Chemosphere 56 (2004) 7180 www.elsevier.

com/locate/chemosphere

Determination of volatile organic compounds in workplace air by multisorbent adsorption/thermal desorption-GC/MS

Chien-Hou Wu a, Chien-Tai Feng a, Yu-Shiu Lo a, Tsai-Yin Lin a, Jiunn-Guang Lo a,b,*
b

Environmental Chemistry Laboratory, Department of Atomic Science, National Tsing Hua University, 101 Sec. 2 Kuang Fu Road, Hsinchu 300, Taiwan, ROC Department of Radiological Technology, Yuan-Pei Institute of Science and Technology, Hsinchu 300, Taiwan, ROC Received 27 February 2003; received in revised form 20 November 2003; accepted 2 February 2004

Abstract Investigation of volatile organic compounds (VOCs) was rst conducted in the air of class-100 cleanrooms at liquid crystal display (LCD) fabrication facilities. Air samples were collected on multisorbent tubes (including Carbopack B, Carbopack C, and Carbosieve S-III) and analyzed using adsorption/thermal desorption coupled with gas chromatographymass spectrometry (GCMS). Optimal conditions lead to average recoveries in the range of 96.298.2%, and method detection limits between 0.38 and 0.78 ppb, under the condition of 1-l sampling volume and 80% relative humidity. The method appears to be accurate, sensitive, simple and well-suited for determining VOC distributions from various stages of LCD manufacturing process and temporal variations of the analyte concentrations. About 15 VOCs were identied in workplace air. The major pollutants such as propylene glycol methyl ether acetate (PGMEA), butyl acetate, and acetone that are commonly used in the opto-electronics industry were detected and accurately quantied with the established method. 2004 Elsevier Ltd. All rights reserved.
Keywords: Volatile organic compounds; Multisorbent adsorption; Gas chromatography; Air analysis; Environmental analysis

1. Introduction Monitoring of volatile organic compounds (VOCs) in ambient air has received substantial attention over the past several years because certain VOCs are known to be hazardous to human health (Roche et al., 1999; Uchiyama et al., 1999; Kuo et al., 2000; Tolnai et al., 2000a). The majority of the methods that have been developed for ambient VOC monitoring have been either canister-

Corresponding author. Address: Environmental Chemistry Laboratory, Department of Atomic Science, National Tsing Hua University, 101 Sec. 2 Kuang Fu Road, Hsinchu 300, Taiwan, ROC. Fax: +88-63-572-5075. E-mail address: jglo@mx.nthu.edu.tw (J.-G. Lo).

based or sorbent-based. The respective advantages and limitations of these two methods have been discussed in the literature (Sunesson et al., 1995; Pankow et al., 1998; Tolnai et al., 2000b). The canister method collects air samples directly with evacuated stainless steel canisters lined with fused silica (Wang et al., 2000; Chang et al., 2001). As the canister method is very costly and it may be dicult to recover quantitatively some polar and low volatility VOCs, a sorbent-based method with adsorption/thermal desorption was chosen for this study. The main advantages of sorbent-based methods are their capabilities for enrichment with high sensitivity, ease of use, and cost-eectiveness (Wolska et al., 1995; Clement et al., 2000). When solid adsorbents are used for sampling in the vast majority of cases, particular precautions should be taken in many factors including ow rate,

0045-6535/$ - see front matter 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2004.02.003

72

C.-H. Wu et al. / Chemosphere 56 (2004) 7180

sample concentration, humidity, temperature, breakthrough, and analyte stability. Although blank levels can be made extremely low (down to sub-ppbv) by meticulous conditioning of the adsorbing material, the danger of artifact formation during storage and exposure may be unavoidable and needs to be considered. Adsorbent tubes are widely used by occupational hygienists to collect VOCs for personal exposure monitoring in the workplace (Chan et al., 1990; Tang et al., 1993; Hallama et al., 1998). Industrial hygiene studies have utilized charcoal or silica gel adsorption tubes and solvent extraction with gas chromatographic analysis for many years (NIOSH, 1994). The main disadvantages are that this method is rather labor-intensive, no universally usable solvent for the quantitative desorption of polar compounds exists, and the waste solvent contamination from the solvent extraction process produces. A recently developed thermal desorption method with new types of solid adsorbents, such as graphitized carbon blacks and carbon molecular sieves, is better in some of the aspects mentioned above (Hallama et al., 1998; Pankow et al., 1998; Clement et al., 2000). EPA Method TO-17 (1997) employs a multisorbent tube for determination of VOCs in ambient air. Tubes with more than one sorbent, packed in order of increasing sorbent strength, can be used to facilitate quantitative retention and desorption of VOCs over a wide range of polar and nonpolar targets. Pollutants of various molecular weights or boiling points are collected in successive adsorption beds, thereby improving the sampling eciency. The method employs thermal desorption to release the compounds to the GC/MS detector for analysis. Depending on the sampled volume, sensitivities in the low and sub-ppb range can be easily achieved for active sampling, so the sampling has preferentially been used in air analysis. Nevertheless, reaction of collected compounds during adsorption or desorption and the consequent production of artifact species is sometimes a problem with this type of sampling. In 2000, notebook computers from Taiwan represented 32% of the worlds output, creating a market for locally manufactured LCD panels. With an increasing number of companies entering the market to produce LCDs and others expanding their capacity, Taiwans demand for LCD raw material has reached a record high. LCD fabrication takes place in class-100 cleanrooms, with ltered air of very low particle density, plus extremely well-controlled temperature and humidity (Zappe, 1995; Chuah et al., 2000). Production requires routinely a diverse range of VOCs in the factories; however, little is known about the quality of workplace air in the so-called class-100 cleanrooms at LCD fabrication facilities. The use of Tenax-TA adsorption for the measurement of toxic organic compounds in workplace air of surface coating industries using thermodesorption/gas chromatography has been evaluated (Yang

and Lo, 1997). The method was found to be reliable for monitoring nonpolar VOCs in the workplace environment. In this study, adsorptive (active) sampling on multibed collection tubes followed by thermal desorption-GC/MS was used. An analytical protocol was developed for the analysis of nonpolar and polar VOCs in workplace air during the LCD manufacturing process.

2. Experimental 2.1. Standard samples All VOCs (acetone, isopropyl alcohol, hexane, 2butanone, 2-butanol, benzene, toluene, butyl acetate, propylene glycol methyl ether acetate (PGMEA), m,pxylene, and o-xylene) were obtained from Chemservice (West Chester, PA, USA) with purity above 99.5%. Gas standards were prepared by injecting a certain amount of the liquid standards containing a mixture of the 12 VOCs into a Model 232 Tedlar (SKC Inc., USA) sampling bag lled with a known amount of helium gas as the dilution gas. In order to simulate the humidity in the workplace air, the dilution gas had been passed through deionized water before lling the sample bag. Water was puried by a Millipore Milli-Q instrument on the basis of reverse osmosis, charcoal adsorption, and UV irradiation. Gas standards were prepared with a relative humidity of 80 5%. The concentration of a VOC in a gas mixture can be calculated from Cppmv 22:4 106  qVl MV  T 273  760 P  1

where Cppmv is the concentration of the VOC sample (ppmv), 22.4 is molar volume of gas at STP (l/mol), q is the density of the liquid sample injected (g/ml), Vl is the volume of the liquid sample injected (ml), M is the molecular weight of the analyte (g/mol), V is the total volume of the dilution gas (l), T is temperature (K), and P is pressure (Torr). The concentration of each VOC in the gas standards was prepared at a level of approximately 100 ppmv by this static method. 2.2. Sorbent collection tube The multibed collection tubes were custom-made of glass (16 cm 4 mm ID and a wall thickness of 2 mm). Each tube contained 0.2 g of Carbopack C, 0.2 g of Carbopack B, and 0.2 g of Carbosieve SIII (Supelco Inc., Bellefonte, PA) in that order. This was the same order in which the tubes were used for sampling. The direction of the multisorbent collection tubes was always reversed during thermal desorption. Carbopack C is a

C.-H. Wu et al. / Chemosphere 56 (2004) 7180

73

moderately strong graphitic carbon adsorbent with a specic surface area of 10 m2 /g. Carbopack B is a stronger adsorbent with a specic surface area of 100 m2 /g. It exhibits a low capacity for water, a low background in thermal desorption, and a generally high thermal stability. Being one of the strongest sorbent materials currently available, Carbosieve SIII has a specic surface area of 830 m2 /g. Its use allows the collection of highly volatile compounds. Like Carbopack C and B, Carbosieve SIII exhibits a low background in thermal desorption and a high thermal stability. Its high surface area, however, translates into a high sorptive capacity for water. Excess water on the capillary column can change the adsorption properties of the stationary phase and cause unpredictable changes in the retention times and, in some instances, irreparable damage to the column. The adsorbents were held in place by a small amount of deactivated quartz wool. Prior to sampling, all tubes were conditioned for 3 h at 350 C using nitrogen gas (>99.999%) at 120 ml/min. In blank experiments, the tubes were found to be stable for at least 15 days.

2.3. Active sampling Air samples were collected from 10 dierent sites of STN-LCD and TFT-LCD manufacturing process at Factories A and B in March 2002, respectively. Active sampling was done using a sequential tube sampler (Perkin Elmer STS 25, Norwalk, CT, USA). The ow rate was regulated by a needle valve and calibrated by a rotameter (Drycal BIOS, USA). The coecient of variation in ow rate (1050 ml/min) was less than 1.6%. The sampled volume is calibrated as the product of the average gas ow through the adsorption tube and the sampling time. Typical sample volumes in the range of 1 l were taken over a period of 1 h at a sampling rate of 1020 ml/min. Before and after sampling, all tubes were plugged tightly with special 1/400 Teon screw caps and stored in a clean plastic container lled with active charcoal at room temperature. To avoid any possible cross contamination, blank experiments were performed between samples.

desorbed at 220 C for 6 min. The direction of the gas ow was always reversed during thermal desorption so that higher boiling compounds were back-ushed easily from the sampling end of the tube. The desorbed analytes were retrapped at )160 C for 6 min by a Teon tube (10 cm 1.6 mm ID) lled with 2 cm deactivated glass beads (Mesh 60/80). Cooling of the internal trap was made possible by purging liquid nitrogen into the trap housing. The thermocouple was attached to the trap tubing to provide temperature feedback. After sample concentration, the trap was ash-heated to 200 C at 50 C/s for 5 min, and the trapped analytes were introduced into a moisture control system to remove water vapor before being injected into GC/MS system. The thermodesorption and cryofocusing conditions would be discussed in the subsequent sections. The valves and the transfer line were kept at 180 C. The trap was cleaned by baking at 200 C for 20 min after desorption. The analytical column was a CP-Wax fused-silica capillary column of 60 m 0.32 mm ID with 0.5 lm lm thickness (Chrompack, Middelburg, The Netherlands). The inlet pressure of the ultra-pure helium carrier gas (>99.9995%) was set at 0.36 bar. Oven temperature were programmed as follows: start at 35 C; ramp at 3 C min1 to 100 C and hold for 5 min; nally ramp at 5 to 160 C min1 and hold for 5 min. The total analysis time was 44 min. The transfer line between GC and MS was kept at 280 C. The mass selective detector with electron ionization was operated in scan mode at a rate of 6.1 scans s1 from 40 to 260 amu. From comparison of both the retention times and major fragments with those of the standards, the peaks of the real samples were identied when they exhibited >85% match (Wu et al., 2002). With the establishment of the calibration curves of the analytes, the concentrations of VOCs could be calculated from their peak areas. 2.5. Analyte stability and breakthrough Storage tests were performed by loading a set of 30 tubes with standard gas as mentioned above. Six of the tubes were analyzed immediately. The remaining tubes were divided into two same groups, stored at room temperature (25 C) and in the refrigerator (3 C), respectively. They were analyzed in sets of three each over the next four weeks to examine analyte stability on the cartridges. The breakthrough volume of a given compound on a given adsorbent is dened as the maximum air volume that can be sampled using a trap made with a given quantity of the adsorbent, before the compound exits the trap (Comes et al., 1993). A direct measurement was adopted by a ame ionization detector (FID) to examine the breakthrough on the tubes. The analytes were passed through the multisorbent tubes using the sample pump

2.4. Analytical instrumentation An automatic thermal desorption cryofocusing injector system (Tekmar 6016 and Tekmar 6000, Cincinnati, OH, USA) was connected by a transfer line maintained at 180 C to a Hewlett-Packard 5890 A GC system equipped with a HP 5971 A mass selective detector (Palo Alto, CA, USA). The automated Tekmar thermal desorption device can process up to 16 samples automatically. The multisorbent tubes were thermally

74

C.-H. Wu et al. / Chemosphere 56 (2004) 7180

(SKC Inc., USA) at a ow rate of 400 ml/min from a sampling bag which contained a certain concentration for each gas standard in a condition of 80 5% RH and at 27 1 C. Data were collected automatically every 15 s. When the concentration of the eluting vapor reached 5% of the applied test concentration, the breakthrough volume was determined. The safety sampling volume is two-thirds of the breakthrough volume. The permissible exposure limit (PEL) values were prepared twice for the measurement of the breakthrough volume. The thermal desorption recovery was calculated as: Recovery Amultisorbent 100% Aempty 2

were identied from their molecular fragmentation and quantied from the peak area of their major fragment ions. Except m-xylene and p-xylene, which were not resolved, all the VOCs could be well separated and identied. In order to attain the optimization of stability and response, the eects of desorption temperature of the adsorbent tube, desorption time of the adsorbent tube, cryofocusing temperature, desorption temperature of the cryofocusing desorber, and desorption time of the cryofocusing desorber have been considered and studied

Table 1 Summary of recovery yield, method detection limit, and repeatability Compound Acetone IPA Hexane 2-Butanone 2-Butanol Benzene Toluene Butyl acetate PGMEA m,p-Xylene o-Xylene
a b

where Amultisorbent is the peak area of the standard sample through a conditioned multisorbent tube and Aempty is the peak area of the standard sample through an empty tube.

Recoverya (%) 96.6 97.3 97.7 96.5 97.4 96.2 96.7 96.8 97.2 97.9 98.2

MDLb (ppb) 0.38 0.78 0.69 0.55 0.75 0.69 0.58 0.43 0.44 0.46 0.45

Repeatabilityc , CV (%) 1.1 2.3 1.6 1.3 1.8 3.2 3.4 2.2 1.4 2.2 2.3

3. Results and discussion 3.1. Chromatogram of the multibed adsorption/desorption analysis The target compounds in this study are widely used in the opto-electronics industry as cleaning solutions or solvents of photoresists. Displayed in Fig. 1 is a total ion chromatogram resulting from the multisorbent sorption analytical method of a mixed standard. Compounds

Three replicate measurements were used. Sampling volume was 1 l. c Seven replicate measurements were used. CV: coecient of variation.

Fig. 1. Chromatogram of the 11 VOCs reference standard acquired with GCMS under the optimized conditions. Chromatographic conditions are described in Section 2.

C.-H. Wu et al. / Chemosphere 56 (2004) 7180 Table 2 Linearity of the 11 VOCs Compound Acetone IPA Hexane 2-Butanone 2-Butanol Benzene Toluene Butyl acetate PGMEA m,p-Xylene o-Xylene
a b

75

Calibration curve formulaa; b Y Y Y Y Y Y Y Y Y Y Y 0:96 10 X 0:95 105 X 2:22 105 X 3:22 105 X 3:72 105 X 0:78 105 X 0:95 105 X 1:14 105 X 0:90 105 X 0:98 105 X 1:04 105 X
5

Correlation coecient (R2 )

6

Linear range (ng) 24713 25737 18529 15442 15455 16479 19565 24713 27810 22651 22651

6:28 10 1:31 106 4:16 106 2:37 106 1:83 106 3:94 105 2:38 106 1:61 106 4:82 106 0:46 106 0:51 106

0.996 0.995 0.998 0.998 0.998 0.997 0.995 0.997 0.998 0.995 0.996

X is quantity of each compound (ng) in a 10-l-sampling bag and Y is peak area count. Seven dierent quantities were used for the calibration curve of each compound.

(see Appendix A). Under the optimization of analytical method, response factors for most compounds were linear over 2 orders of magnitude. 3.2. Breakthrough volume and analyte stability No breakthrough was observed for any of the compounds for an air volume of 5.6 l (relative humidity 80%). For an air volume of 10 l (again at relative humidity 80%), breakthrough to a limited extent was

observed for isopropyl alcohol and acetone. These results are in good agreement with the data in the literature, which suggest that the breakthrough volume of the compounds examined is in order of 101000 l/g adsorbent (Mastrogiacomo et al., 2000; Tolnai et al., 2000b). For practical purpose, a 1-h sampling time at a ow rate of 1020 ml/min was used. The volume used in sampling in the eld was 1 l. Therefore, the collection tube had sucient capacity to adsorb the VOCs at ambient temperature.

Fig. 2. Total ion chromatogram of 1000-ml air sample collected with adsorbent tubes in a LCD manufacturing company.

76

C.-H. Wu et al. / Chemosphere 56 (2004) 7180

Fig. 3. Extracted ion chromatograms for m=z 58 (acetone), 45 (IPA), 75 (trimethyl silanol), 147 (hexamethyl disiloxane), 91 (toluene), 61 (butyl acetate), 72 (PGMEA), 106 (xylene), and 57 (1-butoxy-2-propanol) for a typical air sample collected with adsorbent tubes in a LCD manufacturing company.

The stability of the compounds on the adsorbent was studied. When the duration of storage increased, the recovery was always better for storage at 3 C than at room temperature. The results indicate that storage time should be as short as possible in order to keep the blank build-up level on adsorbents as low as possible (Cao and Hewitt, 1994). The recovery values after 7-day sample storage at 3 C were close to 100% for all compounds

except hexane (94%) and 2-butanone (93%). This proved perfect analyte stability on the multibed collection tubes. 3.3. Quality parameters Table 1 lists the recovery, method detection limit (MDL), and repeatability of the target compounds. To ensure the accuracy of the calibration and the stability of

C.-H. Wu et al. / Chemosphere 56 (2004) 7180

77

the system, the recoveries of the analytes were evaluated. Recovery for the standards from the collection tube was determined from three replicate experiments, using a standard gas containing 2342 ng/l of each compound. The average recoveries of all the analytes ranged from 96.2% to 98.2%. The results indicate that the multisorbent air-sampling system established in this study is well suited for quantitative analysis of the target compounds. The method detection limits (MDL) were calculated as three times the standard deviation of seven replicate measurements at 14 ng, close to blank concentration. The system exhibited capabilities of detection at a ppb level. For the quality assurance/quality control of the calibration curves, the repeatability of each analyte was evaluated from seven measurements. The repeatability expressed in terms of coecient of variation (CV) was found to be in the range of 1.13.4%. The calibration curves were constructed by injecting a certain amount of the gas standards from 10-l sampling bag into the multibed collection tubes, and followed by passing through 300 ml of high-purity nitrogen gas. Table 2 displays the seven concentrations of the standards prepared with the range of 15737 ng, and their respective linearity. Under the optimization of analytical method, all analytes exhibit good linearity over 2 orders of magnitude. The linear correlation coecients, r2 , were better than 0.995. 3.4. Measurements of VOCs in workplace air Exposure to VOCs in the workplace is subject to regulations and has to be monitored regularly. The air samples were collected from the class-100 cleanrooms at two LCD fabrication facilities (Factories A and B). Fig. 2 presents a typical total ion chromatogram for the VOCs in the workplace air. The results showed that 15 compounds have been detected and seven of them could be well quantied with the established method. Ten compounds were identied by direct interpretation of the mass spectra obtained by GCMS analysis. Fig. 3 shows the extracted ion chromatograms for m=z 58, 45, 75, 147, 91, 61, 72, 106, and 57. The peak areas at m=z 58 due to acetone, 45 due to IPA, 75 due to trimethyl silanol, 147 due to hexamethyl disiloxane, 91 due to toluene, 61 due to butyl acetate, 72 due to PGMEA, 106 due to xylene, and 57 due to 1-butoxy-2-propanol were measured. In contrast to the total ion chromatogram in Fig. 2, the peaks due to the seven components of reference standard can be clearly seen. When multiple peaks occur in the single ion chromatogram, the component of interest can be identied by the mass spectra. The temporal variations in the analyte concentrations observed at LCD fabrication Factory A are shown in Fig. 4. In Factory A, the results showed that concentrations of acetone, butyl acetate, and PGMEA were the three highest among those of VOCs detected any time.

Fig. 4. Hourly measurements of major VOCs determined with adsorbent tubes in LCD Factory A from 2223 March 2002.

Other VOCs, such as toluene, IPA, and xylene, were present at 140 ppb levels. Butyl acetate and PGMEA were observed periodically during day and night with an amplitude of 300 ppb. The uctuations in their concentrations were possibly due to improper use, thus resulting in intermittent expulsion during the manufacture of LCD. A typical LCD fabrication process consists of nine stages. In order to locate the sources of VOC emission, measurements were made at the nine places in the class100 cleanrooms. Table 3 shows the contribution of VOCs during various stages of STNLCD fabrication. Acetone and PGMEA were two major components of VOCs emitted during working hours in Factory B. Acetone, a widely used solvent for cleaning, was detected with the highest concentration at UV exposure site. PGMEA, a major solvent for photoresist, was found at the beginning of manufacturing process: glass base cleaning and photoresist coating sites. Most of the contaminants found in Factory A were also detected in Factory B, but with a little dierent concentration prole. Table 4 lists the average and range of VOC concentrations in the workplace air of LCD Factories A and B. As can be seen, the concentration of total VOCs in the workplace air ranged from 10 to 700 ppb, of which over 90% was attributed to acetone, butyl acetate and PGMEA. The results showed that concentrations of VOCs in the class-100 cleanrooms were always below their permissible exposure limit values, but still signicantly higher than those in outdoor air (Chang et al., 1994; Yang and Lo, 1998; Lo, 2000; Doskey and Bialk, 2001; Wu et al., 2003). In particular, trimethyl silanol, a potentially toxic compound, was identied from the air

78

C.-H. Wu et al. / Chemosphere 56 (2004) 7180

Table 3 Concentrations of VOCs in the workplace air of various process stages of STNLCD fabricationa Sitesb 1 2 3 4 5 6 7 8 9
a b

Concentration (ppbv) Acetone 13.5 23.5 695.7 24.5 10.1 5.6 1.5 3.9 5.3 Benzene NDc ND 1.1 2.8 2.1 2.3 1.5 1.2 2.3 Toluene 1.1 ND 2.3 2.5 2.2 1.5 2.3 2.1 2.1 Butyl acetate ND ND ND ND ND ND ND ND ND PGMEA 10.2 23.1 ND ND ND ND ND ND ND m,p-Xylene 1.2 ND 1.1 1.3 1.1 1 ND 5.5 1.2 o-Xylene 1.5 ND 1.2 1.5 1.1 1.5 ND 1.1 1.2

Samples were collected simultaneously in class-100 cleanrooms at a LCD factory during a working hour on 5 March 2002. Nine sampling sites are in the sequence of manufacturing process. The activities carried out at each site were glass base cleaning (site 1), photoresist coating (site 2), UV exposure (site 3), development (site 4), etching (site 5), photoresist stripping (site 6), rubbing and PI print (site 7), spacer sealant (site 8), and hot press and cell attachment (site 9). c ND: Not detectable. Table 4 Average and range of VOC concentrations in the workplace air of LCD fabrication facilities in March 2002 Compound Acetone IPA Benzene Toluene Butyl acetate PGMEA m,p-Xylene o-Xylene Average (ppb) 75.9 6.7 1.5 3.5 237.0 252.7 1.54 1.2 Range (ppb) 1.3695.7 <87.9 <2.8 <10.3 <346.1 <455.2 <2.9 <1.8

work was supported by the National Science Council of the Republic of China.

Appendix A. Optimization of desorption conditions The appropriate desorption temperature was evaluated to ensure that all analytes were completely desorbed from the multisorbent tubes in order to reach the highest sensitivity and to avoid carryover. Fig. A.1 shows the dependence of the analyte recovery on the desorption temperature. Except for the desorption temperature, the thermodesorption conditions of the cryofocusing system were set the same as those in the analytical instrumentation section. It was found that the signals for all analytes increased initially with increasing desorption temperature of the multisorbent collection tubes, and then decreased gradually afterwards when the temperature was above 220 C. The decrease may be due to the decomposition of analytes. Therefore, an optimal desorption temperature of the adsorbent tube was chosen to be at 220 C for subsequent experiments. Fig. A.2 shows the dependence of the analyte recovery on the desorption time. The desorption time was found to have little inuence on the desorption eciency for desorption times. The desorption time of 6 min was enough to ensure total desorption and no peaks appeared in the chromatogram corresponding to the analysis of the adsorbent tube prior to reuse. Fig. A.3 shows the dependence of the analyte recovery on the cryotrap temperature. Usually, a cryofocusing trap is needed before the GC analysis, to provide narrow bands entering the capillary columns, and thus good resolution. Due to the weak adsorption of the glass beads with analytes in the cryotrap tube, the cryotrap temperature was highly correlated with the eciency of the cryofocusing system. Optimum cryotrap-temperature is de-

samples though not quantied. In addition, the VOC pollution was also of prime concern in LCD manufacturing process for high-quality production. 4. Conclusion A method for qualitative and quantitative analysis of volatile organic compounds (VOCs) in the air at LCD fabrication facilities has been developed with good precision and linearity. The thermal desorption method using multisorbent tubes including Carbopack B, Carbopack C, and Carbosieve SIII has been successfully applied to the analysis of VOCs in workplace air. Optimal adsorption/thermal desorption conditions have been determined and analysis can be completed in 44 min. The system has been demonstrated to be capable of measuring the temporal variations in trace levels of VOCs with a short time resolution. Acknowledgements The authors would like to thank all the graduate students for their assistance with the eldwork. This

C.-H. Wu et al. / Chemosphere 56 (2004) 7180

79

Fig. A.1. The dependence of the analyte recovery on the desorption temperature for the multibed adsorbent tube. The peak area has been normalized to the maximum value for each individual compound. Symbols: ( ) acetone; ( ) IPA; (M) hexane; (N) 2-butanone; (}) 2-butanol; (r) benzene; () toluene; (j) butyl acetate; () PGMEA; ( ) m,p-xylene; (O) oxylene. Other conditions as in Section 2.

Fig. A.3. The dependence of the analyte recovery on the cryotrap temperature. The peak area and notation as in Fig. A.1. Other conditions as in Section 2.

Fig. A.4. The dependence of the analyte recovery on the desorption time of the cryofocusing desorber. The peak area and notation as in Fig. A.1. Other conditions as in Section 2. Fig. A.2. The dependence of the analyte recovery on the desorption time for the multibed adsorbent tube. The peak area and notation as in Fig. A.1. Other conditions as in Section 2.

ned as the temperature after which the amount of extracted analyte remains constant. No signicant dierences were observed in MS areas registered when the cryofocusing temperature was below )160 C. As a

conclusion, )160 C was considered as the optimized cryofocusing temperature for further studies. Fig. A.4 shows the dependence of the analyte recovery on the desorption time of the cryofocusing desorber at 200 C. The prole indicates that the desorption time has little inuence on the desorption eciencies for acetone, benzene, and trichloroethylene and the desorption

80

C.-H. Wu et al. / Chemosphere 56 (2004) 7180 of Research and Development, US Environmental Protection Agency, Cincinnati, OH. NIOSH, 1994. Manual of Analytical Methods, Department of Health and Human Services, National Institute of Occupational Safety and Health, Cincinnati, OH. Pankow, J.F., Luo, W., Isabelle, L.M., Bender, D.A., Baker, R.J., 1998. Determination of a wide range of volatile organic compounds in ambient air using multisorbent adsorption/thermal desorption and gas chromatography/ mass spectrometry. Anal. Chem. 70, 52135221. Roche, A., Thevenet, R., Jacob, V., Kaluzny, P., Ferrari, C., Baussand, P., Foster, P., 1999. Performance of a thermally desorbable type-tube diusive sampler for very low air concentrations monitoring. Atmos. Environ. 33, 19051912. Sunesson, A.L., Nilsson, C.A., Andersson, B., 1995. Evaluation of adsorbents for sampling and quantitative analysis of microbial volatiles using thermal desorption-gas chromatography. J. Chromatogr. A 699, 203214. Tang, Y.-Z., Tran, Q., Fellin, P., Cheng, W.K., Drummond, I., 1993. Determination of C1C4 hydrocarbons in air. Anal. Chem. 65, 19321935. Tolnai, B., Gelencser, A., Gal, C., Hlavay, J., 2000a. Evaluation of the reliability of diusive sampling in environmental monitoring. Anal. Chim. Acta 408, 117122. Tolnai, B., Hlavay, J., Moller, D., Prumke, H.-J., Becker, H., Dostler, M., 2000b. Combination of canister and solid adsorbent sampling techniques for determination of volatile organic hydrocarbons. Microchem. J. 67, 163169. Uchiyama, S., Asai, M., Hasegawa, S., 1999. A sensitive diusion sampler for the determination of volatile organic compounds in ambient air. Atmos. Environ. 33, 19131920. Wang, J.L., Ding, W.H., Chen, T.Y., 2000. Source determination of light non-methane hydrocarbons by simultaneous multi-site sampling in a metropolitan area. Chemosphere 2, 1122. Wolska, L., Janicki, W., Namiesnik, J., 1995. Inuence of concentration and sample volume on the recovery of compounds from water following direct sorption on Tenax TA-thermal desorption. Analyst 120, 27812786. Wu, C.H., Lu, J.T., Lo, J.G., 2002. Analysis of volatile organic compounds in wastewater during various stages of treatment for high-tech industries. Chromatographia 56, 9198. Wu, C.H., Lin, M.N., Feng, C.T., Yang, K.L., Lo, Y.S., Lo, J.G., 2003. Measurement of toxic volatile organic compounds in indoor air of semiconductor foundries using multisorbent adsorption/thermal desorption coupled with gas chromatographymass spectrometry. J. Chromatogr. A 996, 225231. Yang, K.-L., Lo, J.-G., 1997. Evaluation of tenax-TA adsorption for the measurement of toxic organic compounds in workplace air using thermodesorption/gas chromatography. Anal. Sci. 13, 337342. Yang, K.-L., Lo, J.-G., 1998. Volatile hydrocarbons (C6C10) measurements at remote sites of Taiwan during the PemWest A experiment (1991). Chemosphere 36, 18931902. Zappe, H.P., 1995. Introduction to Semiconductor Integrated Optics. Artech House, Norwood, MA.

eciencies for the other analytes reach plateau within 5 min. Therefore, the desorption time of 5 min was selected for best recovery and further studies.

References
Cao, X.L., Hewitt, C.N., 1994. Build-up of artifacts on adsorbents during storage and its eect on passive sampling and gas chromatography-ame ionization detection of low concentrations of volatile organic compounds in air. J. Chromatogr. A 688, 368374. Chan, C.C., Vainer, L., Martin, J.W., Williams, D.T., 1990. Determination of organic contaminants in residential indoor air using an adsorptionthermal desorption technique. J. Air Waste Manage. Assoc. 40, 6267. Chang, S.Y., Tso, T.-L., Lo, J.-G., 1994. Determination in situ of chemical solvent vapors in a semiconductor workplace with an open-path FTIR spectrometer. J. Chin. Chem. Soc. 41, 685692. Chang, C.-C., Lo, J.-G., Tsai, C.-H., Wang, J.-L., 2001. Concentration variability of halocarbons over an electronics industrial park and its implication in compliance with the Montreal Protocol. Environ. Sci. Technol. 35, 32733279. Chuah, Y.K., Tsai, C.-H., Hu, S.C., 2000. Simultaneous control of particle contamination and VOC pollution under dierent operating conditions of a mini-environment that contains a coating process. Building Environ. 35, 9199. Clement, M., Arzel, S., Le Bot, B., Seux, R., Millet, M., 2000. Adsorption/thermal desorption-GC/MS for the analysis of pesticides in the atmosphere. Chemosphere 40, 4956. Comes, P., Gonzalez-Flesca, N., Menard, T., Grimalt, J.O., 1993. Langmuir-derived equations for the prediction of solid adsorbent breakthrough volumes of volatile organic compounds in atmospheric emission euents. Anal. Chem. 65, 10481053. Doskey, P.V., Bialk, H.M., 2001. Automated sampler for the measurement of non-methane organic compounds. Environ. Sci. Technol. 35, 591594. Hallama, R.A., Rosenberg, E., Grasserbauer, M., 1998. Development and application of a thermal desorption method for the analysis of polar volatile organic compounds in workplace air. J. Chromatogr. A 809, 4763. Kuo, H.-W., Wei, H.-C., Liu, C.-S., Lo, Y.-Y., Wang, W.-C., Lai, J.-S., Chan, C.C., 2000. Exposure to volatile organic compounds while commuting in Taichung, Taiwan. Atmos. Environ. 34, 33313336. Lo, J.-G., 2000. Report of Air Toxics in Hsinchu Science-Based Industrial Park in 2000, Science Park Administration, Hsinchu, Taiwan. Mastrogiacomo, A.R., Pierini, E., Sampaolo, L., 2000. Sampling and analysis of volatile organic compounds in air using a dualsorbent trap. Chromatographia 52, 345350. Method TO-17, 1997. Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air, Center for Environmental Research Information Oce