At the four vehicle manufacturing plants, 28 personal and 8 area sample pairs were collected using actively pumped coconut-shell charcoal tubes (SKC 226-01, SKC Inc., Eighty Four, PA) and diffusive charcoal badges (SKC 575-001, SKC Inc.). The former represents an active sampling method (i.e., drawing air throughout the media using a pump) and the latter represents a passive sampling method (i.e., air intake by chemical diffusion). All workers sampled at the vehicle manufacturing plants handled the PCBTF-containing materials.

At the three paint manufacturing plants, 64 personal and 26 area sample pairs were collected. Participants were workers who handled PCBTF and workers who did not but were in close proximity to the workers handling PCBTF. The sample size and sampling time for each task are listed in Table III. The sampling times ranged from 15 to 407 min for the vehicle manufacturing plants and 70 to 535 min for the paint manufacturing plants. Two types of sampling pumps, Pocket Pump (SKC Inc.) and Gilian LFS-113 (Sensidyne, Clearwater, FL), were used at sampling flow rates between 20 and 200 ml/min for the active sampling method. The sampling flow rates were adjusted based on anticipated concentrations, previously collected from similar workplaces. Each pump was calibrated before and after sample collection with a DryCal DC-Lite device (BIOS International Corporation, Butler, NJ) to assure the difference between pre- and post-sampling flow rates was within ±5%. The position of passive and active samplers for the personal sampling method was randomized to minimize bias from workers’ handiness (i.e., not always on the left or right of worker’s collar). All field surveys were performed between 2010 and 2012.

All active and passive samples were analyzed with gas chromatography/flame ionization detector according to the NIOSH Manual of Analytical Methods (NMAM) 1026⁽⁷⁾ by the NIOSH contract laboratory. The NIOSH method has suggested a maximum of 25 ppm for a 10 L air sample with a working range between 0.024 and 9.15 ppm (0.178 to 67.8 mg/m³). Yost and Harper⁽⁸⁾ tested passive badges at various loadings in a standard atmosphere chamber in which the test concentrations of the standard atmosphere were confirmed by means of coconut charcoal tubes for time period up to 8 hr. Those loadings were 0.012 mg (0.01×CEL), 0.123 mg (0.1×CEL), 0.505 mg (0.5×CEL), 1.10 mg (1.0×CEL), and 2.09 mg (2.0×CEL). Yost and Harper⁽⁸⁾ showed charcoal tubes and passive badges to have a large capacity covering up to 2 times the CEL and the maximum concentration suggested by the NIOSH method.

The mass concentrations of passive badges were calculated using the average sampling rate of 11.8 ml/min.⁽⁸⁾ From each sampling site, 1–10 field blank samples were collected. In this study, sample results were not adjusted by field blank samples because almost all field blank samples (96% of 56 field blank samples) showed non-detectable masses. The limit of detection was 0.1 – 0.7 μg for both diffusive badge and charcoal tube. The limit of quantitation ranged between 0.5 – 2.5 μg for the diffusive charcoal badge and 0.5 – 3.4 μg for the charcoal tube. Three sample pairs showing at least one of each pair resulted in less than the limit of detection were excluded. None of the samples except for the three sample pairs showed less than the limit of quantitation.

Table III shows air sampling results using the active sampling method at the vehicle and paint manufacturing plants. Overall, the geometric mean of personal exposures (2.1 ppm) at the vehicle manufacturing plants was 3 times higher than that (0.7 ppm) at the paint manufacturing plants, while the geometric means of area exposures were similar for both types of plants. For the combined results of personal and area samples, the geometric mean concentration at the vehicle manufacturing plants was higher (about 2.5 times) than at the paint manufacturing plants. The comparison of log-transformed exposures between the vehicle and the paint manufacturing plants resulted in statistically significant difference (p-value < 0.0001). None of the samples exceeded even one-half the Occidental Chemical Corporation in-house CEL of 25 ppm.

For the vehicle manufacturing plants, all individual measurements per task were less than 10 ppm except for one measurement (12.2 ppm) from an interior refurbishment worker at Plant A (Figure 1). Although the amount of PCBTF handled during the mixing was larger than the other two tasks (interior refurbishment and painting tasks), the differences of geometric mean exposure concentrations between tasks were not substantial. Note that the mixing task could not be compared to other tasks because only one measurement was obtained. For the primer painting task, Plant C showed higher concentrations than Plant B shown in Figure 1. The geometric means of area and personal exposures for all tasks and areas were less than 3 ppm (Table III). The geometric standard deviation (GSD) per task was high (i.e., > geometric mean per task), indicating a wide spread of exposure measurements.

For the paint manufacturing plants, all collected exposure measurements were lower than 8 ppm. Note that personal exposure measurements of lab quality worker (n = 1) at Plant E, cleaner (n = 3) and batch adjuster (n = 1) at Plant F, and mixer (n = 3) and pilot worker (n = 3) at Plant G were merged into the category “miscellaneous” due to small sample sizes (≤ 3). Overall, although slight differences in geometric mean concentrations between tasks were observed, the magnitudes of differences were negligible. The variations of exposure measurements were high for most tasks (i.e., high GSD). It was also noted that the exposures measured at Plant E were higher than at the other two plants when personal and area exposures of pre-batch making and batch-making tasks were compared (Figure 2). Unlike the vehicle manufacturing plants, area geometric mean exposures were higher than the personal exposures when comparison was made by task. At Plant G, workers’ exposures who did not handle the PCBTF compound were about 20–55% lower than those exposures who handled PCBTF during full-shift.

Figure 3 presents side-by-side sample results (i.e., active versus passive) for the vehicle and paint manufacturing plants. Comparison by job task was not performed due to insufficient sample sizes per job task. The slope developed from linear regression method of log-transformed data was 0.929 with adjusted R² = 0.606 for the vehicle manufacturing plants and 1.012 with adjusted R² = 0.788 for the paint manufacturing plants. The slope was 0.990 with adjusted R² = 0.773 for the combined data. All p-values for testing H_o: slope (β) = 1 were > 0.05 indicating statistically no significant differences of concentrations between active and passive samples.

As shown in Table IV, the median of concentration ratios (passive/active) ranged from 0.9 to 1.2 for the personal and area samples. The geometric mean of concentration ratios was 1.0 regardless of the type of workplace and sampling method (e.g., personal versus area). For the paint manufacturing plants, more variation of the concentration ratios (passive/active) was observed from the area samples compared to the personal samples, while the area samples showed less variation than the personal samples for the vehicle manufacturing plants. A strong correlation was observed for all pairs of samples (all correlation coefficients ≥ 0.763), indicating statistical significance (all p-values < 0.0001 except for area samples at vehicle manufacturers (p-value = 0.0014)). No overall statistical differences were observed from the comparison of exposure measurements between the active and passive samples (all p-values > 0.05) regardless of worksite type. Additionally, the separation of personal and area exposures did not yield statistically different results (all p-values > 0.05).

The PCBTF exposure measurements (personal and area combined) were 0.1–12.2 ppm (geometric mean 2.0 ppm, GSD 2.9 ppm) for the vehicle manufacturing plants and 0.1–7.7 ppm (geometric mean 0.8 ppm, GSD 3.1 ppm) for the paint manufacturing plants. All individual measurements represented task-specific exposures but only some tasks extended close to a full 8-hour shift. Regardless of sampling time period, all measurements were considerably lower than a previous manufacturer’s in-house CEL of 25 ppm. The geometric mean exposure was only 8% of the CEL for the vehicle manufacturing plants and 3.2% of the CEL for the paint manufacturing plants. If the CEL is selected as an occupational exposure limit (OEL) value, it is very likely that none of tasks involving handling PCBTF measured in this study would result in the employee being overexposed. Similarly, a low exposure range was also observed in a survey performed at the Occidental Chemical Corporation. They monitored personal exposures from operators working in a cold cleaning machine containing PCBTF to remove grease and dirt from metal parts. The amount of PCBTF used in the machine was about 300 L and a solid metal lid covered the opened top. The observed personal exposures during a 3-week period ranged from 0.008 to 2.6 ppm (average 0.7 ppm, SD 0.9 ppm) for the operators working at the machine approximately 20 min per full-shift (Personal communication).⁽⁹⁾

The PCBTF inhalation exposures were determined by the combined effect of the PCBTF characteristics, amount handled during tasks, surface area of solvent exposed to air, handling method (e.g., spraying or manual paint brush application), and control strategy during the handling of the substance. For example, although large amounts of PCBTF were handled during the batch-making task (up to 3200 L), the partial opening of the batch for adding/transferring materials (i.e., enclosed system) did not result in high exposure to workers (geometric mean of personal exposures for all paint manufacturers = 0.8 ppm). Similarly, the geometric mean exposure of the pilot worker was 0.5 ppm when spraying and testing the PCBTF-containing material under a local-exhaust fume hood. If the pilot work was done without the presence of a local-exhaust hood, the PCBTF exposure would likely have been higher than 0.5 ppm. Although it would be difficult to determine which exposure determinants were more important than others without sensitivity analysis, one of the main reasons for the low exposure ranges is likely to be the relatively low vapor pressure of PCBTF (5.3 mmHg at 20°C) compared to other chemicals such as acetone (180 mmHg at 20°C).

Painter exposures measured at Plant C and exposures of pre-batch maker and batch maker at Plant E were higher than the same tasks of other plants. The different workplace conditions such as ventilation method, room sizes, and tools used to perform the tasks might cause such differences. However, another factor could be the sampling times in comparing to the time of specific use of PCBTF-containing materials. For example, the sampling time of painter at Plant B (85 and 111 min) was about 3 times longer than that at Plant C (32 and 37 min). Communication with the industrial hygienist who collected samples at Plant B confirmed that the painter exposures included the task handling materials with and without PCBTF. Inclusions of sampling time other than the painting task such as painting preparation prior to and cleaning after the painting task would result in lower concentration compared to the painting task handling the PCBTF only (geometric mean: 1.4 ppm at Plant B versus 9.5 ppm at Plant C). In the present study, the sampling time and task time are not necessarily equivalent due to the nature of task performance.

Personal and area measurements showed relative differences between the two worksite types. For the vehicle manufacturing plants, the geometric mean of personal exposure measurements was about 1.3 times higher than that of area measurements, while the geometric mean of area measurements for the paint manufacturing plants was 1.8 times higher than that of personal exposure measurements. This difference was very likely due to the distance of the sampler’s location from the source location. Although the geometric mean of area measurements was higher than that of personal exposure measurements for the paint manufacturing plants, this would not change the general conclusions of this study.

This case study was limited to the range of room temperature from 13°C to 27°C. Tasks at temperatures higher than the temperature range in this study might reveal different findings. Also, most samples were collected less than a full-shift (i.e., 8-hr TWA) and do not necessarily reflect 8-hr sampling.

No statistical differences between active and passive sample measurements were determined. For the vehicle manufacturing plants, the variation of the ratios of passive/active concentrations was higher for the personal exposure measurements than for the area measurements. On the other hand, for the paint manufacturing plants, a higher variation was observed from the ratios of area measurements than the ratios of personal exposure measurements. Higher variation between personal sampling methods may be the result of a worker’s movement and position of the sampler against the source (e.g., one sampler closer to the source than the other one), whereas the area sample pairs were fixed adjacent to each other. There is no clear explanation for the higher variation of the area sampling methods for the paint manufacturers. The findings of this study support the use of a passive sampler as an alternative to the active sampler to assess exposures to PCBTF in the vehicle and paint manufacturing plants.