Evaluation of Physical Capture Efficiency and Disinfection Capability of an Iodinated Biocidal Filter Medium

Poly(styrene–divinylbenzene)-4-(methyltrimethylammonium) triiodide (PMT) has recently been applied onto nonwoven air filter media to purportedly combine filtration and iodine disinfection to achieve enhanced attenuation of viable airborne pathogens without aggravating the pressure drop of the medium. This paper reports and compares the physical capture efficiency and biological removal efficiency of the novel biocidal filter medium. During challenges with inert fluorescent particles, both iodine-treated and untreated media displayed statistically equivalent physical capture efficiencies > 97%, and typically > 99%. The pressure drag (3.2 kPa·s/m) was less than 10% that of a glass fiber HEPA (38 kPa·s/m) medium. Biological disinfection by the media was evaluated using aerosols containing M. luteus and E. coli vegetative bacterial cells. Biological removal efficiency (99.997%) was observed to be two logs greater than inert particle capture. Viable penetration through the biocidal filters was observed in only two of 10 experiments. The results suggest that an antimicrobial-coated filter medium can provide effective protection against airborne pathogens with a significantly lower pressure drop than that imposed by conventional high-efficiency filtration systems. A near-contact mechanism is proposed in which the distances of nearest approach to treated fiber surfaces as a microbe penetrates through the filter define the probability that charge sites on the microbe’s surface will capture enough iodine molecules from triiodide complexes at the surfaces to terminate viability.


INTRODUCTION
Concern excited by press coverage of the anthrax incident in 2001 escalated the emerging threat of bioterrorism into a great concern for national security. Among the various modes available for staging a biological attack, aerosolization is considered the most effective for dispersing biological agents over a wide area in a short time (Kortepeter and Parker, 1999 (Qian et al., 1998;Richardson et al., 2006;Rengasamy et al., 2007) and against aerosols containing MS2 coli phage (Brosseau et al., 1997;Balazy et al., 2006) showed less than 2% penetration at airflow rates near 30 L/min for the most penetrating size (~ 50 nm), but penetration increased to 5% at 85 L/min.
The capture efficiency of filters is known to depend on several factors including the size of the challenging aerosols, the filter fibers, the velocity of airflow through the filter, and the presence or absence of electric charge on the fibers or particles (Hinds, 1999a).
A more efficient technology to remove bioaerosols from air is the High Efficiency Particulate Air (HEPA) filter, which is rated to remove 99.97% of 300-nm particles. According to OSHA,a NIOSH certified N,R,or P100 class respirator is equivalent to a HEPA. Brosseau et al. (1997)  relative humidity and atmospheric dust (Maus et al., 2000). Subsequent re-entrainment from filter media is then possible, e.g., the reentrainment of MS2 virus from filter media described by Richardson et al. (2006). These functional limitations suggest that some augmentation of conventional filtration might lead to better technologies for respiratory protection.
The halogens iodine and chlorine are antimicrobial agents of great importance.
Halogen disinfection is a method of chemical sterilization in which oxidation of cell constituents and halogenation of cell proteins occurs (Prescott et al., 2002). Iodine is widely used as a disinfectant for potable and on-site water treatment, and is known for its stable chemical storage characteristics (Brion and Silverstein, 1999). Elemental iodine (I 2 ) is not highly soluble in water but may be introduced by heat vaporization, crystal dissolution, oxidation of iodide (I -) ion, and release from iodine-containing resins or from the direct addition of high-strength iodine/alcohol solutions or solutions of triiodide (I 3 -) ions (Black et al., 1968). I 2 is more soluble in solutions of iodide ion, forming the "classical" (Cotton et al., 1999) linear (Swenson and Kloo, 2002) I 3 ion, which was reported (Carroll, 1955) to be a much less-effective antiseptic than I 2 in water. The I -+ I 2 -I 3 reaction is reversible (Jones, 1930). Rates of exchange reactions of I 3 with I 2 (Katzin and Gerbert, 1955) and with I - (Ruff et al., 1972) have been measured, and higher homologues have been observed in the solid state, e.g., I 5 - (Teitelbaum et al., 1980) with I 3 and higher homologues (Saenger, 1984;Yu et al., 1996) in the starch-iodine complex and in charge transfer complexes of iodine with polynuclear aromatic hydrocarbons (Teitelbaum et al., 1980) and with metal α-diketonates (Cowie et al., 1979). I 3 and I 5 ions bind to quaternary ammonium anion exchange resins and form stable complexes (Chang, 1958;Berg et al., 1964;Taylor et al., 1970;Brion and Silverstein, 1999).
Nonspecific antimicrobial activity of these complexes has been studied extensively in water (Taylor et al., 1970;Fina et al., 1982), and a patent was issued (Lambert and Fina, 1975) for the preparation and use of a 4trimethylammoniummethyl derivative (PMT, Fig. 1) of poly(styrene-divinylbenzene) as a broad-spectrum, demand-type disinfectant for bacteria in water, in which spontaneous release of iodine was shown to be very slow and deduced to be strongly enhanced when reactive species are encountered. This concept was adopted for water disinfection aboard spacecraft (Marchin et al., 1997). its conceptual mechanism for disinfecting microorganisms.
A patent by Messier (2000)

METHODS
The experiments were carried out in two phases. In Phase I, the PRE was evaluated using fluorescent aerosols. In Phase II, two types of vegetative cells were challenged to assess the filter's vBRE. The same experimental system was used in both phases. The experimental conditions for both phases are summarized in Table 1.  (May, 1972

Experimental set-up
where d d is the mass median diameter of the atomized droplet (~ 3 µm; (May, 1972)) and F ν is the volume fraction of fluorescein in the solution.
In Phase II, nonpathogenic microorganisms for bioaerosol challenges were selected based on several factors. Escherichia coli is a Gramnegative, rod-shaped bacterium that ranges in size from 2 to 3 µm in length and 0.25 to 1 µm in diameter. The strain used was obtained from the Water Reclamation Facility at the University of Florida. The samples obtained were inoculated and maintained on Difco tryptic soy agar and grown at 33ºC prior to sampling. Micrococcus luteus is another frequently used representative bioaerosol (Wake et al., 1997;Li and Lin, 2001;Agranovski et al., 2003). M. luteus cells are Gram-positive, non-motile, nonsporulating, round bacteria normally found in clusters or tetrads. The individual cells are 0.9 to 1.8 µm in diameter (Wake et al., 1997). M. luteus samples were obtained from the University of Florida,

Department of Microbiology and Cell Sciences.
Cells were inoculated on standard nutrient agar (Difco 0001) and maintained at 33ºC prior to sampling. Note that the purpose of this investigation was to characterize the interaction between filtration media and generically delivered aerosols containing viable microbes.
Accordingly we used a simple aqueous dispersant that creates conditions favorable for survival of the organisms in air, rather than the actual or simulated human fluids that one would employ in a study considering factors relating to contagion. Prior studies (Wake et al., 1997;Crook et al., 1998) (Vanderpool et al., 1987). However, a negative interference was observed for the iodine-treated filters. The interference was attributed to oxidation and/or iodination of fluorescein in the rehydrated filter solution, and it resulted in a lower apparent concentration (Naim et al., 1986). A concentration-based interference curve was therefore established to adjust the values measured.

Phase II
To measure the inlet concentration of bioaerosols entering the test system vs. those captured by filtration, two impactors were used in parallel. One impactor contained an iodinetreated filter upstream, whilst the other contained no filter upstream as positive control.
Although the impact of the variation of the microorganism's viability during nebulizer operation is a concern during bioaerosol generation, such a concern is irrelevant to our study as we were comparing two systems in parallel, i.e., the changes were equal. The inlet concentrations were measured by the positive control stream for the first and last five minutes of each experiment using Petri dishes on all six stages of the impactor.

Statistical analysis of data
Three controls (no filter upstream) and three experimental runs were conducted separately for each flow rate tested. A linear least squares method was used to correlate the data (Vining, 1998). This method considers the data points assessed during the experiments and finds the best straight-line equation to represent all combinations of the data. The x-axis represents the control mass fraction measured for each stage for this analysis, and the y-axis represents the difference between the control mass fraction and the penetration mass fraction for each stage.
The efficiencies reported were the slopes calculated based on the least squares methods, and the error ranges reported were the standard errors for the y estimates.

Filter morphology analysis
Microscopic images of the filters before and after sampling were taken using scanning electron microscopy (SEM) to evaluate any differences in morphology due to treatment or use of the filters. Elemental analysis of objects magnified with SEM was performed using energy dispersive X-ray spectroscopy (EDS).

Morphology analysis of filter media
Untreated filters were analyzed via SEM prior to experimentation (not shown). A dense woven structure of long fibers of different thickness was observed by visual analysis.
Iodine-treated filters were also analyzed via SEM prior to experimentation ( Fig. 3a and 3b). Small dark flecks were observed around the outer perimeter of the filter surfaces. The EDS analysis of the small flecks indicated the presence of iodine (Fig. 3c). Similar results were observed for the EDS taken of the filter fibers (not shown), demonstrating the presence of iodine on the fiber surface.

Phase I -physical capture
The mass size distribution of particles produced in the system reaching the point of filtration was determined during control runs using no filter upstream of the impactor. Fig. 4a shows the size distribution at 15 L/min as an example. The shape of the distribution of other flow rates had a similar pattern, while the total mass increased as the flow rate increased. The majority of the fluorescent particles were collected on the downstream filter stage (< 0.65 µm) and 5th and 6th stages (2.1-1.1 and 1.1-0.65 µm, respectively) for each flow rate. The data for stages 1 to 4 were not used due to the low concentrations of detectable particles. The mean corresponding total mass concentrations were 1.33 x 10 7 , 2.14 x 10 7 , and 4.03 x 10 7 µg/m 3 for 13, 15, and 21 L/min, respectively.    PREs greater that 99.8% were observed for the 1.1-2.1 µm particles for all flow rates tested.
The result is of similar magnitude to that found in previous studies using N95 filters (Richardson et al., 2006). The efficiency decreased to less than 99% for particles smaller than 0.65 µm and to 96.84% for the iodinetreated filters tested at 15 L/min. Both treated and untreated filters appeared to perform similarly based on the data, which showed no significant difference. The thicker filters (2 mm) appeared to perform better than the regular iodine-treated filters at 15 L/min. This was expected; however, the improvement was relatively slight because the thin filters (1 mm) already demonstrated high capture efficiency.
Pressure drop (∆p) across the filters was recorded. The system was tested with and without the use of the aerosolized particles to determine how particle accumulation on the filter affected ∆p. The 1-mm thickness filter had an initial ∆p of 0.45 kPa at 15 L/min and 0.57 kPa for 21 L/min. ∆p increased as the particles flowed through the testing filters and were subsequently captured.
Pressure drag (S) is a measure of the filter's aerodynamic resistance to air flow. It can be calculated by dividing ∆p across a filter by filtration velocity (U) as (Noll, 1999): It is worthwhile noting that the initial pressure drag of the iodinated filters was significantly less than the associated pressure drag of the glass fiber HEPA filters tested, 3.2 kPa·s/m vs. 38 kPa·s/m. Lower filter drag is associated with less-labored respiration, which is beneficial when the mobility of the protected person is critical.

Phase II -Biological disinfection
The size distributions of colony-forming units of the challenged bioaerosols detected at the inlet are shown in Fig. 5. As shown, the majority of the bioaerosol particles generated numerous studies (Berg et al., 1964;Black et al., 1968;Marchin et al., 1997). Hence, we can not exclude the possibility that the microorganisms were deactivated in water rather than on the filter. Nevertheless, both from the low level of penetration and from the intensive contact with iodine at the fiber surface, one can expect the possibility of survival on the filter to be extremely low.
The initial pressure drag of the 1.5-mm thickness filter was less than 10% of that of the glass fiber HEPA filter. Another useful criterion for comparing different types of filters is filter quality, F q (Hinds, 1999a), where P is the aerosol penetration through the filter. The iodinated medium's filter quality based on the biological removal was 19.9/kPa, which is higher than the glass fiber filter's value, 1.8/kPa. The high efficiency of biological disinfection, zero viable cells on the filter, low pressure drag and high filter quality together demonstrate that the concept of applying a nonspecifically reactive antimicrobial coating to a less-efficient filter offers some significant advantages over conventional HEPA filtration for removing

DISCUSSION
The iodine treated filter exhibits a 2-log enhancement of vBRE over PRE, which agrees with previously reported observations of MS2 bioaerosols using similar iodine-treated media (Heimbuch et al., 2004;Heimbuch and Wander, 2006). The enhancement is at least partly due to a demand-release mechanism of the material (Hatch et al., 1980). Hatch et al. (1980) proposed three possible mechanisms to explain the phenomenon of demand-release disinfection in water: 1) I 2 release aided by an internal exchange mechanism involving transient formation of higher polyiodide species; 2) hydrolysis of I 2 on the resin to form HOI as the active principle; 3) spontaneous dissociation of I 2 that adheres to the insoluble resin. The scheme in Fig. 6 is drawn by analogy to the reaction of I 3 with HCN to form ICN and iodide (Edwards et al., 1976), which demonstrates displacement of iodine with concurrent oxidation (demand-release). The gas-solid system studied here corresponds to the conditions for X-ray crystallography, in which HOI in air has not been reported. So, one may conclude that mechanism 2 is not important for aerosols passing the reactive medium. Mechanism 3 is also suspect because vapor concentrations of I 2 measured downstream of PMT were reported to be < 0.2 mg/m 3 (Di Ionno et al., 2001;Di Ionno and Messier, 2004), and we have measured similar values. In such a short-contact environment, it is difficult to envision a replenishment mechanism fast enough to deliver a lethal dose of I 2 that would not be identical to mechanism 1. A supportive argument for mechanism 1 also comes from the crystallography cited above and the generalization by Cowie et al. (1979) about polyiodide electronic structure and vibrational spectra: "Molecular I 2 acts as a Lewis acid and coordination to a Lewis base, e.g., I-weakens the I-I bonding. Thus, the I-I distance in I 2 is 2.72 Å while in I 3 it is near 2.92 Å. Molecular orbital calculations on I 3 also show that the highest occupied molecular orbital has I-I antibonding character." The simple canonical image (Fig. 6) of PMT that emerges is mobile lines of I 2 molecules and Iions (Hatch et al., 1980) in which atoms in the I 2 molecule pass into and out of dsp 3 hybridization as complexes form and dissociate-and from which I 2 molecules can readily be displaced.
Whereas solution disinfection is characterized by long residence times, the aerosol-fiber interaction as a particulate organism traverses the filter matrix is fleetingly transient and the observed efficacy of kill requires that the kinetics of transfer of I 2 be extremely fast-which this visualization allows. Accordingly we propose that near-contact transfer of I 2 occurs from the polyiodide complexes as the bacteria pass near the resinbound (nominal) I 3 complex. Fig. 7 displays the concept. A notional probability function for displacement and capture of I 2 -not to be misinterpreted as a concentration gradient of I 2 vapor-from the bound I 3 complex is illustrated by the graduated intensity of color surrounding the fiber (viewed in cross section). Bacteria are almost universally anionic at their surface. When one passes in proximity to the I 3 complex on the polymer surface, the distance of closest approach defines the probability that the negative charge on the microbe surface will displace one or more Iion(s) and capture an equal number of I 2 molecules. Each captured I 2 molecule then reacts with an iodine-sensitive group on the microbe, and if a critical amount of damage is accumulated the organism ceases to function. Thus, disinfection can occur near but without direct contact with filter surfaces, allowing the carcass to penetrate as an inert particle.

CONCLUSION
In this study, the physical capture and biological disinfection efficiency of a novel biocidal filter medium were evaluated. Significant capture (greater than 97%) by the filters was observed for a wide particle size range. In most cases the efficiency was greater than 99%. There was no discernible difference in the PRE between the iodine-treated and untreated filters, suggesting no mechanical impact of the treatment. The iodinated resin filters were not as efficient at physical removal of aerosols as the glass fiber HEPA filter.
Particles containing two types of microorganisms-M. luteus and E. coli, which were dominantly in the 1.1-2.1 µm aerodynamic range-were used in the bioaerosol challenge experiments. A simple aqueous dispersant that creates conditions favorable for survival of the organisms in air was employed, rather than the actual or simulated human fluids that one would employ in a study considering factors relating to contagion. vBREs of these two species, representing Gram-positive and Gram-negative bacteria, respectively, were identical, at approximately 99.997%, almost 5-log attenuation of viable bioaerosols. Indeed, only two of 10 experiments performed showed detectable penetration through the iodinetreated filters. A 2-log enhancement of vBRE over PRE was observed, and we propose a direct mechanism involving displacement and capture of I 2 from the medium. No viable bacterial cells were observed in the vortexing experiments, further supporting the effectiveness of the iodine mechanism to disinfect micro-organisms. The high biological disinfection capacity combined with the low pressure drag and high filter quality demonstrates the novel reactive filter medium to be a viable alternative to conventional filtration for the removal of micrometer bioaerosols. While the filter medium was a disposable unit and was intended for short-term use, information about its effective lifetime is useful and should be investigated in future studies.