1, and 723 cm−1 (aromatic C–H bend-

ng vibrations). There are no significant changes between theALP 0 and HALP 100 samples. This shows that treatment of HALPsing H2O2 does not alter the functional groups of the polymer.

.3.3. EPR spectroscopy—radical propertiesHALP 100 contained stable monomeric radicals, with a pH

ependent concentration, see Fig. 5. At pH 12 under O2, i.e.h = +150 ± 5 mV, a strong EPR signal is detected with a g-valuef 2.0045 ± 0.0002 with a linewidth of �H = 6 Gauss. This EPR sig-al is characteristic of phenolic �-type radicals and bears strongesemblance to the indigenous radicals of natural HA [16,25,28].hus the g-value shows that the molecular identity of the rad-cal bearing moieties is similar in both GA (2.0040) and HALP2.0045) as well as in Leonardite HA (2.0036–2.0047). This cor-oborates our previous suggestion that GA is a good model for theadical properties in HAs [25,28]. Quantitative estimate of the rad-cal concentration, by double integration of the EPR signal, gives

high radical concentration of 8.9 × 1018 spins g−1 for HALP 0 vs..4 × 1018 spins g−1 for Leonardite HA at pH 12 while HALP 100ives only 4.0 × 1017 spins g−1 at the same pH.

The radicals HALP 100 were pH dependent. At pH < 7 a severeignal decrease occurs, see Fig. 5. Importantly, this effect iseversible. After posing the HALP at pH 5, the pH 12 EPR signalan be 100% recovered after raising pH to 12 for 20 min. The pHeversibility is not observed for GA in solution. As we have shown25] GA alone, under a pH cycle between pH 5 and 12, results inn irreversible disappearance of EPR signals [25]. In contrast, theH-reversibility effect is a hallmark of the radical properties ofatural HA [1,16,25] and was analyzed in detail in reference [28].ocal effects such as �-stacking, hydrophobic sequestration mod-

late the stability of the radicals in HAs. The high aromatic andhenolic character of HALP and the high radical content concurith this idea. Thus, both the phenolic groups and the aromatic

nvironment contribute to stabilization of radicals in HALP. This

ochem. Eng. Aspects 389 (2011) 254– 265 259

offers independent evidence that in natural HAs [1,16,25,28,29]the unusual stability of the indigenous radicals in HAs even, i.e.in aqueous solutions under various pH and redox conditions is acombined result of local effects, i.e. phenolic groups and aromaticenvironment [21,28].

3.3.4. 13C-CP-MAS NMRSolid-state 13C NMR spectra for HALP 100 and HALP 0 are dis-

played in Fig. 6. Compared with the NMR spectra for HALP 0, Fig. 6,HALP 100 has much more intense broad feature at 0–60 ppm. Innatural HA this is typically resolved in 13C NMR spectra see forexample Leonardite HA in Fig. 6 and typically attributed to alkylchains [1,27]. This signal is absent in the NMR spectra for GA and PAwhich lack aliphatic structures [17]. This provides strong evidencethat ring-opening reactions are contributing to the HALP forma-tion. This corroborates previous evidence that the polymerizationmechanism of phenolics involves ring-opening reactions [13,30].Pyrogallol was found to be the easier to be ring-cleaved [30] by non-tronite. The present data show that ring opening in HALP can occurwith no catalyst used. A fundamental quantum mechanical analy-sis of this mechanism is still lacking in the literature. So, adjustingthe potential to 100 mV using H2O2 as an oxidizer agent, we obtainHALP 100 that has ∼35% more alkyl groups (Fig. 6 and Table 3).

Importantly in Fig. 6 and Table 3, the phenolic C (140–160 ppm)originating from the ring phenolic groups of GA and/or PA,appear strong in the NMR spectrum of the HALP 0 and HALP 100compared with Leonardite HA, while HALP 0 and HALP 100 con-tain a lower percentage of aliphatic C (0–45 ppm), and aliphaticC–O (45–110 ppm). Thus, although ring-opening reactions occur,certain aromatic-ring phenol structures of GA/PA appear to beincorporated in the HALP. Overall the 13C NMR data demonstratethat HALP 100 possesses more alkyl-C structural units than HALP 0.These alkyl-C are absent in the reacting monomers and are indica-tive of ring opening-reactions occurring in the polymerizationreaction.

3.3.5. Fluorescence spectroscopyLeonardite HA has more intense fluorescence spectrum than

HALP (see Fig. 7), while HALP 100 has more intense fluorescencespectrum than HALP 0. In general, a high fluorescence intensityderives from a higher concentration of fluorescent chromophores,e.g. phenolic, aromatic rings. A low fluorescence intensity couldresult from quenching when the condensation degree of the ringsis increased [1]. Taking into account the NMR, UV–vis and EPR andNICA-Donnan information we consider that data in Fig. 7, reflectthe interplay of higher ring content in HALP 0 whose fluorescenceis quenched due to the higher degree of ring-ring interaction thanin HALP 100.

Overall the present data allow peering into the key-aspects ofthe physicochemical polymerization process of production of ahumic acid like polymer (polycondensate). Starting by a 1:1 mix-ture of GA and PA, a modification of the oxidation potential Ehresults in significant changes in the spectroscopic and macromolec-ular properties of the HALP produced. These can be summarized asfollows:

3.4. Macromolecular properties

(a) Increased Eh results in increased aliphatic character by e.g. 13CNMR shows a 35% in HALP 100 vs. HALP 0.

(b) Increased Eh results in increased Donnan volume and therefore

increased sensitivity of the H-binding to ionic strength effects.Using Eq. (2) from the NICA-Donnan theory, the b-values cal-culated from the fit of the H-binding data, Table 1, give theVD-values listed in Table 2.

260 M. Drosos et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265

Table 4Comparison of carbon concentration (%) for HALP 0 and HALP 100 determined by solid-state CP-MAS 13C NMR spectroscopy vs. calculated based on the structural modelspresented in Figs. 8 and 9C.a

Material Aliphatic (0–110 ppm) Aromatic (110–140 ppm) Phenolic (140–160 ppm) Carboxyl (160–190 ppm) Carbonyl (190–220 ppm)

HALP 100 31.5 29.2 9.8 18.3 11.2Model of HALP 100 (Fig. 9C) 34.5 29.9 11.5 14.9 9.2HALP 0 25.8 23.0 16.9 26.1 8.2Model of HALP 0 (Fig. 8C) 25.6 25.5 17.0 17.0 14.9

a Error: ±0.2.

COO

O

O O O

O

COO

O

O

O

COO

COO

O

O

O

O

O

COO

COO

O

OOH

O

O

O

COO

COO

O

OHO

O

COO

COO

O

O

O

OH

O

COO

COO

O

O

OO

-CO2

O

COO

COO

O

O

O

OO

+

-H+

B

A

Fig. 8. (A) Suggested mechanism for radical polymerization of a 1:1 GA:PA mixture under O2 at alkaline pH. (B). Reaction of a GA radical with the radical of the alkyl-phenolate(B) forming a basic block unit (C). (C) A proposed macromolecular structure for HALP 0 under O2 at alkaline pH.

M. Drosos et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265 261

COO

O CO O

O

O

O

COO

O

O O

O

COO

COO

O

O

O

OOC O

O

O

C

Fig. 8. Continued

O

OOC

COO

O

OO

OOC O

O

O

COO

O

O

O

OOOC

COOO

O

O

OOC

OO

O

O

OOC

COO

O

OOOOC O

O

O

COO

O

O

OOOC

COOO

O

O

OOC

OO

O

COO

further oxidationduring polymerization proces s

a first reactive macromolecul e

a putative structural model for HALP_0

Fig. 8. Continued

262 M. Drosos et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265

O

CHOO

COO

O

O

O

(B)

2

O

CHOO

COO

O

O

O

O

CHOO

COO

O

O

O

(D)

O

COO

COO

O

O

O

O

COO

COO

O

O

O

O

COO

COO

O

OO

O

COO

COO

O

O

O+2OH

(D)

A

B

Fig. 9. (A) Formation of tetramer (D) in the presence of H O . (B) Proton abstraction from the tetramer (D) in the presence of H O . (C) A proposed macromolecular structuref

epHuseo

3

flHtitH

mHdtasHn

m

2 2

or HALP 100 in the presence of H2O2.

Thus the VD-estimates show that HALP 100 has a morexpanded structure. At low ionic strength, 1 kg of HALP 100 is occu-ying 355 L of hydrodynamic volume, while 1 kg of IHSS LeonarditeA or HALP 0 occupies a significantly smaller hydrodynamic vol-me, e.g. 230–250 L. In this context at non-negligible IS = 0.1 theynthetic humic acids – as well as IHSS Leonardite HA – shrink,.g. as predicted by the NICA-Donnan theory, by about tenfold,ccupying a Donnan volume of less than 3 L.

.5. A structural model

The present data reveal that HALP 100 has a significantly moreexible, more expanded three-dimensional structure, e.g. thanALP 0 or IHSS Leonardite HA. E4/E6 ratio and fluorescence spec-

roscopy show that HALP 0 is more condensed than HALP 100. Thiss evidence of structural differences between HALP 100 and HALP 0hat could point to differences in the Donnan volumes, found from-binding.

These macromolecular differences are also reflected in theolecular fingerprints: The present 13C NMR data show thatALP 100 has significantly more aliphatic structure vs. HALP 0. EPRata show that the concentration of the stable radicals is about 20imes lower in HALP 100 vs. HALP 0. The stable radicals in HAsnd HALP are localised on phenolic groups [16,17,25] thus EPRhows that HALP 100 contains less aromatic phenolic groups. The

-binding data confirm this view, i.e. HALP 100 contains less phe-olics vs. carboxylates.

In natural organic matter, large Donnan volumes correspond toore aliphatic structures [1,19], e.g. as exemplified by fulvic acids

2 2

which reportedly have more aliphatic structure, increased b-valuesand VD [19]. Characteristically, in our previous work on comparisonof HA from soils and lignites [21] we found that HAs isolated formsoils are poorer in phenolic/aromatic units and richer in aliphatics,e.g. when compared with lignite HAs including IHSS Leonardite HA[21]. This analysis shows that spectroscopic properties of HALP 100resembles to FA and can be pertinent as a model for soil and aquaticHA, while HALP 0 is a more lignite-like HA model.

3.5.1. Radical polymerization chemistryAt this point it is tempting to discuss possible pathways of rad-

ical polymerization which result in the observed differences in themacromolecular characteristics of HALP 0 vs. HALP 100. The perti-nent information which should be taken into account is: solid-state13C NMR spectra for HALP show that ring-opening reactions of theGA, PA monomers are involved in the HALP formation. EPR spec-troscopy shows that HALP contains stable phenol-based radicals,with pH-dependent concentration. FT-IR spectra show that the useof hydrogen peroxide does not alter the functional groups’ identity.A key feature is that the optimal yield for HALP formation requires1:1 concentrations of GA and PA [17]. Accordingly, we suggest thefollowing possible reactions:

3.5.1.1. Reactions for HALP 0.

(a) The ring-opening step: At pH 10.5, gallic acid forms long-livedradicals, e.g. detected by EPR, while protocatechuic acid doesnot stabilise radicals [25]. Accordingly, in a 1:1 reaction mix-ture, a GA radical may react with PA resulting in an ether type

M. Drosos et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 389 (2011) 254– 265 263

O

COO

COO

O

OO

O

COO

COO

O

O

O

OOC

O

OOC

O

O

O

OOOC

COO

OO

O

O

COO

COO

OO

O

OCOO

OOC

O

OO

a first reactive macromolecule

further oxidationduring polymerization process

O

COO

COO

OO

O

COO

COO

O

O

O

OOC

O

OOC

O

O

O

OOOC

COO

OO

O

O

COO

COO

OO

O

OCOO

OOC

O

OO

OOC

a putativestructuralm o delforHAL P _100

C

Fig. 9. Continued

dimer which undergoes a nucleophilic attack by an OH− (struc-ture A in Fig. 8A). Subsequently, protocatechuic ring-openingcleavage and deprotonation leads to an anionic adduct which isreadily decarboxylated forming a radical of an alkyl-phenolatedimer (B in Fig. 8A). As shown in Fig. 8A a CO2 molecule is pre-

dicted to be released for the formation of adduct B. As statedherein, this CO2 release was observed as bubbles during thepolymerization procedure when the pH was adjusted from 10.5to 1.

(b) Polymerization under O2 [HALP 0]: A macromolecular chain canbe formed by a sequential reaction of a GA radical – which asshown by EPR analysis of the reaction mixture [17] is generatedunder dioxygen – with the radical of the alkyl-phenolate dimer(B) as described in Fig. 8B. The obtained trimeric structure (C)

can be seen as the elementary block unit of the HALP 0. Thisoccurs by a radical polymerization process via proton abstrac-tion of the olefinic double bonds and subsequent addition ofradical fragments. An example of a reactive polymeric structure

2 Physic

3

(

sd

(

(

(

(

(

3

qcfpt

tmssm(aH

rFe1r

64 M. Drosos et al. / Colloids and Surfaces A:

formed by this way is proposed in Fig. 8C: further oxidation ofthis could result in the tentative structural model of Fig. 8C.The NMR data, analyzed in Tables 3 and 4 are consistent to themodel provided, e.g. 25.8% aliphatic C from NMR vs. 25.6% fromFig. 8C.

.5.1.2. Reactions for HALP 100.

(a) The ring-opening step occurs as in the case of HALP 0. However,in the presence of H2O2 the radical of dimer (B) is more abun-dant in the reaction mixture, e.g. vs. the GA radicals. Therefore,the reaction of two radicals (B) would result in the tetramer (D)(Fig. 9A).

b) Polymerization under H2O2 [HALP 100]: In the presence of H2O2,proton abstraction by the OH• radicals from the olefinic doublebonds of the tetramer (D) generates reactive sites (Fig. 9B) forevolution of the radical polymerization reaction. Fig. 9C rep-resents a possible polymeric structure via such radical reactionfollowed by further oxidation, in consistence with the NMR dataprovided in Tables 3 and 4, e.g. 29.2% aliphatic C from NMR vs.29.9% from Fig. 9C.

Accordingly, the structures suggested in Figs. 8C and 9C, canerve as working models for the interpretation of the experimentalata.

1) The structure of HALP 100 (Fig. 9C) contains more aliphaticunits compared to the structure of HALP 0 (Fig. 8C) in accor-dance with the 13C NMR data.

2) The structure in Fig. 9C contains more carboxyl vs. phenolic H-binding sites, which concurs with the H-binding data in Table 1.

3) The structure of HALP 100 (Fig. 9C) contains less phenolicradical-stabilizing hydroxyphenolic units in accordance withthe EPR data.

4) HALP 0 has low fluorescence due to ring-ring proximity, e.g.causing fluorescence quenching [1]. The higher fluorescenceof HALP 100 can be attributed to increased distances betweenthe ring-fluorophores as shown in Fig. 9C avoiding ring-ringstacking, e.g. lower fluorescence quenching.

5) Importantly, the structure in Fig. 9C will be more flexible thusmore prone to be influenced by Ionic Strength effects.

.6. Relevance to humification in natural systems

Natural HA form in time scales of centuries [1], starting fromuite diverse initial “reactants”, i.e. organic biomass [1]. Variousomplex (bio)physicochemical pathways have been hypothesizedor the humification process in nature [1]. In natural systems thesehysicochemical characteristics ultimately prevail through a con-inuum of chemical transformations.

However, despite these apparently diversified conditions, forhe sake of the present analysis three types of natural organic

atter (NOM) structures can be broadly categorized [17,19]: (1)oil-type or aquatic type HAs characterized by more aliphatictructures, (2) lignite-type HAs characterized by more aro-atic/phenolic structure, forming more compacted molecules, and

3) FAs characterized by an extremely low aromatic/phenolic char-cter, and significantly larger Donnan volume per mass unit, thanAs.

The present data show that starting from the same simple basiceactants, e.g. GA and PA, either soil-type/aquatic-type HA-like and

A-like structures or lignite-type HA model structures can be gen-rated, by modifying the redox potential of the polymerization by00 mV. Since this range of redox potentials can occur in natu-al systems, the present data exemplify how by changing just one

ochem. Eng. Aspects 389 (2011) 254– 265

parameter may result in diverse end-products of the humificationprocess.

An intriguing finding of the present wok is the correlation of theoxidation potential with the mass yield of the obtained HALP. Asshown in Fig. 1, the higher the oxidation conditions the lower theHA mass yield. This shows that, in an environment under limitedoxidation conditions the humification is more efficient, e.g. highermass yield, and the structures obtained are more lignite-like. Atmoderate oxidation potentials the humification yield is decreasingand the obtained structures are more fulvic-like. This might be ofrelevance to the significantly lower abundance of FAs in naturalsystems. Strongly oxidizing conditions would result in a minimiza-tion of the humification mass yield. These observations providehighlight on the importance of the oxidation potential in naturalsystems in the evolution of the humification process.

4. Conclusions

In summary the present work, together with the precedentone [17] demonstrate that starting from simple – structurally andphysicochemically pertinent – reactants a careful control of thephysicochemical conditions may result in controllable diversifiedend-products which can be useful models of natural organic matter.

The redox potential (Eh) of polymerization plays a determina-tive role on the physicochemical properties of the HALP as well ason the mass yield. The mass yield drops drastically at increasing Eh.13C NMR data show that HALP 100 has significantly more aliphaticstructure vs. HALP 0. EPR data show that the concentration of thestable radicals is about 20 times lower in HALP 100 vs. HALP 0. Thestable radicals in HAs and HALP are localized on phenolic groups[17,25] thus EPR shows that HALP 100 contains less aromatic phe-nolic groups. The H-binding data confirm this view, i.e. HALP 100contains less phenolics vs. carboxylates.

Overall, HALP 100 has physicochemical properties, and a pre-vailing aliphatic structure, which resemble those of fulvic acidsor soil-type/aquatic-type HAs. HALP 0 has a prevailing aromaticstructure which resembles lignite-like HAs. Ionic Strength hada significant impact on the charge and H-binding properties ofthe HALP 100. Donnan volume VD-values estimates show thatHALP 100 has a more expanded structure. At low ionic strength 1 kgof HALP 100 is occupying 355 L while 1 kg of IHSS Leonardite HA orHALP 0 occupies a significantly smaller volume, e.g. 230–250 L.

A molecular model is suggested for the polymerization reactionsin connection with the observed macromolecular, spectroscopicand H-binding characteristics of the HALPs. Ring opening reactionsare involved in the HALP formation.

Acknowledgment

M.D. was supported by the Bodosakis Foundation.

References

[1] F.J. Stevenson, Humus Chemistry: Genesis, Composition, Reactions, 2nd ed.,John Wiley & Sons, New York, 1994.

[2] J.M. Bollag, J. Dec, P.M. Huang, Formation mechanisms of complex organicstructures in soil habitats, Adv. Agron. 63 (1998) 237–266.

[3] W.G. Sunda, D.J. Kieber, Oxidation of humic substances by manganese oxidesyields low-molecular-weight organic substrates, Nature 367 (1994) 62–65.

[4] Role of soil minerals in transformations of natural organics and xenobiotics inthe environment, in: J.-M. Bollag, G. Stotzky (Eds.), Soil Biochemistry, vol. 6,Marcel Dekker, New York, USA, 1990, pp. 29–115.

[5] H. Shindo, P.M. Huang, Role of Mn(IV) oxide in abiotic formation of humicsubstances in the environment, Nature 298 (1982) 363–365.

[6] J.M. Bollag, Enzymes catalyzing oxidative coupling reactions of pollutants, in:H. Siegel, A. Siegel (Eds.), Metal Ions in Biological Systems, Marcel Dekker, NewYork, NY, 1992.

Physic

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

M. Drosos et al. / Colloids and Surfaces A:

[7] K. Haider, J.P. Martin, Synthesis and transformation of phenolic compounds byEpicoccum nigrum in relation to humic acid formation, Soil Sci. Am. Proc. 31(1967) 766–772.

[8] J.P. Martin, K. Haider, D. Wolf, Synthesis of phenolic polymers by Hendersonulatoruloidea in relation to humic acids formation, Soil Sci. Am. Proc. 36 (1972)311–315.

[9] J.P. Martin, K. Haider, C. Saiz-Jimenez, Sodium amalgam reductive degradationof fungal and model phenolic polymers, soil humic acids, and simple phenoliccompounds, Soil Sci. Am. Proc. 38 (1974) 760–765.

10] T.S.C. Wang, S.W. Li, Y.L. Ferng, Catalytic polymerization of phenolic compoundsby clay minerals, Soil Sci. 126 (1978) 15–21.

11] M.C. Wang, P.M. Huang, Catalytic polymerization of hydroquinone by nontron-ite, Can. J. Soil Sci. 67 (1987) 867–875.

12] T.S.C. Wang, M.M. Kao, P.M. Huang, The effect of pH on the catalytic synthesisof humic substances by illite, Soil Sci. 129 (1980) 333–338.

13] M.C. Wang, P.M. Huang, Abiotic ring cleavage of pyrogallol and the associ-ated reactions as catalyzed by a natural soil, Sci. Total Environ. 81–82 (1989)501–510.

14] M.C. Wang, P.M. Huang, Cleavage and polycondensation of pyrogallol andglycine catalyzed by natural soil clays, Geoderma 112 (2003) 31–50.

15] R.H. Schulten, M. Schnitzer, Chemical model structures for soil organic matterand soils, Soil Sci. 162 (1997) 115–130.

16] N. Senesi, C. Steelink, in: H.B. Hayes, P. MacCarthy, R.L. Malcolm, R.S. Swift (Eds.),Humic Substances II: Search of Structure, Wiley, New York, 1989, pp. 373–408.

17] E. Giannakopoulos, M. Drosos, Y. Deligiannakis, A humic-acid-like polycon-densate produced with no use of catalyst, J. Colloid Interface Sci. 336 (2009)59–66.

18] K. Taguchi, Y. Sampei, The formation, and mineral and CaCO3 association reac-tions of melanoidins, Org. Geochem. 10 (1986) 1081–1089.

19] C.J. Milne, G.D. Kinniburgh, E. Tipping, Generic NICA-Donnan model parame-ters for proton binding by humic substances, Environ. Sci. Technol. 35 (2001)2049–2059.

[

ochem. Eng. Aspects 389 (2011) 254– 265 265

20] D.G. Kinniburgh, W.H. Van Riemsdijk, L.K. Koopal, M. Borkovec, M.F. Benedetti,M.J. Avena, Ion binding to natural organic matter: competition, heterogeneity,stoichiometry and thermodynamic consistency, Colloids Surf. A: Physicochem.Eng. Aspects 151 (1999) 147–166.

21] M. Drosos, M. Jerzykiewicz, Y. Deligiannakis, H-binding groups in lignite vs. soilhumic acids: NICA-Donnan and spectroscopic parameters, J. Colloid InterfaceSci. 332 (2009) 78–84.

22] E.M. Thurman, R.L. Malcolm, Preparative isolation of aquatic humic substances,Environ. Sci. Technol. 15 (1981) 463–466.

23] Y. Chen, N. Senesi, M. Schnitzer, Information provided on humic substances byE4/E6 ratios, Sci. Soc. Am. J. 41 (1977) 352–358.

24] M. Schnitzer, in: A.L. Page, R.H. Miller, D.R. Keeney (Eds.), Methods of Soil Anal-ysis. Part 2, Am. Soc. Agron. and Soil Sci. Soc. Am., Madison, WI, 1982, pp.581–594.

25] E. Giannakopoulos, K.C. Christoforidis, A. Tsipis, M. Jerzykiewicz, Y. Deligian-nakis, Influence of Pb(II) on the radical properties of humic substances andmodel compounds, J. Phys. Chem. A 109 (2005) 2223–2232.

26] O.B. Peersen, X. Wu, I. Kustanovich, S.O. Smith, Variable-amplitude cross-polarization MAS NMR, J. Magn. Reson. A 104 (1993) 334–339.

27] M.A. Wilson, Solid state NMR spectroscopy of humic substances: basic con-cepts and techniques, in: M.H.B. Hayes, P. MacCarthy, R.L. Malcolm, R.S.Swift (Eds.), Humic Substances II: In Search of Structure, Wiley, NY, 1989,pp. 309–338.

28] K.C. Christoforidis, S. Un, Y. Deligiannakis, High-field 285 GHz electron param-agnetic resonance study of indigenous radicals of humic acids, J. Phys. Chem. A111 (2007) 11860–11866.

29] K.C. Christoforidis, S. Un, Y. Deligiannakis, Effect of metal ions on the indige-

nous radicals of humic acids: high field electron paramagnetic resonance study,Environ. Sci. Technol. 44 (2010) 7011–7016.

30] M.C. Wang, P.M. Huang, Pyrogallol transformations as catalyzed bynontronite, bentonite, and kaolinite, Clay Clay Miner. 37 (1989)525–531.