A previous study used a PortaCount Plus to measure the ratio of particle concentrations outside (_{out}) to inside (_{in}) of filtering facepiece respirators (FFRs) worn by test subjects and calculated the total inward leakage (TIL) (_{in}/_{out}) to evaluate the reproducibility of the TIL test method between two different National Institute for Occupational Safety and Health laboratories (Laboratories 1 and 2) at the Pittsburgh Campus. The purpose of this study is to utilize the originally obtained PortaCount _{out}/_{in} ratio as a measure of protection factor (PF) and evaluate the influence of particle distribution and filter efficiency. PFs were obtained for five N95 model FFRs worn by 35 subjects for three donnings (5 models × 35 subjects × 3 donnings) for a total of 525 tests in each laboratory. The geometric mean of PFs, geometric standard deviation (GSD), and the 5th percentile values for the five N95 FFR models were calculated for the two laboratories. Filter efficiency was obtained by measuring the penetration for four models (A, B, C, and D) against Laboratory 2 aerosol using two condensation particle counters. Particle size distribution, measured using a Scanning Mobility Particle Sizer, showed a mean count median diameter (CMD) of 82 nm in Laboratory 1 and 131 nm in Laboratory 2. The smaller CMD showed relatively higher concentration of nanoparticles in Laboratory 1 than in Laboratory 2. Results showed that the PFs and 5th percentile values for two models (B and E) were larger than other three models (A, C, and D) in both laboratories. The PFs and 5th percentile values of models B and E in Laboratory 1 with a count median diameter (CMD) of 82 nm were smaller than in Laboratory 2 with a CMD of 131 nm, indicating an association between particle size distribution and PF. The three lower efficiency models (A, C, and D) showed lower PF values than the higher efficiency model B showing the influence of filter efficiency on PF value. Overall, the data show that particle size distribution and filter efficiency influence the PFs and 5th percentile values. The PFs and 5th percentile values decreased with increasing nanoparticle concentration (from CMD of 131 to 82 nm) indicating lower PFs for aerosol distribution within nanoparticle size range (<100 nm). Further studies on the relationship between particle size distribution and PF are needed to better understand the respiratory protection against nanoparticles.

Engineered nanoparticles are materials deliberately synthesized for specialized applications because of their unique physical and chemical properties. The concentration of nanoparticles released in some work-places is unlikely to pose substantial risk (

The Occupational Safety and Health Administration (OSHA) defines the assigned protection factor (APF) as the workplace level of respiratory protection that a respirator or class of respirators is expected to provide to employees when the employer implements a continuing, effective respiratory protection program (

The WPF for different types of respirators in a variety of workplaces has been described (

A surrogate measurement of the WPF called simulated workplace protection factor (SWPF) in a variety of different laboratory settings has been reported (

Similarly, PF for N95 FFRs worn by subjects was measured as a ratio of particle concentration outside to inside the respirator in a controlled test chamber, using an Electrical Low Pressure Impactor (ELPI) (

Recently, a PortaCount Plus was used to measure the ratio of particle concentration outside (_{out}) to inside (_{in}) of filtering facepiece respirators (FFRs) worn by test subjects. The _{out}/_{in} ratio was used to calculate the total inward leakage [total inward leakage (TIL) = _{in}/_{out}] to evaluate the reproducibility of the TIL test method between two different National Institute for Occupational Safety and Health (NIOSH) laboratories (Laboratories 1 and 2) at the Pittsburgh Campus (

In this study, the originally measured PortaCount _{out}/_{in} ratios were considered as PFs, and the influence of particle size distribution and filter penetration was evaluated. The purposes of this study are to: (i) calculate the GM of PFs and their 5th percentile values for the five N95 FFR models tested on 35 subjects exposed to aerosols of Laboratory 1 (CMD 82 nm) and Laboratory 2 (CMD of 131 nm), (ii) compare both the PFs and the 5th percentile values in the two laboratories for the five FFR models, (iii) evaluate the influence of particle size distribution and filter penetration on PF and 5th percentile value.

This study uses the data collected in a previous study (_{out}/_{in} ratios for respirators were measured, and then converted to _{in}/_{out} ratios (TIL) to evaluate the reproducibility of the TIL test method between two laboratories. This article, however, has used the originally measured _{out}/_{in} ratios. Brief details of that study are described below. For more details, the methods used in that study are available in the above article.

Five N95 FFR models were tested in the study. The manufacturers and models in parentheses are: 3M (Model 8000), 3M (Model 8511), 3M (Model 9210), Kimberly-Clark (Model 170/174), and Sperian–Willson (Model SAF-T-FIT) which were labeled randomly as A, B, C, D, and E.

A particle generator (TSI Model 8026) was employed, as needed, to supplement laboratory particle concentration levels with NaCl aerosol to maintain an ambient aerosol concentration between 0.01 to 5 × 10^{5} particles/cm^{3} required for the PortaCount measurement. Ambient aerosol concentration (particles/cm^{3}) in Laboratory 1 ranged between 1310 and 8740 (average 3010) and in Laboratory 2 ranged between 1370 and 10 100 (average 5410).

Thirty five subjects were tested for PF measurement with each of the five FFR models in both test laboratories. The NIOSH bivariate panel was used for placement of test subjects in specific face length by face width cells (

A PortaCount^{®} Pro+ (Model 8038, TSI, Inc. Shoreview, MN), with the N95-Companion mode turned off, was used to measure the ambient and in-mask particle concentrations to obtain the PF. The accuracy of the PortaCount measurement is ±10% as specified by the manufacturer. Subjects were randomly chosen to start the test in either Laboratory 1 or Laboratory 2, and then completed the test in other laboratory on the same day. Different test operators administered the testing in each of the two laboratories and were blinded to the test results of the subjects in the other laboratory. Subjects performed the eight exercises described in the standard OSHA fit test protocol (OSHA, 1998b). These eight exercises were performed in the following order: (i) normal breathing, (ii) deep breathing, (iii) turn head side to side, (iv) move head up and down, (v) speak out loud (recitation of the ‘rainbow’ passage), (vi) reach for floor and ceiling, (vii) grimace, and (viii) normal breathing. The duration of time for each exercise is about a minute for a total of 8 min for the test. At the end of the test, the subject removed the FFR, redonned the same FFR, and repeated the test two more times consecutively with a 5-min break between the tests.

The PortaCount calculates the PF for each individual exercise (PF_{i}

Two similarly calibrated PortaCounts were used to measure the PF in the two test laboratories. Test data, including test subject and respirator identifiers were recorded by the FitPro Fit Test software (TSI, Inc.) and accessed after the test for analysis. Test data was also recorded manually for immediate review by project personnel and verification.

Two Scanning Mobility Particle Sizers (SMPS, TSI, Inc.) with a long differential mobility analyzer (Model 3081) were used to measure the size distribution of particles in the 10–700 nm size range in the two laboratories. The SMPS was programmed to scan the particle size distribution for 135 s, three times, every hour from 8:00 AM to 4:00 PM Monday through Friday. From the SMPS scans, the average CMD of the laboratory aerosol was obtained. No specific attempt was made to maintain the aerosol size distributions different in the two laboratories. Laboratory 1 and Laboratory 2 are located in two different buildings with additional laboratories, where experiments not related to the work described in the article, were also performed. Particles generated by other processes in the two buildings could have contributed to the particle size distributions in the two laboratories.

Only four N95 respirator models (A, B, C, and D) were tested because model E was not available during the initial part of the study. Instantaneous penetration for FFRs was measured using two ultrafine condensation particle counters (UCPCs) against aerosols in the test Laboratory 2 (^{−1}. The particle number concentration upstream and downstream of the respirator was measured simultaneously after 1 min equilibration time. Percentage penetration was obtained from the ratio of the aerosol concentration downstream to upstream and multiplying by 100. From the penetration values, the filter efficiencies for the four models were assessed.

Thirty-five subjects tested five N95 model FFRs three times (35 subjects × 5 FFR models × 3 donnings) to give a total number of 525 PF tests in each laboratory. The 5th percentile PF value was calculated using the formula GM/GSD^{1.645}. Statistical significance tests were based on a mixed effects analysis of variance (with ‘subject’ as a random effect) using specific contrasts for hypothesis testing. R software was used for this analysis (

Laboratory 2 aerosol size distribution data and penetration data were analyzed using the SigmaPlot^{®} version 11 (Aspire Software International, Ashburn, VA, USA) computer program. The CMD values measured in the two laboratories were analyzed by a two sided ^{−1} flow rates.

Particle distribution was measured on all PF test days (27 days in Laboratory 1 and 32 days in Laboratory 2). The mean CMD was 82 ± 19 nm in Laboratory 1, which was significantly (

Filter penetration for four models (A, B, C, and D) was measured on 5 days only in Laboratory 2. The CMD values on Days 1–4 (108, 136, 124, and 75 nm, respectively) were smaller in the morning than in the afternoon (149, 154, 161, and 171 nm, respectively). In contrast, the CMD value on Day 5 was larger in the morning than in the afternoon (127 versus 87 nm, respectively).

Filter penetration was measured at 30 l min^{−1} on 2 days (Days 1 and 2), and at 85 l min^{−1} on 2 days (Days 3 and 4). Average penetrations at 30 l min^{−1} (Days 1 and 2) as well as at 85 l min^{−1} (Days 3 and 4) were calculated to explain the results better. ^{−1} (left panel) and at 85 l min^{−1} (right panel). Penetration was relatively larger in the morning than in the afternoon at 30 and 85 l min^{−1} for all four N95 FFR models. The penetration values on Day 5 were not combined with the penetrations on other days, because of the opposite trend in penetration in the morning and afternoon at both flow rates (bottom panels). Penetration was relatively smaller in the morning than in the afternoon at both flow rates for all four N95 FFR models. Penetrations in the morning and afternoon were not statistically different on all 5 days. Model B showed significantly (

Data from this study showed an association between PF and particle size distribution. Two models (B and E) tested in the study showed that the GM of PFs in Laboratory 1 was significantly (

The relationship between particle size and PF obtained in our study is corroborated by the findings in previous studies (

Some studies have described a lack of correlation between particle size and WPF values (

Moreover, the measurement of _{in} for larger size particles in a workplace with fewer numbers of larger size (~5 μm) particles may produce inconsistent _{out}/_{in} ratio. This may be a potential source for the lack of correlation between particle size and WPF in studies with mass-based particle concentration measurement, because the loss of a single large particle could underestimate _{in} and may result in an overestimation of the WPF value (_{in} measurement, test equipment and methodologies are important criteria to obtain a strong correlation between particle size and WPF.

The results obtained in the study raises a question why only models B and E showed significantly (

The impact of particle size distribution on PFs can be explained by the filter penetration measured in Laboratory 2. Filter penetration measured in the morning and afternoon on 5 days showed larger penetration values for laboratory aerosol with a smaller CMD values and vice versa (

The limitations of the study include that subjects tested only five N95 models. Additional models need to be tested to get consistent information on the 5th percentile values for N95 FFRs. This study was done in two laboratories with aerosol distribution closer to each other (CMD values 82 and 131 nm, respectively). PF tests against aerosols with CMD values closer to ~50 nm size as well as away from that value such as >300 nm would be important to recognize the significance of PF values for nanoparticles and larger size particles. Further studies are underway in our laboratory to address the respiratory protection against nanoparticles.

The PFs and 5th percentile PFs for two N95 FFR models were smaller in Laboratory 1 for aerosol with a mean CMD of 82 nm than in Laboratory 2 with a CMD of 131 nm indicating an association between particle size distribution and PF values. The smaller CMD (CMD 82 nm) value in Laboratory 1 shows the distribution of relatively higher concentration of nanoparticles than in Laboratory 2 (CMD 131 nm). The results indicate that smaller PFs and 5th percentile values can be expected for particle size distributions within the nanoparticle size range (<100 nm). Filter efficiency was directly related to PF obtained in both laboratories. Further studies on PF for environments with particles <100 nm sizes are needed to evaluate the respiratory protection against nanoparticles.

The authors acknowledge NIOSH colleagues including Raymond Roberge, William Monaghan, and Emanuele Cauda for their useful suggestions and critical review of the manuscript.

FUNDING

National Institute for Occupational Safety and Health.

DISCLAIMER

Mention of commercial product or trade name does not constitute endorsement by the National Institute for Occupational Safety and Health. The findings and conclusions of this report are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health.

The authors declare there are no conflicts of interest in relation to this article.

Schematic of the filtration test set up used for measuring laboratory aerosol filter penetration.

Filter penetration in the morning (AM) and in the afternoon (PM) for four N95 FFR models (A, B, C, and D) against ambient Laboratory 2 aerosol. The CMD was smaller in the morning than in the afternoon on Days 1–4 and showed a reverse trend on Day 5. Top panels show the average penetration values on Days 1 and 2 at 30 l min^{−1} (top left) and on Days 3 and 4 at 85 l min^{−1} (top right). Bottom panels show the penetration values at 30 and 85 l min^{−1} on Day 5.

GM, GSD, and 5th percentile of PF values for the five N95 models tested in the two laboratories

N95 model | Total number of tests | PF | |||||
---|---|---|---|---|---|---|---|

Laboratory 1 | Laboratory 2 | ||||||

GM | GSD | 5th percentile | GM | GSD | 5th percentile | ||

A | 105 | 29.09 | 1.93 | 9.86 | 32.28 | 2.06 | 9.83 |

B^{a} | 105 | 115.05 | 2.36 | 28.02 | 156.37^{b} | 2.54 | 33.74 |

C | 105 | 44.12 | 1.80 | 16.78 | 43.28 | 1.58 | 20.39 |

D | 105 | 39.07 | 1.43 | 21.69 | 36.43 | 1.52 | 18.29 |

E^{a} | 105 | 92.05 | 2.07 | 27.81 | 111.15^{b} | 2.18 | 30.94 |

Significantly different from other models.

Significantly different from Laboratory 1 for this model.