Ann. Occup. Hyg., Vol. 56, No. 2, pp. 207–220, 2012 Ó The Author 2011. Published by Oxford University Press on behalf of the British Occupational Hygiene Society doi:10.1093/annhyg/mer089 Performance Study of Personal Inhalable Aerosol Samplers at Ultra-Low Wind Speeds DARRAH K. SLEETH1* and JAMES H. VINCENT2 1 Rocky Mountain Center for Occupational & Environmental Health, Department of Family & Preventative Medicine, University of Utah, Salt Lake City, UT 84108, USA; 2Department of Environmental Health Sciences, University of Michigan, Ann Arbor, MI 48108, USA Received 31 March 2011; in final form 15 August 2011; published online 10 October 2011 The assessment of personal inhalable aerosol samplers in a controlled laboratory setting has not previously been carried out at the ultra-low wind speed conditions that represent most modern workplaces. There is currently some concern about whether the existing inhalable aerosol convention is appropriate at these low wind speeds and an alternative has been suggested. It was therefore important to assess the performance of the most common personal samplers used to collect the inhalable aerosol fraction, especially those that were designed to match the original curve. The experimental set-up involved use of a hybrid ultra-low speed wind tunnel/calm air chamber and a rotating, heating, breathing mannequin to measure the inhalable fraction of aerosol exposure. The samplers that were tested included the Institute of Occupational Medicine (IOM), Button, and GSP inhalable samplers as well as the closedface cassette sampler that has been (and still is) widely used by occupational hygienists in many countries. The results showed that, down to 0.2 m s21, the samplers matched the current inhalability criterion relatively well but were significantly greater than this at the lowest wind speed tested. Overall, there was a significant effect of wind speed on sampling efficiency, with lower wind speeds clearly associated with an increase in sampling efficiency. Keywords: aerosols; dust sampling conventions; inhalable dust; low wind speed; personal samplers; wind tunnel ticles that will enter into the nose and/or mouth during breathing—is increasingly replacing what used to be referred to as ‘total aerosol’ and therefore continues to be of considerable interest in relation to aerosols found in many workplace environments. Several sampling instruments have been proposed as being applicable to the collection of inhalable aerosols. The first such instrument was the IOM sampler, named after the Institute of Occupational Medicine in Edinburgh, Scotland, where it was developed (Mark and Vincent, 1986). This sampler was designed in the 1980s specifically to collect particles according to the criterion that had been proposed at the time, which was based on experiments with breathing mannequins in wind tunnels for wind speeds upwards of 0.5 m s–1. In the years that have followed, this sampler has become widely regarded as providing a good measure of inhalable aerosol exposure and has been commercially available for INTRODUCTION Personal sampling is widely accepted as providing the most representative measure of human exposure to airborne contaminants. For effective assessment of exposure to aerosols, it is essential to properly understand the performance characteristics of the various sampling instrument options so that the choice may be made to provide the best match with the aerosol fraction of interest. For industrial hygienists, these fractions of interest include the health-related inhalable, thoracic, and respirable fractions, for which widely agreed particle size-selective criteria now exist (ACGIH, 1985; CEN, 1992; ISO, 1992). The coarser inhalable fraction—consisting of all par*Author to whom correspondence should be addressed. Tel: þ1-801-585-3587; fax: þ1-801-581-7224; e-mail: [email protected] 207 208 D. K. Sleeth and J. H. Vincent many years. In the meantime, several other samplers have been proposed to fill this need and they too are commercially available. These include the Button sampler and the GSP sampler (which is equivalent to the CIS or conical inlet sampler). Meanwhile, the plastic closed-face cassette (CFC), nominally intended to collect ‘total aerosol’, is still widely used by occupational hygienists, especially in the USA. These are the samplers that will be the subject of the work described below. Comprehensive laboratory studies of the personal aerosol samplers listed above have been performed in both relatively fast-moving air (wind speeds .0.5 m s–1) (Kenny et al., 1997; Görner et al, 2010) and under calm air conditions (essentially zero wind speed) (Kenny et al., 1999; Görner et al., 2010). However, it is now known that most indoor working environments are characterized by ambient wind speeds in the range generally ,0.2 m s–1, intermediate between the conditions pertaining to the work described above (Berry and Froude, 1989; Baldwin and Maynard, 1998). Previous work from our laboratory to investigate human inhalability at such low wind speeds showed significant differences from what had been found earlier at the higher wind speeds (Sleeth and Vincent, 2009; Sleeth and Vincent, 2011). With that in mind, it was important to test the samplers in question under these more realistic conditions as well. Only limited such research has been previously carried out, inhibited by the inherent difficulty in generating a uniform test aerosol under such conditions, especially for particles toward the upper end of the size range of interest. For those sizes, the effect of gravitational settling on aerosol behavior becomes increasingly influential. However, with respect to the effects of wind speed, it is noted that one recent experimental effort suggested that the IOM, GSP, and Button samplers generally performed similarly at 0.5 and 1 m s–1, with minimal dependence on the external air velocity (Aizenberg et al., 2001). As already indicated in reference to previous work, human aspiration efficiency within the ultra-low range of wind speeds of interest was significantly different from what had been found at the higher wind speeds (Sleeth and Vincent, 2011). Specifically, human aspiration efficiency for aerosols in the range of particle aerodynamic diameter from 7 lm up to 90 lm, as assessed using a breathing mannequin, was found to be significantly greater at the ultra-low wind speed of 0.10 m s–1 compared to the higher values of 0.24 and 0.42 m s–1. With those results in mind, it is therefore important to determine whether personal samplers like those mentioned might reflect corresponding performance characteristics. Specifi- cally, we were interested to know whether these sampling devices might still provide fair measures of what is inhaled, even at reduced wind speeds. In other words, are the observed differences in human aspiration efficiency based on wind speed matched by the samplers in question? This paper describes an experimental study that was conducted to answer that question, namely to carefully evaluate the effectiveness of the various aerosol samplers listed above for collecting inhalable aerosols at ultra-low wind speeds. This work also represents an important extension of our previously reported work to examine changes in the particle size-dependent inhalable fraction at such low wind speeds. EXPERIMENTAL METHODS Wind tunnel and mannequin set-up The experimental set-up consisted of an ultra-low speed wind tunnel and life-sized heated, breathing mannequin system that has been described in detail elsewhere (Schmees et al., 2008a). In brief, the novel wind tunnel design combined a conventional aerosol wind tunnel and a calm air aerosol chamber in terms of both the principles and the modes of operation. In this way, it was able to generate a spatially uniform distribution of aerosols—with respect to both aerosol concentration and particle size distribution—at the low wind speeds of interest. The mannequin, consisting of head, torso, and upper arms, was attached to a computer-controlled breathing machine that enabled the simulation of a representative range of breathing minute volumes and various combinations of breathing modes (i.e. nose or mouth breathing or combinations of the two if desired). Separate pathways were designed for inhalation and exhalation to prevent the possibility of re-entrainment of particles already deposited inside the mannequin nose or mouth. A 47-mm filter holder was situated along the inhalation pathway inside the mannequin head for the collection of particles during inspiration. Any other material that deposited prior to reaching the filter was also recovered by wiping the inside walls with cotton balls impregnated with alcohol. In this way, all particulate material that passed through the opening of the breathing orifice was measured and so the inhalable aerosol to which the samplers were exposed was measured directly. Inhalable aerosol samplers The four different personal sampling devices mentioned earlier were those that were tested, i.e. the IOM, Button, GSP/CIS, and CFC. The first three Performance study of personal inhalable aerosol samplers samplers were chosen because they are increasingly being used around the world to measure personal exposures to the inhalable aerosol fraction and are commercially available to industrial hygienists. The fourth sampler, the CFC, is not specifically described as an inhalable aerosol sampler, but it is the sampler most commonly used by occupational hygienists in many countries for collecting ‘total aerosol,’ and therefore, it is of considerable interest to determine the extent to which it too might collect inhalable aerosol under low wind speed conditions. The IOM inhalable aerosol sampler (SKC, Inc., Eighty-four, PA, USA) was designed with specific regard to the inhalability criterion, which was already widely accepted at the time of its development (Mark and Vincent, 1986). The version of this sampler used in the experiments described here is shown mounted to the mannequin in Fig. 1 (in the far right of the picture). It included a stainless steel cassette insert that held a 25-mm glass fiber filter, and the whole cassette (including the filter) was analyzed gravimetrically. For this, the entire cassette was placed in a desiccator overnight prior to weighing to reduce contributions from moisture uptake, which included both before the experiment was carried out Fig. 1. Experimental set-up for assessing personal sampler performance at ultra-low wind speeds, showing the mannequin with all four personal samplers tested—from left to right: CFC, GSP, Button, and IOM. Not shown is the bag situated on the back of the mannequin torso that held the four sampling pumps. 209 and after the sample was obtained. Fitted caps were placed over the inlet when the samples were inside the desiccator—to prevent dust from settling onto the loaded filters—and care was taken to wipe off the outside of the cassette with a dry cloth prior to weighing. The IOM sampler required a sampling pump operated at a flow rate of 2 l min–1 and the fully assembled sampler was clipped to the lab coat of the mannequin so that the inlet pointed directly outwards from the body, as shown in the picture. The Button sampler (SKC, Inc.) is a newer addition to the group of samplers that have been suggested as matching the inhalable aerosol convention (Aizenberg et al., 2000). It features a hemispherical, multi-orifice inlet configuration and is shown on the center right side of the mannequin picture in Fig. 1. For this sampler, a 25-mm glass fiber filter, which had been desiccated overnight, was weighed on its own before and after sampling. The O-ring that held the filter in place tended to be quite snug against the filter, and so extra care was taken when removing it from atop the filter to prevent sample loss. The pump was operated at 4 l min–1 and the sampler itself was attached to the lab coat, again with the inlet pointing outwards from the body. The GSP conical inlet sampler (BGI, Inc., Waltham, MA, USA) is shown attached to the mannequin lab coat in the center left portion of the photograph in Fig. 1. In this case, the sampler utilized was the plastic CIS version, which is no longer commercially available, but it was expected to perform similarly to the metal version that is currently available. It featured a removable plastic cassette holding a 37-mm glass fiber filter. For these experiments, the filter—again desiccated overnight—was removed from the cassette and weighed separately before and after sampling. The pump connected to this sampler was operated at 3.5 l min–1 and the sampler was clipped to the lab coat with the inlet pointing directly away from the body. Finally, the 37-mm CFC (SKC, Inc.) is shown to the far left of Fig. 1, draped over the mannequin shoulder in the mode of its most common use (i.e. at 45° from horizontal). The version of the CFC sampler used for these experiments consisted of three polypropylene stages that fitted snugly together. A 37-mm glass fiber filter was weighed individually and placed on top of a supportive pad inside the cassette. All pieces of the sampler were kept in the desiccator overnight with the filter before and after sampling to control moisture uptake. Again, care was taken when removing the filter from the sampler to avoid potential sample loss due to mishandling. The sampling pump was operated at 2 l min–1, 210 D. K. Sleeth and J. H. Vincent consistent with the general practical use of this instrument. One concern with the CFC sampler, as noted previously, was that it had been consistently shown to suffer from high levels of internal wall deposits, which are not analyzed when only the filter is weighed (Puskar et al., 1991; Demange et al., 2002). There are ways in which to collect these deposits, such as wipes or specially designed inserts, as is often done by the Occupational Safety and Health Administration (OSHA) and hygienists in some industries. However, it was decided that, for this study, the analysis of the CFC would be for the filter only. In general, this is the way it has historically been done by the majority of industrial hygienists as general practice, and so the results presented here may be informative for interpretation of historical data. Additionally, there is no standard method for collection of wall deposits; for example, there is no mention in the US National Institute of Occupational Safety & Health (NIOSH) Method 0500 (Particles Not Otherwise Regulated, Total). Sampling and analytical procedures During each experiment, all four personal samplers were operating concurrently with the breathing mannequin, as was a thin-walled cylindrical tube operating as an isokinetic reference sampler placed 0.75 m upstream of the mannequin. The body was always heated to skin temperature (33°C) and continuously rotated reciprocally (i.e. through a full 360° and then reversing back again) at 2 r.p.m. in order to achieve orientation-averaged sampling. In order to eliminate any possible biases associated with position, the samplers were moved to the opposite coat lapel for repeat tests. Otherwise, initial determination of sampler placement, that is, on which lapel they were attached, was arbitrary in the actual experiments. The parameters that were studied are listed in Table 1. The first of these is particle size, with aerosols generated from narrowly graded fused alumina (powder grades F1200, F800, F500, F400, F280, and F240; Washington Mills, Niagara, NY, USA) providing test aerosols with mass median aerodynamic diameter from 9.3 to 89.5 lm. For most of the powder grades, the particle size distribution was measured in the wind tunnel directly using a modified Marple-type cascade impactor (Wu and Vincent, 2007) under each set of experimental conditions. The exception was for the two coarsest powder grades (F280 and F240) at 0.10 m s–1, where a nominal particle size was used (Mark et al., 1985). Wind speed covered the ultra-low range of interest (0.10, 0.24, and 0.42 m s–1), and several different mannequin breathing patterns (expressed as the combination of minute volume and mode of breathing) were used (6 l min–1 mouth, 20 l min–1 mouth, 6 l min–1 nose, 6 l min–1 nose–mouth, and 20 l min–1 nose– mouth). For each combination of the experimental conditions tested, at least two repeats were performed, for a total of 180 experiments, with 20 min per sampling period. Each sampler was operated with its own individual personal sampling pump (Model XR5000, SKC Inc.). In a typical occupational hygiene survey in an actual workplace, a personal pump of this type would be clipped to the worker’s belt, usually positioned at the lower back. In our experiments, it was more convenient to place all the pumps together in a small backpack located at the back of the mannequin torso, secured by a single strap that rode across the mannequin chest. This allowed easy rotation of the mannequin with all equipment fully contained. All samplers used either 25-mm (IOM and Button) or 37-mm (GSP and CFC) glass fiber filters which, as already mentioned, were conditioned overnight in a desiccator to reduce moisture content, both before and after sampling. As indicated, the IOM was used with a stainless steel cassette insert that was weighed along with the filter; for all other samplers, only the filter was analyzed. With the exception of the disposable CFC samplers, the IOM, Button, and GSP samplers were washed with soap and water, Table 1. Parameters assessed for inhalable sampler performance study. Samplers Powder grades (MMADa, lm) Wind speeds (m s1) Breathing patterns IOM F1200 (9.6, 9.5, 9.3) 0.10 6 l min1 mouth Button F800 (13.9, 12.8, 12.4) 0.24 20 l min1 mouth GSP F500 (28.8, 32.7, 28.7) 0.42 6 l min1 nose CFC F400 (37.7, 44.3, 40.0) b F280 (74.0 , 62.4, 66.9) 6 l min1 nose–mouth 20 l min1 nose–mouth b F240 (89.5 , 60.1, 63.0) a Mass median aerodynamic diameter measured directly for each combination of powder grade and wind speed. Listed values are for wind speeds 0.10, 0.24, and 0.42 m s–1, respectively. b MMAD is a nominal value from Mark et al. (1985). Performance study of personal inhalable aerosol samplers rinsed and dried, and reused for subsequent experiments. Blanks were obtained for each sampler type once per day, usually during the first experiment of that day, alongside cotton, and filter blanks prepared separately for the mannequin and isokinetic reference sampler. All the recorded sampled mass values were blank corrected. In general, most such corrections were very small, so contributed little to the overall recorded collected mass. These steps basically follow the procedures adopted by most occupational hygienists in practical aerosol sampling. The different pump flow rates required for the personal samplers (i.e. between 2 and 4 l min–1) were calibrated using a primary flow meter both before each test and then again after sampling was completed. If the change in flow rate was .–5% for a given sampler, that sample was deemed unacceptable and not included in subsequent analyses. The average of the two flow rates was used to determine the total volume of air sampled. Taking the change in filter weight (fm2 – fm1) less the change in filter blank weight (fb2 – fb1) and dividing that by the pump flow volume (V) provided the concentration of sampled material (CS), calculated using CS 5 ½ðfm2 fm1 Þ ðfb2 fb1 Þ : V ð1Þ This value was then compared to the reference sampler concentration (CR) measured by the isokinetic sampler. That value was calculated in a similar way but included the addition of wall deposits in the sampling tube, which were determined by taking the change in weight of the cotton wipe (cm2 – cm1) less the change in cotton blank weight (cb2 – cb1). The ultimate calculation for the reference sampler concentration was therefore CR 5 211 by the mannequin (CM), again calculated in a similar manner to equation (2)—including deposits from along the inhalation pathway—to estimate sampling efficiency of the ‘inhalable’ aerosols (AI) as estimated by what was inhaled by the mannequin, thus AI 5 CS : CM Ultimately, a total of 735 personal samples were taken (not including the mannequin aspiration efficiency measurements), with just seven rejected for being unacceptable due to unexplained inconsistency but most likely due to equipment failure or operator error. RESULTS IOM inhalable aerosol sampler Figure 2 shows the results for the aspiration efficiency of the IOM sampler as a function of particle aerodynamic diameter (dae). The graph represents all experimental data averaged across all mannequin breathing conditions and separated out into each wind speed (0.10, 0.24, and 0.42 m s–1, respectively). Each data point represents the mean of at least 10 experiments performed for a single particle size at the same wind speed, with error bars representing 1 SD. The current inhalability convention is also included for the purpose of comparison. Analysis of variance (ANOVA) results, comparing the IOM efficiency based on wind speed, showed that significant differences existed, with sampling efficiency decreasing as wind speed increased (P-value , 0.0001). There were no statistically significant differences based on the breathing pattern of the mannequin (P-value 5 0.1900) or for just the mode of ½ðfm2 fm1 Þ ðfb2 fb1 Þ þ ½ðcm2 cm1 Þ ðcb2 cb1 Þ : V Combining equations (1) and (2) provided an estimate of the sampling efficiency (AS) using AS 5 CS : CR ð3Þ The sampler filter mass concentration was also compared to the inhaled concentration as measured ð4Þ ð2Þ breathing (P-value 5 0.2712). However, sampling efficiency was significantly different based on mannequin breathing rate (P-value 5 0.0180). Such differences could be associated with the increased mixing in front of the mannequin at lower wind speeds and higher breathing flow rates, resulting from the significant and persistent air disturbances that have been seen to be present in the air approaching the mannequin under such conditions (Schmees et al., 2008b). However, we 212 D. K. Sleeth and J. H. Vincent do not have enough information to make any firm conclusions in this regard. From Fig. 2, it can also be seen that, at the lowest wind speed (0.10 m s–1), the IOM sampler consis- Fig. 2. Mean aspiration efficiency of the IOM sampler (AIOM) as a function of particle aerodynamic diameter (dae) for three different ultra-low wind speeds when attached to a heated, breathing, and rotating mannequin. Error bars represent 1 SD and the current inhalability convention is also shown. tently indicated sampling efficiencies that lay significantly above the original inhalability curve. At the two higher wind speeds (0.24 and 0.42 m s–1), the IOM followed the criterion somewhat better, but even so, most of those data still appeared to fall above the curve. Paired t-tests confirmed that, for each wind speed (0.10, 0.24, and 0.42 m s–1), the IOM sampler indicated significantly greater values for sampling efficiency in relation to the convention (P-values: ,0.0001, ,0.0001, and 0.0147, respectively). Next, Fig. 3 shows the IOM sampling efficiency compared directly with what was inhaled by the mannequin. This, after all, is the most appropriate comparison in terms of the relevance of the IOM sampler to inhalable aerosol and it removes any biases that might have been introduced through the reference measurement. In this figure, the solid line represents perfect agreement (AMannequin 5 AIOM) and the various dashed lines—all passing through the origin—represent simple linear regressions of the data for each wind speed separately. Results show that, for a given set of conditions, the IOM sampler consistently collected more than what was Fig. 3. Comparison of the aspiration efficiency for the IOM sampler (AIOM) to the mannequin aspiration efficiency (AMannequin), for all concurrent experiments at wind speeds of 0.10 m s1 (black circles), 0.24 m s1 (grey squares), and 0.42 m s1 (white triangles). The solid line shows perfect agreement and the various dotted lines are linear regressions for 0.10 m s1 (long dashes), 0.24 m s1 (dashes and dots), and 0.42 m s1 (dotted). Performance study of personal inhalable aerosol samplers inhaled by the mannequin, a tendency that was supported by a t-test comparison (all P-values , 0.0001). Numerical values for each of those relationships (including values for all other samplers tested as well) are summarized in Table 2. Button inhalable aerosol sampler Figure 4 shows the results for the mean sampling efficiency of the Button sampler as a function of dae (in this case, only the filter was analyzed and so the results are a measure of ‘sampling’ efficiency instead of ‘aspiration’ efficiency, with the terminology reTable 2. Slope factor for linear regression (forced through the origin) shown as the ratio of sampling efficiency to mannequin aspiration efficiency (AI), for each wind speed tested. Wind speed (m s1) Sampler 0.10 IOM 1.52 (0.36) Button 1.48a (0.38) GSP 1.40a (0.51) CFC 0.35 (0.15) 0.24 0.42 Slope Factor (R2) IOM 1.30 (0.14) Button GSP 1.12 (n/a) 1.10a (0.48) CFC 0.27a (0.19) IOM 1.18 (n/a) Button 1.00 (n/a) GSP 0.92a (0.27) CFC 0.26a (0.09) a Linear regression that is not forced through the origin has intercept that is not significantly different from zero. Fig. 4. Mean sampling efficiency of the Button sampler (AButton) as a function of particle aerodynamic diameter (dae) for three different ultra-low wind speeds when attached to a heated, breathing, and rotating mannequin. Error bars represent 1 SD and the current inhalability convention is also shown. 213 flecting this difference). These data are organized in the same manner as Fig. 2, separated into the different wind speeds studied, with each data point then representing the average efficiency across all mannequin breathing conditions at that particle size. ANOVA results indicated that the Button sampling efficiency was significantly different based on the wind speed inside the wind tunnel (P-value , 0.0001), with sampling efficiency increasing with decreasing wind speed. This is reflected in Fig. 4, which clearly shows that the Button sampler typically indicated higher sampling efficiencies at 0.10 m s–1 compared to 0.24 and 0.42 m s–1. There were no significant differences (at a 5 0.05) with respect to the mannequin breathing pattern, breathing flow rate or mode of breathing. As was seen for the IOM sampler, the lowest wind speed was consistently associated with sampling efficiencies that were greater than the current inhalability convention, with better agreement for the two higher wind speeds. In a general sense, therefore, performance was similar to that of the IOM. However, the results show that agreement between the Button sampler and the inhalable convention was better for the highest wind speed in the range tested here. Although the two lower wind speeds revealed significant differences to the existing convention based on paired t-tests (P-values: ,0.0001 and 0.0021, respectively), at the highest wind speed the Button was not statistically different from that standard (P-value 5 0.1308). Figure 5 shows the relationship between what was collected by the mannequin and by the Button sampler for all concurrent experiments. Overall, a paired t-test showed that the two data sets were significantly different (P-value , 0.0001). After separation based on wind speed, however, it emerged that there was no significant difference at the higher wind speed tested here (U 5 0.42 m s–1) (P-value 5 0.2317). This is reflected in Table 2, where it is seen that, on average, the Button sampler collected 48% more than the mannequin at 0.10 m s–1 and 12% more at 0.24 m s–1. At 0.42 m s–1, however, there was no detectable difference between the Button and the mannequin. GSP conical inlet sampler Figure 6 shows the results for the mean sampling efficiency of the GSP conical inlet sampler at the various wind speeds indicated. That figure is organized in the same way as previously, with the inhalability convention again included for comparison. ANOVA results showed that the sampling efficiency of the GSP was significantly different based on the 214 D. K. Sleeth and J. H. Vincent Fig. 5. Comparison of the sampling efficiency for the Button sampler (AButton) to the mannequin aspiration efficiency (AMannequin), for all concurrent experiments at wind speeds of 0.10 m s1 (black circles), 0.24 m s1 (grey squares), and 0.42 m s1 (white triangles). The solid line shows perfect agreement and the various dotted lines are linear regressions for 0.10 m s1 (long dashes), 0.24 m s1 (dashes and dots), and 0.42 m s1 (dotted). (In this case, the regression for U 5 0.42 m s1 exactly matches the mannequin and so overlaps with the solid line). Fig. 6. Mean sampling efficiency of the GSP sampler (AGSP) as a function of particle aerodynamic diameter (dae) for three different ultra-low wind speeds when attached to a heated, breathing, and rotating mannequin. Error bars represent 1 SD and the current inhalability convention is also shown. wind speed in the wind tunnel (P-value , 0.0001), with sampling efficiency again increasing with decreasing wind speed. There were no statistically sig- nificant differences (at a 5 0.05) for the sampling efficiency of the GSP based on any of the mannequin breathing parameters tested. Similar to what was seen for the performance of the other inhalable aerosol samplers already discussed, the GSP sampling efficiency was generally greater than the inhalability convention at the lowest wind speed as well. For the higher wind speed, however, the GSP performed similarly to the Button, more so than the IOM sampler. Statistically, the GSP performance was significantly different from the current convention for each wind speed (P-values: ,0.0001, 0.0029, ,0.0001, respectively), but here, it was observed that, at the highest wind speed (0.42 m s–1), the GSP under-sampled relative to the convention. Figure 7 shows the direct relationship between the GSP sampling efficiency and the mannequin aspiration efficiency for each of the three ultra-low wind speeds tested. Overall, a paired t-test showed a statistically significant difference (P-value , 0.0001). However, again, when separated out by wind speed, the results for the highest wind speed (0.42 m s–1) Performance study of personal inhalable aerosol samplers 215 Fig. 7. Comparison of the sampling efficiency for the GSP sampler (AGSP) to the mannequin aspiration efficiency (AMannequin), for all concurrent experiments at wind speeds of 0.10 m s1 (black circles), 0.24 m s1 (grey squares), and 0.42 m s1 (white triangles). The solid line shows perfect agreement and the various dotted lines are linear regressions for 0.10 m s1 (long dashes), 0.24 m s1 (dashes and dots), and 0.42 m s1 (dotted). showed no significant difference. Looking at the values in Table 2, the same trend is observed as for the other samplers already discussed—that is, better agreement between the mannequin and sampler is seen with increasing wind speed. Specifically, at 0.10 m s–1, the GSP collected, on average, 42% more than the mannequin, at 0.24 m s–1 it collected 10% more, and at 0.42 m s–1 it collected 8% less. It is worth recalling that, as indicated previously, the plastic cassette that held the filter inside the sampler was not included for gravimetric analysis in the present set of experiments. However, during the experiments, visible deposits of particulate material were observed on the inside walls of the cassette, and these were not evaluated. It is expected, therefore, that inclusion of the cassette in the gravimetric analysis (as was done for the IOM sampler) would generally result in higher sampling efficiency than has been reported here. CFC sampler Figure 8 shows the results for the CFC sampler, with data again organized as for the other samplers tested. ANOVA results indicated that filter-only sampling efficiency was significantly different based on wind speed (P-value , 0.0001) but not significant (at a 5 0.05) based on any of the mannequin breathing parameters. It is clear from that figure that the sampling efficiency of the CFC dropped off quickly with increasing particle size, approaching zero for particles with aerodynamic diameter larger than 30 lm. The only conditions for which the CFC filter provided a sampling efficiency approaching the current inhalability convention were for the smallest particles (aerodynamic diameter below 10 lm) and at the lowest wind speed (0.10 m s–1). These observations were consistent with many previous studies of this device indicated previously, both in the lab and in the field. Overall, the CFC sampler significantly underestimated the inhalable fraction of aerosols—as defined by the currently accepted convention—at all three ultra-low wind speeds (all P-values , 0.0001). It was only slightly better at the lowest wind speed compared to the highest wind speed. Figure 9 shows the direct relationship between what was collected by the CFC filter and the 216 D. K. Sleeth and J. H. Vincent mannequin, respectively. As expected, a t-test comparison of those data revealed statistically significant differences, across all wind speeds as well as for each wind speed separately (all P-values , 0.0001). In contrast to the other samplers tested, differences associated with the CFC were in the opposite direction, indicating that it consistently collected much less than was inhaled by the mannequin. From Table 2, it can be seen that, on average, the CFC typically collected only between 26 and 35% of what the mannequin inhaled. DISCUSSION Fig. 8. Mean sampling efficiency of the CFC sampler (ACFC) as a function of particle aerodynamic diameter (dae) for three different ultra-low wind speeds when attached to a heated, breathing, and rotating mannequin. Error bars represent 1 SD and the current inhalability convention is also shown. The primary objective of this personal sampler performance study was to assess the effectiveness of these aerosol samplers for collecting the inhalable aerosol fraction at ultra-low wind speeds considered to be realistic of a large proportion of workplaces. Therefore, it will be useful to include: (i) a further discussion about the limitations of these results, (ii) a discussion about the relationship between their sampling efficiencies and the current inhalability criterion (which is currently the benchmark for what is expected of inhalable aerosol samplers); and (iii) assessment of how these results relate to the Fig. 9. Comparison of the sampling efficiency for the CFC sampler (ACFC) to the mannequin aspiration efficiency (AMannequin), for all concurrent experiments at wind speeds of 0.10 m s1 (black circles), 0.24 m s1 (grey squares), and 0.42 m s1 (white triangles). The solid line shows perfect agreement and the various dotted lines are linear regressions for 0.10 m s1 (long dashes), 0.24 m s1 (dashes and dots), and 0.42 m s1 (dotted). Performance study of personal inhalable aerosol samplers applicability of a modified convention for defining inhalability at low wind speeds (Aitken et al., 1999; Sleeth and Vincent, 2011). Study limitations First, the results for the sampling efficiency of the IOM, Button, and GSP were much larger than expected for the smallest particles in the lowest wind speed. In those experiments, it would be expected that the efficiency would be at or below one. The possibility exists that some experimental bias was introduced to skew those results, inherent in the wind tunnel design or in the sampling and analytical methods used. Considering the consistent shift in results, it is also possible that the isokinetic reference sampler was producing erroneous results. It is uncertain if this was in fact the case, but it should be pointed out that the experiments were carried out randomly for each particle size. In other words, all the experiments for the smallest particles at the lowest wind speed were not carried out in succession. Therefore, if there were an experimental bias, it would most likely affect the other results as well. An additional possibility is that, at the lowest wind speed, there was a greater amount of deposition (as opposed to actual aspiration) inside the sampler due to the presence of aerosols being injected from overhead. On the other hand, the uniformity was previously tested for those conditions (i.e. smallest particles and lowest wind speed) and shown to be within –5% throughout the wind tunnel, so we are left with the results as they are. Relatedly, an important detail to note in Figs. 2, 4, and 6 is the magnitude of the standard deviations, which appear to be greatest for the largest particle sizes at the lowest wind speed. It is likely that this may be associated with the fact that the spatial distribution of aerosol concentration and wind speed was less uniform under these conditions, as had been noted during our initial set-up and calibration tests (see Schmees et al., 2008a). This may simply represent the physical limitations of studying large aerosols at low wind speeds—similarly high variability in sampling efficiency for large aerosols has been seen by others as well (e.g. Aizenberg et al., 2001). It implies that, not surprisingly, repeatability was more difficult under these conditions, but it may also limit the ability to generate firm conclusions based on these results. Next, as seen in Figs. 3, 5, 7, and 9, it was decided that the linear regressions should be forced through the origin. However, it is interesting to note that, when the regression line was not forced through 217 the origin (data not shown), the relationship is close to the desired 1:1 for many of the inhalable aerosol samplers. This indicates that, at ultra-low wind speeds, increases in mannequin aspiration efficiency were matched by corresponding similar increases in sampling efficiency. But, that said, it is the more acceptable assumption here to force the line to go through the origin, based simply on the undeniable fact that, when no aerosol is present, both the mannequin and the sampler should collect zero particulate material. In the absence of any direct evidence for an offset in the performance of these samplers (e.g. an artifact leading to a consistent bias), the only acceptable approach is the simple linear regression leading to the numbers presented in Table 2. However, for completeness, Table 2 also provides information about those samplers for which the intercept of a linear regression not forced through the origin was not significantly different from zero. And finally, regarding the CFC sampler, a few further remarks are required. It is worth noting some results of other researchers showing that, if worn with the inlet pointing directly outwards (as opposed to facing in the customary 45° downward direction), there was some increase in the sampling efficiency of the CFC sampler (Buchan et al., 1986; Görner et al., 2010). In addition, inclusion of wall deposits inside the plastic cassette of this sampler will increase the overall estimate of efficiency and so provide improved performance in relation to inhalable aerosol. Indeed, the US OSHA have recently revised several aerosol sampling recommended methods, notably for metals analysis, to require that such deposits be included as part of the analyzed sample (Hendricks et al., 2009). It has therefore been suggested that wall deposits might be large enough that, if included routinely in the analysis, would provide better agreement with the inhalability convention. However, it should be pointed out that, from the point of view of the physics of the sampling process (see Vincent, 2007), there is no reason to expect that the CFC should aspirate aerosol with the same efficiency as, say, the IOM sampler over the entire range of inhalable aerosol particle sizes (i.e. up to 100 lm). In the present work, wall deposits were not routinely analyzed and therefore, only speculations can be made regarding their importance. Ultimately, taking into account all that is known, including what has been learned from the present work, we must conclude that, when used in the traditional manner as it was in this study and in many historical exposure assessments, the CFC does not accurately sample the inhalable aerosol fraction. 218 D. K. Sleeth and J. H. Vincent Comparison of personal sampling data to current inhalability criteria As discussed previously, one of the primary applications of the existing inhalable aerosol convention (or ‘inhalability curve’) is to provide a benchmark against which inhalable aerosol samplers may be examined, with respect to how accurately they actually sample the inhalable fraction of aerosols. Therefore, integration of the results from the personal samplers presented here relative to the existing convention is of great interest. First, it will be remembered that all the inhalable aerosol samplers—IOM, Button, and GSP—showed greater sampling efficiencies relative to the mannequin at the lowest wind speed used in this study. Therefore—assuming this mannequin provides an accurate estimate of the inhalable fraction—each sampler would be expected to be greater than the convention under those wind speed conditions as well. However, the performances of the Button and GSP samplers were actually quite similar to the mannequin at the highest wind speed and would therefore be expected to follow the inhalability curve in a similar manner under those conditions. Finally, the sampling efficiency of the CFC was shown to be far too low for it to be of use for sampling the inhalable fraction at ultra-low wind speeds and would therefore not be expected to match the current criteria. In fact, the quantitative comparison between the sampler results and the current convention, as discussed separately for each sampler previously, generally produced these expected relationships. Specifically, each of the inhalable aerosol samplers over-sampled relative to the convention at 0.10 m s–1 and showed increasing agreement with increasing wind speed. Notably, the Button and GSP samplers did not appear to be statistically different from the inhalable convention for the higher wind speeds. In contrast, the CFC significantly under-sampled relative to the convention, based on the statistical analysis discussed previously. Keeping in mind the relatively high intrinsic variability in this work, this new collection of data implies that at ultra-low wind speeds, these commonly used personal aerosol samplers may not always be appropriate for estimating the inhalability convention as defined by the currently accepted criteria. On the one hand, at the highest wind speed tested here, the inhalable samplers do appear adequate for those purposes. However, at 0.10 m s–1, and up to 0.24 m s–1, an alternative inhalability criterion, against which inhalable aerosols samplers may be compared, may be advisable. Comparison of personal sampling data to proposed low wind inhalability criteria As suggested by previous work, a practical transition between ‘calm’ and ‘moving’ air might lie somewhere in the region of 0.2 m s–1 (Sleeth and Vincent, 2011). Therefore, it will be instructive to compare the new personal sampler data at 0.10 and 0.24 m s–1 to the low wind/calm air criterion proposed by Aitken et al. (1999). At 0.10 m s–1, it was shown that all the nominally inhalable aerosol samplers provided greater sampling efficiency relative to the proposed calm air criterion, with the CFC sampler providing lower sampling efficiency relative to that criterion. Again keeping in mind the high variability in the results at this wind speed, it appears that all those samplers were significantly different from the target suggested by the calm air model (IOM, GSP, and CFC: P-values , 0.0001 and Button: P-value 5 0.0002). That indicates that, although the proposed calm air criterion might provide a better definition of true human inhalability at 0.10 m s–1, the personal samplers currently used to measure the inhalable fraction did not tend to match that criterion particularly well at that wind speed in this study. At 0.24 m s–1, agreement with that proposed criterion improved for all the inhalable aerosol samplers but remained poor for the CFC sampler. Although the GSP and Button samplers appear to provide moderately good fits to the calm air model, these data were in fact quite widely dispersed and so a paired t-test found the sampling efficiency of both those samplers to be significantly different from the calm air criterion (P-values , 0.0001 and 0.0035, respectively). On the other hand, the IOM sampler, which also provided somewhat variable data, did match reasonably well with the proposed criterion for calm air at this wind speed (P-value 5 0.7488), despite the fact that it over-estimated exposures relative to the mannequin at that same wind speed. Again, however, the variability in the data does limit the strength of the conclusions. CONCLUSIONS The results shown here should help fill a gap in knowledge about the performance of personal samplers under low wind speed conditions. It is an important finding that the wind speed in the wind tunnel was a significant factor influencing the performance of all four samplers tested (IOM, Button, GSP, and CFC). In contrast, previous studies have suggested that, for air velocities of 0.5 and 1 m s–1, sampling efficiency was not dependent on wind speed Performance study of personal inhalable aerosol samplers (Aizenberg et al., 2001). In general, the results presented here appear to indicate that each of the inhalable aerosol samplers (IOM, Button, and GSP) provide an increasingly accurate estimate of the inhalable aerosol fraction as wind speed increases from 0.1 up to 0.42 m s–1—a range which is more representative of actual workplaces. It was also shown that the breathing pattern of the mannequin (i.e. the combination of minute volume and mode of breathing) and the breathing mode on its own (i.e. nose, mouth, or nose–mouth breathing) had no significant impact on the sampling efficiency of any personal samplers attached to the body. Where the breathing minute volume was looked at as a separate factor, only the IOM sampler was significantly different, although that difference appeared to be only minimal. As a whole, these results suggest that differences in human aspiration do not have any substantial effect on the performance of personal samplers attached to the body. However, it is possible that mounting personal samplers much closer to the nose or mouth (e.g. behind a welder’s face shield as was carried out by Lidén and Surrakka, 2009) could increase the likelihood that such factors may become influential. The performance of each sampler was also assessed relative to the mannequin aspiration efficiency obtained during the same experiments. Again, keeping in mind the large spread in the data, both the Button and the GSP were statistically similar to the mannequin at the highest wind speed (0.42 m s–1) but the IOM remained different from the mannequin at each wind speed. This suggests that at the higher end of the ultra-low wind speeds used in this work, these aerosol samplers may provide reasonably accurate measurements of the inhalable fraction of aerosols. In contrast, at both lower wind speeds (0.10 and 0.24 m s–1), the inhalable samplers were only somewhat more adequate at estimating the inhalable aerosol fraction—as measured here directly by the mannequin. The CFC, however, remained statistically significantly different from the mannequin at each wind speed. All these conclusions presume that the mannequin aspiration efficiency measured here provided an accurate estimate of the inhalable fraction. However, a direct comparison of the results for these personal samplers to the currently accepted inhalability convention—as well as to proposed calm air criteria—was also instructive for more practical purposes. Those results indicate that, at 0.10 m s–1, each of the three inhalable aerosol samplers tested here consistently provided greater sampling efficiency relative to the current inhalable aerosol convention. In contrast, at 219 the highest wind speed of 0.42 m s–1, agreement with the criterion appeared to improve for all three inhalable aerosol samplers, indicating that they may still be appropriate for use at such wind speeds. For the CFC sampler, it was not surprising—based on the results of many other studies discussed previously, as well as what is known about the physics of the aspiration process—that it had poor agreement with the inhalability convention. On the other hand, the alternative criterion suggested by Aitken et al. (1999) for use in calm air/ low air movement environments appeared to provide good agreement with measured sampling efficiency at a wind speed of 0.24 m s–1. However, the personal samplers were still significantly different from that criterion at the lowest wind speed (0.10 m s–1). That suggests that, if current samplers are to be used in ultra-low wind speed environments ,0.24 m s–1, physical modifications to better correlate those sampler measurements to the inhalable aerosol fraction may be necessary. However, the large amount of variability in sampler performance may also explain some of the discrepancy in the data and so attempts to reproduce these results would be advisable. Ultimately, for the most accurate application of these samplers in low wind speed environments, a modified criterion that is specific for the low wind speed regime against which to compare those measurements would need agreement and adoption. In fact, such discussions have become more formalized in recent years (Lidén and Harper, 2006; Sleeth and Vincent, 2011). Regardless, the results presented here indicate that these samplers may generally measure higher aerosol concentrations than what would actually be inhaled by humans at ultra-low wind speeds. In that case, the proposed calm air criterion appears to provide a somewhat more accurate, yet still conservative benchmark for use at ultra-low wind speeds. As a whole, the experiments just described represent an important addition to the knowledge and understanding of common personal samplers in use today. In general, all these findings are consistent with what has been learned about the relative performances of these samplers in other studies, both in laboratory and in field tests. But laboratory assessment of inhalable aerosol samplers at ultra-low wind speeds between 0.05 and 0.5 m s–1—being more representative of typical working environments—had not been performed previously. Recognizing the large variability in the data, future studies utilizing this wind tunnel system will attempt to improve the uniformity and spatial distribution of test aerosols. Ultimately, these results suggest that, as wellestablished inhalable aerosol samplers with 220 D. K. Sleeth and J. H. Vincent consistent agreement to mannequin inhalability, the IOM, Button, and GSP may be useful at some ultralow wind speeds. On the other hand, the CFC sampler is not recommended for measuring the inhalable aerosol fraction at ultra-low wind speeds without modifications, such as inclusion of wall deposits and altering the inlet configuration. Such results should be informative not only for the industrial hygienists performing workplace assessments but also for the standards-setting community concerned with sampling methodologies. Finally, while we have acknowledged repeatedly that the conclusions to be drawn from this work are softened by the considerable variability in the experimental data, it is important to note that experiments like those described were extremely difficult, at the edge of what aerosol scientists have long considered to be nearly impossible. It is difficult to imagine laboratory experiments for the conditions of interest that can achieve lower variability than was achieved here. So, we are pleased to present this work as a significant step forward in our knowledge about sampling for the inhalable aerosol fraction under realistic workplace conditions. FUNDING National Institute for Occupational Safety and Health (5-RO1-OH002987-09). Acknowledgements—This research was conducted while one of the authors (D.K.S.) was at the University of Michigan, where we are grateful to the Department of Environmental Health Sciences and the School of Public Health for the space and resources that were required to perform the work described in the paper. We are especially grateful to the National Institute for Occupational Safety and Health for its support in the form of grant 5-RO1-OH002987-09. REFERENCES Aitken RJ, Baldwin PEJ, Beaumont GC et al. (1999) Aerosol inhalability in low air movement environments. J Aerosol Sci; 30: 613–26. Aizenberg V, Choe K, Grinshpun SA et al. (2001) Evaluation of personal aerosol samplers challenged with large particles. J Aerosol Sci; 32: 779–93. Aizenberg V, Grinshpun SA, Willeke K et al. (2000) Performance characteristics of the button personal inhalable aerosol sampler. Am Ind Hyg Assoc J; 61: 398–404. ACGIH. (1985) Particle size selective sampling in the workplace. Report of the Technical committee on Air Sampling Procedures. Cincinnati, OH: American Conference of Governmental Industrial Hygientists. Baldwin PE, Maynard AD. (1998) A survey of wind speeds in indoor workplaces. Ann Occup Hyg; 42: 303–13. Berry RD, Froude S. (1989) An investigation of wind conditions in the workplace to assess their effect on the quantity of dust inhaled. U.K. Health and Safety Executive Report IR/L/DS/89/3. London, UK: Health and Safety Executive. Buchan RM, Soderholm SC, Tillery MI. (1986) Aerosol sampling efficiency of 37-mm filter cassettes. Am Ind Hyg Assoc J; 47: 825–31. CEN. (1992) Workplace atmospheres: size fraction definitions for measurement of airborne particles in the workplace. European Standard EN 481. Brussels, Belgium Comité Européen de Normalization. Demange M, Görner P, Elcabache JM et al. (2002) Field comparison of 37-mm closed-face cassettes and IOM samplers. Appl Occup Environ Hyg; 17: 200–8. Görner P, Simon X, Wrobel R et al. (2010) Laboratory study of selected personal inhalable aerosol samplers. Ann Occup Hyg; 54: 165–87. Hendricks W, Stones F, Lillquist D. (2009) On wiping the interior walls of 37-mm closed-face cassettes: an OSHA perspective. J Occup Environ Hyg; 6: 732–4. ISO. (1992) Air quality-particle size fraction definitions for health-related sampling. Technical Report ISO/TR/7708– 1983 (E), Revised version. Geneva, Switzerland: International Organization for Standardization. Lidén G, Harper M. (2006) Analytical performance criteria: the need for an international sampling convention for inhalable dust in calm air. J Occup Environ Hyg; 3: D94–101. Lidén G, Surrakka J. (2009) A headset-mounted mini sampler for measuring exposure to welding aerosol in the breathing zone. Ann Occup Hyg; 53: 99–116. Kenny LC, Aitken RJ, Baldwin PEJ et al. (1999) The sampling efficiency of personal inhalable aerosol samplers in low air movement environments. J Aerosol Sci; 30: 627–38. Kenny LC, Aitken RJ, Chalmers C et al. (1997) A collaborative European study of personal inhalable aerosol sampler performance. Ann Occup Hyg; 41: 135–53. Mark D, Vincent JH. (1986) A new personal sampler for airborne total dust in workplaces. Ann Occup Hyg; 30: 89–102. Mark D, Vincent JH, Gibson H et al. (1985) Applications of closely graded powders of fused alumina as test dusts for aerosol studies. J Aerosol Sci; 16: 125–31. Puskar MA, Harkins JM, Moomey JD et al. (1991) Internal wall losses of pharmaceutical dusts during closed-face, 37-mm polystyrene cassette sampling, Am Ind Hyg Assoc J; 52: 280–6. Schmees DK, Wu Y-H, Vincent JH. (2008a) Experimental methods to determine inhalability and personal sampler performance for aerosols in ultra-low wind speed environments. J Environ Monit; 10: 1426–36. Schmees DK, Wu Y-H, Vincent JH. (2008b) Visualization of the air flow around a life-sized, heated, breathing mannequin at ultra-low wind speeds. Ann Occup Hyg; 52: 351–60. Sleeth DK, Vincent JH. (2009) Inhalability for aerosols at ultra-low wind speeds. J Phys Conf Ser; 151: 9012062. Sleeth DK, Vincent JH. (2011) Proposed modification to the inhalable aerosol convention applicable to realistic workplace wind speeds. Ann Occup Hyg; 55: 476–84. Vincent JH. (2007) Aerosol sampling: science, standards, instrumentation and applications. Chichester, UK: Wiley & Sons. Wu Y-H, Vincent JH. (2007) A modified Marple-type cascade impactor for assessing aerosol particle size distributions in workplaces. J Occup Environ Hyg; 4: 798–807.
© Copyright 2026 Paperzz