Performance Evaluation of Full Facepiece Respirators with Cartridges

Full facepiece respirators (FFRs) with charcoal cartridges are very common wearable devices for industrial or military purposes preventing respiratory tract, face, eyes, and mouth from hazards exposure. However, related research for performance evaluation was rare. We aimed to use a self-developed testing system investigating (1) filtration efficiency (FE) of aerosols in charcoal cartridges; (2) breakthrough time of cyclohexane in charcoal cartridges; (3) fit factors (FFs) and (4) protection factors (PFs) of FRRs. The results showed that FE of charcoal cartridges against 0.3 μm of sodium chloride (NaCl) and dioctyl phthalate (DOP) particles under 85 L min were 99.93% and 99.89%, respectively, which did not meet the standard 99.97% regulated for filters such as P100 and HEPA. As for 42.5 L min condition (dual canisters), the FE against NaCl (99.97%) did meet the requirement set for HEPA (99.97%); however, that against DOP (99.95%) failed to meet the standard. In addition, the breakthrough time of 3000ppm cyclohexane was 31 min under 32 L min, which was longer than the 30 min standard described in CNS6636 Z2023 and JIS8152. As to the results of fit test among 21 subjects, only 1 failed to meet the requirement (fit factor = 500) and the average of fit factors was 8342 ± 5892. The mean value of protection factors was 112.3 ± 54.6 while 3 users had lower PFs than the assigned protection factor (APF = 50) regulated by Occupational Safety and Health Administration (OSHA). APFs obtained for particles less than 100 nm were smaller than the APF standard regulated by OSHA, indicating APF standards may be overestimated for such particle size. Therefore, users should be aware of potential risks caused by such substances in nano-sized scale (e.g., nanoparticles or viruses) when wearing FFRs.


INTRODUCTION
Respirators, mostly for industrial, medical or military purposes, are often used to protect hazardous air pollutants that could compromise health, including dust, chemical vapors, fumes, or bioaerosols from being inhaled in human bodies through airway.There are two major types of respirators with different functions such as air-purifying respirators (APRs) and atmosphere-supplying respirators (ASRs).According to OSHA's Respiratory Protection Standard 29 CFR 1910.134 (US Department of Labor, 1998a), APRs have filters, cartridges, or canisters that remove contaminants from the air by passing the ambient air through the air-purifying element before it reaches the user.ASRs supply clean breathing air directly to the user from a source other than the air surrounding the user and include supplied-air respirators (SARs) and self-contained breathing apparatus (SCBA) units.Although ASRs has better removal efficiency against hazardous air pollutants than APRs, the mobility and convenience of ASRs is far less than APRs due to the requirement of an air supply tank/unit.Therefore, the use of APRs is more extensive and well accepted.
All APRs are tight-fitting and available in four forms: mouth bit respirators, quarter masks, half facepiece respirators (HFRs), and full facepiece respirators (FFRs).Among them, HFRs and FFRs are the most popular forms.HFRs cover the nose and mouth with the lower sealing surface resting under the chin while FFRs cover the eyes, nose, and mouth, and usually extend from the hairline to below the chin.FFRs are made of elastic materials (e.g., silicone, rubber or PVC) and thus can fit the edge of the mask tightly with user's face along with adjustable head straps.Furthermore, FFRs contain a lens for vision.So in addition to the function preventing from inhaling hazardous substances, FFRs can also be used to protect eyes or face from contaminations caused by irrigating or erosive substances.However, several factors including odor filtration, strap adjustment, filtration of particles, and ease of breathing may affect the satisfaction levels of respirator wearers.Discomfort, difficult to breath and poor fitting were principal factors preventing users from wearing respirators (Gutierrez et al., 2014).To solve above issues, the Army used the new generation of M50 respirator (equipped with dual canisters, a wide monocular glass, and light weight, low breathing resistance as well as ~60 Pa pressure drop under 30 L min -1 air intake) developed by US Army as a reference and developed an exclusive military respirator (X103) to reduce breathing resistance and physiological stress.
Hazardous air may be inhaled into human body under the circumstances of poor protection caused by filter and/or canister malfunctions, poor fitness between the face and the mask, and the leakage of exhalation valve.According to National Institute for Occupational Safety and Health (NIOSH) 42 CFR 84 (NIOSH, 1995), dust-proof filters are rated as N95, N99, N100, R95, R99, R100, P95, P99 and P100; where FFP1, FFP2 and FFP3 are filter categories determined by EU Standards "EN 149" (CEN, 2001), and RS1, RS2, RS3, RL1, RL2, RL3, DS1, DS2, DS3, DL1, DL2 and DL3 by Japan International Center for Occupational Safety and Health (JICOSH, 2000).In NIOSH 42 CFR 84 Guide, N indicates "not resistant to oil", so filters rated as "N" are not applicable to workplaces involving oily aerosols; R indicates "resistant to oil", so filters rated as "R" are applicable to workplaces involving oily and non-oily aerosols, though regular filter replacement is required; and P indicates "oil proof", so filters rated as "P" are also applicable to workplaces involving oily and non-oily aerosols, but the properties and replacement intervals of the filters are determined by manufacturers.In US Guide, non-oily aerosol "sodium chloride (NaCl)" and oily aerosol "dioctyl phthalate (DOP)" are often used to determine filtration efficiencies, and the number 95, 99 or 100 indicates minimum filtration efficiencies (95%, 99% or 99.97%, respectively).Filtration efficiency (FE) refers to the efficiency of screening the most penetrating particle size (MPPS) (mass median diameter = 0.3 µm) under 85 ± 2 L min -1 air intake (to simulate heavy workload and intensive breathing).In EU Standards "EN149: 2001", FF refers to filtering facepiece, P to particulate filtration potentials, and numbers such as 1, 2 or 3 to corresponding filtration efficiencies (80%, 94%, and 99%, respectively) using NaCl aerosols under 95 L min -1 air intake.In JICOSH Criteria, R refers to "replaceable", D to "disposable", S and L to "aerosol types" (including solid aerosols (NaCl) and liquid aerosols (DOP)), and numbers such as 1, 2, or 3 corresponding to 80%, 95% or 99% FEs under 85 L min -1 air intake.According to regulations for disposable dust-proof respirators "CNS 14755 Z2125" (BSMI, 2011) established by Bureau of Standards, Metrology & Inspection, MOEA, R.O.C. (similar to JICOSH), there are 3 classes of respirators (D1, D2 and D3) that are applicable to environment containing non-oily aerosols.D1, D2 and D3 refer to filtration efficiencies 80-95%, 95-99%, and ≥ 99% against NaCl under 85 ± 4 L min -1 air intake.
The canister of respirator primarily uses adsorption or chemical reaction to remove hazardous gases.The protection levels of respirators mostly depend on the adsorption or chemical reaction capacities of adsorbents in the canister.Therefore, the types of adsorbents may vary depending on the types of hazardous gases to achieve effective protection.Because Bureau of Standards, Metrology & Inspection, MOEA, R.O.C. adopted the testing criteria "JIS T8152" (JSA, 2012) from Japan and established standards for respirator canister test "CNS6636 Z2023" (BSMI, 2013) for present days, these two standards were quite similar.That was, different types of gases were tested based on individual requirements for different classes of canisters under the same conditions (air intake: 30 L min -1 , temperature: 20 ± 2°C and humidity: 50 ± 5%).Similarly, EU Standard "EN 14387" (CEN, 2004) also performs breakthrough time tests for different gases under the condition of air intake: 30 L min -1 , temperature: 20 ± 2°C and humidity: 70%.In addition, canisters can be classified as Type A (certain organic gases), Type B (certain inorganic gases), Type E (sulphur dioxide and other acidic gases) and Type K (ammonia and organic derivatives) based on the types of removed gases.According to the filter capacity, Type A, B, E, K filters are further classified into three categories: Class 1 (< 1000 ppm), Class 2 (< 5000 ppm), and Class 3 (< 10000 ppm).In 2004, NIOSH published standards of filter capacities against chemical, biological, radiological, and nuclear (CBRN) substances in full facepiece air purifying respirators (NIOSH, 2004a).The classification and testing criteria for canisters/cartridges were described in Sections involving canister test challenge and test breakthrough concentrations.The service life test of canister in NIOSH CBRN APR was performed under environment of 64 L min -1 air intake at room temperature (25 ± 5°C) with two relative humidity (25 ± 5% and 80 ± 5%, respectively).All canisters/cartridges approved by NIOSH were tested by using reagents with known concentrations under previously mentioned conditions.The classifications of canisters/cartridges can be divided into CAP1 (15 min), CAP2 (30 min), CAP3 (45 min) and CAP4 (60 min) depending on breakthrough time.Hence, the breakthrough time of CAP1 canister/cartridge must be at least for 15 min, when that of CAP2, CAP3 and CAP4 should be at least for 30, 45 and 60 min, respectively.
In addition to penetration through filter materials, the protection levels of respirators are also related to faceseal leakage (Grinshpun et al., 2009;Mukhametzanov et al., 2016).Usually regular assessments for faceseal leakage consist of a qualitative fit test (QLFT) and quantitative fit test (QNFT) (US Department of Labor, 1998a).QLFT refers to the fitness of respirator depending on the responses of individuals to test agents, while QNFT refers to the fitness of respirator depending on the amount of leakage into the respirator measured numerically.Because the results of QLFT are prone to be influenced by users' physical conditions and subjective consciousness plus the operation of QNFT instrument is rather easy and fast, the results of QNFT are extensively applied to the assessment of respirator fitness.Fit factors (FFs) of tight-fitting half facepieces should be ≥ 100 when those of tight-fitting full facepieces should be ≥ 500.However, it is insufficient to indicate proper protection at workplace only using FF assessment during working.Because real workplace protection factors were very difficult to measure or obtain, NIOSH proposed a concept of total inward leakage test in 2004 (NIOSH, 2004b) to evaluate the sum of air leakage from the filter, faceseal leaks, or other places of the respirator (e.g., exhalation valve or crack) in respirator wearers.Our previous studies also used the same concept and successfully evaluated performance of N95 filtering facepiece respirators, surgical masks and FFP series respirators (Lee et al., 2008;Reponen et al., 2009;Lee et al., 2016).There were only studies investigating breakthrough time of hazardous gases, particle penetration through filters, or percentages of exhalation valve leakage in FFRs.Very few were involved in FFRs, not to mention studies regarding protection of FFRs performed with human subjects.This study aimed to investigate (1) filtration efficiency (FE) against aerosols; (2) breakthrough time of cyclohexane; (3) fit factors (FFs); and (4) protection factors (PFs) of FFRs.The study results will help us understand better the protection provided by FFRs and use it as references for future respirator selections.

METHODS
The performance tests of FFRs can be divided in four major parts, including breakthrough time of organic gas in charcoal cartridges, penetration efficiency of aerosols in charcoal cartridges, fit factors and protection factors of FFRs.Required instruments and testing procedures are described as follows:

Breakthrough Time of Organic Gas in Charcoal Catridges
The system built for breakthrough time measurement of charcoal cartridges against organic gas is indicated in Fig. 1.The system used nitrogen (99.99%,TOYO Gas, Taiwan) as a carrier gas and three mass flow controllers (MFC) (Type 8711, Burket, Germany) (flow rates are 5, 30 and 100 L min -1 ) to control production of organic gas and dilution of dry air as well as humidified air.Organic gas production device referred to the setting evaporating 500 mL of cyclohexane (99.9%,Merck, Germany) from a 20 L-serum bottle at 40°C in a thermostatic sink (BU550D, Yihder, Taiwan) to produce cyclohexane vapour.Air humidifier consisted of two 1 Laeration bottles along with a thermostatic tank (BU-410D, Yihder, Taiwan), which produced water vapour by warming up an aeration bottle and distilled water at constant temperature (25°C).The cyclohexane and water vapour were then carried by nitrogen to a mixing chamber and mixed with dry air to dilute cyclohexane concentrations to 3000 ± 300 ppm at 25 ± 2°C, 70 ± 5% with a flow rate of 32 L min -1 .After thorough equilibrium in the mixing chamber, a flame ionization detector (FID) (2005A, China Chromatography Co., Taiwan) was used to determine upstream concentrations of cyclohexane in the charcoal cartridge.When the concentration reached up to stable 3000 ppm, a pneumatic valve (F02-001-04, Fortelice International Co., Taiwan) was then activated to direct cyclohexane vapour into a charcoal cartridge.Subsequently, a thermo-Fig.1. System for breakthrough time measurement of charcoal cartridges against organic gas.hygrometer (HT800, Maxthermo, Taiwan) and a differential pressure gauge (Magnesense II, Dwyer, USA) in the cartridge holder were used to monitor temperature, relative humidity and pressure drops of the charcoal cartridge.Another FID was then used to determine downstream cyclohexane concentrations in the charcoal cartridge.These measurements could be real-time displayed on a computer screen to monitor the changes of all parameters and obtain breakthrough time of cyclohexane running through charcoal cartridge.We also purchased standard cyclohexane gas from the manufacturer (3137 ppm, Jing De Gases Co., Taiwan) to prepare different concentrations of cyclohexane gas for making a calibration curve and calculate the concentrations of cyclohexane vapours produced by the system during the study.In addition, the breakthrough concentration of cyclohexane vapour in charcoal cartridges was set as 5ppm.

Penetration Efficiency of Aerosols in Charcoal Cartridges
The system built for penetration efficiency measurement of charcoal cartridges against aerosols is indicated in Fig. 2. Added 50 mL of NaCl solution (20g NaCl/L DI H 2 O) or DOP alcohol solution (0.05 mL DOP/L alcohol) in a six-hole Collison nebulizer (CN25, BGI Inc., Waltham, MA, USA).The solution was aerosolized in the Collison nebulizer by nitrogen gas at a flow rate of 12.5 L min -1 and then mixed with dry air at a flow rate of 87.5 L min -1 .Since laboratorygenerated particles might carry high electrical charges, the entire airflow of 100 L min -1 was directed through a charge equilibrator (3.6 µCi of 241 Am) to achieve the Boltzmann charge equilibrium.As a result, aerosols running through charge equilibrator could be directed into a test chamber.In the test chamber, a charcoal cartridge was fixed with a cartridge holder, and an electrical low pressure impactor (ELPI) (Dekati Ltd., Finland) was used to determine upstream and downstream aerosol concentration as well as size distribution in the charcoal cartridge.After installation of a filter stage, the ELPI measures the numerical concentration of particles in an aerodynamic size, ranging from Da = 0.007 to 10.01 µm, in 12 channels.In this study, we used eight channels with geometric mean (GM) diameters of 0.0141, 0.039, 0.072, 0.122, 0.205, 0.320, 0.489, and 0.771 µm, in the particle size range of 0.007-0.958µm.Not only the diameters of nanoparticles and microparticles are within this limit, but the sizes of infectious microorganisms such as viruses and bacteria fall within this range.
Because the ELPI was expensive, we only had one ELPI to measure particle concentrations.The stability of particle concentrations in the test chamber was important when the calculation of penetration efficiency in the charcoal cartridge is the percentage of particle concentration inside the charcoal cartridge to that outside of the charcoal cartridge.Therefore, only when the upstream particle concentration of charcoal cartridge in the test chamber had been stabilized, and the upstream particle concentration was reliable enough to be documented.In this study, the ratio of upstream to downstream particle concentrations (range 0.007-0.958µm) Fig. 2. System for penetration efficiency measurement of charcoal cartridges against aerosols.
was 1-1.19 without the existence of charcoal cartridge.Although there were no large differences, all penetration efficiencies were expressed after being calibrated to deliver more accurate values.In addition, the penetration efficiency test was performed under 15 (air intake to simulate tidal volumes during light workload indicated by ICRP (1994)), 30 (air intake to simulate tidal volumes during medium workload indicated by ICRP (1994)), 42.5 (air intake of two cartridges indicated by NIOSH (1996)), and 85 L min -1 (air intake of one cartridge to simulate tidal volumes during heavy workload indicated by NIOSH (1996)) air intake (when the effective diameter in charcoal cartridge was 9 cm, the corresponding surface velocities were 22.3, 11.1, 7.9 and 3.9 cm s -1 ).

Fit test and Total Inward Leakage Test of FFRs
This study enrolled 21 male students (aged 20-28) from Feng Chia University.All subjects were required to perform fit tests and total inward leakage (TIL) tests.The fit test is an exam to investigate the tight-fit of face and FFRs to help respirator users choose adequate respirators preventing from hazardous substance damages at workplace while the TIL test can be used as an indicator of overall protection when wearing respirators.All subjects were nonsmokers and inexperienced respirator users.Human testing in this study had been approved by the Research Ethics Committee of China Medical University & Hospital, Taichung, Taiwan (approval number CMUH102-REC1-102) and each test subject provided written informed consent.A researcher informed the subjects that they could demand suspension of the experiment if they experienced any discomfort.To assure that respirator itself did not cause any physical or physiological damages or burdens to respirator wearers, all subjects had to pass the medical clearance evaluation to ensure that they were free of allergies and any cardiovascular or respiratory tract diseases.In addition, face shape measurements using a tape and calliper were required to be done three times for each individual subject.The standard deviation should fall within 5%.The results of face size measurements in 21 subjects are listed in Table 1.Both water drinking and smoking were prohibited 30 min before fit test and TIL test, and all male subjects were not allowed to grow a beard.Each subject was trained to wear the tested respirator by guidance from a researcher.A user seal check was performed to ensure that an adequate seal was achieved when the respirator was put on the subject's face.Subjects performed the US's Occupational Health and Safety Association's (OSHA) fit testing exercises, including normal breathing, deep breathing, turning head side to side, moving head up and down, talking, grimace, bending over, and returning to normal breathing (US Department of Labor, 1998b).Fit test was performed in a 19.5 m 3 of subject test chamber with controllable and stabilized particle concentrations (10 4 particles cm -3 ).Detailed layout, particle generation and control methods of subject test chamber were described in our previous study (Lee et al., 2016), and a PortaCount® Plus (TSI Inc., St. Paul, MN, USA) was used to conduct fit factors determination.
The FFR (X103) used in this study was the most up to date generation of respirator, which was developed by using the new generation of military respirator M50 from the US Army as a reference.X103 respirators not only improved issues such as high breathing resistance, poor visual field and heavy weight, but also mitigated the physical and psychological burdens of the user and thus greatly increased the mobility and working efficiency of the user.
In addition, the TIL test was used as an indicator of protection factors (PFs) of FFRs.The same as fit test, the PF of TIL refers to the ratio of particle concentrations collected from outside and inside of the respirator.We developed a novel personal sampling system (Fig. 3) to collect particle concentrations from inside and outside of the respirator by using previous studies as references (Lee et al., 2004;Lee et al., 2005;Lee et al., 2008).The prototype of this sampling system had been used in our previous study to measure PFs of EN-specified FFP respirators and surgical masks (Lee et al., 2016).The novel model was equipped with two sampling lines that could be used to collect NaCl particles inside and outside of the respirator, respectively.Each sampling line consisted of an adaptor, a 1/2" Tygon tubing (Tygon Tubing, Fisher Scientific and Pittsburgh, PA, USA), a three-way valve (LEGRIS Inc., France), a Nafion dryer (Models PD-50T-12, Perma Pure, NJ, USA) and an ELPI.The TIL test also required a subject test chamber equipped with stable particle concentrations (~10 5 particles cm -3 ) and the novel personal sampling system to complete PF measurements.During the test, each subject should perform 2 min of each of the following steps including normal breathing, deep breathing, head side to side, head up and down, talking, bending to toe, grimace and normal breathing to allow sample collections within the respirator.The average particle concentration at minute 2 (to eliminate the residual influence caused by previous action) was considered as faceseal leakage due to motions.Two minutes of particle concentrations outside of the respirator should also be collected before and after the activities, and the average was considered as the average particle concentration outside of the respirator during activity.Please refer to the study of Lee et al. (2016) for detailed PF calculation.

Data Analysis
Data were organized and managed using Microsoft Excel 2016, and the plots were made by SigmaPlot 10.0.The data analysis was performed using Kruskal Wallis test and the Spearman correlation model provided by SPSS 12.0 for Windows (SPSS Inc., USA) software.P-values of < 0.05 were considered significant.The difference in fit factors and PFs among exercises was examined by the Kruskal Wallis test.The Spearman correlation coefficients were obtained to examine the association between protection factors, fit factors and facial dimensions.

Breakthrough Time of Cyclohexane Vapor in Charcoal Cartridges
Fig. 4 presents the average breakthrough time of charcoal cartridge against cyclohexane vapor in three tests.The average upstream cyclohexane concentration determined using a 139 min-breakthrough time test of charcoal cartridges against organic gas at 26.5 ± 1.4°C, relative humidity of 70.5 ± 6.2%, 31.8 ± 0.3 L min -1 air intake and pressure difference of 98 ± 2 Pa was 2897 ± 168 ppm.So if the breakthrough concentration was set as 5ppm, the average breakthrough time from triplicate tests was 31 ± 2 min, which was longer than that indicated in CNS6636 (BSIM, 2013) as well as JIS8152 (JAS, 2012) Standards (30 min).For NIOSH APR CBRN (NIOSH, 2004a) Standard (10ppm), the test results of charcoal cartridges against organic gas in this study could be classified as CAP2 level (30 min < breakthrough time < 45 min).But because the air intake, temperature, and relative humidity used for this study were higher than that regulated in CNS6636 or JIS8152 Standards (air intake: 30 L min -1 ; temperature: 20°C; relative humidity: 50%), the breakthrough time of charcoal cartridge might be longer than that obtained in this study (31 min) if the criteria of CNS6636 or JIS8152 were applied to the test.Previous studies showed that the service life of activated carbon sorbent was closely related to the types and concentrations of organic gas; environmental temperature, relative humidity and air intake; as well as the types and load of aerosols.Tanaka et al. (1999) indicated that organic vapor with high boiling point and large molecular weight had longer breakthrough time due to strong adsorption.In addition, Nelson et al. (1976a) showed that in a single pure organic vapor environment, the higher the organic vapor concentration, the shorter the breakthrough time.However, even though Nelson et al. (1976b) also proved that the higher environmental temperature, the poorer carbon adsorption.However, the impact of temperature was limited.That was, the breakthrough time of organic vapor only decreased about 1-10% when the temperature was increased by ~10°C.On the contrary, relative humidity was found to be an effective factor influencing breakthrough time of charcoal cartridges due to water vapor's "competitiveness" against organic vapor to activated carbon sorbent.In other words, the higher the relative humidity, the shorter service life the charcoal cartridges, especially when relative humidity was 65% or above.But the impacts were reduced if relative humidity was lower than 50%.Nelson et al. (1972) also discovered that under steady-state and different pulsating flows, the breakthrough time of charcoal cartridge against organic vapor remained unchanged and the effective service life of charcoal cartridge might be inversely in proportion to flow rate, although the workload or relative humidity was rather high.Swearengen et al. (1988), Nelson et al. (1976c) and Yoon et al. (1992) all proved that the lower adsorption, the higher penetration efficiency of the organic vapor in the environment of more than two organic vapors; that was, the downstream concentration of organic vapor was increased with time.In addition, organic vapor with higher affinity (low polarity) toward activated carbon sorbent could replace that with lower affinity (high polarity) due to the differences of competitiveness caused by vapor polarities (affinities), so the downstream organic vapor concentration was usually found higher than that in upstream.As a result, with the existence of more than two vapor mixture, the effective service life of charcoal cartridges was entirely depending on the type of organic vapor with the lowest adsorption (or the shortest breakthrough time).Kuo et al. (2013) revealed that when a charcoal cartridge was loaded with liquid aerosols, the adsorption capacity of charcoal cartridge against organic gas was diminished along with shorter breakthrough time.On the contrary, if a charcoal cartridge was loaded with solid aerosols, the adsorption capacity of charcoal cartridge against organic gas was unchanged, but the breakthrough time might be decreased with the increase of aerosol load.Therefore, according to OSHA requirements, the employer should set up a respirator replacement schedule if no end-of-service-life indicators (ESLI) were available (US Department of Labor, 1998a).

Penetration Efficiency of Aerosols in Charcoal Cartridges
The penetration efficiency for charcoal cartridge against NaCl and DOP aerosols is presented in Figs.5(a) and 5(b), respectively.The results in Fig. 5(a) show that the penetration efficiency for charcoal cartridge against NaCl aerosols was increased with the increase of air intake, but decreased with the increase of particle size (less significant).Huang et al. (2013) also found that aerosol penetration through filter media increased with increasing face velocity.According to the NIOSH (1996) requirements for P100 filters or the Department of Energy (2015) requirements for HEPA filters, the filtration efficiency against the most penetrating particle size (MPPS) (0.3 µm) should be at least 99.97%.However, our results showed that under 15 L min -1 , 30 L min -1 , 42.5 L min -1 and 85 L min -1 air intake, the filtration efficiencies for charcoal cartridges against NaCl aerosols were 99.98%, 99.97%, 99.97, and 99.93%, respectively, which did not meet the requirements of NIOSH (1996) for P100 filters or the requirements of DOE ( 2015) for HEPA filters.Because the test system and aerosol measurement device used in the study are different from those required for the NIOSH and DOE certification protocols, the filtration results obtained from the study may not be predictive of those received using the respective certification test methods.These results cannot be directly compared with one another and should be used with concern.However, our results still indicate that respirator wearers might be exposed to greater risks or damages under larger air intake or environment containing smaller particulate matters (e.g., nanoparticles or viruses).Fig. 5(b) shows the penetration efficiency for charcoal cartridge against DOP aerosols, which was increased with air intake (the same as results observed in NaCl aerosols), but it was not significantly affected by particle size; i.e., the filtration efficiencies of charcoal cartridge against DOP aerosols under 15 L min -1 , 30 L min -1 , 42.5 L min -1 and 85 L min -1 air intake were 99.99%, 99.97%, 99.95% and 99.89%, respectively.The results indicated that the filtration efficiency of charcoal cartridge against NaCl was higher than that against DOP.Rengasamy et al. (2010) found that the penetration of P100 filters increased after treating P100 filters with isopropanol.This indicated that some P100 filters are electrostatic.DOP aerosols have been proved to reduce electric charges on electrostatic filters due to charge neutralization (Biermann et al., 1982) or ionic conduction through an oil film on the filter fibers (Tennal et al., 1991).Moreover, Rengasamy et al. (2013) discovered that the particle penetration efficiency (such as NaCl) for P100 filters loaded with DOP at lower amount was higher.The results indicated the phenomenon of filter degradation in DOP loaded respirators.Their study also found that the MPPSs of P100 filters were approximately 40-80 nm, which is similar to those found for CE-marked filters (Serfozo et al., 2017).Although the impacts of particle size on penetration efficiency were insignificant in this study, the penetration efficiencies with particle size of 55 nm and 94 nm were slightly higher than that with other particle sizes.Mostofi et al. (2012) studied the penetration efficiency for N95 filtering facepiece respirators against NaCl particles using ELPI and scanning mobility particle sizer (SMPS).The results showed that even though the trend was similar, particle size distributions measured by two different instruments were different since SMPS determined electrical mobility diameter (MPPS = 40 nm) and ELPI determined aerodynamic diameter (MPPS = 70 nm).Price et al. (2014) also revealed that the measures of small particles (< 100 nm) and large particles (> 1 µm) using ELPI were slightly different from those determined by other instruments, which were mainly caused by various measurement principles, particle bounce, particle deagglomeration as well as low size resolution.

Fit Factors and Protection Factors of FFRs
Total of 21 subjects were enrolled in this study to conduct fit test and total inward leakage test of FFRs.Although most of the users had smooth breath at rest, most of the wearers complained about the compression force and tightness around nose bridge, forehead and chins caused by FFRs, and the sense of smothering and muggy atmosphere was increased with wearing time.In addition, faceseal leakage surrounding the chin and cheeks was easily found in subjects with smaller faces due to fixed size of FFRs.
The results of fit test showed that the average of fit factors (FFs) of FFRs was 8342 ± 5892, and the qualified rate of fit factors ≥ 100 among 21 subjects was 100%, while that of fit factors ≥ 500 was 95.2% (only one subject failed).The results proved that FFRs tested in this study did not fit all wearers in order to provide perfect faceseal.Fig. 6 indicates the impacts of movements on FFRs fit testing results.To investigate whether residual aerosols inside the FFRs affect fit testing results, we performed two consecutive fit tests on 21 subjects.During the 1 st test, all subjects were asked to perform normal breathing without any actual movements, but they were required to complete all movements indicated in OSHA fit test during the 2 nd test.The test results are presented in Fig. 6.We found that from the 1 st to the 4 th normal breathing (i.e., head up and down (HUD) in OSHA fit test), the fit factor was increased with time, which indicated that residual aerosols inside FFRs were diminished with time.Therefore, the influence of residual aerosols on FFRs fit factors disappeared after the 4 th normal breathing.That was, the fit factor was not stabilized until the 4 th normal breathing.The results proved that residual aerosols inside the FFRs not only could affect fit factors, but also underestimated fit tightness of FFRs.Because four movements last for four minutes, the fit test is recommended to start after the respirator is put on four minutes later in order to obtain an accurate fit factor.In addition, according to the results of the 2 nd fit test, talking and bending over could easily cause faceseal leakage due to the increase of aerosol concentrations (caused by droplets) within FFRs while talking and the movements of FFRs caused by gravity while bending over.Interestingly, fit tightness of FFRs returned after discontinuing those actions.The results of Kruskal Wallis Statistical test also showed that physical activities did not have significant impacts on FFRs fit factors (p = 0.604).We also performed Spearman correlation coefficient statistical analysis to observe the effects of facial dimension on FFRs fit tightness.The results show in Table 2 indicating that there were no significant relationships between the two factors (p > 0.05).
We also performed total inward leakage test on 21 subjects to determine FFRs protection factors by measuring particle concentrations and size distribution of NaCl aerosols inside and outside the FFRs using our new developed personal sampling system.The study results are presented in Table 3 and Fig. 7.Only one subject failed to pass fit test and the data with or without that subject did not show  significant differences, hence we used a sample size of 21 to prepare all tables and plot all figures.Table 3 shows PFs of FFRs, which indicated that the average PF of FFRs against 0.028-1.161µm aerosols in 21 subjects were 109.7 ± 54.5, and the geometric mean was 99.5 ± 1.3.We found that the PFs among all subjects (n = 21) were all larger than the assigned protection factor (APF) of 10 required for HFRs, but the PFs in 14.2% of subjects (n = 3) were smaller than that of 50 for FFRs.In addition, the 5 th percentile PFs of FFRs (= 34.4) in this study was also smaller than APF of 50 for FFRs, thus the tested FFRs used in this study failed to meet the PF Standard for FFRs established by OSHA.Further intensive protective properties of FFRs are thereby required.In regards to the effect of particle size, 42.9% and 100% of subjects had smaller PFs of FFRs than APF of 10 or APF of 50 Standards when wearing FFRs against particles of 0.028-0.055µm.In addition, 0% and 61.9% of them had smaller PFs when wearing FFRs against particles of 0.055-0.094µm.In regards to wearing FFRs when compared to particles larger than 0.094 µm, all subjects had larger PFs than APF of 10 or APF of 50 Standards.We found that PFs of FFRs against particles less than 0.386 µm were increased with particle size.It is worth noting that the PFs of FFRs against particles below 0.094 µm were lower than APF of 50, but PFs of FFRs against particles larger than 0.094 µm were higher than APF of 50 regulated in OSHA.In summary, particle sizes did affect APF of FFRs, especially for FFRs against particles less than 100 nm (usually indicating viruses and nanoparticles).Fig. 7 shows the effect of particle sizes between 0.028-1.61µm on PFs of FFRs.The results indicated that when particle size was between 0.028-0.386µm, the PFs of FFRs were increased with particle size; but when particle size was between 0.386-1.61µm, the PFs of FFRs almost remained unchanged.For particles of 0.386-1.61µm, interception and impaction dominate particle penetration through filter and leaks.At low air intake, particle motion is gentle, resulting in low size dependence on PFs.When subjects performed light activities such as normal breathing at the beginning, deep breathing, head side to side, head up and down, talking, bending over, grimace and normal breathing at the end, their tidal volume (air intake) was between 8.4-16.9L min -1 (Lee et al., 2015), plus sampling air flow of 10 L min -1 , total air flow within FFRs should be around 18.4-26.9L min -1 .Based on the study results published by Chen and Willeke (1992), air flow of 18.4-26.9L min -1 indicated that the air flow through filter material and faceseal leakage fell in regime of laminar flow.Based on fit testing results, faceseal leaks created in our study were quite small (average fit factor > 500), hence our results were expected to be similar with those with smaller leaks published by Chen and Willeke (1992).The filter used for FFRs in this study was HEPA, therefore we expected that our results could be the same in Chen and Willeke's (1992) HEPA study.That was, with particle size between micrometers and submicrometers, the effect of particle size on PFs of FFRs was not pronounced because numbers of particles penetrating through faceseal leaks was increased under low air intake.For particles of 0.028-0.386µm, the MPPS usually occurs at particle size  The boxplots show the following: dots (from bottom) represent 5% and 95% percentiles; horizontal lines (from bottom) represent 10%, 25%, 50%, 75%, 90% percentiles.APF refers to the assigned protection factor for that particular mask.
< 100 nm, resulting in a decrease in PFs with decreasing particle size.In the literature review published by Shaffer and Rengasamy (2009), the MPPS in NIOSH certified respirators (P100, N99, and N95) or EU certified respirators (FFP1, FFP2, FFP3) were 200 nm or less.And in all respirators containing charged filters, the MPPS was less than 100 nm.In our study, the MPPS was between 40-60 nm, so the penetration efficiency of aerosols was increased in the smaller particle size range.As a result, the PFs were increased with particle size for particles less than 386 nm.Moreover, the smallest PFs of FFRs used in this study were between 28-55 nm, which were in accordance with previous findings.Vo et al. (2015) conducted a simulated workplace protection factor (SWPF) study using half elastomeric masks and half filtering facepiece respirators along with N95 and P100 filters.The results showed that the PFs against particles of 10-100 nm were larger than those against particles of 100-400 nm.Although such results were completely opposite to our or Grinshpun's et al. (2009) findings, all three studies used NaCl as testing aerosols, which may have hydroscopicity under very humid environment (RH > 70%) and thus led to aerosol aggregation, the increase of large particle concentrations and the decrease of small particle concentrations, and eventually resulted in PFs increase.However, the same as Grinshpun's et al. (2009) study, we performed expiratory air dehydration and thereby eliminated that possibilities of particle size and concentration changes due to hydroscopicity.The geometric mean (GM) of PFs of FFRs in this study was 97, which was higher by 3.6-4.9folds than that indicated in EN-specified FF1 (GM = 19.6),FFP2 (GM = 27.1) or FFP3 (GM = 26.7),and higher by 57 folds than that of surgical masks (GM = 1.7) (Lee et al., 2016).It was also higher by 4.5 and 40 folds than that of N95 filtering facepiece respirators (GM = 21.5) or that of surgical masks (GM = 2.4) in another study (Lee et al., 2008).The above results indicated that the protection factors of full masks against aerosols were better than half masks or surgical masks.Previous farm studies showed that the protection factors of N95 filtering facepiece respirators against particles not only were increased with particle size, but also varied depending on the types of pollutants (e.g., dust, fungi, bacteria, endotoxin, (1→3)-beta-D-glucan) (Lee et al., 2005;Cho et al., 2011), which indicated that both determination methods (e.g., sensitivity, working principle) as well as particle properties (e.g., shape, size) affected the results of PFs.
After Spearman correlation coefficient analysis, we realized that the menton-pronasal length (r = 0.495, p = 0.022) rather than other facial dimensions and FFs was significantly correlated with PFs.The results are presented in Table 2.Although face features such as bigonial breadth, face width, face length and nose protrusion might affect N95 respirator fit (Zhuang et al., 2005), since we used full facepiece respirators instead, the respirator almost covered that entire face and the contact surface between the respirator and the face was rather large and flat, respirator fit was not affected as much as that of half facepiece respirators.Furthermore, the moderate correlation coefficient (r = 0.427, p = 0.054) between FFs and PFs also indicated similar results in studies involving EN-specified FFP respirators (r = 0.378) and surgical masks (r = 0.482) (Lee et al., 2016).Because fit test results may reflect the PF of FFRs, we suggest that users should receive a fit test before selecting a respirator in order to determine whether the selected respirator could provide sufficient protection.

CONCLUSIONS
In general, FFRs not only provide excellent protection against organic vapor or aerosols, but also are great fits to most of the users with different face shapes.FFRs could provide sufficient protection to almost all subjects although our study results did not show significant correlations between facial dimension and protection factors of FFRs.Due to limited enrolled subjects, a further extensive study consisting larger sample size, both genders as well as larger populations is required to help us understand specific factors of face characteristics that may affect PFs of FFRs in order to reduce health risk of respirator users caused by hazardous air substances.In addition, the FFRs are not applicable to environment containing aerosols less than 100 nm (size of nanoparticles and viruses).Therefore, FFRs should be more cautiously used with the existence of aforementioned matters.Finally, issues such as the compression force and tightness around wearer's nose bridge, forehead, and chins caused by FFRs, and the sense of smothering or discomfort should be considered when using, modifying or redesigning FFRs in the future.

Fig. 6 .
Fig. 6.Impacts of movements on FFRs fit factors (NB: normal breathing at the beginning of fit testing; DB: deep breathing; HS: head side to side; HUD: head up and down; Bending: bending over; NB_end: normal breathing at the end of fit testing).

Table 1 .
Facial dimensions of 21 enrolled subjects.

Table 2 .
Relationships between face dimensions, fit factors and protection factors.

Table 3 .
Results of FFRs protection factor measurements.