Durana et al.

1176 Journal of the Air & Waste Management Association Volume 52 October 2002

ISSN 1047-3289 J. Air & Waste Manage. Assoc. 52:1176-1185

Copyright 2002 Air & Waste Management Association

TECHNICAL PAPER

ABSTRACTThis paper presents results of the processing and valida-tion of data collected by an automatic gas chromato-graph (AGC). This system was used to monitor 62 volatileorganic compounds (VOCs) in urban air in the BasqueCountry, Spain. The nonpolar compounds (C2–C10) iden-tified—paraffins, olefins, aromatics, and chlorinatedcompounds—accounted for 88% of the mass of total non-methane hydrocarbons (TNMHCs) in ambient air. Theevaluation of linearity, precision, detection limits (DLs),and stability of retention times (RTs) indicates that theequipment is suitable for measuring ambient air auto-matically for prolonged periods (6 months). The calibra-tion of the equipment using response factors calculatedon the basis of the effective carbon number (ECN)showed variations of over 10% for acetylene, isoprene,and n-hexane. The results provided by the automaticchromatograph correlated significantly with simulta-neous results from other widely used techniques for de-termining VOCs in ambient air: (1) portable GC,equipped with photoionization detector (PID), and (2)active adsorption on Tenax-TA followed by thermal des-orption and chromatographic analysis.

INTRODUCTIONPrecise information on the levels of volatile organic com-pounds (VOCs) in the air is essential for monitoring ofphotochemical air pollution in a particular area and forevaluation of air quality, because these compounds areprecursors of photochemical oxidants and fine particu-late matter1-3 and are also potentially toxic in themselves.The current trend in air-quality monitoring networks isto use automatic nonattended systems that can operatecontinuously or semicontinuously for long periods. Theautomatic measurements of total VOCs (CH4 excluded)in these air-quality monitoring networks are routinelyperformed and give a picture of the evolution of volatileorganic matter concentrations. However, these networksdo not provide any information about the reactivity ofthe atmosphere, O3 formation potential, or toxicity. There-fore, there is a move toward speciation of individual VOCanalysis as a source of essential information for air-qual-ity monitoring and management in urban and industrialareas.4,5 This is especially true near sites where very reac-tive or toxic compounds are involved. Depending on thesources in the area, the number of VOCs to be monitoredvaries and, therefore, careful design of both the instru-ments and monitoring strategies is required.

To study the temporal evolution of VOCs in ambientair, the use of automatic gas chromatography systems hasgrown rapidly in recent years.6-8 An automatic gas chro-matograph (AGC) is able to provide frequent (usuallyhourly) analyses of VOCs for long (days to months) peri-ods. The system consists of (1) a sample preconcentrationsystem; (2) a GC with nonspecific detectors, such as a flameionization detector (FID), an electronic capture detector(ECD), or a photoionization detector (PID); and (3) dataacquisition software and hardware for the automated con-trol of sampling and GC operation.

Earlier projects in the Basque Country, Spain, weredesigned to lay down the technical and scientific basesfor defining strategies to monitor and control VOCs and

Online Hourly Determination of 62 VOCs in Ambient Air:System Evaluation and Comparison with Another TwoAnalytical Techniques

Nieves Durana, Marino Navazo, Lucio Alonso, José A. García, Juan L. Ilardia,M. Carmen Gómez, and Gotzon GangoitiDepartment of Chemical and Environmental Engineering, School of Engineering,University of the Basque Country, Bilbao, Spain

IMPLICATIONSThis work has been carried out in a framework set up bythe Basque Government to establish the technical andscientific bases for a monitoring and control strategy forVOCs and photochemical oxidants in the autonomouscommunity of the Basque Country. This paper presentsresults of general interest obtained through the implemen-tation and validation process of the data generated withan AGC system and a comparison with another two ana-lytical techniques. The sampling and analysis techniqueswere set up and applied under field conditions through-out various measurement campaigns in selected urban,rural, and industrial settings.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

Volume 52 October 2002 Journal of the Air & Waste Management Association 1177

photochemical oxidants.9,10 In 1997, an onlineanalyzer of VOCs was installed to attempt system-atic monitoring of these compounds. Among otherachievements, these projects prepared the way forthe proper choice of technique and equipment forthe uninterrupted measurement of VOCs in theurban area of Bilbao, Basque Country.

This paper presents results of general interestfrom the implementation and validation processof the data generated with the AGC system. Thesystem was optimized for automatic analysis,identifying each hour for seven consecutivemonths, in a semi-unattended way, 62 nonpolarhydrocarbon (HC) concentrations in ambient air,from C2–C10 with sub-parts-per-billion sensitivity.It is well known that these systems produce ex-tremely large and complex data sets whose han-dling, processing, and validation procedures needconsiderable effort before they can be turned intoa consistent and reliable database. This papershows the results provided by the automatic chro-matograph along with simultaneous findingsusing (1) the portable GC equipped with a pho-toionization detector and (2) active adsorption onTenax-TA and subsequent thermal desorption and GCanalysis. These sampling and analysis techniques wereset up and applied under field conditions throughoutvarious measurement campaigns in selected urban,rural, and industrial settings.11

EXPERIMENTAL METHODSOnline VOC Analysis

Analyses were performed using an AGC (VOC Ozone Pre-cursor Analyzer System, from Perkin Elmer). The system(AGC-FID) was optimized for the automatic measurementof first 27 (1998 and 1999) and then 62 nonpolar O3

precursors (Table 1) in urban air, for a nominal range ofC2–C10 hydrocarbons and sensitivities down to parts pertrillion in volume (10–6 ppmv). The main elements areas follows:

(1) Thermal desorption unit (ATD400) fitted with anair sampler accessory. This system is an adapta-tion of the standard ATD400 that includes a spe-cial means for direct introduction of ambientsample air. The dehumidification device used is asemipermeable membrane (Nafion, PermapureInc.). Nafion drying devices have shown demon-strated losses of certain polar VOCs (amines, ke-tones, alcohols, and some ethers). Nafion driersalso have caused rearrangement of several monot-erpenes (α-pinene and β-pinene) but have no ef-fect on the recovery of isoprene.12 The unit is ableto carry out automatic online sampling and sample

preconcentration using a small pump, a mass flowmeter, and a cold trap at –30 ºC with Peltier cells.Liquid nitrogen is not required.

(2) GC. Separation of the target components isachieved through the use of a dual-column sys-tem with column switching (Figure 1). The samplefirst enters into the BP1 column. The first com-pounds to elute from this column are switched tothe PLOT column. The system includes two FIDs(one for each column) for signal detection.

Operating conditions for the GC are as follows: car-rier gas, He; FID temperature, 250 °C; chromatographiccolumns, BP1 (50 m × 0.22 mm i.d. with layer thicknessof 1 µm) and PLOT Al2O3/Na2SO4 (50 m × 0.32 mm i.d.with layer thickness of 5 µm); and oven temperature pro-gram: 45 °C (15 min), 5 °C/min up to 170 °C, 15 °C/minup to 200 °C (6 min). We have noted, especially in thePLOT column, that after a period of operation some peakstend to co-elute (1,3-butadiene and propyne; 3-methyl-1-butene and cyclopentene). It is necessary to bear thisin mind to quantify and complicate data validation.

VOCs in samples were identified and quantifiedweekly using a certified standard mixture with 27 com-pounds (Table 2) prepared by the U.K. National PhysicalLaboratory by gravimetric methods. Compounds chosenfor the standard were selected for reactivity reasons, fortheir abundance in ambient air, and because they are typi-cal pollutants in urban areas.13 In addition, other com-pounds not included in this standard were identified by

1. Ethane2. Ethylene3. Propane4. Propylene5. i-butane6. n-butane7. Acetylene8. trans-2-butene9. 1-butene10. i-butene11. cis-2-butene12. Cyclopentane13. i-pentane14. n-pentane15. 1,3-butadiene16. Propyne17. 3-methyl-1-butene18. Cyclopentene19. trans-2-pentene20. 2-methyl-2-butene21. 1-pentene

43. 2-methyl-heptane44. 3-methyl-heptane45. n-octane46. Tetrachloroethene47. Ethyl benzene48. m + p-xylenes49. Styrene50. o-xylene51. n-nonane52. i-propyl-benzene53. n-propyl-benzene54. m-ethyl-toluene55. p-ethyl-toluene56. 1,3,5-trimethyl-benzene57. o-ethyl-toluene58. 1,2,4-trimethyl-benzene59. n-decane60. 1,2,3-trimethyl-benzene61. m-diethyl-benzene62. p-diethyl-benzene

Table 1. List of the 62 identified and quantified VOCs with the AGC-FID system (by elution order).

Note: The 27 compounds in bold were initially measured in 1998 and 1999.

22. cis-2-pentene23. 2,2-dimethyl-butane24. 2,3-dimethyl-butane25. 2-methyl-pentane26. 3-methyl-pentane27. Isoprene28. n-hexane29. Methyl-cyclopentane30. 2,4-dimethyl-pentane31. Benzene32. Cyclohexane33. 2-methyl-hexane34. 2,3-dimethyl-pentane35. 3-methyl-hexane36. Trichloroethene37. 1-heptene38. 2,2,4-trimethyl-pentane39. n-heptane40. Methyl-cyclohexane41. 2,3,4-trimethyl-pentane42. Toluene

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

1178 Journal of the Air & Waste Management Association Volume 52 October 2002

means of individual standards and calibrated by responsefactors for these compounds.14

Since September 1997, a routine operation has mea-sured ambient air pollutants hourly, using aspirated airthrough a sampling stainless-steel manifold. A samplingflow of 15 mL/min is collected by the ATD400 for 40 min.VOCs are collected in a cryogenic trap at –30 ºC with solidadsorbents able to accumulate C2–C10 VOCs. Later, thetrap is heated and carrier gas pushes the HCs to the chro-matographic columns.

Grab Sampling and In Situ AnalysisA portable GC equipped with a photoionization detec-tor (PGC-PID) (Photovac 10SPlus) was employed. This

procedure is included in the U.S. Environmental Protec-tion Agency (EPA) Environmental Monitoring MethodsIndex (AOB VA-008-1) as a reference method for directVOC measurement in ambient air. Sample air automati-cally is pumped without concentration into the injec-tion loop (1 mL). The PGC-PID was programmed to carryout an analysis every 30 min. Chromatography operat-ing conditions were as follows: carrier gas, N2; PID lamp,10.6 eV; chromatographic column, capillary 0.53-mmCp-Sil 5 CB preconfigured with a 1-m precolumn and 9-m analytical column for automatic injection precolumnbackflush operation; and oven temperature, 40 ºC.

A multicomponent gaseous standard prepared by di-lution of methanol stock solutions in a gas bag was em-ployed for quantitative analysis. A commercial standardmade up of a mixture of chlorinated and aromatic HCs inmethanol—trichloroethene, tetrachloroethene, benzene,toluene, ethyl benzene, m + p-xylene, and o-xylene—wasused. The detection limits (DLs) (signal-to-noise ratio 3:1)were 0.1 ppbv for benzene, toluene, trichloroethene, andtetrachloroethene, and 0.5 ppbv for compounds with longerretention times (RTs). The coefficient of variation (CV) wasbetter than 10% for all compounds except for xylenes, whichwent up to 23% for low concentration levels (1.7 ppbv).10

Integrated Sampling andLaboratory Analysis

The method used was based on the active sampling andadsorption in tubes filled with the adsorbent material Tenax-TA, subsequent thermal desorption and concentration in acryogenic trap, and then chromatographic analysis. The useof adsorbent tubes involves a single measurement for each

sampling period and, thus, gives a cumulativefigure that enables a mean value for this periodto be calculated. These sampling conditions15

enable nonpolar organic compounds, such asC6 and higher aliphatic, olefinic, and aromaticHCs, and halogenated HCs less volatile than1,1,1-trichloroethane, to be determined.

Stainless-steel ATD tubes (90 × 5 mm)filled with 250 mg of Tenax-TA (60-80 mesh;Chrompack) were conditioned for 1 hr at300 ºC with a He flow of 100 mL/min. Asequential sampler at ambient temperaturewas used for the capture (STS25, Perkin Elmer),then the tubes were analyzed in the AGC-FID.The Tenax-TA tubes were heated at 250 ºC for10 min with He (10 mL/min), and the des-orbed compounds were collected in a cryo-genic trap at –30 ºC with solid adsorbents. Thetrap was then heated at 325 ºC and the com-pounds were conducted toward the analyti-cal column. The sampling (flow rate and

Table 2. Compound list of the certified standard mixture: concentration (ppbv), coefficients of determination(r2), and DLs.

Conc. DL Conc. DLCompound ppbv r2 ppbv Compound ppbv r2 ppbv

Ethane 21.8 0.9991 0.18 cis-2-pentene 26.3 0.9986 0.11Ethylene 15.1 0.9984 0.16 2-methyl-pentane 10.3 0.9994 0.04Propane 16.9 0.9997 0.04 3-methyl-pentane 25.4 0.9986 0.09Propylene 26.5 0.9993 0.20 Isoprene 41.6 0.9900 0.11i-butane 13.7 0.9976 0.08 n-hexane 20.6 0.9877 0.12n-butane 6.3 0.9978 0.07 Benzene 13.6 0.9980 0.09Acetylene 21.0 0.9985 0.32 n-heptane 33.2 0.9991 0.36trans-2-butene 11.2 0.9983 0.07 Toluene 31.8 0.9980 0.091-butene 23.2 0.9986 0.08 Ethyl benzene 22.5 0.9967 0.05cis-2-butene 20.0 0.9985 0.08 m + p-xylenes 17.4 0.9988 0.05i-pentane 50.6 0.9987 0.14 o-xylene 7.8 0.9954 0.05n-pentane 17.8 0.9988 0.05 1,3,5-trimethyl-benzene 3.1 0.9936 0.021,3-butadiene 19.0 0.9987 0.06 1,2,4-trimethyl-benzene 5.7 0.9948 0.02trans-2-pentene 48.2 0.9988 0.12

Figure 1. Chromatograms corresponding to an ambient air sample(07h UTC, May 28, 2001); top: PLOT column—light VOCs; bottom:BP1 column—less volatile VOCs. The numbers correspond to thecompounds in Table 1.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

Volume 52 October 2002 Journal of the Air & Waste Management Association 1179

sample volume) and operating conditions of the AGC-FID described previously also were used here.

RESULTS AND DISCUSSIONEvaluation of the AGC-FID System

Results of laboratory tests on several analytical param-eters, carried out to define the application range of theinstrument for various compounds, are summarized inthe next sections.

Linearity. For the linearity test, six mixtures were preparedusing different dilutions of the certified standard with pureN2 (no dilution and dilutions of 2/3, 1/3, 1/6, 1/20, and1/50). Results show a linear response over this test con-centration range, with r2 coefficients for all compoundsbetween 0.99 and 1.00 (see Table 2).

Precision. Ten replicate measurements of six standards withdifferent concentrations were analyzed (Figure 2). Onlythose compounds that present concentrations below 1ppb or close to their DL show a CV greater than 10%.

DLs. The values were determined by examining the re-sults of 10 replicate runs using a dilution sample of 1/50(see Table 2). The DL was calculated by multiplying theStudent t test figure by the SD of the analysis.12 As such,DL = t s, where t corresponds to the Student t with 99%confidence and n–1 analysis for a given standard, and s isthe SD of the data.

RT Stability. RT stability, which directly affects compoundidentification, needs to be established so an automatic chro-matography system can operate unattended for lengthyperiods of time, especially at remote sites. Variations incompounds may be caused by several factors affecting thecolumn elution process, such as gas flows, oven tempera-ture, humidity, and so on. To evaluate RT variability for

the AGC-FID, the absolute RTs of compounds in standardswere monitored in a measurement campaign (April–October). Absolute deviation for each individual responsetime was always within the range of ±0.5 min and, moreimportantly, deviation showed the same trend for all com-pounds. This facilitated the automatic correction of theidentification windows once the species used for refer-ence in each column were identified correctly. RT vari-ability was greatest for the PLOT column (Figure 3).

Data ValidationThe intensive measurement of VOCs over a relatively longperiod, using automatic chromatography techniques, gen-erates a large amount of data. In the measurement cam-paigns, the sampling of ambient air and its analysis tookplace every hour throughout the day, which amounts to1488 pieces of data on average hourly concentrations of62 VOCs. The data validation scheme shown in Figure 4refers to the process undertaken with data stored in thechromatographic archives before introducing them intothe final database.

A protocol of routine checks is followed daily: the in-formation with which the result archives are stored (date,hour, kind of sample, and name) is checked and the chro-matograms from the previous day are reviewed visually.This visual review begins with the appearance of the chro-matogram, the shape and resolution of the peaks andbaseline, and then proceeds to a more thorough examina-tion of the correct identification and quantification of the62 VOCs. Any problem detected during this review is dulyrecorded, and corrective actions in the method of control,acquisition, storage, detection, identification, quantifica-tion, or data generation of the system are taken.12

After this first validation level, the data are importedto a spreadsheet that checks their internal consistency and

Figure 2. CV vs. concentration measured during repeatability analysis.Figure 3. RT variation for compounds that are separated in the BP1column using analysis of a certified standard.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

1180 Journal of the Air & Waste Management Association Volume 52 October 2002

identifies those values that are unrepresentative of theambient air conditions when compared with the overalldata. The internal consistency test includes the verifica-tion of the compliance with the lower limits set for theconcentration of compounds16 and of the specific limitsof the monitoring site, based on the relationship betweencompounds observed historically and graphics of the

data.17 The time series are laid out graphically to identifyatypical values, excessive differences between adjacentvalues, and incorrect identifications of peaks or long pe-riods with unusually high or low concentrations. Disper-sion graphs among pairs of compounds with habituallyclose (r2 > 0.98) correlation also are drawn to identify atypi-cal figures. After this thorough validation, the data are

Figure 4. Validation scheme.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

Volume 52 October 2002 Journal of the Air & Waste Management Association 1181

introduced into the historic database for statistical stud-ies that then are also checked. This second level of valida-tion means that the database is more robust, because anyanalytical errors will have been corrected,18 and the be-havior patterns of each VOC, which have to be establishedto monitor and manage air quality in urban and indus-trial areas, are properly established.

Percentage of Total Non-MethaneHydrocarbons Identified

An automatic processing method has been developed bywhich not only the 62 compounds are identified andquantified, but also the total non-methane hydrocarbons(TNMHCs) are calculated by integration of all chromato-graphic peaks (identified or not) in the C2–C10 range, ex-pressed as ppbC. TNMHC concentration, according to thenotation used in PAMS (photochemical assessment moni-toring stations),19 is calculated by integrating all peaksfrom the PLOT column, using the response factor of pro-pane, and all peaks from the BP1 column, using the re-sponse factor of benzene, and then adding thecontributions of both channels. Table 3 shows the resultsgrouped by families of compounds and expressed as per-centages in ppbC. The most abundant are alkanes andaromatics, without important variations throughout theyears. Regarding chlorinated compounds used as indus-trial and commercial tracers, although trichloroethene andtetrachloroethene have been measured only during 2000(when the technique was extended to 62 compounds), itis important to indicate their contribution is up to 1%.

When only 27 compounds included in the certifiedstandard are identified and included (see Table 2), ~65%of the total VOCs can be determined. When the 62 com-pounds identified in the 2000 campaign are included, 88%of the TNMHCs can be determined. These results are con-sistent with, though somewhat higher than, those reached

at this kind of station in the United States,18 where incities such as Fresno, San Diego, and Los Angeles, CA, 50,70, and 75%, respectively, of the TNMHCs can be identi-fied when the 55 species in the PAMS are included. In ourcase, the unidentified fraction contains polar species, of-ten oxygenated, such as carbonyls. Figure 5 shows thescatterplots of the portion identified with 27 and 62 com-pounds analyzed against the corresponding TNMHC val-ues (in ppbC). In both cases, it is possible to deduce theconcentration of TNMHCs from the hourly figures, be-cause the relationship between the two is linear, with co-efficients of determination above 0.9 with 4032 pairs offigures (hourly averages).

Comparison between Two Calibration MethodsThe routine quantification of a high number of com-pounds can be a thorny problem because of the difficultyin achieving a calibration standard for numerous com-pounds at low concentrations and because of its high cost.The current trend in air-quality monitoring networks isto use a primary calibration standard with a single com-pound (propane) or a mixture of two compounds (pro-pane and benzene) to determine the concentration of thevarious VOCs on the basis of response factors.6,12 Thismethod of calibrating equipment is used because the FID

Table 3. Annual averages of VOCs grouped by families (percent in ppbC).

Family 1998 1999 2000

Alkanes 46 43 47Alkenes + alkynes 14 16 14Aromatics 40 41 38Trichloroethene + tetrachloroethene – – 1TNMHCS nonidentified (%) 35a 35a 12b

a27 compounds identified; b62 compounds identified.

Figure 5. Relationship between the sum of the concentrations of the 27 VOCs identified and the totals. On the right, the same is shown with the62 compounds determined at present.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

1182 Journal of the Air & Waste Management Association Volume 52 October 2002

gives out a signal that is a linear function of the numberof carbon atoms in the compound.

In this work, two calibration methods are compared:one using a primary calibration standard of two compo-nents, one for each column; the other using the certifiedstandard of 27 compounds. The comparative study was per-formed on the basis of 64 calibrations (along one singlerun of 1551 hours) by means of a certified standard. Thebest results were found when the following were adoptedas reference compounds in the PLOT column: propane,n-pentane, or trans-2-pentene; and when benzene, tolu-ene, or ethyl benzene were used in the BP1 column. Forthe propane and benzene compounds, commonly used asthe primary calibration standard in this type of equipment,it was found that the differences in quantification wereabove 10% in three compounds: acetylene, isoprene, andn-hexane (16.2, 12.5, and 12.1%, respectively), whereas theremaining compounds had differences under 6%. Acety-lene and isoprene are, in fact, the two compounds thatshow larger bias (39.66 and 19.26%, respectively) in theintercomparison of VOC measurements between labora-tories.13 Stability problems have been reported for thesecompounds in the standard. Nevertheless, such problemshave not been observed in our certified standard mixture.Table 4 shows the theoretical effective carbon number(ECN) for each compound and the experimental figures,on the basis of the 64 calibrations found for propane (PLOTcolumn) and benzene (BP1 column). The greater the dif-ference between the theoretical ECN adopted and the ex-perimental ECN obtained, the greater the differences inthe quantification of the compound.

The divergences in the results of the two calibrationmethods arise from the assumption that the FID responsedepends exclusively on the ECN. This number, extractedfrom the bibliography, ignores the configuration of eachparticular chromatographic system. In practice, as the re-sult of possible interactions of each compound with thechromatographic system, the experimental ECN may varyfrom the theoretical ECN adopted. Results improve con-siderably if the theoretical ECN is replaced by its experi-mental value, but it is impractical because this parametermay vary each time there is a modification or adjustmentto the equipment. Thus, it is better to use a certified stan-dard with the compounds of interest, although this wasnot possible for the 62 compounds studied.

Comparison between the Results with theAGC-FID and the PGC-PID

Figure 6 shows, by way of example, the toluene concentra-tion figures obtained by these two techniques (243 hr ofoperation in parallel). The AGC-FID was set up to providea result every hour using a 40-min sampling time, whilethe PGC-PID was taking two instantaneous samples spaced30 min in the same period. To compare results, an averagefor the two PGC-PID measures was used. The hourly con-centration data with the two measuring systems operatingsimultaneously are fairly similar, although some differencesare found. These probably originate in the type of averag-ing (sampling) used in the techniques, the lower precisionof the PGC-PID and, to a lesser extent, the different chro-matographic resolution of the columns used.

The results with both techniques correlate significantlyfor every compound. Toluene, m + p-xy-lene and tetrachloroethene correlate bestwith r2 = 0.93, 0.90, and 0.93, respectively.The remaining compounds, whose concen-trations range between 0 and 3 ppbv, havelower correlation figures (r2): 0.75 for ben-zene and ethyl benzene, 0.65 for o-xylene,and 0.60 for trichloroethene. The latter hasthe worst correlation, probably because inmost data it occurs in concentrations be-low 1 ppbv (97% of the data determinedwith the PGC-PID), which is the level atwhich the equipment is less precise.

Comparison between the DirectSampling and Preconcentration

Results in Adsorbent Tubes ofTenax-TA

Results obtained using two sampling tech-niques are shown: direct sampling andAGC-FID analysis and sampling using ad-sorbent tubes of Tenax-TA, and thermal

Table 4. Theoretical and experimental ECN.

ECN ECNExperimental Experimental

ECN Reference ECN ReferenceCompound Theoretical Propane Compound Theoretical Benzene

Ethane 2.0 2.1 cis-2-pentene 4.9 4.9Ethylene 1.9 2.0 2-methyl-pentane 6.0 5.5Propane 3.0 – 3-methyl-pentane 6.0 6.1Propylene 2.9 2.9 Isoprene 4.8 4.2i-butane 4.0 4.0 n-hexane 6.0 6.7n-butane 4.0 4.2 Benzene 6.0 –Acetylene 2.6 2.2 n-heptane 7.0 7.4trans-2-butene 3.9 3.8 Toluene 7.0 7.21-butene 3.9 3.8 Ethyl benzene 8.0 8.0cis-2-butene 3.9 3.6 m + p-xylenes 8.0 7.8i-pentane 5.0 4.9 o-xylene 8.0 7.9n-pentane 5.0 5.0 1,3,5-trimethyl-benzene 9.0 9.81,3-butadiene 3.8 3.6 1,2,4-trimethyl-benzene 9.0 8.5trans-2-pentene 4.9 4.9

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

Volume 52 October 2002 Journal of the Air & Waste Management Association 1183

desorption and analysis in the same equipment. To com-pare the two work procedures, 12 simultaneous samplesof ambient air were taken. In the case of Tenax-TA ad-sorption, the concentration figures were corrected by theresults obtained with blanks.

Under normal sampling conditions (40 min with aflow rate of 15 mL/min) and operation, the quantitativedetermination of compounds on the basis of n-pentanewas possible. As Figure 7 indicates, there is good concor-dance between the results with both procedures, thoughthe mean figures obtained through direct sampling areslightly higher than those obtained in adsorbent tubes. Itis in this latter case that there may be loss of componentsin the concentration and thermal desorption stages be-fore the chemical analysis and also in the process of stor-age, because of lack of stability of the components retainedon the adsorbent.

The results with the two systems of sampling corre-late closely, at about r2 > 0.9 for all compounds. Threecompounds are exceptions: 2,2-dimethylbutane,n-propylbenzene, and 1,2,3-trimethylbenzene, whichshow low correlation between techniques and have lowerconcentrations with the use of Tenax-TA. The greatest di-vergences between the two techniques occurred in com-pounds at low concentrations (under 1 ppbv), and thesealso were present in the blank. In these cases, the preci-sion of the equipment is lower.

Although the two sampling systems provide similarresults and correlate closely for the conditions of sam-pling and operation used, in the case of active samplingwith adsorbent tubes, thorough control of the flow rateand careful handling of the tubes are absolutely essen-tial to avoid breakthrough and other problems arisingfrom the operations of filling, conditioning, capture,freight, degradation of the polymer adsorbent in the pres-ence of O3 and other reactive gases, conservation, andanalysis.20,21 Factors that may affect the volume of satu-ration considerably are temperature, sampling flow, rela-tive humidity of the air, the concentration of pollutants,

the physicochemical properties of the adsorbent and analyte,and the presence of other components in the mixture.22,23

CONCLUSIONSIn this work, a technique based on gas chromatographyhas been evaluated for online determination of 62 VOCsin ambient air during extended periods of continuous op-eration. The evaluation of analytical parameters—linear-ity, precision, DL, and stability of RTs—indicates thesuitability of the equipment for measuring ambient airautomatically for long periods. Absolute divergence foreach individual response time is always within the rangeof ±0.5 min and shows the same tendency for all com-pounds. This makes it easier for the identification win-dows to be corrected automatically. These systems generatea large amount of extremely complex data, wherein han-dling, processing, and validation require a huge effortbefore a consistent and reliable database can be achieved.

The 62 VOCs identified—paraffins, olefins, aromat-ics, and chlorinated compounds—make up 88% of theTNMHCs. The calibration of equipment using responsefactors reports major benefits. However, it should be notedthat if the response factors used are based on the theo-retical ECN data published, some compounds may havedifferences above 10%, as is the case of the compoundsacetylene, isoprene, and n-hexane.

To facilitate comparisons between the results of widelyused techniques for determining VOCs, parallel measure-ments with an automatic and a portable chromatographwere made. The concentration figures from the two mea-suring systems operating simultaneously were fairly close,although there were significant differences. These prob-ably arose from the type of averaging (sampling) used ineach of the techniques and, to a lesser degree, from thedifferent chromatographic resolution of the columns usedand different DLs.

Direct sampling and ACG-FID analysis also was com-pared with sampling through adsorbent tubes of Tenax-TA, subsequent thermal desorption, and analysis in thesame equipment. With the sampling and operating con-ditions used, simultaneous quantification of compoundsbased on n-pentane was possible. There was good concor-dance between the results from the two procedures. How-ever, in the active sampling with tubes, a thorough controlof the flow rate and the tube-handling operations wereessential to avoid breakthrough or other problems arisingfrom the filling, conditioning, capture, transport, conser-vation, and analysis operations.

It is necessary to know the advantages and limitationsof each technique when combining the results from the dif-ferent instruments into one data set. The AGC-FID systemallows a continuous, integrated, and automatic samplingand analysis without problems of storage and degradation

Figure 6. Hourly concentrations with the PGC-PID and the AGC-FIDfor toluene.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

1184 Journal of the Air & Waste Management Association Volume 52 October 2002

of the sample, but is limited to only one location. The PGC-PID system allows grab sampling (instantaneous sampling)every 30 min. This provides daily evolution of compounds

in several locations, but it has limitations, such as alower chromatographic resolution. And finally, the in-tegrated sampling with adsorption in Tenax allows a

Figure 7. Mean concentrations in simultaneous measurements of ambient air through direct sampling of air and preconcentration in adsorbenttubes of Tenax-TA.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14

Durana et al.

Volume 52 October 2002 Journal of the Air & Waste Management Association 1185

noncontinuous, integrated sampling in several loca-tions, but it is limited by possible problems arisingmainly from the conservation and degradation of thesample. These three measurement techniques can beespecially useful when employed as complementarytools to design air-quality monitoring networks; theyenable the real conditions of each area to be studiedand, in line with the results, allow the best location formore sophisticated equipment (such as the automaticVOC measurement equipment) to be chosen.

ACKNOWLEDGMENTThe financial support of the Basque Government(Viceconsejería de Medio Ambiente) for this research isgratefully acknowledged.

REFERENCES1. Chan, C.C.; Lin, S.H.; Her, G.R. Student’s Exposure to Volatile Or-

ganic Compounds while Commuting by Motorcycle and Bus in TaipeiCity; J. Air & Waste Manage. Assoc. 1993, 43, 1231-1238.

2. Odum, J.R.; Jungkamp, T.P.W.; Griffin, R.J.; Flagan, R.C.; Seinfeld,J.H. Science 1997, 276, 96-99.

3. Tran, N.K.; Steinberg, S.M.; Johnson, B.J. Volatile Aromatic Hydro-carbons and Dicarboxylic Acid Concentrations in Air at an UrbanSite in the Southwestern U.S.; Atmos. Environ. 2000, 34, 1845-1852.

4. Demerjian, K.L. A Review of National Monitoring Networks in NorthAmerica; Atmos. Environ. 2000, 34, 1861-1884.

5. Derwent, R.G.; Davies, T.J.; Delaney, M.; Dollard, G.J.; Field, R.A.;Dumitrean, P.; Nason, P.D.; Jones, B.M.R.; Pepler, S.A. Analysis andInterpretation of the Continuous Hourly Monitoring Data for 26C2–C8 Hydrocarbons at 12 United Kingdom Sites during 1996; Atmos.Environ. 2000, 34, 297-312.

6. Farmer, C.T.; Milne, P.J.; Riemer, D.D.; Zika, R.G. Continuous HourlyAnalysis of C2–C10 Non-Methane Hydrocarbon Compounds in Ur-ban Air by GC-FID; Environ. Sci. Technol. 1994, 28, 238-245.

7. Oliver, K.D.; Adams, J.R.; Daughtrey, E.H.; Mcclenny, W.A.; Yoong,M.J.; Pardee, M.A. Technique for Monitoring Ozone Precursor Hy-drocarbons in Air at Photochemical Assessment Monitoring Stations:Sorbent Preconcentration, Closed-Cycle Cooler Cryofocusing and GC-FID Analysis; Atmos. Environ. 1996, 30, 2751-2757.

8. Parrish, D.D.; Fehsenfeld, F.C. Methods for Gas-Phase Measurementsof Ozone, Ozone Precursors and Aerosol Precursors; Atmos. Environ.2000, 34, 1921-1957.

9. Durana, N.; Navazo, M.; García, J.A.; Ilardia, J.L.; Olaso, A.; Alonso, L.Determination of Volatile Organic Compounds in Urban Areas Us-ing Discontinuous and Semicontinuous Techniques (in Spanish);Información Tecnológica 1998, 9, 53-60.

10. Alonso, L.; Durana, N.; Navazo, M.; García, J.A.; Ilardia, J.L. Determi-nation of Volatile Organic Compounds in the Atmosphere Using TwoComplementary Analysis Techniques; J. Air & Waste Manage. Assoc.1999, 49, 916-924.

11. Alonso, L.; Navazo, M.; Durana, N.; García, J.A.; Ilardia, J.L.; Gómez,M.C. Measurement of Atmospheric Organic Compounds in the Basque Au-tonomous Community: Measurement Techniques and Results (in Spanish);Servicio Central de Publicaciones del Gobierno Vasco: Vitoria, Spain,1999; p 180.

12. Technical Assistance Document for Sampling and Analysis of Ozone Pre-cursors; EPA/600-R-98/161; U.S. Environmental Protection Agency:Research Triangle Park, NC, 1998.

13. Pérez-Ballesta, P.; Field, R.A.; De Seager, E. The Second EC Intercomparisonof VOC Measurements; Report EUR 17757 EN; European Commission,Joint Research Center: Ispra, Italy, 1998.

14. Scanlon, J.T.; Willis, D.E. Calculation of Flame Ionization DetectorRelative Response Factors Using the Effective Carbon Number Con-cept; J. Chromatogr. Sci. 1985, 23, 333-340.

15. Compendium of Methods for the Determination of Toxic Organic Com-pounds in Ambient Air; EPA-625/R-96/010b; U.S. Environmental Pro-tection Agency: Research Triangle Park, NC, 1997.

16. Parrish, D.D.; Trainer, M.; Young, V.; Goldan, P.D.; Kruster, W.C.;Jobson, B.T.; Fehsenfeld, F.C.; Lonneman, W.A.; Zika, R.D.; Farmer,C.T.; et al. Internal Consistency Tests for Evaluation of Measurementsof Anthropogenic Hydrocarbons in the Troposphere; J. Geophys. Res.1998, 103 (D17), 22339-22359.

17. Bricker, A.B.; Farmer, C.; Milne, P.; Riemer, D.; Zika, R. Validation ofSpeciated Non-Methane Hydrocarbon Compound Data Collectedduring the 1992 Atlanta Intensive as Part of the Southern OxidantsStudy (SOS); J. Air & Waste Manage. Assoc. 1995, 45, 591-603.

18. Spatial and Temporal Characteristics of California PAMS and Long-TermTrend Site VOC Data (1990–1997); STI-998241-1883-FR; U.S. Environ-mental Protection Agency: Research Triangle Park, NC, 1999.

19. PAMSGRAM; AMTIC PAMS; U.S. Environmental Protection Agency,Office of Air Quality Planning and Standards, 1997; Vol. 5. Availableat www.epa.gov/ttn/amtic/pamsgram.html.

20. Pellizzari, E.D.; Demian, B.; Krost, K. Sampling of Organic Compoundsin the Presence of Reactive Inorganic Gases; Anal. Chem. 1984, 56,793-798.

21. Walling, J.F.; Bumgarner, J.E.; Driscoll, D.J.; Morris, C.M.; Riley, A.E.;Wright, L.H. Apparent Reaction Products Desorbed from Tenax Usedto Sample Ambient Air; Atmos. Environ. 1986, 20, 51-57.

22. Ciccioli, P.; Brancaleoni, E.; Cecinato, A.; Di Palo, C.; Brachetti, A.;Liberti, A. Gas Chromatographic Evaluation of the Organic Compo-nents Present in the Atmosphere at Trace Levels with the Aid ofCarbopack B for the Preconcentration of the Sample; J. Chromatogr.1986, 351, 433-449.

23. Woolfenden, E. Monitoring VOCs in Air Using Sorbent Tubes Followedby Thermal Desorption-Capillary GC Analysis: Summary of Data andPractical Guidelines; J. Air & Waste Manage. Assoc. 1997, 47, 20-36.

About the AuthorsLucio Alonso is a professor in the Department of Chemicaland Environmental Engineering at the University of theBasque Country, Spain. Nieves Durana, Marino Navazo,José A. García, Juan L. Ilardia, and Gotzon Gangoiti areassistant professors at the School of Engineering of Bilbao,University of the Basque Country, and members of the At-mospheric Research Group. M. Carmen Gómez is a doc-toral student. Correspondence should be addressed to Dr.Nieves Durana, Department of Chemical and Environmen-tal Engineering, School of Engineering, Alameda de Urquijos/n, 48013 Bilbao, Spain; fax: 34-94-6014179; e-mail:iapdujin@bi.ehu.es.

Dow

nloa

ded

by [

Uni

vers

idad

Del

Pai

s V

asco

] at

06:

05 1

6 Ja

nuar

y 20

14