Determination of Short-term Exposure with a Direct Reading Photoionization Detector

Transcription

1 Ann. occup. Hyg., Vol. 48, No. 1, pp , British Occupational Hygiene Society Published by Oxford University Press DOI: /annhyg/meg079 Determination of Short-term Exposure with a Direct Reading Photoionization Detector P. POIROT 1 *, I. SUBRA 1, F. GÉRARDIN 1, V. BAUDIN 2, S. GROSSMANN 3 and M. HÉRY 4 1 Département Ingénierie des Procédés; 2 Service Epidémiologie en Entreprise; 3 Département Polluant et Santé; 4 Direction Scientifique, INRS, BP 27, F Vandoeuvre Cedex, France Received 21 October 2002; in final form 12 June 2003 The use of a direct reading photoionization detector (PID) to determine short-term solvent exposures is described in the present paper. To assess the relevance of such a total exposure evaluation it was necessary to compare it with the real concentration of pollutants. This comparison was made by measuring in parallel with the PID determination the concentration of each pollutant using a standard technique, i.e. sampling on charcoal tubes and subsequent analysis by gas chromatography. Laboratory tests showed that the linearity of the answer of the PID is good for many compounds and for a mixture of these compounds. Similar tests were carried out for painters in workplaces with the same good correlations (determination coefficient r 2 close to 1) between the PID response and the real concentration of the pollutants measured on the sampling tubes. The use of PID also allowed determination of the exposure profile of the workers and comparison of the short-term exposure to the corresponding limit values. Many cases of the short-term limit values being exceeded were revealed by use of the PID, although very few cases of the long-term limit values have been found by the usual sampling (charcoal tube) and analytical (gas chromatography) methods. Keywords: photoionization detector; short-term exposure; solvent; limit values INTRODUCTION To protect workers from diseases such as irritation, chronic or irreversible tissue damage and narcosis, organizations responsible for occupational safety and health have laid down specific short-term exposure limit values. In the case of the American Conference of Governmental Industrial Hygienists (ACGIH), whose values are widely referred to throughout the world, the corresponding values are the TLV-STEL (threshold limit value short-term exposure limit), which should be checked over a period of 15 min, and the TLV-C (ceiling), for which instantaneous monitoring is recommended whenever possible (ACGIH, 2002). These short-term exposure limit values are the counterpart of the well-known long-term TLV-TWA (time-weighted average). Comparing the exposure of workers to these shortterm limit values is often difficult and time *Author to whom correspondence should be addressed. Tel: ; fax: ; pascal.poirot@inrs.fr consuming work for industrial hygienists (Kumagai and Matsunuga, 1995, 1999): determining the most appropriate monitoring time may be difficult and the occupational hygienist is often obliged to increase the number of sampling operations; the detection limits of the analytical methods are sometimes poor and in some cases a 15 min sampling may be too short for the conventional analytical techniques. However, the use of direct reading photoionization air detectors (PIDs) permits, under certain conditions, the evaluation of occupational exposure to organic solvent mixtures, taking into account possible short duration peak concentrations. A PID has recently been used to measure the benzene exposure of a truck driver during the loading of gasoline (Drummond, 1997), the TWA concentration being determined in parallel by sampling on a charcoal tube. Assuming that the response of the PID is linear, the author can estimate the work site conditions with a correlation of 75

2 76 P. Poirot et al. exposure and activity. Another study of short-term exposure to solvent mixture was performed on construction painters (Coy et al., 2000). The principle of simultaneous sampling on charcoal tubes and PID was also adopted: the data were highly correlated. As part of an epidemiological study devoted to the neurological effects of solvents among apprentice painters (Poirot et al., unpublished results), our laboratory was asked to assess the exposure of house painters. This assessment involved evaluating the mean exposure during a workshift as well as the exposure profile and took into account the variations in the exposure level over the workshift. Special emphasis was put on the exposure peaks. At the same time, a study of occupational hazards among workers employed in the reclamation of an industrial site led us to suspect the occurrence of short but high exposure to a broad range of organic solvents. In both cases the mixture of pollutants was complex and the quantity of each product sampled over a short period was small, making the analysis difficult and time consuming. The use of a direct reading PID to determine these short-term exposures is described in the present paper. MATERIALS AND METHODS As the measurement principle of the PID is based on the determination of all photoionizable gases and vapors, it does not distinguish between the individual pollutants present in the atmosphere. Consequently, the concentration is measured in the form of a composite index. To assess the relevance of such a total exposure evaluation it was necessary to compare it to the real concentration of pollutants. This comparison was made by measuring the concentration of each pollutant using a standard technique, i.e. sampling on charcoal tubes and subsequent analysis by gas chromatography, in parallel with the PID determination. Photoionization detector A direct reading PID (2020 PE Photovac) was used. The energy of the UV lamp was 10.6 ev. The equipment was calibrated with a 100 p.p.m. calibration gas (isobutylene). In use, the concentration is measured every second, but a 1 min averaging interval was set because the data logger can only store 1000 entries. In these conditions the monitor can operate for 16 h before the data logger is filled. After sampling the data were transferred to a personal computer. Sampling and analysis on charcoal tubes Two-section coconut charcoal tubes (SKC ) were used at a sampling rate of 0.1 l/min (Gilian LFS 113 sampling pumps). The analysis was performed by gas chromatography. In both cases (house painters and site remediation), the charcoal was desorbed in 1 ml of carbon disulfide with a low benzene content (CS 2, Aldrich ). The pollutants were different in the two experiments: petroleum spirit, i.e. a petroleum cut from C 8 to C 12 for the painters, and a mixture of industrial solvents for the site reclamation (chlorinated hydrocarbons, ketones, naphtenic hydrocarbons, etc.), but the chromatographic conditions were the same. A wide bore capillary column (60 m long, 0.75 mm i.d. Supelco SPB1 column) was used with flame ionization detection (FID). Analysing petroleum spirits is always difficult on account of the large number of components contained in the mixture. When it is not possible to obtain a reference sample of the petroleum spirit used, a qualitative analysis is carried out on a tube sampled in the work atmosphere to determine which petroleum spirit currently available in the laboratory is analytically closest to the sample to be determined. This petroleum spirit is then taken as the reference and used to plot the calibration curve. For the quantitative analysis, about a dozen peaks evenly distributed over the chromatogram are chosen amongst the major ones. It was checked that the results obtained by this method do not differ significantly from the results that would have been obtained with calibration with the petroleum spirit itself considered. Tests in the laboratory The proportionality of the response of the PID according to the concentration of the product tested was checked in the laboratory. These tests were carried out successively for two different petroleum spirits, p-xylene, m-xylene, styrene and ethylbenzene, in controlled atmospheres. They consisted of sampling simultaneously on a charcoal tube and by a PAM sampling inlet, both placed in the cell of generation benches where the spatial and temporal homogeneity had been tested previously (Gradiski et al., 1978; Castel et al., 1998). The sampling time ranged from 5 to 120 min according to the concentration of the product generated in the bench cell. Tests in workplace atmospheres The same type of comparison (PID versus charcoal tube) was carried out during the house painting and industrial site reclamation operations. The sampling devices were worn by volunteers for 3 8 h during their workshift. These comparisons were in fact complementary to the principal aim of the study which was, as mentioned previously, to determine the exposure profile of the workers and to compare the short-term exposure to the corresponding limit values. As few products have been assigned short-term exposure limits, such as a TLV-STEL or a TLV-C, the comparison was made to the excursion limits defined by the ACGIH:

3 Direct reading photoionization detector 77 Table 1. Limit values of solvents encountered during the study (expressed in mg/m 3 ) France: VME a Assigned by analogy with 1-methoxy-2-propanol. UK: Long-term exposure limit ACGIH: TLV- TWA France: VLE UK: Short-term ACGIH: TLVexposure limit STEL n-hexane Cyclohexane Ethanol Isopropanol Isobutanol n-butanol Acetone Methyl ethyl ketone Ethyl acetate n-butyl acetate Methoxy propanol acetate 1-Ethoxy-2-propanol 425 a Methylglycol acetate Ethylglycol Ethylglycol acetate Butylglycol Butylglycol acetate Butyldiglycol Dichloromethane Dichloroethane Trichloromethane Trichloroethane Trichloroethylene Perchloroethylene Benzene Toluene C petroleum C6 C12 hydrocarbons Stoddard solvent 525 Xylene Ethylbenzene C9 C12 aromatic hydrocarbons Excursions in worker exposure levels may exceed 3 times the TLV-TWA for no more than a total of 30 minutes during a work-day, and under no circumstances should they exceed 5 times the TLV-TWA, provided that the TLV- TWA is not exceeded. (ACGIH, 2002) For exposure of the house painters, the TLV-TWA of Stoddard solvent was taken into account, i.e. 525 mg/m 3 (100 p.p.m.). To determine the exposure of the house painters to other mixtures of solvents and that of the workers employed in industrial site reclamation, where the atmosphere was polluted by various solvents, the long-term threshold limit of the mixture was considered exceeded when the formula recommended by the ACGIH (I TWA = C 1 /TLV 1 + C 2 /TLV C n /TLV n, with C n the atmospheric concentration of component n and TLV n its TLV- TWA) exceeded 1. In the same way an I 3TWA index was considered to take into account the above mentioned ACGIH recommendation for excursion limits. When this index (I 3TWA = C 1 /3TLV 1 + C 2 /3TLV C n /3TLV n ) exceeds 1, the short-term threshold limit of the mixture is considered exceeded. The comparison to the short-term limit values established for solvents corresponds to a particular case in that, as can be seen in Table 1, most shortterm limits established by the ACGIH or the French Department of Employment for these compounds have not been established at 3 times the corresponding TLV-TWAs, but rather at levels ranging from 1.25 to 2 times this long-term value. A second index, I STEL, was then calculated from the ACGIH

4 78 P. Poirot et al. formula with TLV = TLV-STEL (or French VLE, roughly equivalent to the American TLV-STEL) or TLV = 1.5 TLV-TWA in the absence of a TLV- STEL. If the ACGIH and French Department of Employment lists are considered, this value of 1.5 is actually the most common ratio between a TLV- STEL and the corresponding TLV-TWA, and it is an intermediate value in the range generally encountered. This second index I STEL was then considered exceeded when its value was >1. RESULTS Tests in the laboratory For some compounds, the response factors (RF) (real concentration/pid response) of the PID were provided by the apparatus manufacturer, but for complex mixtures such as Stoddard solvent (whose composition varies depending on the origin) it was necessary to determine this RF. Experiments were carried out in the laboratory to check (or determine) this factor and to verify whether the response of the PID to a mixture was the sum of the responses to the different constituents of the mixture. The linearity of the response of the PID was tested between 0 and 100 p.p.m. for Stoddard solvent and between 0 and 500 p.p.m. for toluene, methyl ethyl ketone, ethylbenzene and for a 1/1/1 (w/w/w) mixture of these last three compounds. The results are summarized in Table 2. The linearity is good for all these compounds and mixtures and the RFs are close to the values provided by the apparatus manufacturer. The RF obtained for the toluene/methyl ethyl ketone/ ethylbenzene mixture in the experimental chamber was compared to the theoretical RF calculated from the RFs obtained for each of the components, taking into account their relative weights in the composition of the vapor (which could be calculated from their relative molecular weights as the entire mixture was vaporized). The two results are similar. With y the real concentration (p.p.m.) and x the PID response (p.p.m.), the equations are: y 1 = 0.65x for the measurement of the three-component mixture in the experimental chamber; y 2 = 0.68x 0.05 for the theoretical value established from the RFs of the PID to each of the components tested separately in the experimental chamber. This corresponds to values of y 1 = 68 p.p.m. and y 2 = 69 p.p.m. for x = 100 p.p.m., y 1 = 136 p.p.m. and y 2 = p.p.m. for x = 200 p.p.m. and y 1 = 340 p.p.m. and y 2 = p.p.m. for x = 500 p.p.m. From these results it can be considered that if the relative composition of an atmospheric mixture remains constant, the PID can be used to watch the concentration development over time. Tests in workplace atmospheres Similar tests were carried out for painters in workplaces with the same good correlation (determination coefficient r 2 close to 1) between the PID response and the real concentration of the pollutants measured on the sampling tubes. These results are summarized in Table 3. The correlation was established between Table 2. Comparative response factors (RF) determined by the PID manufacturer and from measurements in an INRS laboratory Compound RF according to manufacturer RF calculated from measurements in the laboratory (determination coefficient r 2 ) Petroleum spirit (A) No data y = 0.61x (r 2 = 0.99) Toluene (B) y = 0.5x y = 0.53x (r 2 = 0.99) Methyl ethyl ketone (C) y = 0.8x y = 0.91x 3.90 (r 2 = 0.99) Ethylbenzene (D) y = 0.5x y = 0.51x (r 2 = 0.99) 1/1/1 (w/w/w) B/C/D mixture No data y = 0.65x (r 2 = 1) x = response of PID (p.p.m.); y = concentration of pollutant (p.p.m.). Table 3. Correlation between pollutant concentration measured in workplace atmospheres and response of the PID Compound No. of values Pollutant concentration = f (response of PID) Exposure index = f (response of PID) Petroleum spirit (E) 15 y = 0.58x Petroleum spirit (F) 4 y = 0.67x Polyurethane paint (1) 8 z = x Epoxy paint (2) 11 z = x Determination coefficient (r 2 ) x = response of PID (p.p.m.); y = pollutant concentration (p.p.m.); z = index of exposure (C 1 /LV 1 + C 2 /LV C n /LV n ); 1, the solvents in the polyurethane paint are methyl ethyl ketone, toluene, methoxy propanol acetate, xylene, ethylbenzene and C 9 aromatic hydrocarbons; 2, the solvents in the epoxy paint are isopropanol, methyl ethyl ketone, toluene, n-butyl acetate, xylene and ethyl benzene.

5 Direct reading photoionization detector 79 Fig. 1. Exposure of a painter during polyurethane floor painting with a roller. Fig. 2. Exposure of a painter during epoxy floor painting with a roller. the PID response and either the Stoddard solvent concentration or the exposure expressed in the form of an I TWA, as defined previously for mixtures (equivalent correlations would have been obtained with the exposure expressed in the form of an I 3TWA or an I STEL ). The corresponding components of the paints are defined in Table 3. Two examples of the exposure profile of the painters are given in Figures 1 and 2. The case described in Fig. 1 involves the painting of a floor with a polyurethane paint exposing the worker to methyl ethyl ketone, toluene, methoxy propanol acetate, xylene, ethylbenzene and C 9 aromatic hydrocarbons. Figure 2 describes the exposure of a worker using an epoxy paint on a floor (isopropanol, methyl ethyl ketone, toluene, n-butyl acetate, xylene and ethylbenzene). In both cases the PID values were recorded on a 1 min integration basis. Although the comparison with the TLV-TWA does not highlight any excess over the limit value expressed in the form of an I TWA, the use of the PID shows several intervals during which the I STEL is exceeded. On the other hand, if the TLV-STELs (or their French equivalents the VLEs) are given up on behalf of the I 3TWA level, no excursion over this reference short-term index over 1 is registered. As part of the epidemiological study mentioned previously (Poirot et al., unpublished results), a total of 125 personal samples on charcoal tubes were taken to assess the mean exposure of the painters during a workshift. The exposure profile was also recorded in parallel for 43 of these 125 samples in parallel by means of the PID, i.e. the probe of the PID and the inlet of the charcoal tube were situated in the imme-

6 80 P. Poirot et al. Table 4. Respective exceedings of exposure indices established from short-term and long-term limit values Activities exposing to Charcoal tube measurements No. of Comparison to I TWA samples No. of samples Maximum time >1 averaged value PID measurements No. of Comparison to I STEL samples No. of samples with instantaneous index value >1 Maximum instantaneous value Comparison to I 3TWA No. of Maximum samples with instantaneous instantaneous value index value >1 Petroleum spirit paints and varnishes Indoor emulsion paints Indoor primers Indoor polyurethane floor paints Indoor epoxy floor paints Neoprene glue Floor vitrification diate vicinity and the measurements performed over the same period. For only 7 of the 11 painting activities studied was at least one sample taken with a PID. The comparative occurrences of excesses over the limit value, depending on whether the long-term limit or the short-term limit is considered, are shown in Table 4. Only the 7 activities for which both charcoal tube and PID measurements were taken are considered in this table. Although exceeding the I TWA index is rare, 15 of the 43 samples taken with the PID reveal at least one instantaneous (in fact determined for a 1 min integration time) value of the I STEL > 1. This number decreases drastically if the limit is set at I 3TWA. It is worth noting in the case of the painters that the exposure is due to a defined product with a relatively constant composition (even if it is likely to vary slightly over time because of differences in the vapor pressures of the different components of the paint). For the polluted soil reclamation work the qualitative and quantitative variations in the atmospheric composition are higher because the pollution is not homogeneous over the entire area treated and the correlation is therefore poorer (r 2 = 0.87). To show this variability, the data from which this correlation was calculated are given in Table 5: they (the samples on charcoal tubes and the concentration recordings with the PID) were collected in the immediate vicinity of a sifting machine through which materials polluted by organic solvents were riddled. Although the composition of the atmosphere is likely to vary significantly from one sample to another for the same pollutant, reflecting the differences in the composition of the soil pollution, the number of products in the mixture is high enough to minimize the influence of this variability. In fact, the most significant contribution to the exposure index is due to trichloromethane, the limit value of which is low enough (25 mg/m 3 ) to make the atmospheric concentration of this compound a major influence. An example of the concentration profile recorded with the PID near the sifting machine is shown in Fig. 3. The variations in this particular profile are also representative of the five other recordings used to determine the correlation between the results of the charcoal samplings and the results of the PID (Table 5). Simultaneously, personal samples were taken on charcoal tubes on workers whose work was to avoid blocking of the sifting machine, which led them to intervene in the middle of the most polluted atmosphere, over the sifter. The exposure of these workers is summarized in Table 6. The pollutants detected are the same as in the area samples taken in parallel with the PID measurements, and the contribution of trichloromethane to the exposure index is also the most important. An example of such a profile is given in Fig. 3. DISCUSSION The choice of the reference limit value is especially important in the case of lack of short-term limit values. Depending on whether the equivalent of the TLV-STEL is fixed at 3 (in conformity with the official philosophy of the ACGIH) or 1.5 (according to its practice in the case of solvents) times the TLV- TWA, the number of excursions above the limit value will vary greatly. In the two cases of painters described in Figs 1 and 2, only methoxy propanol acetate and C 9 hydrocarbons were not assigned a short-term limit value (TLV-STEL or VLE). Their contribution to the pollution index being limited

7 Table 5. Atmospheric concentrations measured on the site remediation from which the correlation between PID response and charcoal tube sample was established Sample Concentration (p.p.m.) Methyl Trichloroethanethanethylenbutylketonethylenbenzentoluenbenzene Isobutanol Dichloro- Trichloro- Benzene Trichloro- Methyliso- Toluene Perchloro- Ethyl- Xylene Ethyl- Trimethyl- Exposure ethyl ketonemethane index <dl 0.2 <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl 0.8 <dl 0.1 <dl <dl Direct reading photoionization detector 81

8 Table 6. Exposure of workers in a polluted setting measured by sampling on a charcoal tube Sample Concentration (p.p.m.) Methyl ethyl ketone Trichloromethane Benzene Isobutanol Dichloroethane Trichloroethane Trichloroethylene Methylisobutylketone Toluene Xylene Perchloroethylene Ethylbenzene Ethyltoluene Trimethylbenzene <dl <dl <dl <dl <dl <dl <dl <dl <dl 0.1 <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl <dl 1.4 <dl 0.1 <dl Exposure index 82 P. Poirot et al. dl, detection limit.

9 Direct reading photoionization detector 83 Fig. 3. Concentration profile near the screen during remediation of the polluted site. (~10% of the total index for each), neglecting the possibility of using the I STEL and using only the I 3TWA comparison level would constitute, in our opinion, a very poor use of the limit value system. In these cases (methoxy propanol acetate and C 9 hydrocarbons), we preferred determining an arbitrary short-term value at 1.5 times the long-term value, which represents the most common ratio between the corresponding shortterm and long-term limit values for the compounds present in these workplace atmospheres. Whatever the answer to this debate of choice of reference limit value, the use of the PID can be helpful in the appreciation of exposure at the workplace. Considered in aggregate with exposure index values ranging from 0.05 to 0.83 (Table 5), the exposure of workers involved in the remediation of the polluted area examined in this paper could be considered as lower than the exposure limit values. However, examination of the concentration profiles recorded on charcoal tubes during work near the sifting machine revealed peaks 8 10 times higher than the mean value measured during the shift (Fig. 3). As these concentration profiles are representative of the exposure of workers intervening directly in the sifting machine, i.e. in the most polluted zone, it can be suspected that the short-term exposure of these workers may have exceeded the short-term limit values defined previously (I STEL or I 3TWA ). In reality, with personal exposure index values ranging from ~0.44 to 0.83 measured for the workers in charge of preventing blockage of the sifter, exposure peaks reaching levels 3.5 (0.44 8) to 8 ( ) times the I TWA are likely to have occurred. In terms of occupational health and safety, such information is important to ensure protection of the workers, either by installing a ventilation system on the sifter or by issuing the workers with a respirator when they have to intervene in this sifting machine. The limits of the method must nevertheless be kept in mind. For example, although trichloromethane cannot be detected by the PID as the energy of the lamp used (10.6 ev) is too low, its contribution to the exposure index is important. Insofar as it can be assumed that the concentration of trichloromethane varies in the same way as other detectable vapors, this non-detection is not a major drawback, provided that the relative concentrations of trichloromethane and detectable compounds are not too unbalanced in favor of trichloromethane. Another requirement for the relevance of establishing a correlation between index of exposure (in the case of a mixture of solvents) and PID response is that the relative concentrations of the different compounds of the mixture must not vary significantly during the measurement. From the results obtained in this study (in particular from the results of Table 3, where the petroleum spirits are themselves a mixture of various compounds with different vapor pressures), it looks likely that with determination coefficients close to 1, a good estimation of the actual exposure can be performed with the PID. The duration of the sampling (ranging from 3 to 8 h) probably minimized the influence of the differences in the vapor pressures of the various compounds, since the time was long enough to allow a stable atmosphere to settle and the activity of the workers did not change throughout the day. Whatever the case, there is an obvious contradiction in deducing an instantaneous value (PID measurement) from a time-weighted average one (charcoal tube measurement): given the measurement uncertainties (location of the sampling probe, analytical uncertainties, temperature differences from one day to the next, etc.), we consider that the information provided by the PID must mainly be considered as an indication which should be confirmed by other measurements (possibly done with the PID). The uncertainty on short-term measurements performed by other methods, such as charcoal tubes for example, is also great and they are not

10 84 P. Poirot et al. always so easy to use (determination of the pertinent moment to sample for example). CONCLUSION This paper shows the advantages and limitations of using a PID to assess the exposure of workers over short periods. Although the composition of the working atmosphere was constant enough (qualitatively and quantitatively) to ensure the relevance of the PID results for the two activities considered in this study, this property should be verified in all cases. However, because of the non-specificity of the response of the apparatus for mixtures, a number of measurements by a specific method (e.g. charcoal tubes with chromatographic analysis) should be carried out in parallel with the PID measurements to determine the relative concentration of the components of the mixture. This standardization should be completed at the beginning of the experiment and checked at regular intervals during the period when the PID is used. Under these conditions the PID is an excellent tool to show possible developments in the exposure of a worker. At the present stage of our research, it is not yet possible to determine the optimal conditions under which the technique developed in the present paper will show the best quality cost ratio. Given the rather good correlation coefficients obtained in various situations, it looks likely that four or five pairs of samples (PID + charcoal tubes) are enough provided that they are taken at concentrations ranging across the exposures encountered in the studied tasks and that the relative concentrations of the different compounds in the workplace atmosphere are relatively stable. Using the PID to compare short-term limit values is also a time consuming and difficult task, as the examples described in this paper show, but this difficulty is not directly related to the principle or the performance of the apparatus. Independently of the specificity problems mentioned above, this study has revealed contradictions between the TLV-STEL system and the excursion limit system, at least for pollutants such as solvents for which the TLV-STEL (when this exists) is generally established at 1.5 times the corresponding TLV-TWA. REFERENCES ACGIH. (2002) Threshold limit values for chemical substances and physical agents and biological exposure indices. Cincinnati, OH: American Conference of Governmental Industrial Hygienists. Castel B, Lefevre C, Lhuillier F, Delcourt, J. Sandino JP. (1998) Détermination du benzène par échantillonnage passif: essais interlaboratoires. Cah Notes Documentaires; 171: Coy JD, Bigelow PL, Buchan RM, Tessari JD, Parnell JO. (2000) Field evalaution of a portable photionization detector for assesing exposure to solvent mixtures. Am Ind Hyg Assoc J; 61: Drummond I. (1997) On-the-fly calibration of direct reading photoionization detectors. Am Ind Hyg Assoc J; 58: Gradiski D, Bonnet P, Raoult G, Magadur JL. (1978) Toxicité aiguë comparée par inhalation des principaux solvants aliphatiques chlorés. Arch Mal Prof; 39: Kumagai S, Matsunuga I. (1995) Changes in the distribution of short-term exposure concentration with different averaging times. Am Ind Hyg Assoc J; 56: Kumagai S, Matsunuga I. (1999) Within-shift variability of short-term exposure to organic solvent in indoor workplace. Am Ind Hyg Assoc J; 60: