Eleven firefighters serving in the Cincinnati, Ohio metropolitan area, including ten males and one female, were recruited to participate as subjects. Prior to recruitment, the investigators held a series of meetings with the Cincinnati District Fire Chief to develop an appropriate recruitment strategy. Recruitment flyers were sent to the area fire stations. Included in the flyer was a description of the study, the tasks to be performed by the subjects and a list of the minimum requirements to qualify in the study, such as having recently not sustained any bodily injury, being able to provide and operate their turnout gear and being clean shaven prior to the test (no beard, mustache, etc.). The study received an approval from the University of Cincinnati Institutional Review Board.

Small, medium and large sized elastomeric half-mask respirators (Model: 6000 Series, 3M, Minneapolis, MN, USA) were offered to the subjects. Individuals were requested to select the same respirator size they would use on the job. While being worn by the subject, the respirator was visually examined to verify that it was not only suitable to the subjects’ facial dimensions but also provided a comfortable and snug fit. The respirator chosen for this study was of the same model that was tested in the recently published study(¹); this model is commonly used by firefighting personnel as well as by workers in other occupational environments such as foundry operators, fiber glass gunners and laminators, and shipyard workers.(^8-10) The respirator was equipped by two new P100 pancake-shaped filters (Model: 2091, 3M), which were attached onto the half-mask prior to each testing session. Before and after each test, respirators were cleaned with disposable Kimwipes dampened with isopropyl alcohol. Additionally, respirators were thoroughly inspected for any damages that it could have incurred during previous subject testing. While the subjects were experienced respirator users, for consistency, each firefighter was shown how to don the respirator and adjust the straps to ensure a suitable fit. After adjustments were made, a subject performed a positive pressure user seal check. Any faceseal leakages that were revealed during the positive pressure user seal check were remediated by adjusting the straps. Subject testing was initiated once the researchers verified that the mask was well-adjusted so that no faceseal leakage was identified.

During each test, a subject was asked to wear their turnout gear, which included boots, protective pants, and a jacket. Oxygen tanks and hard hats were not included as they would potentially interfere with Tygon sampling tubes during testing. Subjects were evaluated to ensure that they were cleanly shaven. Additionally, each subjects’ face was assessed to ensure no signs of moisture on their facial surface existed. While wearing the half-mask respirator, a subject entered a 24.3-m³ exposure test chamber(¹¹) and performed a series of five different activities representative of those executed during an emergency response situation such as the fire overhaul. These activities included: (1) stepping up and down a stepladder, (2) crawling back and forth, (3) squatting, (4) bending and touching their toes and (5) picking up and moving an object. Each activity was performed for 2 min. As a subject was performing each activity, he/she was asked to occasionally turn their head from side to side and nod their head up and down. After the final activity, subjects were asked to remain stationary and breathe normally for 1.5 min. Thus, the total subject testing time was 11.5 min.

Wood combustion aerosol was generated in the test chamber for approximately 15 min, following by a waiting period of 10 min, which allowed for the stabilization of the challenge aerosol in the chamber prior to subject testing.

The total and size-resolved aerosol concentrations outside (C_out) and inside (C_in) of the tested respirator were measured over the entire 11.5-min testing period. The aerosol measurement was performed using a prototype PAMS, an instrument developed at the National Institute for Occupational Safety and Health (NIOSH).(¹²) It is a battery-operated scanning mobility spectrometer, which neutralizes the sampled particles to a steady-state charging status using a dual-corona bipolar charger(¹³), then separates the particles according to their electrical mobility size in a differential mobility analyzer (DMA), followed by optical detection and counting with a condensation nuclei counter. The instrument is capable of real-time measurement of aerosol size distribution in the range of approximately 10 to 863 nm.(¹⁴) Given the typical size range of combustion particles, we focused primarily on the sizes between 20 and 200 nm. In addition to the PAMS, a P-Trak (TSI Inc., St. Paul, MN, USA) was used in parallel to monitor the ambient particle concentration in real time.

Figure 1 presents the experimental setup. The outside (ambient) aerosol concentration was a subject for natural decay; therefore, the PAMS and P-Trak readings were acquired continuously during the test and an integrated value of C_out was determined. Similarly, an integrated value of C_in was obtained. The 2-min testing time was sufficient to complete three full scanning cycles for the PAMS. Each of these three cycles comprised of the eight channels making up the particle size range of interest. Upon completion of the given subject test, the SWPF was calculated as C_out/C_in. – size selectively and size integrated (based on the total concentration) for the PAMS data and in terms of total concentration for the P-Trak.

Figure 2 presents the size-resolved concentrations (C_out) of combustion-generated aerosol particles measured in the chamber during the 11 tests involving Subjects 1 through 11. The figure reveals a consistent pattern; the distributions covered primarily a size range of 20–200 nm expected for the tested combustion particle sizes.(¹⁵) In the quoted study 95% of particles generated by wood combustion under similar conditions were found to be within the referred size range. As seen from Figure 2, most of the curves reached their peaks at sizes slightly above 100 nm. The test with Subject 7 produced a different ambient particle distribution with the peak occurring at a higher level than in other tests. This may be associated with a different relative humidity in the chamber on the day of that test, which could cause more intense combustion of the material and affect the post-combustion aerosol decay.

The particle size distributions measured inside the respirator were found to have consistent patterns (data not shown); the “inside” distributions covered about the same size range as the ambient aerosol measured outside the respirator. Among the eleven subjects tested, the total counts inside the respirator ranged from 192 to 1,276 particles during the entire reading time. In some particle size channels no particles were detected. For example, when testing Subject 1, the “inside” aerosol concentration was so low that particles were counted only in four PAMS channels. These low counts occurred because of high protection factor of the tested elastomeric respirator. Considering such low aerosol concentration levels and a short measurement time per particle size channel, the uncertainty of counting in PAMS, estimated from Poisson statistics, is high. This uncertainty could be reduced by increasing the “per-channel” measurement time; however, we did not choose to do so in the present investigation because it would require substantially longer overall measurement period (to scan over the entire particle size range).

Figure 3 presents the SWPF values calculated for all tested firefighters based on the total concentration in the entire size range of interest. The “total” SWPF ranged from 4,222 (minimum) to 35,534 (maximum) with values falling primarily in a range from ~1,000 (25 percentile) to 26,000 (75 percentile) and a median value of approximately 15,000 and a geometric mean (GM) value of 16,180. The SWPF results from this study fall within the fit factor (FF) data range reported by He et al.(¹) for the same respirator (also shown in this figure). The latter reported GM of 4,779 and a geometric standard deviation (GSD) of 9.1 (referred to as a non-modified respirator(¹)). The difference in exercises between the present simulation study that aimed at determining the SWPF and the fit testing conducted by He et al. to obtain FF as well as different number of the tested subjects (11 versus 25, respectively) can explain the differences seen in Figure 3.

Figure 4 presents the size-resolved SWPF data. Each point represents the GM value obtained for a specific particle size with the error bars representing the corresponding GSD calculated for 11 subjects. The increasing trend was found statistically significant (t-test, p<0.05). It was unclear whether the observed trend was due to the filter penetration, or the faceseal leakage, or some other reason. To confirm there were no measurement artifacts, a separate experiment was conducted using a P100 filter housed in a filter holder under a cyclic flow produced by a breathing simulator (Koken Ltd., Tokyo, Japan). Results from this test confirmed that the particle penetration through a P100 filter was in agreement with the conventional filtration theory. By ruling out an unusual effect related to the filter, we surmise that the trend seen in Fig. 4 can be attributed to one of the following effects. The first one is associated with the removal of particles from the respirator cavity during exhalation. Smaller particles subjected to diffusional motion may more likely be “trapped” inside the facepiece and not fully exit through the valve with effluent air. The second may be associated with small particles generated by subjects during exhalation. Both effects would artificially increase C_in and thus decrease the SWPF. It is acknowledged that the trend found in this human subject study was not observed in our earlier manikin-based investigation(¹⁶) involving similar conditions (an elastomeric half-mask respirator fitted with two P100 filters; wood combustion particles). Thus, the finding presented in Fig. 4 is likely associated with the specifics of human breathing that may not be easily replicated when testing respirators on a headform. At the same time, a human subject study of elastomeric respirators published earlier(¹⁷) did reveal an increase of WPF with an increasing particle size, although it was reported for larger particles (0.7–10 µm).