ArticlePDF Available

Viral Penetration of High Efficiency Particulate Air (HEPA) Filters (PREPRINT)

  • Air Force Civil Engineer Center, Tyndall Air Force Base, United States

Abstract and Figures

High Efficiency Particulate Air (HEPA) filters are the primary technology used for particulate removal in individual and collective protection applications. HEPA filters are commonly thought to be impenetrable, but in fact they are only 99.97% efficient at collecting the most-penetrating particle (approx. 0.3 micrometer). While this is an impressive collection efficiency, HEPA filters may not provide adequate protection for all threats: viruses are submicron in size and have small minimum infections doses (MID50). Thus, an appropriate viral challenge may yield penetration that will lead to infection of personnel. However, the overall particle size (agglomerated viruses and/or viruses attached to inert carriers) will determine the capture efficiency of the HEPA filter. Aerosolized viruses are commonly thought to exist as agglomerates, which would increase the particle size and consequently increase their capture efficiency. However, many of the threat agent viruses can be highly agglomerated and still exist as submicron particles. We have demonstrated that MS2 coli phage aerosols can penetrate Carbon HEPA Aerosol Canisters (CHAC). At a face velocity of 2 cm/sec, a nebulized challenge of approx. 105 viable plaque forming units (PFU) per liter of air results in penetration of approx. 1 - 2 viable PFU per liter of air. We are currently investigating the particle size distribution of the MS2 coli phage aerosol to determine if the challenge is tactically relevant. Preliminary results indicate that 200-300-nanometer particles account for approx. 7.5% of the total number of particles. Our aim is to characterize multiple aerosol conditions and measure the effects on viable penetration. This study will expand our knowledge of the tactical threat posed by viral aerosols to HEPA filter systems.
Content may be subject to copyright.
Brian K. Heimbuch
Applied Research Associates
P.O. Box 40128
Tyndall Air Force Base, FL 32403
C. Y. Wu
Department of Environmental Engineering Sciences
University of Florida
Gainesville, FL
Joseph D. Wander
Air Force Research Laboratory
DISTRIBUTION STATEMENT A: Approved for public release;
distribution unlimited.
Standard Form 298 (Rev. 8/98)
Prescribed by ANSI Std. Z39.18
Form Approved
OMB No. 0704-0188
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,
gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of
information, including suggestions for reducing the burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188),
1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any
penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number.
19b. TELEPHONE NUMBER (Include area code)
Journal Article PREPRINT
01-JAN-2006 -- 31-DEC-2007
Viral Penetration of High Efficiency Particulate Air (HEPA) Filters
* Heimbuch, Brian K.; Wu, J. D.; Wander, Joseph D.
* Applied Research Associates, P.O. Box 40128, Tyndall Air Force Base, FL 32403
+ University of Florida, Department of Environmental Engineering Sciences,
Gainesville, FL
Air Force Research Laboratory
Materials and Manufacturing Directorate
Airbase Technologies Division
139 Barnes Drive, Suite 2
Tyndall Air Force Base, FL 32403-5323
Distribution Statement A: Approved for public release; distribution unlimited.
Ref AFRL/RXQ Public Affairs Case # 09-141. Submitted for publication in the Applied Research Associates, Inc technical journal.
Document contains color images.
High Efficiency Particulate Air (HEPA) filters are the primary technology used for particulate removal in individual and collective protection
applications. HEPA filters are commonly thought to be impenetrable, but in fact they are only 99.97% efficient at collecting the most-penetrating
particle (~0.3 micrometer). While this is an impressive collection efficiency, HEPA filters may not provide adequate protection for all threats:
viruses are submicron in size and have small minimum infections doses (MID50). Thus, an appropriate viral challenge may yield penetration that
will lead to infection of personnel. However, the overall particle size (agglomerated viruses and/or viruses attached to inert carriers) will determine
the capture efficiency of the HEPA filter. Aerosolized viruses are commonly thought to exist as agglomerates, which would increase the particle
size and consequently increase their capture efficiency. However, many of the threat agent viruses can be highly agglomerated and still exist as
submicron particles. We have demonstrated that MS2 coli phage aerosols can penetrate Carbon HEPA Aerosol Canisters (CHAC). At a face
velocity of 2 cm/sec, a nebulized challenge of ~105 viable plaque forming units (PFU) per liter of air results in penetration of ~1 -2 viable PFU per
liter of air. We are currently investigating the particle size distribution of the MS2 coli phage aerosol to determine if the challenge is tactically
relevant. Preliminary results indicate that 200-300-nm particles account for ~7.5% of the total number of particles. Our aim is to characterize
multiple aerosol conditions and measure the effects on viable penetration. This study will expand our knowledge of the tactical threat posed by
viral aerosols to HEPA filter systems.
pathogens, airborne, spores, aerosol, filtration, viral, infectious, influenza
Joseph Wander
Viral Penetration of HEPA Filters 1
Heimbuch*, B.K1, C.Y. Wu2, and J.D. Wander3
1Applied Research Associates, ESD, Tyndall Air Force Base, FL 4
2University of Florida, Dept. of Environmental Engineering Sciences, Gainesville, FL 5
3Air Force Research Laboratory, RXQL, Tyndall Air Force Base, FL 6
Abstract 8
High-efficiency particulate air (HEPA) filters are the primary technology used for 10
particle removal in individual and collective protection applications. HEPA filters are 11
commonly thought to be impenetrable, but in fact they are only 99.97% efficient at 12
collecting the most-penetrating particle (~ 0.3 micrometer). While this is impressive 13
collection efficiency, HEPA filters may not provide adequate protection for all threats: 14
viruses are submicron in size and have small median infectious doses (MID50). Thus, an 15
appropriate viral challenge may yield penetration that will lead to infection of personnel. 16
The overall particle size (agglomerated viruses and/or viruses attached to inert carriers) 17
will determine the capture efficiency of the HEPA filter. Aerosolized viruses are 18
commonly thought to exist as agglomerates, which would increase the particle size and 19
consequently increase their capture efficiency. However, many of the threat agent viruses 20
can be highly agglomerated and still exist as submicron particles. We have demonstrated 21
that MS2 coli phage aerosols can penetrate carbon–HEPA aerosol canisters (CHACs). At 22
a face velocity of 2 cm/sec a nebulized challenge of ~105 viable plaque-forming units 23
(PFU) per liter of air results in penetration of ~12 viable PFU per liter of air. We are 24
currently investigating the particle size distribution of the MS2 coli phage aerosol to 25
determine if the challenge is tactically relevant. Preliminary results indicate that 200–300 26
nm particles account for ~7.5% of the total number of particles. Our aim is to characterize 27
multiple aerosol conditions and measure the effects on viable penetration. This study will 28
expand our knowledge of the tactical threat posed by viral aerosols to HEPA filter 29
Introduction 32
Biological Warfare/Terrorism is defined as actual or threatened deployment of biological 34
agents to produce casualties or disease in man or animals and damage to plants or 35
material. It is actually much farther reaching than that because contamination of 36
infrastructure, which does directly affect individuals, is a concern due to the extensive 37
and costly clean up required. The potential of biological weapons was demonstrated early 38
in world history (Hawley 2001) starting in the 14th century when plague-infected 39
carcasses were catapulted into enemy cities in an effort to spread the disease. Also, 40
during the French and Indian war in 1754–1767, British soldiers provided American 41
Indians with smallpox- contaminated blankets and handkerchiefs. These events predate 42
Louis Pasteur’s discovery that infectious diseases are caused by microorganisms, and 43
clearly root biological agents as mans first attempt at creating a Weapon of Mass 44
Destruction (WMD). Once microorganisms were linked to human disease, it did not take 45
long for purified microbes to be used as weapons. It is well documented that many 46
countries, including the United States, had extensive bioweapons programs (Gronvall 47
2005, Frischknecht 2003). Perhaps the most feared was that of the Soviet Union. Human 48
history is littered with many examples of microbes being deployed as acts of war and 49
terrorism, the most recent documented example being the attack on the Hart Building in 50
2001. This single act of bioterrorism clearly demonstrated the potential threat that 51
biological agents pose as a weapon of terror. 52
Biological agents are classified into four unique categories: vegetative bacterial cells, 54
spores, viruses, and toxins; viruses are the primary concern in this report. Although the 55
viral warfare agents are diverse and cause a variety of diseases, their physical properties 56
are similar (Woods 2005): all contain a nucleic acid core surrounded by a protein coat; 57
most also contain a lipid membrane, and are termed enveloped. Viruses are submicron 58
particles, ranging in size from ~25–400 nm (Hogan 2005, Kowalski 1999) and the 59
median infectious dose (MID50) for all the threat agent viruses is very low. While 60
absolute figures are not available, most believe that the MID50s are less than ten virions 61
(Woods 2005). The combination of small size and low infectious doses raises concern 62
that high-efficiency particulate air (HEPA) filters may not adequately protect individuals 63
from viral WMD. 64
HEPA filters are commonly used in individual and collective protection applications and 66
are very efficient at removing particulate matter from the air. They are rated to be 99.97% 67
efficient at collecting the nominal most-penetrating particle (0.3 µm) (Lee 1980). 68
Although this collection efficiency is impressive, it is not absolute; depending on 69
conditions, 0.03% of matter at the most penetrating size does penetrate the HEPA filter. 70
For most applications the HEPA is adequate, but tolerance for viral penetration is very 71
low, and thus only a few penetrating virions may be enough to cause disease. For viruses 72
to be efficient at penetrating HEPA filters they must remain as submicron particles. Most 73
agree that viruses will not occur as singlets when dispersed in an aerosol; rather, they will 74
agglomerate or attach to inert materials that will increase the particle sizes (Stetzenbach 75
1992). It is important to note, however, that many of the threat agent viruses (e.g., SARS, 76
EEV) can be significantly agglomerated and still fall into the most-penetrating range. 77
Most of the research on bioaerosols has focused on naturally occurring biological 78
aerosols. The research has demonstrated that a majority of particles in biological aerosols 79
are greater than 1µm in size (Stetzenbach 1992), and thus would not be a threat to 80
penetrate HEPA filters. It should be noted that the technology used in these studies is not 81
able to effectively measure bioparticles smaller than 500 nm. Therefore, the abundance of 82
particles that would be most efficient at penetrating HEPA filters was not properly 83
quantified. Studies of naturally occurring particulate aerosols (non-biological) 84
demonstrate that nanometer-size particles are actually abundant (Biswas and Wu 2005). 85
Weaponized viruses are clearly different from naturally occurring biological aerosols and 87
the particle size for viral weapons is not clearly defined. From a weapons standpoint, it 88
would be advantageous to create smaller particles, because they would remain 89
aerosolized longer. But in addition to creating small particles one must preserve the 90
viability of the viruses. The methods used to produce and protect viruses from 91
environmental stress may dictate creating larger particles. It is unclear if weaponized 92
viruses have been created that are submicron in size. This uncertainty has fueled 93
speculation that viruses may indeed be a threat to penetrate HEPA filters. 94
The study of viral penetration of HEPA filters dates back to the development of HEPA 96
filters by the Department of Energy (DOE) in the 1950s (Mack, 1957). Since that time 97
more than 20 published studies have used a variety of experimental techniques to 98
quantify viable penetration of HEPA filters. A comprehensive review of these studies 99
edited by Wander is due to be published in 2010. Six studies (Decker 1963, Harstad 100
1967, 1969, Roelants 1968, Thorne 1960, and Washam 1966) were published in the 101
1960s; all were chamber tests aimed at determining the viable filtration efficiency of the 102
media and/or devices. The most elegant of these studies were carried out by Harstad, who 103
observed that the principal route of penetration is filter defects (pinhole leaks, media 104
breaks due to pleating, etc.) and not through the medium itself. The next 30 years 105
produced only eight research articles, six chamber tests (Bolton 1976, Dryden 1980, Eng 106
1996, Leenders 1984, Rapp1992, and Vandenbroucke–Grauls1995), and two studies that 107
used an animal model (Burmester 1972, Hopkins1971) to assay the protection provided 108
by HEPA filters. The turn of the 21st century saw a renaissance of interest in research on 109
viral penetration of HEPA media—a total of seven articles were published in seven years. 110
Research on active processes for air purification (reactive/antimicrobial media, heat, 111
energetic light, etc.) that kill microbes rather than just capture them was the main driver 112
for these studies (Heimbuch 2004, Lee 2008, Ratnesar 2008, and RTI 2006). Dee et al
(2005, 2006a, 2006b) also performed three studies using a swine model to determine the 114
effectiveness of HEPA filters 115
The review of all research studies dating back to Mack’s report in 1957 reveals a 117
common theme: HEPA filters provide HEPA-level performance (> 99.97% efficiency), 118
which was duly noted by the authors. Many of these authors could also have concluded 119
that their studies demonstrated that viable viruses penetrate HEPA filters at levels that 120
may cause disease. The purpose of this report is to reanalyze the issues surrounding viral 121
penetration of HEPA filters, and to shed new light on the potential for penetration. 122
Furthermore, the protection afforded by the carbon HEPA aerosol canister (CHAC) is 123
also specifically addressed. We demonstrated (Heimbuch 2004, Figure 1) in previous 124
studies that viable MS2 coli phage can penetrate CHACs. However, these studies did not 125
discriminate between penetration due to viruses passing through the HEPA medium and 126
due to viruses bypassing the medium through defects in the canisters. In this study, the 127
viral simulant MS2 coli phage was used to challenge both flat-sheet HEPA material and 128
CHACs. Both viable penetration and total penetration were measured. In addition, 129
particle size distribution and filtration velocity were varied to measure what effect each 130
had on total and viable penetration. 131
Materials and Methods 133
Microorganisms: MS2 coli phage (ATCC 15597-B1) stock solutions were prepared by 135
infecting 100 mL of the Escherichia coli host (ATCC 15597) that was grown to mid-log 136
phase in special MS2 medium (1% tryptone, 0.5% yeast extract, 1% sodium chloride, .01 137
M calcium chloride, 0.002% thiamine). The infected culture was incubated overnight @ 138
37ºC/220 rpm. Lysozyme (Sigma, L6876) was added to a final concentration of 50 139
µg/mL and the flask was incubated for 30 minutes at 37ºC. Chloroform (0.4%) and 140
EDTA (.02 M) were then added and the culture was incubated for an additional 30 141
minutes at 37°C. Cell debris was removed by centrifugation at 10,000 X g, then the 142
supernatant was filtered thorough a 0.2-μm filter and stored at 4ºC. A single-layer plaque 143
assay was performed according to standard procedures (EPA) to determine the MS2 titer, 144
which typically is ~1011 plaque-forming units (PFU)/mL. For aerosol studies, the MS2 145
coli phage was diluted in either sterile distilled water or 0.5% tryptone to a concentration 146
of ~108 PFU/mL. 147
Aerosol Methods: The BioAerosol Test System (BATS, Figure 2) is a port-accessible 149
aerosolization chamber communicating with a temperature/humidity-controlled mixing 150
plenum and thence to a sampling plenum supplying a homogeneous aerosol to six 151
sampling ports. Three six-jet Collison nebulizers (BGI Inc, Waltham, Mass.) deliver 152
droplets at the source that are ~2 µm mass median diameter into the mixing plenum to 153
create the bioaerosols. Air is drawn into a central vacuum line along a path from the 154
sampling plenum through lines of PVC tubing (Excelon® RNT, US Plastics, Lima, 155
Ohio). Each path runs through a test article and thence through one AGI-30 all-glass 156
impinger (Chemglass, Vineland, N.J.) filled with 20 mL of 1X phosphate buffer 157
saline/0.001% antifoam A (Sigma, A6457). The volume of air passing in each path is 158
controlled by a rotameter (Blue–White 400, Huntington Beach, California, or PMR1-159
101346, Cole–Parmer, Vernon Hills, Illinois). At the end of the sampling path, the air 160
exhausts through a conventional HEPA filter and the vacuum pump that drives the air 161
movement. Each sampling port is able to accommodate test articles as large as 6 inches 162
(15 cm) in diameter. 163
The BATS was configured three separate ways depending on what was being tested 165
(Figure 3). In each case, the total flow through each port of the BATS was set to 85 liters 166
per minute (LPM). The environmental conditions for all tests were ~22°C and 50% 167
relative humidity. For flat-sheet HEPA testing, a portion of the flow was split off the 85-168
LPM flow and directed through the HEPA material (Lydall; Manchester, Conn.; part 169
number 4450HS) that was compression seated and glued into swatch holders (Figure 3). 170
For CHAC tests the entire 85-LPM flow was drawn though the CHAC, but only 12.5 171
LPM was collected in the AGI-30 impinger (Figure 3). For each test a portion of the 172
flow was directed through a model 3936 Scanning Mobility Particle Sizing Spectrometer 173
(SMPS) (TSI Inc, Shoreview, Minn.) that was configured to analyze particles with a 174
diameter of 10 nm – 415 nm. The sample flow through the SMPS was 0.6 LPM with a 175
sheath flow rate of 6 LPM. 176
Viable enumeration of MS2 coli phage was achieved by performing a plaque assay on the 178
collection fluid from each AGI-30 impinger. One mL of solution from each impinger was 179
mixed with 1 mL of log-phase E. coli grown in special MS2 medium. This solution was 180
then mixed with 9 mL of semi-solid medium (special MS2 medium + 1% agar) that had 181
been incubated at 55°C. The solution was poured into sterile Petri dishes and allowed to 182
solidify. The plates were incubated at 37ºC overnight, then plaques were counted. The 183
total collected phage for each impinger was determined using the following formula: 184
Total PFU = counted PFU x dilution-1 x impinger volume 186
Experimental Plan: At each condition tested in this study, six samples were challenged 188
with MS2 coli phage over two days of testing: three samples and one positive control 189
were analyzed each day. After the filters were seated into the swatch holders they were 190
initially leak checked by challenging with an aerosol of 100-µm beads for 5 minutes. 191
After the leak test the BATS was loaded with MS2 coli phage and equilibrated for 15 192
minutes prior to starting the challenge. The challenge comprised four 15-minute intervals, 193
in which new impingers were installed after each interval. The SMPS incrementally 194
analyzed penetration for each of the four swatch holders (three filters and one positive 195
control) for 12.5 minutes of each 15-minute challenge period. 196
Explanation of flow rates and face velocity: The coupon samples used for this study 198
were all 4.7-cm diameter circles, resulting in a surface area of 17.34cm2. The flow rate 199
through each filter was 2 LPM, 4 LPM, 6 LPM, or 8 LPM. Face velocities were 200
calculated using the following formula: 201
Face velocity (cm/sec) = flow rate (cm3/sec) ÷ surface area (cm2) 203
The resulting face velocities were numerically equal to the flow rate (i.e., 2 LPM rate = 2 205
cm/sec face velocity, 4 LPM flow rate = 4 cm/sec face velocity, etc). For the CHAC the 206
entire surface area of the pleated HEPA filter was taken into account when calculating the 207
face velocity. The CHACs used in this study contained 750 cm2 of HEPA medium that 208
was tested at a flow rate of 85 LPM. The resulting face velocity, using the above formula, 209
was 2 cm/sec. 210
Results 212
Size distribution of MS2 aerosols in the BATS: The SMPS analysis of MS2 aerosols 214
created in the BATS revealed that the number mode diameter was ~35 nm and the mass 215
mode diameter was ~ 151 nm (Figure 4). Both are composed of distributions that span the 216
entire data collection range of the SMPS. By number, the fraction of particles that fall 217
into the most-penetrating range for HEPA filters (100–300 nm) was only 7.5%. The 218
curve for the mass distribution is not complete, but if we assume the curve is 219
symmetrical, a reflection around the midpoint indicates that only 94% of the curve is 220
represented by the data. The correction reveals that the amount of mass in the 100–300 221
nm range is 58%. Both number distribution and mass distribution of particles have been 222
used by researchers for determining filter efficiency, but it is unclear which is more 223
appropriate. For this analysis, the mass distribution specifies a much more stringent 224
challenge for HEPA filters than does the number distribution. 225
Particulate penetration of flat sheet HEPA filters: The SMPS analysis (number and 227
mass distributions) of the MS2 aerosols confirmed that the particle distributions and 228
overall challenge levels for each flow rate were similar (Figure 5). This indicates a high 229
degree of repeatability in the experimental setup. Penetration of particles through the 230
HEPA filter increased as flow rate increased (Figure 5). This indicates the HEPA filter 231
becomes less efficient with increasing flow rate, as expected in size regions in which 232
diffusional capture mechanisms dominate. At the low challenge concentrations 233
(beginning and end of curves) the penetration data disappeared into the background and 234
thus were not meaningful. When particle penetration experiments are done for HEPA 235
filters, the particle challenge concentration is orders of magnitude greater than what can 236
be created for biological challenges. Thus the signal-to-noise ratio is much larger. 237
Analysis of penetration efficiency demonstrates that the most-penetrating particle (MPP) 238
at the higher velocities is ~ 135 nm (Figure 6). The lower flow rates have limited overall 239
penetration and an MPP size can not be discriminated. The MPPs for HEPA filters are 240
commonly believed to be 300 nm, but it is actually closer to 200 nm (Lee 1980). The 241
smaller MPP observed in this study is likely due to the higher flow velocities used in this 242
study. 243
Viable MS2 penetration of flat-sheet HEPA filters: The viable MS2 penetration data 245
indicate that as the flow rate increases, penetration through the HEPA also increases 246
(Figure 7); this is in perfect agreement with the SMPS data. The difference in viable 247
penetration increased ~1 log10 order of magnitude as the flow rate doubled. The increase 248
in average penetration between the 2-cm/sec and 4-cm/sec velocity was just shy of the 1 249
log10 mark; this may be attributed to the overall low number of plaques detected for the 2 250
cm/sec assay. Also, the addition of the 4-LPM purge may have added additional 251
variability. The overall viable penetration values are lower than what is reported for the 252
particulate data. The reason for this is unclear, but viable assays are complex in 253
comparison to the SMPS analysis. The SMPS measures all particles regardless of 254
whether or not they are viable or even contain a virus. The viable assay measures only 255
viable MS2 particles. The differences in penetration between the assays indicate that 256
viable MS2 is not evenly distributed across the entire particle size distribution. 257
Particle penetration of CHACs: The penetration of particles through the CHAC tracked 259
most closely with the HEPA penetration data at 2 cm/sec (Figure 5). This was expected 260
because the test flow rate of 85 LPM through the CHAC provides a velocity of 2 cm/sec 261
through the CHAC HEPA filter. Analysis of the filtration efficiency (Figure 6) 262
demonstrates that penetration through the CHAC also follows the penetration observed 263
for flat- sheet HEPA material at velocities of 2 cm/sec and 4 cm/sec. The overall 264
penetration was very low and a determination of MPP size was not possible. 265
Viable MS2 penetration of CHACs: MS2 penetration of the CHAC canister was lower 267
than through any of the flat-sheet HEPA materials tested (Figure 7 and Table 1). The 268
penetration most closely resembled that at 2 cm/sec velocity through the HEPA, as was 269
expected due to similar face velocities, but the total measured penetration was only 1/7 of 270
that through the flat sheet HEPA medium. The decrease in penetration through the CHAC 271
was likely due to the presence of the carbon bed. The carbon bed adds more surface area 272
for the aerosol to travel through, which could mechanically trap the MS2 particles. 273
However, the SMPS analysis demonstrated the particle collection efficiency of the 274
CHAC was very similar to the collection efficiency of the HEPA at the same velocity (2 275
cm/sec) (Figure 6). Thus, other mechanisms must be responsible for the viable reduction. 276
One possibility is that the additive ASZM-TEDA (AntimonySilverZinc–Molybdenum–277
Triethylenediamine) in the carbon bed is exerting a biocidal effect on the bacteriophage. 278
ASZM-TEDA is added to the carbon to prevent microbial growth and it may have 279
virucidal activity as well. 280
Particulate penetration of 0.5% tryptone nebulization solution: The addition of 282
tryptone (0.5%) to the nebulization fluid significantly shifted the size distribution of 283
particles to the right (Figure 8). The number mode diameter shifted to ~89 nm and the 284
mass mode diameter shifted to ~300 nm; the percentage of particles, by number, that fell 285
into the 100–300 nm size range also increased by 28.5%. The mass curve was not 286
complete, and thus the fraction of particles in the 100–300 nm size range could not be 287
definitively calculated. However, if we assume the curve to be symmetrical the mass 288
present in the 100–300 nm size range is 43%, a decrease of 15% over what is observed 289
for MS2 suspended in water. The overall numbers of particles generated by MS2 290
nebulized in 0.5% tryptone and MS2 nebulized in water were not significantly different. 291
The reason for this is that the output of droplets from the Collison nebulizer is constant 292
regardless of what is being nebulized, so the addition of tryptone to the nebulizer did not 293
affect the rate of generation of particles but rather altered the composition of the droplets. 294
The increase in dissolved solids in each droplet produced by the Collison thus 295
dramatically increased the total mass, with the net result that the MS2 coli phage was 296
significantly loaded with protein. Delivery of the extra mass caused the HEPA filters to 297
load with tryptone and they become more efficient over time (Figure 9). Filter loading 298
was not observed for MS2 suspended in water, and penetration remained constant during 299
our experiments. 300
Viable MS2 penetration of 0.5% tryptone nebulization solution: The addition of 302
tryptone to the nebulizer did not positively or negatively influence the viability of MS2 303
coli phage (Figure 10): both conditions of delivery yielded approximately the same 304
concentration of viable MS2, but the addition of tryptone caused a significant decrease in 305
penetration of MS2 coli phage through the HEPA filter over the entire sampling times 306
(Figure 10). The initial decrease in viable penetration (Figure 10) was likely caused by 307
the shift in particles away from the most penetrating size (Table 2). The mass distribution 308
showed a 15% decrease in particles in MPP size, but the number distribution showed an 309
increase of 28.5% MPP size. It would appear that the mass distribution is more relevant 310
than the number distribution for determining viable penetration by MS2. Viable MS2 311
penetration also decreased over time and tryptone loading of the HEPA filter was likely 312
responsible. No pressure drop measurements were made, but an increase in pressure loss 313
with time would have been expected. 314
Discussion 316
Data presented in this report conclusively demonstrate that viable viruses can penetrate 318
HEPA filters. This should not be surprising given the fact that HEPA filters are rated to 319
be only 99.97% efficient at collecting 0.3-µm particles. Hence, given a sufficient 320
challenge, penetration is a mathematical expectation. The penetration is small relative to 321
the challenge, and for most particulate challenges this minimal penetration is not 322
problematic. Viruses, however, pose a unique problem because very few virions are 323
required to cause an infection (MID50 < 10 PFU). This problem is further exacerbated 324
because viruses are very small (25–400 nm), so individual viruses, and aggregates of 325
viruses fall into the MPP range of HEPA filters. The data in this report were gathered 326
from carefully controlled laboratory experimentssuch an approach was necessary to 327
evaluate viable penetration efficiency of HEPA filters. The tactical relevance of these 328
data is a more-challenging problem because no criteria are available to determine that the 329
BATS challenge is—or is not—representative of a biological attack. To determine if viral 330
penetration of HEPA filters is a potential concern, four characteristics of viral aerosols 331
must be considered: 1) Filtration velocity (flow rate), 2) Virus concentration, 3) Duration 332
of the biological attack, and 4) Particle size. Each of these characteristics (discussed 333
below) will significantly impact viral penetration of HEPA filters, and ultimately 334
determine that HEPA filters do or do not provide “complete protection” against 335
respiratory infection by airborne viruses. 336
The concentration of viruses created during a biological attack is not known. The 338
concentration will likely vary depending on distance from the distribution source. The 339
measured concentration of viruses for this study was only 104–105 PFU per liter of air. 340
These concentrations are not excessively high and are likely lower than what would be 341
generated during a biological attack. The duration of time that this concentration can be 342
maintained is also an important parameter, as it directly relates to time of exposure. 343
While there is no clear answer to this question, we do know that the penetration data 344
observed in this study were approximately linear over time. Therefore we can predict that 345
penetration occurs instantaneously. This may be surprising to some but HEPA filters are 346
an “open system” that contains holes. The SMPS analysis of HEPA penetration, which 347
was measured over the duration of the challenge, confirms that particle penetration 348
occurs instantaneously during a challenge. These data indicate that, given an appropriate 349
challenge, an infective dose of viruses could be delivered in a matter of seconds 350
following a challenge. 351
Flow rate and ultrafine particle penetration are directly related. As flow rate increases, 353
penetration near and below the MPP size will increase. HEPA filters are commonly rated 354
for a face velocity of 3.5 cm/sec to maintain the 99.97% collection efficiency and 355
maximum pressure drop ratings. (Liu 1994, VanOsdell 1990). Our study confirms this, 356
demonstrating that the 4-cm/sec velocity is the cutoff for obtaining HEPA performance 357
for particle penetration. Viable MS2 coli phage penetration also increases with flow rate, 358
with a significant increase in penetration at the higher velocities. For individual 359
protection applications, the National Institute for Occupational Safety and Health 360
(NIOSH) recommends a testing flow rate at 85 LPM; that equates to a 2-cm/sec filtration 361
velocity for CHACs. However, breathing is more complex than simply testing at a 362
uniform flow rate. Cyclic breathing will obviously allow penetration only during 363
inhalation, and the most penetration will occur during peak flow velocities. Anderson et 364
al (2006), demonstrated that maximum peak flows for average males range from 125 365
LPM to 254 LPM depending on work load (light to heavy). Peak flow was cyclic and 366
accounted for ~ ½ the total time tested. This indicates that an average male can inhale 367
particles at velocities greater than the rated velocities for HEPA filters. 368
The particle size distribution for this study was very small and may not be representative 370
of a viral weapon attack; only 7.5% of the particles by number fell into the most-371
penetrating range. In an effort to shift the particle distribution to the right, tryptone was 372
added to the nebulization fluid. This generated more particles (by number) in the most- 373
penetrating range (Figure 8, Table 2), but the net result was a decrease in viable 374
penetration (Figure 10). The result is counterintuitive, but if one considers the mass data, 375
which showed a decrease in particles in the MPP size range (Table 2), then a decrease in 376
viable penetration would be expected. Furthermore, the addition of tryptone caused a 377
decrease in the production of particles with diameters ranging from 10 nm–100 nm 378
(Table 2). Diffusional capture, which becomes less efficient as velocity increases, is 379
responsible for collecting particles in this size range. The comparison of aerosolization of 380
MS2 in tryptone solution vs. water was done only at 8 cm/sec velocity; thus the 381
efficiency of diffusional capture was reduced, resulting in more penetration for the water 382
aerosolization, but not significantly impacting the tryptone aerosolization. These 383
combined factors contributed to a 2-log decrease in penetration of viable MS2 virions. 384
The viable penetration was further decreased over time, as a result of tryptone loading the 385
HEPA filter and increasing the efficiency of the filter. The SMPS data clearly shows the 386
time-based increase in filter efficiency for the tryptone aerosolization, but not for the 387
water aerosolization (Figure 9). 388
The distribution of MS2 virions among inert particles is an important parameter that will 390
affect viable penetration of HEPA filters. During nebulization, MS2 virions should be 391
evenly distributed throughout the particle distribution regardless of the composition of 392
the nebulization fluid. In practice nebulization is a harsh process that is known to kill 393
microorganisms (McCullough 1998, Reponen 1997, Mainelis 2005). Viability of the 394
microorganisms will also be reduced once the water has evaporated from the droplet. 395
These factors may have contributed to the reduction of viable MS2 coli phage penetration 396
of the HEPA, during the tryptone aerosolization (assuming that larger particles will be 397
more likely to contain viable virions). Tryptone is reported to protect viruses from 398
desiccation during aerosolization (Dubovi 1970), but our data indicate that aerosolization 399
from tryptone solutions and from water delivered the same amount of viable MS2 coli 400
phage (Figure 10). Therefore, one cannot assume that a proportionally greater number of 401
viable MS2 virions are present in larger particles. Unfortunately technology is not 402
available to determine real-time distribution of viable microorganisms within a particle or 403
distribution of particles. Collection of MS2 in impingers, as was done for this study, can 404
reveal only the viable MS2 virions per collection period, but does not provide 405
information on particle size. 406
Summary 408
HEPA filters are designed to allow penetration of < 0.03 % of challenging 0.3-µm 410
particles. Viruses are simply particulate matter that will penetrate HEPA filters with the 411
same efficiency as inert aerosols. This was clearly demonstrated in this study. What is not 412
clear is the relevance of this finding to biological attack scenarios involving 413
weaponization of viruses. Biological aerosols are complex, and many factors must be 414
considered. The data in this report both support and refute the scenarios required for viral 415
penetration of HEPA filters. One of the key elements that is difficult to quantify is the 416
term “weaponization.” Can viruses be prepared for tactical deployment so that they 417
penetrate HEPA filters efficiently and still remain infectious? The answer to this question 418
is not readily available, but the capability is not completely unlikely. A thorough 419
examination of past biological weapons programs might provide some answers, but those 420
data are hard to obtain and if available, still may not provide clear answers because 421
historical bioweapon research appears to have assumed no respiratory protection. In the 422
absence of those data, the certain way to know if HEPA filters provide adequate 423
protection would be to create tactically relevant biological aerosols and determine their 424
penetration efficiency through the HEPA filters. As a complicating factor, this type of 425
research leads to a conundrum that many face in biological defense applications: the 426
research is crucial to determine if a protection gap exists, but the research might also lead 427
to conditions that could defeat the HEPA filter. This issue notwithstanding, basic research 428
is needed to develop a better understanding of how viruses and other microbes behave in 429
aerosols. In particular, the distribution of viruses, both viable and nonviable, among inert 430
particles in aerosols is not well understood. Data generated from this type of research will 431
help solve biological defense questions, but they will also further basic understanding 432
about and control of the spread of infectious diseases. 433
References 435
Anderson, N. J., P. E. Cassidy, L.L. Janssen, and D.R. Dengel. 2006. "Peak Inspiratory 437
Flows of Adults Exercising at Light, Moderate, and Heavy Work Loads." J. 438
Internat. Soc. Resp. Prot. 23: 53–63. 439
Biswas, P. and C. Y. Wu. 2005. "Nanoparticles and the Environment –A Critical Review 440
Paper." J. Air & Waste Management Association 55: 708–746 441
Bolton, N.E., T.A Lincoln. J.A. Otten, and W.E. Porter. 1976. “A Method for 442
Biological Testing of Containments Systems for Viral Agents.” American 443
Industrial Hygiene Association Journal: 427–431. 444
Burmester, B.R. and L. Witter. 1972. “Efficiency of Commercial Air Filters Against 445
Marek's Disease Virus.” Applied Microbiology. 23: 505–508. 446
Decker, H.M., L.M. Buchanan, B.H. Lawrence, and K.R. Goddard. 1963. “Air Filtration 447
of Microbial Particles.” Am. J. Public Health Nations Health. 53: 1982–1988. 448
Dee, S., L. Batista, J. Deen, and C. Pijoan. 2005. “Evaluation of an Air-Filtration System 449
for Preventing Aerosol Transmission of Porcine Reproductive and Respiratory 450
Syndrome Virus.” Canadian Journal of Veterinary Research. 69: 293–298. 451
Dee, S., L. Batista, J. Deen and C. Pijoan. 2006a. “Evaluation of Systems for Reducing 452
the Transmission of Porcine Reproductive and Respiratory Syndrome Virus by 453
Aerosol.” Canadian Journal of Veterinary Research. 70: 28–33. 454
Dee, S., J. Deen, J.P. Cano, L. Batista and C. Pijoan. 2006b. “Further Evaluation of 455
Alternative Air-Filtration Systems for Reducing the Transmission of Porcine 456
Reproductive and Respiratory Syndrome Virus by Aerosol.” Canadian Journal of 457
Veterinary Research. 70:168–175. 458
Dryden, G.E., S.R. Dryden, D.G. Brown, K.C. Schatzle, and C. Godzeski. 1980. 459
Performance of Bacteria Filters.Respiratory Care. 25: 1127–1135. 460
Dubovi, E. J., and T. G. Akers. 1970. “Airborne Stability of Tailless Bacterial Viruses S-461
13 and MS-2.” Applied Microbiology 19:624–628 462
Eng, K.S., K.C. Hofacre, P.M. Schumacher, R.T. Heckner, and T.L. Forney. 1996. 463
“Filtration Efficiency Assessment of HEPA Filters Against a Bioaerosol 464
Challenge, vol. ERDEC-CR-217.” Edgewood Research Development and 465
Engineering Center; US Army Chemical and Biological Defense Command, 466
Aberdeen Proving Ground, Md. 467
EPA. Environmental Protection Agency .1984. “USEPA Manual of Methods for 468
Virology.” US Environmental Protection Agency, Research and Development, 469
600/4-84-013, USEPA, Cincinnati, Ohio, USA 470
Frischknecht, F. 2003. "The Fistory of Biological Warfare. Human Experimentation, 471
Modern Nightmares and Lone Madmen in the Twentieth Century." EMBO Reports 472
4 Spec No: S47–52. 473
Gronvall, G. K. 2005. "A New Role for Scientists in the Biological Weapons 474
Convention." Nature Biotechnology 10: 1213–1216. 475
Harstad, B. J. and M. E. Filler. 1969. "Evaluation of Air Filters with Submicron Viral 476
Aerosols and Bacterial Aerosols." Am Ind Hyg Assoc J. 30 [3]: 280–290. 477
Harstad, B. J., H. M. Decker, L.M. Buchanan, and M.E. Filler. 1967. "Air Filtration of 478
Submicron Virus Aerosols." Am J Public Health Nations Health 57 [12]: 2186–479
2193. 480
Hawley, R. J. and E. M. Eitzen. 2001. "Biological Weapons—A Primer for 481
Microbiologists." Ann. Rev. Microbiol. 55: 255–353. 482
Heimbuch, B.K., E. Proudfoot, J. Wander, G. Laventure, R. McDonald, and E. Burr. 483
"Antimicrobial Efficiency of Iodinated Individual Protection Filters." Proceeding of 484
the Scientific Conference on Chemical and Biological Defense Hunt Valley, Md. 485
(ECBC-SP-20). 486
Hogan, C. J., E.M. Kettleson, M.-H. Lee, B. Ramaswami, L.T. Angenent, and P. Biswas. 487
2005. "Sampling Methodologies and Dosage Assessment Techniques for 488
Submicrometre and Ultrafine Virus Aerosol Particles." J. Appl. Microbiology 99 489
[6]: 1422–1434. 490
Hopkins, S.R. and L.N. Drury. 1971. “Efficacy of Air Filters in Preventing Transmission 491
of Newcastle Disease.” Avian Disease. 15: 596–603. 492
Kowalski, W. J., W. P. Bahnfleth, and T.S. Whittam.1999. "Filtration of Airborne 493
Microorganisms: Modeling and Prediction." ASHRAE Transactions: Research 105 494
[2]: 4–17. 495
Lee, Jin-Hwa; C-Y Wu, C. Lee, D. Anwar, K. Wysocki, D. Lundgren, S. Farrah, J. 496
Wander, and B. Heimbuch. 2009. “Assessment of Iodine-Treated Filter media for 497
Removal and Inactivation of MS2 Bacteriophage Aerosols.” Journal of Applied 498
Microbiology (in press) 499
Lee, K. W. and B. Y. H. Liu .1980. "On the Minimum Efficiency and the Most 500
Penetrating Particle Size of fibrous Filters." J. Air Pollution Control Assoc. 30 [4]: 501
377–381. 502
Leenders, G.J.M, A.C. Bolle, and J. Stadhouders. 1984. “A Study of the Efficiency of 503
Capturing Bacteriophages from Air by Ultrafilter Yype UltrapolymembranePF-PP 504
30/3 (0.2 mm HF).” Dairy Science Abstracts. 46:233–243. 505
Liu, B.Y. H.1994. "Fundamentals of Air and Gas Filtration." Presented at the Air 506
Filtration: Basic Technologies and Future Trends Conference, Sponsored by the 507
American Filtration and Separations Society, Minneapolis, Minnesota, October 5–6, 508
1994. 509
Mack, W. M. 1956. “Development of a Biological Active Sub-Micron Method for 510
Testing Filtration Efficiencies of Gas Masks and Canisters.” Armed Services 511
Technical Information Agency: Department of the Army, Fort Detrick, Md., 512
Contract # DA-18-064-404-CML-162, Project # 4-11-05-013. 513
Mack, W. M. 1957. “Development of a Biological Active Sub-Micron Method for 514
Testing Filtration Efficiencies of Gas masks and Canisters.” Armed Services 515
Technical Information Agency: Department of the Army, Fort Detrick, Md., 516
Contract # DA-18-064-404-CML-2561, DA-18-064-404-CML-32, DA-18-064-517
404-CML-162. 518
Mainelis, G., D. Berry, J. R. An, M. Yao, K. DeVoe, D. E. Fennell, and R. Jaeger. 2005. 519
“Design and Performance of a Single-Pass Bubbling Bioaerosol Generator.” 520
Atmospheric Environment. 39:3521–3533 521
McCullough, N. V., L. M. Brosseau, D. Vesley, and J. A. Vincent. 1998. “Improved 522
Methods for Generation, Sampling, and Recovery of Biological Aerosols in Filter 523
Challenge Tests.” American Industrial Hygiene Association Journal 59:234–241. 524
Rapp, M.L., T. Thiel, and R.J. Arrowsmith. 1992. “Model System Using Coli Phage 525
phiX174 for Testing Virus Removal by Air Filters.” Applied and Environmental 526
Microbiology. 58: 900–904. 527
Ratnesar–Shumate, S.; C.Y Wu, J. Wander, D. Lundgren, S. Farrah, J.–H Lee, P. 528
Wanakule, M. Blackburn, and M.F Lan. 2008. “Evaluation of Physical Capture 529
Efficiency and Disinfection Capability of an Iodinated Biocidal Filter Medium.” 530
Aerosol and Air Quality Research. 8: 1–18. 531
Reponen, T., K. Willeke, V. Ulevicius, S. A. Grinshpun, and J. Donnelly. 1997. 532
“Techniques for Dispersion of Microorganisms into Air.” Aerosol Science and 533
Technology. 27:405-421. 534
Roelants, P, B.Boon and W. Lhoest. 1968. “Evaluation of a Commercial Air Filter for 535
Removal of Viruses from the Air.” Applied Microbiology. 16: 1465–1467. 536
RTI. 2006. “Test Report No. RTI-TRE-0209289-4 Device Tested: M98 HEPA 537
Element. “ Report prepared for EDGEWOOD CML BIOL CTR under Contract 538
No. W911SR-04-C-0040 Government POC: David Reed. Dated 14 Dec 2006. 539
Stetzenbach, L. D.1992. “Airborne Microorganisms.” Encyclopedia of Microbiology, 540
Academic Press. 1. 541
Thorne, H.V. and T.M. Burrows. 1960. “Aerosol Sampling Methods for the Virus of 542
Foot-and-Mouth Disease and the Measurement of Virus Penetration Through 543
Aerosol Filters.” Journal of Hygiene, Cambridge. 58: 409–417. 544
Vandenbroucke–Grauls, C.M.J.E., K.B. Teeuw, K. Ballemans., C. Lavooi. P.B. 545
Cornelisse, and J. Verhoef. 1995. “Bacterial and Viral Removal Efficiency, Heat 546
and Moisture Exchange Properties of Four Filtration Devices.” Journal of 547
Hospital Infection. 29: 45–56. 548
VanOsdell, D.W., B.Y.H. Liu, K.L. Rubow, and D.Y.H. Pui 1990. "Experimental-Study 549
of Submicrometer and Ultrafine Particle Penetration and Pressure Drop for High-550
Efficiency Filters." Aerosol Science and Technology 12: 911–925. 551
Wander, J.D (Editor), “Respiratory Protection Provided by HEPA Filtration Systems: A 552
Subject Matter Expert Consensus Report. AFRL-RX-TY-TR-2009-4570.” Air 553
Force Research Laboratory, Tyndall AFB, Fla. (In review, distribution statement 554
pending.) 555
Washam, C.J., C.H. Black, W.E. Sandine., and P.R. Elliker. 1966. “Evaluation of Filters 556
for Removal of Bacteriophage from Air.” Applied Microbiology. 14:497–505. 557
Woods, J. B. L. C., MC, USAF.2005. USAMRIID's Medical Management of Biological 558
Casualties Handbook. United States Army Medical Research Institute of Infectious 559
Diseases, Fort Detrick, Maryland. 560
0.5 12345678
Time (hours)
Viable MS2 in effluent
Cumulative PFU
Figure 1: MS2 challenge (103-107PFU/L of air at 85 LPM) of CHAC (n= 21) in BATS
Figure 2: The BioAerosol Test System (BATS) is a Port-Accessible Aerosolization Chamber That is Capable of
Safely Generating and Containing BSL-2 Biological Aerosols.
85 L/min
Swatch Holder SMPS Sampling Port
Impinger Flow Meter
2 LPM - 8 LPM
Flow Meter
Swatch Holder
85 L/min
SMPS Sampling Port
Flow Meter
Flow Meter
Flow Meter
HEPA Filter
Flow Meter
SMPS Sampling Port
Flow Meter
12.5 L/min
Flow Meter
72.5 L/min
Figure 3: Three Test Configurations for Challenging Flat-
sheet HEPA Material and CHACs with MS2 Coli Phage: The
overall design allows for airflow downstream of the test
article both to be analyzed by the SMPS and to be Collected
in an all-glass impinger, allowing for assessment of viable
penetration. 3a) The airflow through the BATS was 85 LPM
and a split stream of either 2 LPM, 4 LPM, 6 LPM or 8 LPM
was directed through the flat-sheet HEPA material. 3b) Purge
air (4 LPM) was fed to the impinger to deliver an net 6 LPM
to maintain collection efficiency (2 LPM through the HEPA
filter plus 4 LPM purge). 3c) A CHAC was fixed to the
BATS and the total airflow of 85 LPM was drawn through
the canister.
Figure 4: SMPS Analysis of MS2 Aerosolized in Water Using the BATS
400 Number
Particle Size (nm)
dN/dlogp, cm3
Figure 5:SMPS Analysis of MS2 Coli Phage Challenge of Flat-Sheet HEPA and
CHAC [(a) Number , (b) Mass]
10 100 1000
2 cm/s ec - HEPA
4 cm/s ec HEPA
2 cm/sec Pos Ctrl
4 cm/sec Pos Ctrl
6 cm/s ec HEPA
6 cm/sec Pos Ctrl
8 cm/s ec HEPA
8 cm/sec Pos Ctrl
2 c m/sec CHAC
2 c m/sec CHAC Pos Ctrl
Particle Size (nm)
dN/dlogp, cm
10 100 1000
2 cm/sec HEPA
4 cm/s ec HEPA
2 cm/sec Pos Cntrl.
4 cm/sec Pos Cntrl
6 cm/sec HEPA
6 cm/sec Pos Cntrl
8 cm/sec HEPA
8 cm/sec Pos Cntrl.
2 c m/sec CHAC Pos Ctrl
2 c m/sec CHAC
Particle Size (nm)
dM/dlogp, µg/cm
Figure 6: Filtration Efficiency of Flat-Sheet HEPA Challenged with MS2 Coli Phage
[(a) Number , (b) Mass]
050 100 150 200 250 300
.9997 2 cm/sec HEPA
4 cm/sec HEPA
6 cm/sec HEPA
8 cm/sec HEPA
2 cm/sec CHAC
Particle Size (nm)
Filtration Efficiency
050 100 150 200 250 300
2 cm/sec HEPA
4 cm/sec HEPA
6 cm/sec HEPA
8 cm/sec HEPA
Particle Size (nm)
Filtration Efficiency
Figure 7: MS2 Challenge of Flat Sheet HEPA and CHACViable Enumeration
8 cm/sec Pos Cntrl
8 cm/sec HEPA
6 cm/sec Pos Cntrl
6 cm/sec HEPA
4 cm/sec Pos Cntrl
4 cm/sec HEPA
2 cm/sec Pos Cntrl
2 cm/sec HEPA
2 cm/sec + 4 cm/sec Pos Cntrl
2 cm/sec + 4 cm/sec HEPA
2 cm/sec CHAC Pos Cntrl
2 cm/sec CHAC
Samples and Face Velocity
Vialbe MS2 in effluent
per Liter of Air
Figure 8: SMPS Analysis: Filtration Efficiency of Flat-Sheet HEPA Challenged with MS2
Aerosolized in 0.5% Tryptone and Water
6000 Number (water)
Mass (water)
Number (tryptone)
Mass (tryptone)
Particle Size (nm)
dN/dlogdp, cm
dM/dlogdp, µg/cm
Figure 9: SMPS Analysis of Flat-Sheet HEPA Challenged with MS2 Aerosolized in
0.5% Tryptone and Water
050 100 150 200 250 300
8 cm/sec (water) - average
8 cm/sec (water) - 15 min
8 cm/sec (water) - 30 min
8 cm/sec (water) - 45 min
8 cm/sec (tryptone) - 15 min
8 cm/sec (tryptone) - 30 min
8 cm/sec (tryptone) - 45 min
Particle Size (nm)
Filtration Efficiency
050 100 150 200 250 300
Particle Size (nm)
Filtration Efficiency
Figure 10: Viable Enumeration of Flat-Sheet HEPA Challenged with
MS2 Aerosolized in 0.5% Tryptone and Water
8 cm/sec Pos Cont. (water)
8 cm/sec HEPA (water)
8 cm/sec Pos Cont. (tryptone)
8 cm/sec HEPA (tryptone)
Vialbe MS2 in effluent
per Liter of Air
020 40 60 80
1068 cm/sec HEPA (water)
8 cm/sec Pos Cont. (water)
8 cm/sec Pos Cont. (tryptone)
8 cm/sec HEPA (tryptone)
Time (minutes)
Vialbe MS2 in effluent
per Liter of Air
Table 1: MS2 Challenge of Flat-Sheet HEPA and CHACs
Sample Face Velocity Collection Flow Rate Average Lower 95% CI Upper 95% CI
Flat Sheet HEPA 2 cm/sec 2 LPM (+4 LPM into impinger) 99.9979% 99.9973% 99.9985%
Flat Sheet HEPA 4 cm/sec 4LPM 99.9951% 99.9941% 99.9961%
Flat Sheet HEPA 6 cm/sec 6 LPM 99.9888% 99.9871% 99.9905%
Flat Sheet HEPA 8 cm/sec 8LPM 99.9626% 99.9571% 99.9681%
CHAC 2 cm/sec 85 LPM 99.9997% 99.9996% 99.9999%
Table 2: Particle Size Distribution of MS2 Aerosolized in Water and 0.5% Tryptone
Particle Size Diameter Water 0.5% Tryptone Water 0.5% Tryptone
10 nm–100 nm 92% 62% 26% 5%
100 nm–300 nm 7.5% 36% 58% 43%
> 300 nm 0.1% 2% 15% 52%
*Data were corrected to account for the entire curve, which was not collected by the SMPS (see fig 8)
Number Distribution Mass Distribution*
... It is of brief significance to mention other technologies such as the use of ultraviolet C (UVC) radiation alongside HEPA filtration, as UVC has been used to disinfect equip-ment which may have come into contact with viruses such as SARS-CoV-2 [28,29], however, little literature exists on how effective commercial air purification devices are at delivering a threshold dose. ...
Full-text available
In dentistry, fallow time is a period which allows for airborne pathogens to settle out of the air and mitigate the risk of airborne infection transmission to dental professionals and staff. The current recommendation is a one-hour period.
... Moreover, even when using the 0.5 efficiency suggested by NHS Scotland [16] to allow for this issue of recirculation and imperfect mixing, it is clear that the performance of an air purifier is far more reliable than that of natural ventilation which relies of course on several different factors, such as ambient and atmospheric conditions, and may at times be far lower than the 6 ACH value needed to make an hour's fallow period valid. In addition to this, studies which measured specifically the penetration of virus particles through HEPA filters found reassuring viral retention values of between 99.996% and 99.97% [27] [28], although this did vary with environmental conditions, for example, when considering efficacy in infection control it is clear that other factors such as airflow and humidity are of great importance [23] [27] [28], however, if we operate on the assumption that natural ventilation is being used as is the current standard of practice, the use of HEPA filtration has been shown to be an effective method to create an adjunctive reduction in airborne particle concentration and a reduction in fallow time. ...
Full-text available
In dentistry, fallow time is a period which allows for airborne pathogens to settle out of the air and mitigate the risk of airborne infection transmission to dental professionals and staff. The current recommendation is a one hour period. Aims and Objectives We aim to show that air purification devices are an undervalued adjunctive measure to mitigate the risk of airborne transmission of pathogens, and so reduce fallow time. Methods Using a computational fluid dynamics software, we created a virtual dental surgery, and simulated a ten-minute aerosol generating procedure. We then modelled air flow in the room with no ventilation, and then the same room with an air purifier at a throughput of 430m3 h-1, and subsequently the room with one open window providing 6 ACH and no purifier. The particles released were monitored and their behaviour and airborne time measured and collated. Results and conclusions The room with no ventilation had a total particle number at 600s of 4.5million, which required 8400s to reduce by 99%. With an open window providing 6 ACH, we obtained a value of 2500s for a 99% reduction in airborne particles, and a similar peak particle volume. Conversely, when using an air purifier throughout the procedure, the peak particle number was ten times lower than the peak number without ventilation, or with an open window, and after the particle injection 99% airborne particle reduction was achieved 60s. Our findings suggest that the use of an air purifier greatly reduces the total particle volume in the air during the aerosol generating procedure, and the fallow period needed. The values found with 6 ACH and an open window are corroborated in other literature. The use of air purification could greatly reduce the risk of infection transmission in a dental surgery.
... Certain researchers express concerns about the HEPA filtering capacity under certain situations. According to a 2009 study, "HEPA filters may not provide adequate protection for all threats: viruses are submicron in size and have small median infectious doses [62]." ...
In the context of the COVID-19, this article reveals the potential of architecture and urbanism in the prevention and control of epidemics and in playing an active role in human health. The historical approach shows that the same space-controlled measures against pandemics were used for centuries to combat leper or plague: quarantine, isolation and confinement. The fight against tuberculosis led, from the 1830s, to the hygiene movement which facilitated current principles for a healthy architecture regarding sunlight and ventilation. In the 1920s, hygienic concepts constituted the foundation for modernist architecture and urbanism. With the advent of antibiotics, in the 1940s, medicine was emancipated from architecture. In the 1970s, the criticism of the social modernist shortcoming led to the New Urbanism or Urban Village movements and environmental issues to Green Architecture and Urbanism. The paper investigates how the present pandemic confirms the last decades warnings and the previous concerns about the correspondence between population density and mortality rates. The article examines the linkages between scale in the built environment, epidemiology and proxemics. The goal is to determine the place of architecture and urbanism in social resilience management during pandemics. Solutions for health engaged architecture and urbanism are indicated at different scales: object scale—hygiene; people scale—distancing and isolation; interior spaces—air control by ventilation, filtering and humidifying; residential—intermediate housing, public spaces between buildings—the key for social interactions; working—telecommuting, size and dispersion; shopping—proximity and downscaling; transportation—walking, bicycling, shared mobility and robo-taxies; and higher scale-mixed use neighborhoods. Architectural certifications such as BREAM and LEED may need to implement similar guidelines for public health. Healthy building movements like Fitwel and WELL Building Standard have already taken steps to foster healthy urbanism, and LEED for Neighborhood Development addresses health related issues. In the context of the COVID-19 and the concern of future pandemics, research in these areas will need to be expanded.
Full-text available
We describe and analyze a new particle generator that utilizes a bursting bubble principle and eliminates carrier fluid reuse. In this Liquid Sparging Aerosolizer (LSA), a suspension of particles or microorganisms is pumped at a flow rate of 0.2–2mLmin−1 to the top surface of a porous stainless-steel disk where it forms a thin suspension film. Filtered air is then sparged through the disk into the film causing it to break into bubbles that subsequently burst, releasing particles into the air. The released particles are then captured by the sparging air stream and are carried away. Particles that impinge the glass vessel and liquid droplets not captured by the air stream drain to the bottom of the vessel and play no further role in the aerosolization process. We tested the LSA with disks of different pore sizes (0.2, 0.5, 2.0 and 10.0μm) and different air flows (2–30Lmin−1) through the porous disks while generating polydisperse and monodisperse particles. Our tests showed that the use of 0.5 and 2.0μm porosity disks resulted in the highest output of PSL particles in the desired size range, i.e., comparable to bacterial size. Each pore size seemed to have an optimal air flow rate; the produced aerosol concentration increased with increasing suspension delivery rate. The LSA also demonstrated stability of output concentration when aerosolizing particles over extended periods of time. In addition, the size distribution of injury-sensitive Pseudomonas fluorescens bacteria virtually did not change during 90min of continuous aerosolization by the LSA. In fact, there was no (0%) viability loss, whereas the bacterial spectrum produced by a Collison nebulizer changed significantly over 90min and there was a 50% loss in viability. The results indicate that the new instrument could be used to generate particles for the evaluation of pathogen collection methods, inhalation and other studies where extended delivery of stable and undamaged biological aerosols is required.
Full-text available
The effect of relative humidity (RH) on the airborne stability of two small bacterial viruses, S-13 and MS-2, was studied. Poorest recovery of S-13 was obtained at 50% RH. Humidification prior to aerosol sampling significantly increased the recovery of S-13 at RH deleterious to the airborne virus. A commercial preparation of MS-2 suspended in a buffered saline solution showed a rapid loss of viability at RH above 30%, whereas a laboratory preparation containing 1.3% tryptone showed high recoveries at all RH studied. Dilution of the commercial MS-2 into tryptone broth conferred stability on the airborne virus. Humidification prior to sampling significantly reduced the viable recovery from aerosols of commercial MS-2, whereas the laboratory preparation was unaffected.
Many commercially available devices initially developed for dispersion of biologically inert particles have been adopted for aerosolization of microoganisms in laboratory settings. However, these dispersion devices are not always adequate for microbial particles, as they do not simulate natural release into air. Wet dispersion methods are appropriate for viruses and most bacteria, whereas dry methods are more suitable for most fungal and actinomycete spores. Characteristics of the resulting aerosol are dependent on the dispersing shear forces and the sensitivity and agglomeration of the tested microorganisms. Consequently, each microbial group may need a specific dispersion technique. The following devices have been developed and tested in this study: the bubbling aerosol disperser, the agar-tube disperser, and the swirling-flow disperser. Testing included the evaluation of both physical and microbiological characteristics of aerosolized microorganisms. Each of the dispersers has shown several advantages over commercially available ones. When used for the dispersion of bacteria from the liquid suspension, the bubbling aerosol disperser was found to produce considerably fewer amounts of microbial fragments and much lower levels of microbial metabolic injury than the commercially available Collison nebulizer. Fungal spores dispersed from their colonies by the agar-tube disperser were found to have a more stable aerosol concentration and a lower fraction of agglomerates than achievable by conventional powder dispersion. The swirling-flow dispersion technique was used for aerosolization of actinomycetes because the agar-tube disperser could not provide a stable concentration of these spores due to their smaller size. The tests have shown that new methods minimize the changes of properties of the microorganisms during their aerosolization in the laboratory.
Short-term (15-min-duration) and long-term (5- to 6-day-duration) test procedures have been developed for determining the efficiency of the removal of bacteriophage phi X174 by air-sterilizing filters. These procedures were sensitive enough to measure a 10(8)-fold reduction in the number of bacteriophage. A filter commonly used in industrial air sterilizations (Domnick-Hunter Bio-X borosilicate glass) effected a 10(8)-fold removal of viable phage in both short-term and long-term tests. A prototype low-flux, hollow-fiber membrane gave similar results; however, a prototype high-flux, hollow-fiber membrane removed only about 99.999% of the bacteriophage in short-term tests.
Air filters of various efficiencies were evaluated as to their ability to prevent transmission of Newcastle disease virus from infected chickens to susceptible chickens. The results indicate that the virus in natural aerosols of mucus secretions readily penetrated roughing and medium-efficiency filters. As exposure time increased, these aerosols began to penetrate high-efficiency filters. Complete removal of virus aerosols for 42 days was accomplished by only one type of high-efficiency filter medium.
The effectiveness of a commercial absolute air filter for removal of viruses from air was tested with an actinophage, under the usual conditions of a laminar airflow clean room. A new method of dry phage dispersion is described. The filter showed an average reduction of 99.996% of airborne actinophage.
Velocity, aerosol particle size, aerosol charge, and exposure to high humidity were found to affect the performance of air filters. Filter papers and the DOP scan-tested filter units (HEPA filters) fabricated from these papers were evaluated with submicron T1 bacteriophage aerosols having a number median diameter (NMD) of 0.12 micron and with Bacillus subtilis var niger spore aerosols with a NMD of 1 micron. Penetration of submicron phage aerosols through filter papers increased markedly with an increase in velocity. Neutralizing phage aerosol with bipolar air ions resulted in 2- to 5-fold increases in penetration. Penetration of bacterial spore aerosols through the filter papers was essentially zero. Phage aerosol penetration through HEPA filters averaged 0.00095%. Spore aerosol penetration was 0.00005%. Exposure to high humidity (>95%) resulted in roughly a 3-fold increase in HEPA penetration. HEPA filters can be expected to provide excellent protection against submicron viral aerosols as well as against bacterial areosols, but the installations must be commensurate with the high performance of this type of filter for full efficiency.
Glass wool, nonabsorbent cotton, fiberglass filter medium, and a commercial absolute filter were tested for effectiveness in removing aerosolized bacterial viruses under low flow rate (1 ft(3)/min) and high flow rate (10 to 25 ft(3)/min) air-flow conditions. Special equipment was designed for measurement of filter efficiencies under the two air-flow conditions. Under low air-flow rate test conditions, glass wool was only 98.543 to 99.83% efficient, whereas cotton (five layers), fiberglass medium (three layers), and the commercial absolute filter were at least 99.900, 99.999, and 99.999 efficient, respectively. Glass wool and cotton were not used under higher air-flow conditions because they were difficult to assemble in leak-tight filters. The commercial absolute filter and fiberglass medium (three layers) were at least 99.990 and 99.999% efficient, respectively, under the higher air flow conditions. A stainless-steel filter of simple design and fitted with three layers of fiberglass medium was found to be greater than 99.999% efficient in removing high concentrations (20,000 to 70,000 plaque-forming units per cubic foot) of aerosolized bacteriophages from air moving at a low flow rate (1 ft(3)/min). Use of this filter on pressure-vacuum tanks in the fermentation industry is suggested. Several other uses of such a filter are proposed.
Four devices used for filtration of microorganisms and/or for heating and moistening the ventilated air during mechanical ventilation were evaluated. This evaluation included measurement of bacterial and viral removal efficiency, heat and moisture exchange properties, dead space and air flow resistance. The devices included: Pall BB50T and DAR Sterivent (filtration devices); DAR Hygrobac and Gibeck Humid-vent [heat and moisture exchangers (HMEs)]. The two devices which are primarily conceived as filters, had the highest bacterial and viral removal efficiency (titre reduction of 10(5)-10(6) for bacteria and of 10(4)-10(5) for viruses), while removal efficiencies of the HME devices were lower: titre reduction of 10(4) for bacteria and 10(1)-10(3) for viruses. As expected, heat and moisture output of HMEs was better than that of filters. In mechanical ventilation, dead space and air flow resistance are important properties of devices, which might disturb efficient ventilation. There were only minor differences in dead space and air flow resistance. Resistance to airflow in the HMEs was increased by nebulization of medication (mesna) unlike that of the filters.
Several kinds and brands of bacteria filters are commercially available for use in anesthesia and respiratory therapy applications. Clinical experience of high airflow resistance, ruptured media, failure to retain visible dust particles, and lack of consistent performance statements or warranties by manufacturers about their bacteria filters prompted a study of the performance of 13 different filters. The filters were challenged by mineral oil droplets, Serratia marcescens and Excherichia coli bacteriophages T4 and T7, tobacco smoke, nebulized india ink, dioctylphthalate smoke (DOP), and water. Results showed that viable bacterial passed through some filters, many filters were unable to retain visible ink or tobacco smoke particles, and resistance to airflow was increased two-fold or more in many filters when the filters were laden with 10 ml of water. Conflicting data resulted from two different types of DOP testing machines. There was a wide variation in performance among the different brands of filters; variable results also were seen within a given brand. Five brands of filters met the federal DOP standards for HEPA filters, but the wide variation in DOP testing results with two different kinds of DOP machines indicates a need for better standards. The DOP 0.3-micron bubble test is the most readily available nontoxic test to rate filtration efficiency; however, the DOP efficiency rating cannot be used to equate equivalent performance against infectious organisms.