Filtration of Bioaerosols Using Fibrous Air Filter Media.

Filtration, which protects against the intrusion and spread of airborne microorganisms in indoor environments, is recognized as one of the primary means to control IAQ. Filtration of particles has been studied extensively, but the filtration of microorganisms remains a little-understood phenomenon (Washam 1966). As the unique characteristics of microorganisms in the submicron size range become better understood, they are receiving increased attention. The test plan proposed in the ASHRAE RP-909 final report (Karin and Hanley 1996) describes an integrated system for air filters for evaluation of antimicrobials and proposes a test protocol supported by rigorous quality assurance practices, but further research on the standard test rig and bioaerosol infiltration model was not conducted. The experimental apparatus and test protocol were developed to measure the efficiencies of the surgical mask and respirator using a microbial aerosol challenge (Brosseau et al. 1993). M. chelonae bioaerosol, Dioctylphthalate (DOP), and polystyrene latex (PSL) were used in the above system, and the results were compared. Filter collection of DOP was linearly related to that of both mycobacterial and PSL sphere aerosols ([r.sup. = 0.99), demonstrating that an inert aerosol may be available to predict the collection of bioaerosols (Chen et al. 1994). A filter certification system was also created for the filtration efficiencies of unloaded N95 respirators (an N-series filter that is at least 95% efficient). The reported efficiencies of respirators were compared with those of dust/mist (DM) and dust/fume/mist (DFM) against bacteria with a size and shape similar to those of Mycobacterium tuberculosis. The results in all cases showed the filtration efficiencies are 99.5% or higher (Qian et al. 1998).

Some analyses indicated that existing filter models could be used to predict the filtration efficiency of bacteria and spores at their log mean diameters rather than at their arithmetic mean diameters (Kowalski et al. 1999). The filtration performance for airborne microorganisms was studied by addressing the critical aspects of filter sizing and a method for predicting a filter's performance against airborne microorganisms (Kowalski and Bahnfleth 2002). But the studies on theoretical bioaerosol filtration are insufficient, and most of the suggested models cannot predict the filtration performance of filter media accurately.

On the other hand, although scientists have made great effort in the study of filtration of bioaerosols, at present no universal test rig is available or proposed for bioparticle efficiency testing of filter media. The air filter media investigated in previous studies do not span the range from low to high efficiency particulate air filter (HEPA) media. The filter media chosen in this investigation covered all media made of glass fiber that are commonly used in central air-conditioning systems. Commercial filters are only labeled by their nonbiological efficiencies, even when the intended application involves bioaerosol filtration, such as for the food industry, pharmaceuticals, and hospitals. There are no international standards of efficiency tests for fibrous filter media with bioaerosol and other challenge particles.

To fill this void, a test system was designed, and the efficiencies of several filter media samples were determined for bioaerosols. The bioaerosol filtration efficiency of a particular fibrous medium depends on the aerosol characteristics of the microorganism. The newly suggested microorganism, Serratia marcescens (S. marcescens), is an approximately rod-shaped bacterium whose colonies appear bright red in color below a certain temperature and is thus easily distinguishable from other markers as a microbiological marker.

SYSTEM DESIGN The experimental test rig (Figure 1) was designed to provide a stable and reproducible bioaerosol with a constant concentration over the sampling period. The rig controlled the sampling flow rate and accommodated various sampling methods. The system consists of three parts: a bioaerosol generation section, a test filter section including sample mounting, and a sampling section. A Collison nebulizer was used for nonbioaerosol and bioaerosol particle generation, while the optical particle counter (OPC) and the Andersen sampler were the detectors used in the sampling section. Bioaerosols were produced by atomizing the bioparticle solution in a six-jet Collison nebulizer at a pressure of 0.05 MPa (7.25 psi). The airflow rate and the pressure were recorded by a rotameter and pressure gauge. The high-concentration bioaerosol from the Collison nebulizer was mixed with dried, filtered air in the mixing chamber to produce a diluted and stable bioaerosol. The concentration remained stable [ or -]10% and did not vary with changes in the airflow rate. The pumps used for aerosol generation and dilution air maintained the experimental rig under positive pressure to prevent environmental background microorganisms from entering the system.

[FIGURE 1 OMITTED] The test filter medium mounting assembly was designed according to EN1822-3 to keep the filter medium and inner system airtight (CEN 1998).

It was driven by compressed air from the laboratory (not illustrated in Figure 1). In this setup, the exposed effective filtration area of the medium was .sup. [mm.sup. (15.5 [in.sup.), and the filter medium face velocity was 0.053 m/s (0.17 ft/s) at the system airflow rate of 32 L/min (1.13 cfm). The bacteria and nonbiological particles were sampled by the Andersen sampler and OPC, respectively, from the upstream and downstream sampling locations. The OPC used for DOP particle counting was a MetOne A2400 particle counter with 6 particle size channels (0.3, 0.5, 0.7, 1.0, 2.0, 5.0 [micro]m), and the sampling flow rate was 28.3 L/min (1 cfm). During the sampling of the HEPA filter media, a diluter (not illustrated in Figure 1) was used to test the particles upstream. The pressure drop of the filter medium was measured by the pressure gauge.

A six-stage Andersen sampler was used for bioaerosol sampling. It operated at the rated airflow 28.3 L/min (1 cfm). Under the conditions in these experiments, the sampling segment flow was turbulent.

Aspiration theory predicts the diffusion and deposition loss of particle with aerodynamic diameters up to about 1 [micro]m do not significantly affect the sampling efficiency (Andersen 1958). In comparison, the inertial collision could be significant and would require attention in the design of the sampling lines. Sampling lines with a large curvature radius would minimize this loss. Hence, sharp bends were avoided in the system (Cheng and Wang 1981). After sampling, the microorganisms captured on the nutrient agar growth medium were incubated at the proper temperature for 24 hours then counted using the positive-hole method. To form the filter medium efficiency, bioaerosol concentrations upstream and downstream were calculated using Equation 1:

C(cfu/[m.sup.) = [[N(cfu)x/[t(min)x 28.3(L/min)]] (1) where C = the concentration of the bioaerosols N = the number of variable particles t = the sampling duration The system and the samplers were sterilized with 75% alcohol before each experiment, and any residue was purged by filtered air to keep the bioaerosols from affecting the sterilizing process. The temperature and humidity of the environment and the system were measured and kept constant at 26[degrees]C [ or -] 1[degrees]C (78.8[degrees]F [ or -] 33.8[degrees]F), 50% RH [ or -] 20%. HEPA filters of 99.999% efficiency were used in the system to produce clean air for the experiment and to prevent the bioaerosols from leaking out of the apparatus. Filters also protected operators from exposure to bacteria in the exhaust air.

METHODS Bioaerosol Two microbiological markers were used in the experiments:

Escherichia coli (E. coli) and Serratia marcescens (S. marcescens). E.

coli is a rod-shaped bacterium with a length of 1.0-3.0 [micro]m and a width of 0.3-0.8 [micro]m, and S. marcescens resembles rod-shaped bacteria and has a length of 0.9-2.0 [micro]m and a width of 0.5-0.8 [micro]m. Figure 2 shows their SEM images. E. coli is not appropriate for aerosol tests, because it is rarely found in natural air (Karin and Hanley 1996). However, this bacterium is still used in experiments because of its safety for humans and its representative aerodynamic diameter for filtration mechanism research. To avoid the confusion caused by the similar appearance of E. coli colonies and bacteria colonies found in air, Serratia marcescens was also used in the experiment. They produce prodigiosin at 25[degrees]C-28[degrees]C (77[degrees]F-82.4[degrees]F); thus, S. marcescens colonies appear bright red, making them easily distinguishable from the detected bacteria colonies. Although S. marcescens has been found to be pathogenic to some people and, hence, is no longer recommended in schools for tracking bacterial movement, it was used in these experiments because appropriate safeguards were implemented to isolate the system interior from the laboratory air, as mentioned above.

[FIGURE 2 OMITTED] The indicating bacteria used in the experiments were incubated for at least five generations to maintain viability. According to ASTM Standard F2101-01 (ASTM 2001), a high-concentration suspension that is incubated in the orbital shaker for 48 hours should be diluted to about 5 x .sup.cfu/mL (8.20 x .sup. cfu/[in.sup.). The suspension was simply made with sterilized peptone water, which provided some nutrients to maintain the viability of the bacteria in the solution. A modified six-jet Collison nebulizer aerosolized the bacterial suspension into a polydisperse aerosol. The working pressure of the atomizer was about 0.05 MPa (7.25 psi), and the corresponding flow rate was about 6.67 L/min (0.236 cfm), with an aerosol generation rate of 0.1 mL/min (3.53 x .sup.- cfm). Since approximately 95% of water droplets contain only one bacterium, and the droplets evaporate quickly in a glass vessel, the size of the particles is close to that of the individual bacteria. The bacterial particles are mainly collected at stages 4-6 (0.65-3.3 [micro]m) of the Andersen sampler, and only a small portion of the particles impinge and stay at the first three stages. The aerodynamic diameters of E. coli and S. marcescens are about 0.87 [micro]m (Chen and Li 2005) and 1.14 [micro]m (Kowalski et al. 1999), respectively, which agree with the cut sizes of the sampler above.

For each measurement, the system was allowed to stabilize for at least five minutes before each sampling run. Generally, the sampling time ranged from 20-30 seconds for upstream sampling and 5-10 minutes for downstream sampling. At these durations, the plates did not become overloaded with sampled bacteria.

Air Filter Medium Four types of fibrous air filter media (A, B, C, and D) made in China were used in the experiments. The detailed specifications are presented in Table 1. The classifications of the filter media are based on EN779 and EN1882 (CEN 2002; CEN 1998). Types A and B are medium efficiency filter media to remove all particles more than 1 [micro]m, while Type C and D are HEPA media to remove small particles more than 0.3 [micro]m. The classification of the HEPA filter is according to the filtration efficiency at the most penetrating particulate size (MPPS).

Figure 3 shows their SEM images. Since these four glass fibrous filter media are commonly used in air-conditioning systems, it is hoped the results of the experiments will be useful for common air-treatment applications.

[FIGURE 3 OMITTED] RESULTS AND DISCUSSION The four filter media were tested at the system airflow rates of 10, 15, 20, and 25 for 32 L/min (0.35, 0.53, 0.71, and 0.88 for 1.13 cfm) with artificial bacterial aerosol and inert aerosol (e.g., DOP).

The particle size specific efficiencies were compared for analysis. In the experiment, the sampling time of the upstream and downstream were 30 seconds and 5 minutes, respectively, for the bioaerosol filtration, whereas the sampling time for the DOP filtration was 1 minute at the sampling flow rate of 2.83 L/min (0.1 cfm) both upstream and downstream.

The average particle count efficiencies of the filter media with DOP aerosol under the medium face velocity 0.053 m/s (0.174 ft/s) (airflow rate 32 L/min .13 cfm]) are shown in Table 2.

Bioaerosol Filtration Efficiency The filtration air velocity of a HEPA filter in typical commercial installations is about 0.53 m/s (1.74 ft/s) (CEN 1998; IEST 2005). In order to minimize the error from the testing velocity, the four filter media were all tested at the same media velocity of 0.53 m/s (1.74 ft/s) (airflow rate 32 L/min .13 cfm]) with E. coli and S. marcescens bioaerosols. The efficiencies are shown in Tables 3 and 4. Every efficiency noted is the average of three measurements under the same condition. Although a greater number of replications is preferred, because of the temporal instability ([ or -]10%) of the bioaerosol concentration, measurements needed to be completed in 30 minutes.

From the test results above, these two test bioaerosols provide similar filter efficiency trends for each of the filter media tested.

Samples A and B have relatively high efficiencies for bioaerosols. This means the medium efficiency air filters are suitable devices to remove biological particles from air in air-handling units (AHUs). Both C and D have high efficiencies close to 100% for removing bioparticles from the air.

All the efficiencies measured with E. coli are lower than those measured with S. marcescens, especially for HEPA media C and D. In measurements, bacteria from background environmental air, whose colonies are similar to those of E. coli, can easily be collected and counted as E. coli, causing errors. This has a more serious effect on downstream sampling than upstream sampling and, thus, results in a lower measured efficiency, especially for the higher efficiency samples. The experimental removal efficiencies of media C and D are 100% for S.

marcescens bioaerosol, and the test results have fewer errors because S.

marcescens is a good microbiological marker against the environmental background bacteria.

Bioaerosol removal efficiencies at 32 L/min (1.13cfm) for both C and D are 100%, and they are presumed higher at lower flow rates. Hence, measurements at 10, 15, 20, and 25 L/min (0.35, 0.53, 0.71, and 0.88 cfm) were made only for samples A and B. The average removal results of S. marcescens are shown in Table 5.

Table 5 shows that the efficiencies of the two filter media vary between 80%-85% and 97.5%-99.5%, respectively; however, no obvious variation relationship between efficiency and airflow rate can be noted.

The small variation is due to two reasons. First, aerodynamic diameters of the two microorganisms are 0.87 [micro]m and 1.14 [micro]m. For these sizes, diffusion is not the main filtration mechanism. Hence, particle residence time changing with air velocity hardly contributes to the particle removal. Secondly, though the dominant filtration mechanism for particles of this range is inertial collision, with low velocity ([less than or equal to]0.0533 m/s [[less than or equal to]0.175 ft/s]), as indicated in this study, the particle bounce is not considerable (Phillips et al. 1996), so the bacterial particle momentum is not very distinct. For these reasons and the existence of errors, variations in filter efficiencies are small.

Comparison of the Bioaerosol and DOP Filtration Bacterial particles are mostly collected at stages 4-6 (0.65-3.3[micro]m) of the Andersen sampler, since the aerodynamic diameters of both E. coli and S. marcescens are about 1 [micro]m. When testing with the DOP and OPC, particle count numbers from 0.5-2.0 [micro]m were used to compare with the concentrations of bioaerosols, because their geometrical even diameter is 1 [micro]m.

A theoretical model was employed (Dhaniyala and Liu 1999a, 199b) to predict the filtration efficiencies.

[eta] = 1.6[(-c]/[Ku]).sup./]P[e.sup.[-2/][C.sub.D] 0.6(-c]/[Ku])([R.sup./ R])[C.sub.R] (2) [C.sub.D] = 1 0.388Kn[(-c]/[Ku]Pe).sup./] (3) [C.sub.R] = 1 [.966Kn]/R] (4) Ku = -/lnc-/ c-[[c.sup./ (5) Kn = [[lambda]]/[d.sub.f]] (6) where [eta] = efficiency of the filter medium c = fiber volume fraction, 0.080 for A and 0.071for B filter medium Ku = Kuwabara's hydrodynamic factor Pe = Peclet number, a function of velocity Kn = Knudsen number R = interception parameter [C.sub.D], [C.sub.R] = slip correction factors [lambda] = mean free path of the gas molecules (= 0.06542 pm at 25[degrees]C [degrees]F]) [d.sub.f] = fiber diameter, 4 [micro]m for A and 3 [micro]m for B filter medium Figures 4 and 5 show the theoretical results of filter medium efficiency for 1 [micro]m particles compared with those of testing using DOP (0.5-2.0 [micro]m) and bioaerosol (S. marcescens) as challenges, respectively. No obvious relationship between efficiency and velocity can be found. The tested efficiency results do not significantly deviate from the theoretical value, which indicates that the theoretical model is relatively reliable.

[FIGURE 4 OMITTED] [FIGURE 5 OMITTED] The efficiency for bioaerosols is larger than that for DOP and the theoretical efficiency. The reason may be the integrity and the shape of the bacteria. Considering the integrity of the bacteria, when they collide with the fibers, they are not broken up into smaller particles.

Hence, these particles are effectively intercepted by inertial collision. DOP droplets are not as strongly held together as bacteria.

Hence, one large droplet may split into small ones inside the depth of the filter medium. These small particles may then more readily be re-entrained in the airflow and penetrate easily through the filter medium. This may explain why the two efficiencies are closer to each other at low velocities than at high velocities. Another possible reason for the efficiency difference lies in the shape of the bacteria used.

Rod-shaped bacteria may be intercepted much more easily.

Due to the potential risks from the viability of microorganisms, it is necessary to consider that bioparticles or their fragments may re-entrain into the air and penetrate through the filter media. Any bioaerosol that passes through the air filter may have adverse effects on IAQ; hence, other methods, such as sterilization, in addition to filtration, may be required.

CONCLUSION A test rig was developed for testing filter media with both bioaerosol and nonbioaerosol challenges. The test system includes generators and sampling devices. It is possible to label the bioaerosol removal efficiency of different kinds of filter media by the test system.

Two microbiological markers were employed in the test to make the effect of the environmental background microorganisms clear and to distinguish the factors that influence the filtration efficiency.

Similar efficiency trends were obtained with the two microbiological aerosols, which demonstrates the test data was reliable. However, the efficiency of the filtration on E. coli bioaerosol is lower than that of S. marcescens. The reason for this may be the errors in counting the E.

coli colonies when background environmental microorganisms from the air have the same colony appearances. Therefore, S. marcescens is recommended in the filtration test for its great advantages in distinguishing a target colony from the background microorganisms.

The results show a higher efficiency for bioaerosol than for DOP or the theoretical efficiency. The results also indicate that medium efficiency air filters are suitable for filtering biological particles in AHUs. The F8 medium efficiency air filter is the best for most bioaerosol removal. The monotonic relationship between the higher efficiency of DOP and that of bioaerosols suggests the filter efficiency measured with DOP particles of 1 [micro]m may be useful in predicting the removal efficiency of bioaerosols for the filter medium.

ACKNOWLEDGMENTS We would like to express our thanks for the funding from China National Key Technologies R&D Program: 2008BAI62B01.

REFERENCES Andersen, A.A. 1958. New sampler for the collection, sizing, and enumeration of viable airborne particles. Journal of Bacteriology 76:471-84.

ASTM. 2001. F2101-01, Standard Test Method for Evaluating the Bacterial Filtration Efficiency (BFE) of Medical Face Mask Materials, Using a Biological Aerosol of Staphylococcus aureus. USA: American Society for Testing and Materials Designation.

Brosseau, L.M., S.K. Chen, D. Vesley, and J.H. Vincent. 1993.

System design and test method for measuring respirator filter efficiency using mycobacterium aerosols. Journal of Aerosol Science 25(8):1567-77.

CEN. 1998. EN1822, High Efficiency Air Filters (HEPA and ULPA)--Part3, Testing Flat Sheet Filter Media. Belgium: European Committee for Standardization.

CEN. 2002. EN779:2002E, Particulate Air Filters for General Ventilation--Determination of the Filtration Performances. Belgium:

European Committee for Standardization.

Chen, P.S., and C.S. Li. 2005. Bioaerosol characterization by flow cytometry with fluorochrome. Journal of Environmental Monitoring 7(10):950-59.

Chen, S.K., D. Vesley, L.M. Brosseau, and J.H. Vincent. 1994.

Evaluation of single-use masks respirators for protection of health care workers against Mycobacterial aerosols. American Journal of Infection Control 22(2):65-74.

Cheng, Y.S., and C.S. Wang. 1981. Motion of particles in bends of circular pipes. Atmospheric Environment 15:301-306.

Dhaniyala, S., and B.Y.H. Liu. 1999. Investigations of particle penetration in fibrous filters--Part I. Experimental. Journal of the IEST 42(1):32-40.

Dhaniyala, S., and B.Y.H. Liu. 1999. Investigations of particle penetration in fibrous filters--Part II, Theoretical. Journal of the IEST 42(2):40-6.

IEST. 2005. IEST-RP-CC001.4, HEPA and ULPA Filters. USA: Institute of Environmental Sciences and Technology.

Karin, K.F., and J.T. Hanley. 1996. Determine the efficiency of antimicrobial treatments of fibrous air filters. ASHRAE RP-909 Final Report, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., Atlanta.

Kowalski, W.J., W.P. Bahnfleth, and T.S. Whittam. 1999. Filtration of airborne microorganisms: Modeling and prediction. ASHRAE Transactions 105(2):4-17.

Kowalski W.J., and W.P. Bahnfleth. 2002. Airborne-microbe filtration in indoor environments. HPAC Engineering 74(1):57-69.

Phillips, B. Alan, W.T. Davis, and M. Dever. 1996. Investigation of the effect of a topically applied tackifier in reducing particle bounce in a melt-blown air filter. Filtration and Separation 3(10):933-39.

Qian, Y., K. Willeke, S.A. Grinshpun, J. Donnelly, and C.C. Coffey.

1998. Performance of N95 respirators: Filtration efficiency for airborne microbial and inert particles. American Industrial Hygiene Association Journal 59(2):128-32.

Washam C.J. 1966. Evaluation of filters for removal of bacteria phages from air. Applied Microbiology 14(6):497-505.

Junjie Liu, PhD Ruiying Qi Quanpeng Li Guiyuan Han Jiancheng Qi, PhD Received February 22, 2008; accepted May 27, 2009 Junjie Liu is an associate professor, Ruiying Qi, Quanpeng Li, and Guiyuan Han are graduate students of built environment engineering, School of Environmental Science and Technology, Tianjin University, and Jiancheng Qi is a research fellow and vice director of the National Biological Protection Engineering Center, Tianjin, China.