1. Introduction

2. Reagents
   
   The following TICs were jointly selected by the Marines Corps and by OSHA.
   
   TICs

3. Air Samplers
   
   Air samplers included project and non-project samplers. Project samplers were those that the Marine Corps selected for testing. Non-project samplers were selected for testing by OSHA SLTC.
   
   Project Samplers

4. Apparatus 

Figure 1. Perkin Elmer TurboMatrix ATD.

Thermally-desorbed samples were analyzed using a Perkin Elmer (Norwalk, CT) TurboMatrix ATD (equipped with internal standard addition option) connected to the electronic-pressure controlled volatiles interface inlet of an Agilent Technologies (Wilmington, DE) 6890 Series GC system and an Agilent 5973 Network Mass Selective Detector (MSD). The ATD (automatic thermal desorber) and GC carrier gas was helium. Thermal- desorption tubes were loaded on the ATD carrousel-tray so that the end with the groove was upward. The groove identifies the front of the tube. ATD conditions: thermal-desorption tubes containing Tenax TA were desorbed at 275°C for ten min following a 1-min ambient purge. Tubes containing Chromosorb 106 were desorbed at 225°C, and Carbopack B at 350°C for the same times. The focusing trap was flash heated from -30°C to 300°C and maintained at the upper temperature for 2 min. The GC transfer line temperature was 225°C and the valve temperature was 225°C. The inlet split flow was 22 mL/min, the desorb flow 58 mL/min, the internal standard tube load flow 23 mL/min, and the internal standard loop flow 1.3 mL/min. GC conditions: a Restek (Bellefonte, PA) Rtx-5 capillary GC column (30-m × 0.25-mm i.d. × 0.25-µm df) was used for this work. The GC column was temperature programmed from 40°C (following a one-min hold) at 20°C/min to 230°C. The GC column was operated in the constant flow mode at 0.8 mL/min. The GC inlet temperature was 230°C, and the inlet split ratio was usually 67 to 1. MSD conditions: thermal auxiliary 2 temperature was 280°C, MS source was 230°C, MS quad was 150°C. The MSD was operated in the full scan mode from 24 to 350 AMU. Typically, a delay of 2 min was used to allow air and solvent used to prepare analytical standards to clear the MSD before the filament was energized. Electron multiplier (EM) voltage was automatically set by the MSD software via an "autotune" performed each week of operation. A photograph of the ATD is shown in Figure 1.

SKC CMS, SKC 575-002, and 3M 3520 samplers were analyzed (after desorption with carbon disulfide/DMF solution for one hour) using a Hewlett Packard (Avondale, PA) 5890 GC equipped with a Restek Rtx-5 capillary column (60-m × 0.32-mm i.d. × 1.5-µm df), an automatic liquid injector, and an FID. The GC column was temperature programmed from 40°C (following a one-min hold) at 6°C/min to 190°C. The FID was maintained at 250°C and the injector at 220°C. The GC column flow was 4.0 mL/min hydrogen. The inlet split ratio was 50 to 1. The FID gases were 35-mL/min hydrogen, 415-mL/min air, and 30-mL/min auxiliary nitrogen.

Test samples were collected from dynamically generated test atmospheres using an apparatus constructed from stainless steel. The apparatus consisted of either a large (cross-section area: 161 cm²) or a small (cross-section area: 40 cm²) exposure chamber designed to permit exposure of a large number of diffusive samplers to test atmospheres. The chambers were of different size to permit sample exposure at significantly different face velocities. A stainless-steel manifold was connected in-line and following the exposure chambers to permit collection of active samples.

Humid air for use with controlled-test atmospheres was generated using a Miller-Nelson Research (Monterey, CA) Model HCS 501 Flow-Temperature-Humidity Control System. This system was equipped with a 500-L/min mass flow controller. 

Relative humidity and temperature of the test atmospheres were monitored with an Omega (Stamford, CT) Digital Thermo-Hygrometer meter and probe. The probe was calibrated by the manufacturer. Pressure within the exposure chambers was monitored with an Omega meter and pressure transducer. The transducer was calibrated to read ambient barometric pressure before each run. Ambient barometric pressure was measured with a Princo (Southampton, PA) NOVA mercury barometer.

Dilution air flow rates (50-360 L/min) were measured with an Equimeter (Rockwell International, Pittsburgh, PA) No. 750 gas meter. Equimeter readings at several different flow rates were compared to those of a Singer (Philadelphia, PA) DTM 115 gas meter (which had been tested by the local natural gas distributor and found to be accurate) that was connected in series before the Equimeter. Both meters gave very similar readings.

The TIC mixture was metered into the system with an Isco (Lincoln, NE) 100 DM syringe pump equipped with a cooling/heating jacket and an insulation cover package. The pump was operated in the constant flow mode. The temperature of water in the cooling/heating jacket was maintained at 23°C with a Forma Scientific (Marietta, OH) Model 2006 CH/P Bath and Circulator.

TIC vapors were generated by pumping the liquid mixture into a heated glass tube where it evaporated into the dilution air stream. This vapor generator consisted of a 10-cm length of ¼-inch diameter glass tubing with a small hole in the side. The hole was just large enough for 1/16-inch diameter tubing to be inserted. The glass tubing was placed inside a ½-inch stainless steel Swagelok (Solon, OH) tee wrapped with heating tape that was heated when electricity was applied with a variable-voltage transformer. The 1/16-inch tubing entered the third port of the tee through an adaptor and was inserted about ⅛-inch (approximately in the center) into the glass tubing through the small hole. Solvent was pumped through the 1/16-inch tubing and into the glass tubing. The liquid flow rate was slow enough (0.3 to 20 µL/min) so that it did not accumulate in the heated evaporation tube. The entire dilution air stream passed through the tee and swept generated vapors into the apparatus.

Figure 2. Diffusive sampler exposure apparatus with small exposure chamber.

The following is a description of the arrangement of the apparatus that was placed in a walk-in hood. Liquid from which vapors were to be dynamically generated was pumped with a precision Isco syringe pump (an identical pump and a small solvent mixing tee was available when its use was desired) into a heated glass tube where it evaporated. The generated vapors were swept from the glass tube with dilution air. Stainless steel tubing (½-inch o.d.) connected with stainless steel Swagelok fittings was used to transfer the test atmosphere.  The dilution air was humidified (if desired) using a Miller-Nelson Flow-Temperature-Humidity controller. The vapor/dilution air mixture then passed into a 3×24-inch stainless steel mixing chamber. The test atmosphere next passed through ½-inch ball valves where it could be either diverted to waste, or directed into the exposure chamber. An additional ball valve allowed the chamber to be purged with room air. The transfer tubing diameter was increased from ½ inch to 1 inch at this point using a Swagelok adaptor attached to the chamber inlet. Tube and fitting diameter was increased to 1 inch after this fitting to help reduce any increased pressure to ambient. The 1-inch o.d. chamber inlets have small stainless steel deflectors to help insure that the test atmosphere completely fills the sampling chambers. Stainless steel screens were placed inside the chamber for the same purpose. This design should cause air flow through the chamber to be somewhat turbulent. Face velocities of the test atmospheres were calculated by dividing the volumetric flow of each atmosphere by the cross-sectional area available for air flow in each chamber. The cross-sectional area available for air flow was the cross-sectional area of each chamber reduced by the cross-sectional areas of the samplers. A photograph of the apparatus with the small sampling chamber is shown in Figure 2. Exposure chambers are used with removable doors (not shown in the photograph) to completely seal them when used with test atmospheres.

5. Experimental
   
   This laboratory research was performed partially based on techniques and tests described in OSHA methods development protocols^1,2. The TIC levels studied in this work were jointly selected by the Marine Corps and by OSHA. They are not associated with any military or OSHA exposure standard.
   
   Preparation of Samples

Figure 3. Ultra sampler.

SKC Ultra samplers were prepared for analysis by prying open the back of the sampler with a screwdriver to reveal the small end of the built-in aluminum funnel containing the sorbent. The small end of the funnel is sized to fit inside the thermal-desorption tube. The Tenax TA was then carefully transferred into the back of a clean, empty thermal-desorption tube using the funnel. The front of a thermal-desorption tube is the end with the groove and the back is the opposite end. A gauze screen was then carefully placed on top of the Tenax TA with the aid of a 3/16-inch glass rod and a small screwdriver. The tube was not completely filled with the sorbent and an empty space of approximately ½ inch remained in the end of the tube. A retaining spring was placed into this space, and then seated on top of the gauze screen and slightly below the end of the tube using the small end of the funnel to press the spring into position. A photograph of an SKC Ultra sampler is shown in Figure 3.

Figure 4. GoreSorber sampler.

GoreSorber samplers were prepared for analysis by transferring one of the PTFE-like cartridges into the end of a clean, empty thermal-desorption tube and then pushing it forward with a 3/16-inch glass rod. The other cartridge can then be placed in the same tube, or analyzed separately if desired. Both GoreSorber cartridges were usually analyzed in the same thermal-desorption tube in this work. A retaining spring is placed into the end of the tube, and then seated slightly below the end of the tube with the funnel from an SKC Ultra sampler. A photograph of a GoreSorber sampler and a thermal-desorption tube is shown in Figure 4. The tube endcaps shown are used to seal the samples for analysis with the ATD.

Figure 5. PE Tenax TA sampler.

PE samplers require no preparation for analysis. A photograph of one of these samplers is shown in Figure 5. A diffusive sampling cap with membrane is shown in the upper right section of the figure. A gauze screen and retaining spring is shown in the right center. The vial shows the amount of sorbent contained within the tube. PE samplers are thermally preconditioned by the vendor and can be reused numerous times after thermal reconditioning.

CMS active sampling tubes were prepared for analysis by placing each section of sorbent into separate 2-mL glass vials and then adding 1 mL of 99/1 carbon disulfide/DMF solution. These vials were allowed to stand one hour before analysis, and were shaken by hand several times during this time.

SKC 575-002 samplers were prepared for analysis by adding 2 mL of 99/1 carbon disulfide/DMF solution through one of the two sampler ports designed for that purpose. The sampler was then shaken on an SKC Sorbent Extractor (also designed for that purpose) for one hour before analysis.

3M 3520 OVMs were prepared for analysis by removing the two charcoal wafers, placing each wafer into a separate 4-mL glass vial, adding 2 mL of 99/1 carbon disulfide/DMF solution to each vial, and then desorbing the sample on a tube rotator for one hour before analysis. The wafers were removed from the sampler to eliminate any possibility of desorption solvent loss from in situ desorption.

Analytical Standards

Media standards for thermally-desorbed samples were prepared by spiking 5-µL aliquots of a series of diluted TIC mixtures onto thermal-desorption tubes containing the same medium used for sampling. Standards used for SKC Ultra and GoreSorber samplers were spiked on the back of the thermal-desorption tube. Clean, room air was drawn through these tubes at 50 mL/min for about 10 seconds immediately after they were spiked to help ensure that the spiked TICs resided on the sampling medium. This air entered the thermal-desorption tube at the end spiked. The source of the media desorption tubes used to prepare standards for SKC Ultra and GoreSorber samplers was SKC Ultra samplers that had been previously analyzed, and then reconditioned for 10 min at 275°C using the ATD tube-conditioning feature. These thermal-desorption tubes were reused approximately three or four times before recycling. Recycling thermal-desorption tubes was accomplished by removing the retaining spring, the gauze screen, and the sorbent from previously analyzed samples. The empty tubes were washed twice with methyl alcohol and then air-dried overnight before reuse.

Analytical standards for PE samplers containing Chromosorb 106 or Carbopack B were prepared using reconditioned PE samplers containing the same medium used for sampling. PE samplers were thermally reconditioned by heating them at the same temperature used for desorption for 10 min. Standards for PE Tenax TA samplers were prepared using reconditioned SKC Ultra sampling medium. Standards for all PE samplers were prepared by liquid spiking the TIC mixture on the front of the tube. 

Only limited work was performed to prepare standards using GoreSorber cartridges. Liquid spiking the cartridges would require a needle hole to be punctured in the PTFE-like cartridge wall and this would cause standards to be significantly different than samples. Some success was obtained by placing the cartridge in a glass vial, spiking the liquid standard into the vial, and allowing the vial to stand overnight at 40°C before transfer to the thermal-desorption tube and subsequent analysis. This technique was not fully developed. 

Analytical standards for solvent-desorbed samples were prepared by injecting microliter volumes of diluted TIC mixtures into the same volume of 99/1 carbon disulfide/DMF desorption solution used to desorb the samples.

MSD Calibration

GC/MSD calibration curves were prepared for each TIC (for each sample set) by first calibrating the MSD software in terms of mass per sample with a single standard. MSD response was plotted against actual TIC mass per standard using an electronic spreadsheet. Sample results were calculated using the calibration-curve equation obtained for each TIC. The calibration range, in most cases, was one-half to two times the expected sample concentration. The MSD is saturated when approximately 50 ng of a TIC reaches the detector and accurate quantitation becomes impossible after this point. A combination of GC injector split and/or ATD split should be employed to reduce mass reaching the MSD when samples contain sufficiently high TIC levels.  Alternatively, MSD EM voltage could manually be reduced to prevent premature saturation. All MSD results were calculated without internal standard correction.

GC/FID calibration curves were also prepared using an electronic spreadsheet for samples analyzed by solvent desorption. A Waters Associates Millennium Chromatography Manager data system was used to measure FID response. Dodecane was employed as an internal standard in the analysis of these samples.

MSD Detection Limits

Detection limits for the studied TICs were determined using the procedure described in OSHA methods development guidelines³. A series of standards was prepared with the highest giving a MSD response approximately ten times the MSD baseline noise. These standards were analyzed and MSD response plotted against mass injected on-column. The detection limit was defined as the mass equivalent to three times the SEE (standard error of estimate) of the regression line. SEE is the dispersion of data about a regression line, and it is mathematically similar to standard deviation for a data set.

Desorption Efficiency

Desorption efficiency studies can reveal adverse interactions between TICs and the sampling media. Desorption efficiencies for components of the TIC mixture were determined by liquid spiking the samplers with aliquots of the TIC mixture. No desorption efficiency work was performed for PE samplers because these spiked tubes would be identical to analytical standards. The spiking technique was routine except for GoreSorber samplers. The GoreSorber cartridges were removed from the outer pouch, one of the two cartridges was liquid spiked by penetrating the PTFE-like shell with a syringe needle, and then both were placed in a sealed glass vial. All spiked samples were allowed to equilibrate overnight before analysis. SLTC experience has shown that desorption efficiency results from wet sampling media can occasionally be different than from dry media. This is unlikely for Tenax TA because that sorbent is known to have low affinity for water. Desorption from wet SKC Ultra samplers was, however, tested to a limited degree. Wet media were prepared by sampling a clean atmosphere containing 80% relative humidity for four hours with dry samplers. Wet media were liquid spiked with aliquots of the TIC mixture after exposure to humid air. Dry samplers were as received from the vendors.

Desorption efficiency studies were performed using both wet and dry samplers for CMS sampling tubes, SKC 575-002 samplers, and 3M 3520 OVMs.

Sampling Rate

Sampling rates for the components of the TIC mixture were determined for project, non-project, and control samplers. Sampling rates were determined at ambient temperature and pressure, but all sampling rates presented in this work are expressed at 760 mmHg and 25°C. Three of each type of sampler was exposed to controlled test atmospheres for increasing time periods in the small exposure chamber for these experiments. The exposure times ranged from two to as long as 16 hours. The relative humidity of the test atmospheres was approximately 80% at ambient temperature. The face velocity of test atmospheres through the small exposure chamber was approximately 0.4 m/s. The concentration of each TIC in the test atmospheres was approximately 4 mg/m³. Sampler orientation was parallel to the flow direction of the test atmosphere, except for PE samplers which was perpendicular. This orientation of PE samplers was necessitated because of space limitation within the small chamber.

Long-Term Sampling Capacity

A long-term sampling capacity experiment was performed with SKC Ultra, GoreSorber, and PE Tenax TA samplers. Twenty-one of each sampler were placed in the large exposure chamber, and three of each sampler were removed at approximately eight-hour intervals over a total exposure time of 54 hours. The relative humidity of the test atmospheres was approximately 80% at ambient temperature. The face velocity of the test atmosphere through the large exposure chamber was approximately 0.1 m/s. The concentration of each TIC in the test atmosphere was approximately 0.4 mg/m³. Sampler orientation was parallel to the flow direction of the test atmosphere, except for PE Tenax TA samplers which was perpendicular.

Reverse Diffusion

A reverse diffusion experiment was performed by exposing SKC Ultra, GoreSorber, and PE Tenax TA samplers (six of each) to a test atmosphere containing approximately 4 mg/m³ of each TIC for four hours. The relative humidity of the test atmosphere was approximately 80% at ambient temperature and the face velocity was about 0.4 m/s. Following the four-hour exposure time, the samplers were removed and the chamber was flushed with clean, humid air. Then, three of each sampler was replaced in the chamber and exposure continued for an additional four hours with clean, humid air. The analytical results for samples exposed to the test atmosphere and then additionally to clean air were compared to results for samples exposed to only the test atmosphere.

Storage Stability

SKC Ultra and GoreSorber storage stability samples were simultaneously prepared by sampling a test atmosphere with 33 of each sampler. Thirty-three PE Tenax TA samplers were used to sample another, but similar, test atmosphere. The concentration of the test atmospheres was approximately 4 mg/m³ for each TIC, and the exposure times were four hours. The relative humidity of the test atmospheres was approximately 80% at ambient temperature and the face velocity was about 0.4 m/s. Three of each type of sampler was analyzed on the day they were generated, and the remaining 30 of each type were split into two sets of 15 each. One set of each type of sampler was stored in a laboratory oven maintained at 40 °C to simulate hot ambient environments, and the other set in a refrigerator at -4°C to simulate cold environments, and to test if refrigeration could stabilize any TIC that might degrade at the higher temperature. The samplers were all sealed using original packaging materials as received from vendors. Three of each set of each type of sampler were removed from storage and analyzed at approximately three-day intervals.

Factor Test

A 16-run factor test was performed using a modified version of the test discussed in the NIOSH protocol for evaluation of diffusive samplers⁴. This test is performed primarily to observe the degree with which sampling rates change under the environmental conditions of the test. These conditions are significantly altered following a prescribed regimen for each run of the test. The pooled RSD of the sampling rates is called sampling rate variation⁵ (SRV) by OSHA SLTC, and it has been established as a measure of sampling error for diffusive samplers. SRV is analogous to the often-cited ±5% sampling pump error used to estimate sampling error for active samplers.

Six factors were identified in the NIOSH protocol as having the potential to affect sampler performance. These factors are analyte concentration, exposure time, face velocity, relative humidity, interferant, and sampler orientation. Sixty-four experimental runs (26) would be required to fully evaluate combinations of each factor at two levels. NIOSH recognized that this would be an excessive number of tests, and has devised a 16-run fraction of the full factorial that is capable of revealing any of these factors having a significant effect, free of two-factor interactions, on sampler performance. Some two and three-factor interactions can also be screened by this design. The test is based on the comparison of each factor effect to experimental error so that the significance of that effect can be determined.

Experimental design and conditions are shown in Table 1. Interferant was provided by the components of the TIC mixture, for example, if EtBz was being examined then Bz, TCA, TMB, LIM, UND, and DDVP were the interferants, and the levels were declared either high or low. Each of the samplers has a slightly different cross-sectional area and this difference will slightly affect the face velocity to which that sampler is exposed. The following calculation is applicable only to high face velocity (1.5 to 2.1 m/s) experiments performed in the small exposure chamber. Experiments at low face velocity were performed in the large chamber, and the differences in sampler cross-sectional area are small compared to chamber cross-sectional area. Face velocity shown in Table 1 is for SKC Ultra samplers. Multiply the value by 0.90 to calculate parallel orientation face velocity for GoreSorber samplers, and by 1.05 for perpendicular orientation. Exposure chamber size limitation necessitated that only perpendicular orientation be used for PE Tenax TA samplers. PE sampler orientation was either upward so that the test atmosphere first encountered the diffusion cap, or the sampler was inverted so that the test atmosphere first encountered the back endcap. Upward orientation was designated perpendicular orientation, and inverted designated parallel. Multiply by 0.91 to calculate parallel orientation face velocity for PE Tenax TA samplers, and by 0.86 for perpendicular orientation.

Precision and Accuracy

The NIOSH acceptability criterion (published in their diffusive sampler evaluation protocol⁶) for accuracy is that the method provide results within ±25% of the reference value at the 95% confidence level over the range 0.5 to 2 times the target level of the method. NIOSH specified which data to test against the criterion and those specifications were followed as closely as possible with data obtained in this work. Relative standard deviations (RSD) obtained from the analysis of three replicate samples of each type exposed at 2, 4, and 7 mg/m³ for each TIC were pooled after first examining them for homogeneity at the 95% confidence level with the Cochran Test. RSDs from samples exposed at 0.4 mg/m³ were pooled separately to evaluate precision and accuracy at lower levels.

OSHA methods acceptance criteria require that candidate sampling and analytical methods provide sample results at least 75% (±25%) of the method target level at the 95% confidence level, and that uncorrectable bias be less than 10%. OSHA evaluates precision and bias using data from the storage stability test.

Precision and accuracy data were evaluated for SKC Ultra, GoreSorber, and PE Tenax TA samplers.

Packaging Integrity

An experiment was performed to determine if manufacturer’s packaging is sufficient to prevent contamination of unused samplers placed in contaminated environments. Two each, SKC Ultra, GoreSorber, and PE Tenax TA samplers, unopened and as received from the vendors, were placed in the exposure chamber and allowed to remain undisturbed for 130 days. The samplers were exposed to test atmospheres for approximately 120 hours during the 130 days. The TIC concentrations of test atmospheres varied, but the average was about 4 mg/m³ for each TIC.

Concentrations of Test Atmospheres

Previous OSHA SLTC work⁷ has shown that use of active samplers as an independent means to determine concentrations of test atmospheres gave more consistent experimental results than did calculation of concentrations from generation parameters. This is because of normal day-to-day variations in the operation of test equipment. 

Theoretical concentrations of test atmospheres were calculated from the flow rate at which the liquid TIC mixture was pumped into the vapor generator, the TIC concentrations of the liquid mixture, and the dilution air flow rate. Actual concentrations of test atmospheres were determined from the analytical results of CMS sampling tubes. The CMS tube sampling rate was 0.01 L/min, and four samples were taken simultaneously with diffusive samples for every run. Average CMS results for the 16-run factor test, for example, were 100% of theoretical with a pooled RSD of 15%. Comparison of sampling rates for SKC 575-002 and 3M 3520 control
samplers determined in this work with sampling rates published in the literature allows confirmation of the validity of this practice. A comparison of sampling rates is presented in Table 21.

6. Results and Discussion
   
   MSD Calibration
   
   MSD calibration data provides a means to estimate the contribution of the analytical procedure to overall method imprecision. Calibration was performed with MSD response at the following selected ions: Bz 78.1, EtBz 91.1, TCA 83.0, TMB 105.1, LIM 68.1, UND 57.1, DDVP 109.0, tol-d₈ 98.1 m/z.

The concentration of standards used for MSD calibration is shown in Table 2. MSD response data obtained over this mass range is shown in Table 3. The
combined GC and ATD split was 135 to 1.

The ATD automatic internal standard addition feature was activated for this calibration work, but MSD response data were not corrected for the internal standard. Internal standard data are presented for precision information only. No results obtained by thermal desorption presented in this report were corrected for internal standard. One reason for not using the internal standard option was that the uptake of tol-d₈ was about 115% for SKC Ultra samplers, about 60% for GoreSorber samplers, and about 105% for PE Tenax TA samplers, when compared to the uptake of standards.  Use of an internal standard would obviously bias analytical results. Analytical standards would have to be prepared using blank samplers (including the PTFE-like shell for GoreSorber cartridges) in order to successfully employ the internal standard addition feature. Another reason was that the internal standard addition mechanism did not function properly until the work was fairly well advanced. Tol-d₈ was used in gas form after dilution by the vendor to 0.1% by volume in nitrogen gas. The ATD employs a 0.5-mL loop for internal standard addition, therefore, theoretically 2 µg of tol-d₈ was injected with each standard.

The individual RSDs in Table 3 were tested for homogeneity with the Cochran Test and, with the exception of tol-d₈ data for std no. 4, found to be homogeneous. The homogeneous RSDs were pooled and the result is shown in Table 3. These pooled RSDs are similar to those obtained for FID calibration data which are typically 1 to 2%. Instrument calibration was not a significant source of analytical error.

An example calibration curve for TMB is shown in Figure 6. SEE (standard error of estimate) shown in Figure 6 is equivalent to 87 ng over the calibrated range. Calibration is not performed using total-ion MSD response; it is accomplished using MSD software for optimum response and selectivity. A total-ion chromatogram for std no. 4 is shown in Figure 7.

MSD Detection Limits

MSD detection limits are shown in Table 4. These amounts are mass injected on-column and not mass per sample. The detection limits are based on precision of the analysis, and on spectral quality of the resultant mass spectra.

An example of plotted data used to determine MSD detection limit for TMB is shown in Figure 8. A total-ion chromatogram from the analysis of a standard spiked at approximately the MSD detection limits is shown in Figure 9.

Media standards for project samplers were prepared using the sampling medium from previously analyzed Ultra samplers that have been thermally reconditioned. Analysis of reconditioned Ultra samplers gives clean chromatograms. SKC conditions the Tenax TA to be packaged within Ultra samplers by solvent extraction and subsequent heating to remove the solvent. The sorbent is not thermally conditioned at a sufficiently high temperature, and the analysis of a sampler blank reveals many artifact peaks. An SKC representative has stated that they intend to begin thermally conditioning the sorbent to reduce artifacts. Analysis of blank GoreSorber samplers also shows a significant number of artifacts⁸. Total-ion chromatograms from the analysis of SKC Ultra and GoreSorber blank samplers are shown in Figures 10 and 11. Chemicals identified on the SKC Ultra blank include trioxane,1-methyl-2-propyl acetate, and 2-ethyl-1-hexanol. Present on the GoreSorber blank are 2-ethyl-1-hexanol, 1-decene, and butylated hydroxytoluene. Attaining the detection limits in Table 4 for blank and field samples could be challenging because of co-eluting artifacts and other species. For example, satisfactory MSD spectra for EtBz and for TCA were not obtained for blank SKC Ultra samplers spiked at approximately the detection limits. Almost 2 ng of each TIC was required to obtain good spectra, and this required spectral subtraction of the interfering species. Good spectra were obtained for the other TIC components spiked on a blank SKC Ultra sampler at the stated detection limits. This data emphasizes the need for samplers with low artifact levels.

Table 5 Desorption Efficiency From SKC Ultra Samplers (%)    Bz EtBz TCA TMB LIM UND DDVP dry 102.5 100.4 97.0 100.6 99.7 98.4 99.2 wet 92.3 93.1 94.1 94.3 95.6 95.7 99.7

Desorption Efficiency

Average desorption efficiencies for SKC Ultra samplers are shown in Table 5. The mass range studied was 13 to 32 µg for each TIC per sample. Wet samplers were prepared by exposing them to clean, humid air for four
hours prior to spiking. Some results from wet samplers were lower than from dry samplers for unknown reasons. These results were not used to correct SKC
Ultra sampler results presented in this report because it was assumed that desorption was 100%. This assumption was supported by the fact that a second desorption of several SKC Ultra samples showed no TICs present in significant amounts.

Average desorption efficiencies for GoreSorber samplers are shown in Table 6. The mass range studied was 13 to 30 µg for each TIC per sample. A study using wet samplers was not performed. These results were not used to correct GoreSorber results presented in this report because it was assumed that desorption was 100%. This assumption was again supported by the fact that a second desorption of several GoreSorber samples showed no TICs to be present in significant amounts. The low desorption efficiency obtained for Bz was probably caused by the relatively low affinity of Tenax TA for Bz, and its subsequent loss through the hole punctured in the PTFE-like shell of the cartridge by liquid spiking the sample.

Average desorption efficiencies for the front section (150 mg) of SKC Anasorb CMS sampling tubes are shown in Table 7. The mass range studied was 10 to 165 µg of each TIC per sample. Wet samplers were prepared by drawing clean, humid air through the samplers at 0.1 L/min for four hours prior to spiking. All CMS results presented in this report were corrected using the appropriate desorption efficiency. Results for DDVP desorbed from wet samplers were low and would necessitate use of another solvent or desorption technique in routine analytical work because a minimum recovery of 75% is usually required for NIOSH and OSHA methods. The reason for low DDVP wet results is unknown but could be due to hydrolysis. The low desorption was deemed adequate for this work because samples were analyzed immediately after generation. Results for TCA from wet CMS were lower than from dry CMS, but they were greater than 75% and, therefore, adequate for conditional use.

Average desorption efficiencies for SKC 575-002 samplers are shown in Table 8. The mass studied was 13 µg of each TIC per sample. Wet samplers were prepared by exposing them to clean, humid air for four hours prior to spiking. All SKC 575-002 samplers in this study were exposed to humid air and, therefore, analytical results were corrected using only the wet desorption efficiencies. A chromatographic interference prevented the analysis of DDVP. This co-eluting interference was identified as cyclohexyl isothiocyanate by GC/MS. The low wet desorption efficiency obtained for TCA would necessitate use of another solvent or desorption technique in routine analytical work, however, it was also deemed adequate for this work because these samples were analyzed immediately after generation. 

Average desorption efficiencies for 3M 3520 OVM charcoal wafers are shown in Table 9. The mass studied was 26 µg of each TIC per sample. Wet samplers were prepared by exposing them to clean, humid air for four hours prior to spiking. All 3M 3520 OVM samples in this study were exposed to humid air and, therefore, analytical results were corrected using only the wet desorption efficiencies. The low wet desorption efficiencies for TCA and DDVP would necessitate use of another solvent or desorption technique in routine analytical work, however, they were again deemed adequate for this work because these samples were analyzed immediately after generation.

Sampling Rate

Sampling rates were determined by exposing samplers to test atmospheres containing approximately 4 mg/m³ of each TIC (about 1 ppm) for increasing time intervals. 

Sampling rates were calculated with the following equation and then converted to mL/min:

ambient sampling rate = average mass collected/(concn of test atm × sampling time)

Mass was corrected for desorption efficiency only for SKC 575-002 and 3M 3520 OVM samplers.

Experimental sampling rates were determined at ambient temperature (T_amb) and barometric pressure (P_amb), and were converted to their equivalent at 760 mmHg and 298K with the following equation:

sampling rate_{760 mmHg, 298K} = sampling rate_amb(298/T_amb)^3/2(P_amb/760)

Sampling rates for the components of the TIC mixture were determined for SKC Ultra and GoreSorber project samplers. Sampling rates were also determined for SKC Ultra RSR, PE Tenax TA, PE Chromosorb 106, and PE Carbopack B non-project samplers. They were also determined for the 3M 3520 and SKC 575-002 samplers that were used as controls. DDVP was not thermally desorbed from PE Carbopack B samplers at its recommended maximum desorption temperature. A chromatographic interference that eluted at the same time as DDVP prevented its determination in SKC 575-002 samplers. Relative standard deviations (RSD) were calculated for the three samplers of each type that were exposed for each time interval. RSD was excessive in some cases, but all data were retained because of the small sample size and also to preserve the integrity of precision results. Part of this work was to test and evaluate the analytical method.

Sampling rates for SKC Ultra and SKC 575-002 samplers should be similar providing the sampling medium is adequate. This is because the SKC Ultra sampler design is based on the SKC 575 Series of samplers. The main difference is that the SKC Ultra sampler contains Tenax TA and the SKC 575-002 sampler contains carbon-based Anasorb 747. The greatest disagreements are for Bz and TCA, and to a lesser degree, EtBz. Bz and EtBz differences are due to low affinity of Tenax TA for Bz and EtBz as evidenced by the reverse diffusion experiment results presented in Table 25. The reason for the TCA difference is unknown, but the presence of water may be a contributing factor. Comparison of sampling rates was excellent for TMB, LIM, and UND. These results indicate that Tenax TA is not suitable for more volatile TICs.

Sampling rates for PE Tenax TA, PE Chromosorb 106, and PE Carbopack B samplers should also be similar if the sampling medium is adequate. The only difference in these samplers is the sampling medium. Except for Bz, PE Tenax TA and PE Carbopack B sampling rates were similar. The difference for Bz is due to the low affinity of Tenax TA for Bz. Sampling rates for PE Chromosorb 106 samplers were higher than were the other two for unknown reasons.

Sampling rates for some TICs and some of the samplers tested in this report are published in the literature⁹ and are presented in Table 20 for comparison with those obtained in this current work (CW). Literature (Lit) values cited in Table 20 for SKC 575-001/2 samplers were determined for the SKC 575-001 sampler which is similar to the SKC 575-002 sampler tested in this study. The difference is that the SKC 575-001 sampler contains 350 mg of coconut-shell charcoal and the SKC 575-002 sampler contains 500 mg of Anasorb 747. The 3M 3500 OVM is similar to the 3M 3520 OVM, except the 3M 3520 OVM contains two charcoal wafers. Lit sampling rates cited in Table 20 for PE samplers were obtained using diffusive sampling caps without membranes. It is interesting to note that Lit Chromosorb (Chrom) 106 sampling rates are higher than those for Tenax TA which generally supports results obtained in this work. Blank spaces in Table 20 mean that no Lit data were found. Except for TCA, the comparison of sampling rates is quite good.

OSHA SLTC has previously determined Bz sampling rates for SKC 575-002 and 3M 3520 samplers¹⁰. They were 17.1 and 34.3 mL/min, respectively. OSHA SLTC has also determined the EtBz sampling rate for SKC 575-002 samplers to be 13.8 mL/min¹¹.

Lit values from Table 20 were divided by CW values for the Lit comparison, and values determined from previous work were divided by CW values for the OSHA comparison. Both comparisons are presented in terms of percent in Table 21. No comparison data were available for UND and DDVP. Except for TCA, the agreement between Lit, OSHA , and CW sampling rates for TICs that were tested is good. Low desorption efficiency from wet carbon-based sampling media obtained in this work is a possible reason for the TCA difference. Many laboratories do not investigate desorption from wet media. Low desorption would cause the calculated sampling rate to increase because analytically determined mass would also increase.

The overall good comparison of sampling rates for control samplers with sampling rates available in the literature values support the techniques used to in this work.

Long-Term Sampling Capacity

The possibility of long-term sampling was investigated by exposing 21 each SKC Ultra, GoreSorber, and PE Tenax TA samplers to a test atmosphere containing approximately 0.4 mg/m³ of each TIC (about 0.1 ppm) for up to 54 hours. Three of each type of sampler was removed for analysis at approximately eight-hour intervals. Results in terms of sampling rates for the long-term sampling experiment are shown in Tables 22-24.

Storage Stability

Sampling rates used to calculate storage stability sample results were taken from Tables 18-19 and were converted to their equivalents at sampling site temperature and pressure. Results from the three samples analyzed on the designated days are presented as percent of the concentrations of the test atmosphere. GoreSorber sampler results for Bz on Day 0 and subsequent samples were very low, less than 25% of the expected mass, and are not presented because no meaningful recovery data could be calculated. Storage stability data were examined statistically, and those results are presented following graphical representation of the storage data.

Table 32 Statistical Data for GoreSorber Sampler Ambient Temperature Storage Stability Test TIC equation of line SEER(±%) SEEp (±%) 95% conf int (%) EtBz Y = -0.785X + 95.1 4.83 18.34 35.9 TCA Y = -0.875X + 98.5 6.38 18.81 36.9 TMB Y = -0.803X + 100.3 5.88 18.64 36.5 LIM Y = -0.794X + 96.4 6.66 18.90 37.0 UND Y = -0.640X + 102.6 7.13 19.07 37.4 DDVP Y = -0.600X + 119.4 9.97 20.31 39.8 Table 33 Statistical Data for GoreSorber Sampler Refrigerated Temperature Storage Stability Test TIC equation of line SEER(±%) SEEp (±%) 95% conf int (%) EtBz Y = -0.403X + 95.7 4.54 18.26 35.8 TCA Y = -0.326X + 97.7 6.38 18.81 36.9 TMB Y = -0.373X + 99.8 5.31 18.47 36.2 LIM Y = -0.155X + 96.2 7.23 19.11 37.5 UND Y = -0.132X + 100.6 7.87 19.36 37.9 DDVP Y = -0.0452X + 113.4 8.96 19.83 38.9 Table 34 Ambient Temperature Storage Stability for PE Tenax TA Samplers day Bz EtBz TCA TMB 0 110.8 112.4 110.8 97.0 102.1 98.7 96.7 91.4 89.7 88.1 103.7 105.7 5 107.5 102.5 115.7 108.7 112.1 112.1 123.1 117.8 119.6 123.3 125.3 133.1 9 109.1 94.2 110.8 105.4 103.7 107.1 102.0 98.5 100.2 117.4 105.7 107.6 14 107.5 107.5 90.9 117.1 105.4 88.7 112.5 110.8 93.2 105.7 113.5 86.1 21 99.2 110.8 104.1 100.4 107.1 93.7 103.8 110.8 89.7 113.5 117.4 97.9 26 94.2 104.1 104.1 97.0 107.1 100.4 110.8 112.5 107.3 105.7 113.5 107.6 Table 34 (Continued) Ambient Temperature Storage Stability for PE Tenax TA Samplers day LIM UND DDVP 0 93.7 95.8 112.9 105.6 110.3 103.3 92.0 99.4 84.5 5 117.1 121.4 123.5 110.3 115.0 131.4 67.1 106.9 116.8 9 112.9 119.2 119.2 105.6 103.3 105.6 87.0 96.9 116.8 14 129.9 100.1 91.6 117.4 124.4 105.6 89.5 106.9 96.9 21 108.6 115.0 91.6 103.3 119.7 103.3 129.2 94.4 111.8 26 93.7 110.7 102.2 119.7 122.1 110.3 101.9 87.0 54.7 Table 35 Refrigerated Temperature Storage Stability for PE Tenax TA Samplers day Bz EtBz TCA TMB 0 110.8 112.4 110.8 97.0 102.1 98.7 96.7 91.4 89.7 88.1 103.7 105.7 5 114.1 107.5 109.1 108.7 110.4 113.8 116.1 114.3 117.8 119.4 113.5 121.3 9 100.8 112.4 115.7 107.1 115.4 112.1 100.2 103.8 116.1 113.5 131.1 119.4 14 124.0 119.0 104.1 110.4 118.8 102.1 116.1 112.5 98.5 117.4 123.3 111.6 21 119.0 107.5 104.1 110.4 105.4 110.4 107.3 110.8 105.5 105.7 103.7 117.4 26 104.1 99.2 92.6 107.1 100.4 105.4 112.5 98.5 109.0 111.6 103.7 113.5 Table 35 (Continued) Refrigerated Temperature Storage Stability for PE Tenax TA Samplers day LIM UND DDVP 0 93.7 95.8 112.9 105.6 110.3 103.3 92.0 99.4 84.5 5 104.3 110.7 123.5 124.4 110.3 117.4 109.4 109.4 101.9 9 117.1 136.3 136.3 112.7 115.0 115.0 124.3 114.3 99.4 14 115.0 123.5 112.9 110.3 126.8 103.3 96.9 111.8 104.4 21 110.7 106.5 108.6 112.7 100.9 108.0 96.9 89.5 119.3 26 115.0 102.2 112.9 122.1 103.3 105.6 134.2 141.7 64.6 Table 36 Statistical Data for PE Tenax TA Samplers Ambient Temperature Storage Stability Test TIC equation of line SEER(±%) SEEp (±%) 95% conf int (%) Bz Y = -0.350X + 109.7 6.30 13.60 26.7 EtBz Y = -0.130X + 105.2 7.18 14.03 27.5 TCA Y = 0.205X + 102.5 10.61 16.06 31.5 TMB Y = -0.0515X + 110.1 12.26 17.19 33.7 LIM Y = -0.293X + 112.5 12.23 17.17 33.7 UND Y = 0.195X + 109.6 8.69 14.86 29.1 DDVP Y = -0.0711 X + 97.5 18.34 21.94 43.0 Table 37 Statistical Data for PE Tenax TA Sampler Refrigerated Temperature Storage Stability Test TIC equation of line SEER(±%) SEEp (±%) 95% conf int (%) Bz Y = -0.333X + 113.5 7.39 14.14 27.7 EtBz Y = 0.0716X + 106.6 6.04 13.48 26.4 TCA Y = 0.247X + 103.4 8.86 14.96 29.3 TMB Y = 0.0663X + 111.6 10.01 15.67 30.7 LIM Y = 0.0485X + 112.6 11.88 16.92 33.2 UND Y = -0. 0813X + 112.5 7.77 14.34 28.1 DDVP Y = 0.435X + 99.8 18.27 21.89 42.9