Open Access
Add to BibSonomy
Abstract
We present evidence-based design principles for three different UV-C based decontamination systems for N95 filtering facepiece respirators (FFRs) within the context of the SARS-CoV-2 outbreak of 2019–2020. The approaches used here were created with consideration for the needs of low- and middle-income countries (LMICs) and other under-resourced facilities. As such, a particular emphasis is placed on providing cost-effective solutions that can be implemented in short order using generally available components and subsystems. We discuss three optical designs for decontamination chambers, describe experiments verifying design parameters, validate the efficacy of the decontamination for two commonly used N95 FFRs (3M, #1860 and Gerson #1730), and run mechanical and filtration tests that support FFR reuse for at least five decontamination cycles.
© 2020 Optical Society of America
1. INTRODUCTION
A. Statement of Need
As the numbers of coronavirus disease 2019 (COVID-19) cases globally continue to grow at alarming rates, demand for personal protective equipment (PPE) continues to outstrip production, resulting in PPE shortages worldwide. Among the shortages of various types of PPE, the shortage of disposable N95 filtering facepiece respirators (FFRs), commonly known as N95 masks, is especially severe [1,2]. The tight seal to the face, when properly fitted, and the high filtration efficiency of the filter material make N95 FFRs a suitable choice for use in healthcare settings to protect workers from airborne transmission of infectious agents [3]. Due to recent supply chain limitations, many medical facilities are faced with difficult decisions around how to best ration their supply of N95 FFRs. For some, their strategy has included the decontamination and reuse of these single-use respirators [4,5]. Indeed, this practice is listed by the Centers for Disease Control and Prevention (CDC) as a crisis capacity strategy for when N95 supplies are low [3]. Such large-scale decontamination and reuse of N95 FFRs is unprecedented, and there is an urgent need for practical guidance on how to do so safely, effectively, and without compromising the performance N95 FFRs.
B. Rationale for the Use of UV-C
The use of ultraviolet germicidal irradiation (UVGI) in healthcare settings has been well-documented in the literature and is used for room-level decontamination at many institutions. In these UVGI systems, UV-C irradiation is used to inactivate Clostridium difficile (C. diff), methicillin-resistant Staphylococcus aureus (MRSA), and vancomycin-resistant Enterococci (VRE) in hospital rooms [6]. This method works by generating dimers in DNA and RNA of exposed microorganisms, rendering them incapable of replication [7]. Within the context of this paper, “decontamination” will refer to, at minimum, a 3-log reduction of SARS-CoV-2 analogues (i.e., 99.9% inactivation), per FDA guidelines for N95 Tier 3 bioburden reduction [8]. Like most N95 decontamination procedures, this high-level disinfection method should be viewed as a risk mitigation strategy rather than complete sterilization.
The effectiveness of viral inactivation by UV-C depends on the delivered UV-C dose, which is a function of exposure time and irradiance, as well as the UV-C source wavelength, the ability of the microorganism to resist UV-C degradation, and the surface structure of the objects being decontaminated. In general, N95 respirators consist of several different layers of material with different optical properties. Filtration primarily takes place in two inner, electrostatically charged layers of spun polypropylene, which trap small aerosols by electrostatic attraction. Larger particles or droplets are filtered by mechanical blockage. The spun material is loosely packed in order to minimize the pressure drop across the material during user respiration. The filter layers are sandwiched between inner and outer layers, which mechanically support the spun material and provide additional protection against larger droplets as well as fluid-resistance. The N95 designation implies that the FFR filters at least 95% of particles with a diameter of 300 nm. For SARS-CoV-2 contamination of N95 FFRs, a UV-C irradiation dose of $1.0 {-} 1.2\;{{\rm J/cm}^2}$ or higher at both surfaces of the respirator is recommended to achieve decontamination [9]. N95 FFRs can be exposed to higher levels without apparent degradation, but damage is evident at levels exceeding ${100}\;{{\rm J/cm}^2}$. These extremely high doses have been shown to cause a breakdown of the polypropylene fibers and a decline in respirator filtration and structural integrity and should be avoided [10]. At ${1}\;{{\rm J/cm}^2}$ UV-C exposure dose, the number of reuse cycles is limited by mechanical wear and/or soiling of N95 FFRs, rather than cumulative UV-C exposure.
Some major advantages of UV-C over other decontamination methods include rapid throughput, ease of use, low electrical power requirements, absence of toxic or dangerous chemicals, and relatively simple overall device design and construction. UV-C methods are considered a decontamination process (3-log viral reduction) and not a sterilizing process (reaching at least 6-log viral inactivation and spore destruction). Therefore, it is strongly recommended that processed FFRs be indexed and returned to the initial user.
C. Scope of the Paper
In this paper, we present evidence-based design principles for three different UV-C decontamination systems of N95 FFRs. The approaches were created with consideration for use in resource-constrained settings such as LMICs and rural healthcare facilities. The three designs provide easily constructed, cost-effective solutions, using generally widely available components, with sufficiently high throughput to meet the demands of hospitals, clinics, and first responders who face shortages of N95 FFRs.
An additional aim of the paper is to illustrate optical models and testing procedures that facilitate the design of instruments specifically for the application of N95 FFR decontamination. We want to stress that each manufacturers’ N95 design should be considered unique with distinct optical and mechanical properties. The tests that we report here were primarily performed on the model #1860, manufactured by 3M. The biological and optical testing results provided in this paper establish practical solutions to decontamination problems; however, the data we present here is specific to the 3M 1860. Sufficient design information is provided in this report and in the supplementary materials (Supplement 1, sections 1-6) so that a reasonably skilled person with an optics background can modify, construct, validate, and operate the decontamination chambers using locally available resources (Supplement 1, sections 1-4).
We first present the general optical design methods for the decontamination chambers. We then describe in some detail the optical and mechanical properties specific to the 3M 1860 N95 FFR that are essential to consider for proper decontamination and to determine the suitability for reuse. These properties will likely differ depending on the FFR manufacturer and model. Finally, we describe the biological testing performed to demonstrate the level of decontamination achieved.
2. GENERAL DESIGN PRINCIPLES
UV-C photons, generally defined as having wavelengths in the range of 200–280 nm with germicidal activity peaking at around 265 nm, have been shown to be particularly effective for decontamination of bacterial, viral, and fungal pathogens. N95 FFRs have an approximately hemispherical dome shape and an ideal decontamination chamber would illuminate the N95 FFR uniformly from all directions. Any chamber design needs to take into account the geometry of the most common UV-C sources: mercury UV-C lamps are cylindrical in shape and UV-C LEDs are essentially point sources. Disadvantages of UV-C LEDs include the need to produce diffuse and uniform exposure, much higher cost, and much lower efficiency than mercury discharge lamps. We have chosen to use UV-C lamps in our design since these sources are more widely available, can currently supply higher average powers, and are in general more cost-effective than UV-C LEDs. A typical emission spectrum from a low-pressure mercury UV-C is shown in Fig. 1.
Fig. 1. Typical emission spectrum of a low-pressure mercury discharge germicidal lamp. The dominant emission is at 254 nm (${\sim}{85}\%$), with lower power emission (${\sim}{15}\%$) in the UV-B, UV-A, and visible spectral regions. (Fig. 1 originally published by Schmid J., Hoenes K., Rath M., Vatter P., and Hessling M., in “UV-C inactivation of Legionella rubrilucens,” licensed under Creative Commons Attribution 4.0 License.)
The lamps we use are specifically chosen to be non-ozone producing lamps with rapid start ballast power supplies, similar to those used in common fluorescent lamps (e.g., Philips model 30WT8). The appropriate choice of power supply ballast is particularly important to provide the longest possible lifetime for the UV-C lamp. The design objective can then be formulated as follows: given a desired throughput of N95 FFRs per day, design a containment chamber with sufficiently intense interior UV-C light sources such that the irradiance onto the N95 FFRs is uniform within 20%, a reasonable practical target, on the front, back, and curved side surfaces of the N95 FFRs. This level of uniformity can be predicted using our ray trace models and verified by direct measurement using a properly designed irradiance meter (e.g.,OPHIR model PD300RM-UV). The UV-C units discussed here met this objective and provided a UV-C dose of at least ${1}\;{{\rm J/cm}^2}$ to N95s in approximately 5 minutes exposure time.
Fig. 2. Horizontal UV-C-decontamination chamber. The masks rest on a horizontal metal mesh support. A thin wire mesh (1-inch hexagon) was chosen as a tray to allow UV-C exposure on both sides, while minimizing interference due to shadowing. (Note: 1 in = 2.54 cm.)
Fig. 3. Vertical Cabinet with two front doors. The cabinet is a standard metal storage cabinet approximately 2 m high that was chosen to provide sufficient height to accommodate upper and lower banks of UV-C lamps. The upper and lower banks are offset to fit into the cabinet and have a small region of overlap, which does not appear to materially impact the vertical intensity profile in the cabinet. The CAD model illustrates a supporting frame containing 28 masks in 7 rows of 4. We estimate up to 40 masks could be loaded per UV-C exposure procedure in this design.
Fig. 4. Cylindrical Design chamber. The chamber can hold up to 4 masks in the central support frame.
After some preliminary studies and experiments we selected three configurations (also see Table 1):
- The Horizontal Cabinet: A rectangular cabinet in which N95 FFRs are illuminated from above and below (Fig. 2.)
- The Vertical Cabinet: A rectangular cabinet with doors in which the N95 FFRs are placed in a vertical plane and illuminated from front and behind (Fig. 3.)
- The Cylindrical Design: A cylindrical can in which the N95 FFRs are suspended from a frame located on the axis of the cylinder (Fig. 4.)
Tradeoffs needed to be made regarding chamber size, N95 FFR throughput, cost, availability of local materials, safety, ozone levels, reliability, fluence levels, and isotropy of illumination. The cylindrical geometry of the UV-C lights dictates mounting them on a flat or cylindrical surface, with coatings that reflect and diffuse emitted light to create an isotropic irradiance in the exposure area. Aluminum foil is a low-cost and widely available material that can be draped over or bonded to the interior chamber walls, providing higher uniformity and irradiance due to an approximately 73% reflectivity at 254 nm UV-C wavelengths [11]. The N95 FFRs themselves diffusively reflect, scatter, and transmit UV-C illumination. Careful consideration must be given to account for the effect of the N95 FFRs on the overall irradiance level inside the cabinets or container, in particular to avoid shadowing of adjacent N95s. As this information is not readily publicly available, we carried out an investigation to obtain these data for OpticStudio by Zemax (www. Zemax.com) modeling by measuring N95 optical parameters directly. The N95 FFRs should be geometrically arranged to avoid any obvious obstruction of light by adjacent masks, straps, or suspension systems.
3. OPTICAL MODELS OF THE DESIGNS
OpticStudio modeling is a well-established optical design tool and widely available in high-income countries, but less so in LMICs. We therefore decided to apply a first-order estimate of the design using a simplified analytical approach, without resort to OpticStudio modeling [12]. These first-order calculations are a reasonably close approximation to using OpticStudio tools, but do not include the effects of reflective walls in the enclosure. This approach would allow someone with an optics background to quickly estimate the performance of alternative lamp placements in a variety of chamber geometries and arrive at a useful estimate of the illumination levels necessary for decontamination of N95 FFRs.
A. Geometrical Optical Models for UV-C Linear Sources
Most common UV-C sources are cylindrical lamps with a length 20 to 50 times their diameter emitting approximately 85% UV-C at 254 nm and about 15% in the UV-A, UV-B, and visible portion of the spectrum. These sources are best modeled as cylindrical, linear diffuse-emitting light sources. Of the available linear models, the view factor model seems to be the most accurate at estimating irradiance for a given point of interest. The view factor model assumes that the UV-C source is a homogeneous cylinder. Figure 5 provides a visual aid for the formulas corresponding to the view factor model shown below. Once the irradiance is computed, exposure time is deduced by requiring that the total irradiance flux equals ${1}\;{{\rm J/cm}^2}$. Total irradiance at any point inside the decontamination chamber is computed by superposition of irradiance from all incident sources, projected onto the measurement surface. The following equation only applies to a point of interest located at the end points of the lamps and does not factor in the effects of reflectivity of various surfaces inside the chamber. To compute the view factor for a point located at any position along the lamp (i.e., not coplanar with an end of the lamp), it must be divided into two segments with the power split proportionally amongst those segments.
The full view factor equation is described fully and is summarized in the supplementary materials (Supplement 1, section 6) [12]. The equation can be simplified significantly with a few assumptions, providing a simple estimate of the direct irradiance from an array of UV-C lamps at a specified distance directly across from the array. In the limits where the cylinder length is much greater than the distance to the illumination point and the diameter of the lamps, the view factor model effectively reduces to the emission expected from an infinitely long cylindrical source. The irradiance can be estimated from the Keitz equation [12]. Let ${P}$ be the tubular lamp UV power ($\textit{W}\;$), ${L}$ is the tube length (cm), ${D}$ is the distance to the N95 mask (cm), and ${R}$ is the tube radius (cm). When ${L}\!/\!{R}\; \gt \;{20}$ and ${D}\!/\!{R}\; \lt \;{15}$ the irradiance is approximately
The irradiation time is ${t} = {\rm dose}/\!{E}$, where dose is the required ${J}\!/\!{{\rm cm}^2}$ to inactivate the virus (typically ${1}\;{{\rm J/cm}^2}$). In other words, for all instances where ${L}\!/\!{R}\; \gt \;{20}$ and ${D}\!/\!{R}\; \lt \;{15}$, the full view factor equation as shown in the supplementary materials (Supplement 1, section 6) can be simplified down to Eq. (1) to within 5% accuracy. The contributions of multiple lamps can be accounted for by simply adding the levels predicted by Eq. (1) for each lamp location.
Equation (1) provides a simple first-order estimate of the irradiance from an array of cylindrical UV-C sources without including the effects of reflections in the chamber. In practice, reflections from walls and other surfaces need to be taken into account, as well as detailed considerations of geometry to account for a proper calculation of irradiance. In particular, proper modeling of irradiance incident on the mask needs to take into consideration the shape of the mask and its optical properties such as absorption, reflection, and transmission coefficients. We include these effects in the OpticStudio modeling results and compare our models against experiments for the three cabinet designs described above.
Fig. 6. Dissecting and measuring the transmission through each layer of two N95 FFRs, the 3M 1860S, and the Gerson 1730. This shows the importance of illuminating the respirators from both sides.
Fig. 7. Measured and simulated intensity profiles within the Vertical Cabinet. The measurements were made with the doors closed and no masks in the chamber (Supplement 1, section 2). The optical model was run with both empty and fully loaded chambers (40 FFRs) to illustrate the impact of the additional absorption due to the presence of the FFRs in the chamber. (Note: 1 in = 2.54 cm.)
B. Ray Trace Simulation Models
The formula in Eq. (1) can be used to reasonably estimate direct irradiance from a UVGI lamp array, but we recommend that design models use computational optical ray tracing to achieve the most accurate results [12]. The software used by the authors for the models in this paper is OpticStudio, and the simulations were run in purely non-sequential mode. Details of the optical model are discussed in [13]. The following assumptions were used for the OpticStudio modeling, based on experimental data:
- Aluminum foil: 73% reflectivity: 30% specular, 70% diffuse
- Paint on walls: 7% reflectivity: 30% specular, 70% diffuse
- Bare walls: 7% for non-foil (30% specular, 70% Lambertian for both instances)
- Lamp fixture reflectivity: 80% (100% specular)
- N95 FFR: CAD model that is approximately hemispherical in shape with 1% transmission through all layers, 94% absorbance, and 5% reflectivity: 0% specular, 100% diffuse
1. Characterizing N95 FFR Optical Properties
N95 FFRs trap viral particles within a polypropylene filter layer, not at the N95 surface where UV-C is incident. We examined whether sufficient UV-C penetrates N95s to inactivate virus trapped within the filter. To the best of our knowledge, the detailed optical properties of an N95 mask have not been measured previously. For optical modeling purposes and to assess the potential for decontamination below the surface of N95s, we need to know the optical transmission, reflection, and absorption coefficients for the N95 FFR. Therefore, we set out to investigate the optical properties of N95 FFRs and developed OpticStudio models to design close to isotropic fluence levels incident on batches of N95 FFRs. Using an integrating-sphere spectrophotometer we measured diffuse transmission spectra of the layers and approximated the reflectivity of the N95 FFR, which compared well with experiments, using the parameters noted in Section 3.B. It is important to note that the optical properties of the N95 FFRs listed in Fig. 6 are specific to these versions and that other types of N95 FFR masks will likely have different optical characteristics.
Fig. 8. Measured and simulated intensity profiles within the Cylindrical Design. There is approximately a two-fold change in the intensity profile. The decontamination exposure time should be determined by the minimum level of intensity exposure (Supplement 1, section 3). (Note: 1 in = 2.54 cm.)
Fig. 9. OpticStudio ray trace model and measured intensity profiles of the Horizontal Cabinet. (Note: 1 in = 2.54 cm.) (Described further in Supplement 1, section 1.)
As mentioned, the geometric shape of an N95 FFR is roughly hemispherical and thus its internal and external surfaces are oriented at a variety of angles, making it a unique problem for optical modeling. To better characterize the behavior of an N95 FFR in the presence of UV-C light, all four layers of a single N95 FFR (3M 1860S) were carefully dissected and the diffuse UV transmission through each layer was measured Fig. 6. The results of this study showed that the transmission of UV-C light through all layers of the N95 FFR was 1%-2%. With illumination from both sides, the front three layers combined transmit 6.2% the UV-C light and the back most layer transmits 15.3% of the light. This effectively means that the core of the respirator receives 21.5% of the UV-C dose delivered at the surface and that all layers receive at least this level of exposure, provided the respirators are illuminated from both sides. Another N95 FFR was tested (Gerson 1730), and the UV transmittance measured within its layers was similar at 18.8%. We developed an optical model of the N95 FFR based on a 3D CAD reconstruction and included the geometrical and optical properties of the N95 FFR in our ray trace simulations.
Fig. 10. Biological inactivation of Bacillus pumilus spores. A 6-log reduction is shown after seven days of incubation. The change in color from purple to yellow is due to a chemical indicator that detects a change in the pH of the growth media, which occurs due to any bacterial growth during the seven-day incubation period.
2. Ray Trace Model Results
Ray tracing results using the design parameters noted above were compared against calculations and direct measurements for all three decontamination chambers. The results are shown in Figs. 7–9. Details of these designs and the measurement procedures used to characterize their performance are contained in supplementary materials (Supplement 1, sections 1-3).
The data were found to be in good agreement with the ray trace models, and variations between the model and measurements can likely be attributed to imperfections in the non-smooth, foil-lined walls, and the reproducibility of the placement of the power meter inside the closed cabinet during the measurements. We note that our models indicate that the addition of masks reduces the intensity by about 15% compared to empty containers. Furthermore, the simulation with the aluminum foil-lined walls saw a two-fold or greater increase in irradiance over the model simulating simple paint, consistent with our measurements, further emphasizing the utility of using aluminum foil as a simple means to increase radiant intensity on the masks.
4. OPERATIONAL SYSTEM PERFORMANCE
A. Bacillus Pumilus Spore Testing
The UV-C sensitivity of various microorganisms including bacteria, bacterial spores, mold spores, and viruses is well studied in the literature [12]. The following biological inactivation study uses a highly UV-C resistant bacterial spore, Bacillus pumilus (PM-106, Crosstex), as a surrogate for SARS-CoV2. These spores are generally used for radiation sterilization processes and are notoriously resistant to ultraviolet light [14]. In one study, a certain strain of Bacillus pumilus required a dose of nearly ${0.350}\;{{\rm J/cm}^2}$ to achieve 4 log inactivation [15]. In the biological tests used in this paper, the bacteria spore strips are loaded with ${2.{8 \times 10}^6}$ C.F.U. by the manufacturer, which results in a more than 6 log inactivation measurable with this assay. The spore strips were exposed to UV-C radiation, up to ${1}\;{{\rm J/cm}^2}$, flat or perpendicular to the UV-C light source to mimic the different surface contours of the mask. The strips are placed in growth media containing a pH indicator dye and incubated for two to seven days. Any residual viable bacteria will grow in the media resulting in a change in the media pH, which causes a concomitant change in color of the pH indicator from purple to yellow. The positive control (no UV-C radiation) displayed a color change to yellow while spores treated with UV-C doses as low as ${200}\;{{\rm mJ/cm}^2}$ up to ${1}\;{{\rm J/cm}^2}$ did not show a color change, implying a 6-log bacteria spore inactivation at those doses. The test was repeated under the same conditions with the bacteria spore strips “sandwiched” between the layers of two N95 FFRs, effectively simulating an entire respirator on each side of the strips. After a single UV-C surface dose of ${1}\;{{\rm J/cm}^2}$, all locations showed a 6-log inactivation of the bacteria spores. Both experiments were repeated three times with identical findings (Fig. 10).
B. Mechanical N95 FFR Testing
N95 FFRs use a non-woven layer of electrostatically charged melt blown polypropylene-based filters to capture airborne particles [16]. The N95 classification refers to the masks’ ability to filter ${\ge} 95\%$ of non-oil-based particles of any size greater than 0.300 µm in size, while also being able to filter out smaller particles as well via the electrostatic charge on the polypropylene fibers. Filtration of larger particles (diameters of several microns and higher) occurs through trapping or mechanical blockage of the particles within the melt-blown filter layers. For reference, the diameter of a coronavirus is approximately 0.12 µm; the size of a typical bacterium is a cylinder roughly 1 µm in diameter and several microns long. Typical droplet sizes range from 10 µm to a millimeter and aerosols are generally considered to range from several hundred nanometers to 10 µm. The N95 FFR contours to the human face, sealing as tightly as possible around the nose and mouth and necessitating all inhaled air to flow through, rather than around, the respirator. The mask is held in place by two elastic bands, placed around the top of the head and neck, which supply the pressure required to seal the mask to the face of the wearer.
Table 1. Decontamination Chambers Investigated in This Research
Table 2. Results of the Filtration Test of an N95 FFR (3M 1860S) Show No Significant Mask Deterioration for Up to five cycles of ${1}\;{\rm J/cm}^{2}$a
The balance between achieving suitable N95 FFR decontamination without compromising intrinsic fit and filtration properties is a very delicate one. Many different approaches already exist for general-purpose biological decontamination in healthcare settings, but virtually all of them compromise respirator integrity. Therefore, it is important that any new approach for N95 FFR decontamination include a proper assessment of intrinsic mask properties to make sure that they have not been compromised. The approaches outlined in this paper have proven to have no significant deviation for fit, filtration, or strap elasticity through five cycles of decontamination for the common N95 FFR model (3M 1860S) that was tested, as shown in Table 2.
Fig. 11. Graph showing that during decontamination cycles, the ozone levels are below the limit of detection of the sensor in the Horizontal Cabinet.
5. OPERATIONAL CAUTIONS
A. UV Exposure Warning
Exposure to UV-C irradiation carries human health risks, particularly to the skin and eyes. Painful acute conjunctivitis can result from doses ${\gt}{5}\;{{\rm mJ/cm}^2}$, which is less than 1% of the target dose for mask decontamination of ${1}\;{{\rm J/cm}^2}$. Proper device design and operator precautions are therefore required to avoid UV-C exposure to the eyes. Skin exposure of ${\gt}{10}\;{{\rm mJ/cm}^2}$ results in mild redness, desquamation, and pigmentation. There is no evidence in humans of UV-C-induced skin cancer, but this is a possible long-term hazard [17].
B. Ozone
A common concern surrounding the use of UV-C is the production of ozone. Ozone can be generated by UV photons in the range of 180–220 nm. Some versions of low-pressure mercury lamps are specifically designed using quartz materials that also transmit vacuum ultraviolet light at 185 nm, which can generate significant amounts of ozone. Ozone can pose an additional health hazard to the operator, particularly when generated and accumulated in closed cabinet designs where ozone can be trapped during the decontamination procedure. Therefore, the low-pressure mercury lamps used for decontamination should be non-ozone producing.
Fig. 12. Graphs showing that the UV-C lamps quickly eliminate any ozone in the cabinet that might be generated during the decontamination process.
To confirm the absence of ozone exposure to the end user for standard germicidal lamps used in most UV-C decontamination devices, an ozone detector (Aeroqual model 300 with 0.5 ppm detector) was placed inside the Vertical Cabinet and measurements were taken during a typical decontamination procedure lasting 10 min. During this operation no ozone was detected within the Vertical Cabinet. The Horizontal Cabinet achieved similar results. In both designs, under normal use, the ozone levels were below the limit of detection of the sensor (0.001 ppm), and well below the minimum acceptable exposure levels over an 8-h period (0.070 ppm) as defined by the EPA [18]. We additionally measured ozone levels outside the cabinet and inside the Horizontal Cabinet before the unit was switched on as well as inside during the UV-C cycle and right afterwards. There was significant reduction in ambient ozone levels after the experiment. The baseline ozone level was about 0.007 ppm prior and 0.000 during and after cycle (Fig. 11).
Ozone absorbs 254 nm UV-C and undergoes photochemical decomposition and recombination to yield molecular oxygen, which explains our measurements. To further confirm this explanation and the absence of ozone exposure to the end user for standard germicidal lamps used in most UV-C decontamination devices, the ozone detector and an ozone-generating lamp (OZN-1. R & M Supply, Inc. Perris, CA) were placed inside the Vertical Cabinet and the ozone generator was allowed to run for about 30 s. The ozone-generating lamp was then turned off and the decay over time was measured and graphed, Fig. 12. The time constant was found to be 2500 s with all 16 UVGI lamps turned off but only 33 s with the UVGI lamps turned on. This confirms that the UVGI lamps used are effective at photodissociation of the ozone. Therefore, it appears that a closed cabinet design is likely superior to flowing air through the chamber or having an open design where the ozone can diffuse out of the irradiation area.
C. Strap Decontamination
Reports have demonstrated that under certain conditions, there may be residual virus on N95 FFR straps post UV-C exposure, likely due to portions of the N95 FFR straps being shielded from the UV-C light. This suggests the need for supplementary decontamination of the straps (9). Therefore, it is recommended to wipe down N95 FFR straps with a compatible disinfectant after completing a UVGI exposure cycle [19]. Examples of common compatible disinfectants include hydrogen peroxide, 70% isopropyl alcohol, or ethanol [20]. If this additional step is employed, extra caution should be used to avoid touching the N95 FFR facepiece as common disinfectant chemicals can degrade N95 FFR function [21].
D. Soiling
Soiling has been found to reduce UV-C inactivation efficacy of both MS2 bacteriophage from N95 FFRs [22] and C. difficile spores from glass and plastic surfaces [23]. Materials deposited on the respirator surface such as sebum, cosmetics, blood, and sunscreen may block UVGI light, hindering decontamination. As a general note, cosmetics, skin creams, or other barriers should not be worn during respirator use. Any N95 FFR found to be soiled with such products, or externally contaminated with visible blood or body fluids, should be discarded and not undergo decontamination.
E. Temperature and Humidity
High humidity can decrease UV-C efficacy on generic surfaces [24] and on the surfaces of N95 FFRs [22]. Considering that N95 FFRs are being worn for lengthy periods of time to extend their use, it is not unreasonable to assume that the constant breathing and sweat they are exposed to throughout the day will moisten them. However, while the effect of humidity on microorganism survival is significant for bacterial species, it is practically negligible for viruses [12]. Furthermore, the humidity over time was measured within the Vertical Cabinet and was found to be low, presumably due to the higher internal temperature of the cabinet (Fig. 13). The mild heat produced by the UV-C fixtures was enough to functionally act as a drying step for the N95 FFRs, but not too much as to cause damage to the respirators. However, further investigation on this matter is warranted to assess the viability of microorganisms on the surface of N95 FFRs exposed to UV-C in humid conditions.
Fig. 13. Temperature and humidity data over several cycles of the Vertical Cabinet. Neither the temperature nor the humidity approaches the thresholds that have been documented to damage N95 FFRs. The chamber was cycled 150 times, with the lamps on for 320 s and then off for 320 s, with no observable change in the lamp peak intensity. The data shown was taken after approximately five cycles. For the Horizontal Cabinet, the temperature rise with sustained operation at 50% duty cycle was similar, less than 5 deg C. The graphs indicate less than 10% change in the lamp intensity during a typical decontamination cycle, primarily due to the increase in temperature inside the chamber.
6. CONCLUSIONS AND FUTURE WORK
We present in this report evidence on the efficacy of properly designed UV-C instruments for decontamination of N95 FFRs. However, there is still a need for further research regarding the inactivation specifically of SARS-CoV2 droplets deposited on respirator surfaces. We are currently conducting studies to measure direct viral and spore inactivation with microorganisms directly inoculated on the surface and within the layers of an N95 FFR. It should be noted that ultraviolet germicidal activity hinges directly on the medium that the microorganism resides on. Given the global shortage of N95 FFRs and equivalents, the ability to run a large-scale study with a statistically significant number of samples would be difficult to conduct at this time.
Finally, it is critical to understand that any decontamination procedure is a risk mitigation strategy, currently warranted in critical FFR shortage situations, not a long-term solution to the shortage of PPE. Anyone interested in duplicating the devices described in this manuscript should follow certification guidelines recommended by the relevant regional regulatory authorities.
Funding
National Institutes of Health (T32 DK7573-29).
Acknowledgment
We thank our fellow members of the N95DECON.org collaboration who have provided very useful reviews of the current literature on decontamination procedures. We would like to thank Manu Prakash for his leadership in organizing the N95DECON collaboration and for his enthusiastic support and encouragement of this research.
Disclosures
ME has ownership in a small amount (10 shares) of 3M stock.
See Supplement 1 for supporting content.
REFERENCES
1. M. L. Ranney, V. Griffeth, and A. K. Jha, “Critical supply shortages—the need for ventilators and personal protective equipment during the Covid-19 pandemic,” New Engl. J. Med. 382, e41 (2020). [CrossRef]
2. Center for Disease Control and Prevention, “Recommended guidance for extended use and limited reuse of N95 filtering facepiece respirators in healthcare settings,” 2020, https://www.cdc.gov/niosh/topics/hcwcontrols/recommendedguidanceextuse.html.
3. Centers for Disease Control and Prevention, “Strategies for optimizing the supply of N95 respirators,” 2020, https://www.cdc.gov/coronavirus/2019-ncov/hcp/respirators-strategy/index.html.
4. B. Kaiser and T. Hsu, “Where thousands of masks a day are decontaminated to battle the virus,” New York Times, 13 April 2020.
5. G. Kolata, “As coronavirus looms, mask shortage gives rise to promising approach,” New York Times, 22 March 2020.
6. K. C. Jelden, S. G. Gibbs, P. W. Smith, A. L. Hewlett, P. C. Iwen, K. K. Schmid, and J. J. Lowe, “Comparison of hospital room surface disinfection using a novel ultraviolet germicidal irradiation (UVGI) generator,” J. Occupational Environ. Hyg. 13, 690–698 (2016). [CrossRef]
7. N. G. Reed, “The history of ultraviolet germicidal irradiation for air disinfection,” Public Health Rep. 125, 15–27 (2010). [CrossRef]
8. U.S. Department of Health and Human Services, “Recommendations for sponsors requesting EUAs for decontamination and bioburden reduction systems for surgical masks and respirators during the Coronavirus disease 2019 (COVID19) public health emergency,” 2020, https://www.fda.gov/media/138362/download.
9. B. Heimbuch and D. Harnish, “Research to mitigate a shortage of respiratory protection devices during public health emergencies,” 2020, https://www.ara.com/sites/default/files/MitigateShortageofRespiratoryProtectionDevices_3.pdf.
10. W. G. Lindsley, S. B. Martin Jr., R. E. Thewlis, K. Sarkisian, J. O. Nwoko, K. R. Mead, and J. D. Noti, “Effects of ultraviolet germicidal irradiation (UVGI) on N95 respirator filtration performance and structural integrity,” J. Occupational Environ. Hyg. 12, 509–517 (2015). [CrossRef]
11. R. Randive, “Using UV reflective materials to maximize disinfection,” 2020, https://www.klaran.com/images/kb/application-notes/Using-UV-Reflective-Materials-to-Maximize-Disinfection---Application-Note---AN011.pdf.
12. W. Kowalski, Ultraviolet Germicidal Irradiation Handbook: UVGI for Air and Surface Disinfection (Springer, 2009).
13. J. P. Wilde, T. M. Baer, and L. Hesselink, “Modeling UV-C irradiation chambers for mask decontamination using Zemax OpticStudio,” Appl. Opt. 59, 7596–7605 (2020) [CrossRef] .
14. D. A. Newcombe, A. C. Schuerger, J. N. Benardini, D. Dickinson, R. Tanner, and K. Venkateswaran, “Survival of spacecraft-associated microorganisms under simulated Martian UV irradiation,” Appl. Environ. Microbiol. 71, 8147–8156 (2005). [CrossRef]
15. L. Link, J. Sawyer, K. Venkateswaran, and W. Nicholson, “Extreme spore UV resistance of bacillus pumilus isolates obtained from an ultraclean spacecraft assembly facility,” Microb. Ecol. 47, 159–163 (2004). [CrossRef]
16. L. Liao, W. Xiao, M. Zhao, X. Yu, H. Wang, Q. Wang, S. Chu, and Y. Cui, “Can N95 respirators be reused after disinfection? How many times?” ACS Nano 14, 6348–6356 (2020). [CrossRef]
17. J. Cavallo and V. A. DeLeo, “Sunburn,” Dermatologic Clinics 4, 181–187 (1986). [CrossRef]
18. EPA United States Environmental Protection Agency, “Eight-hour average ozone concentrations,” 2020, https://www3.epa.gov/region1/airquality/avg8hr.html#:∼:text=Based%20on%20extensive%20scientific%20evidence,over%20an%208%2Dhour%20period.
19. D. Mills, D. A. Harnish, C. Lawrence, M. Sandoval-Powers, and B. K. Heimbuch, “Ultraviolet germicidal irradiation of influenza contaminated N95 filtering facepiece respirators,” Am. J. Infection Control 46, e49–e55 (2018). [CrossRef]
20. EPA United States Environment Protection Agency, “List N: disinfectants for use against SARS-CoV-2 (COVID-19),” 2020, https://www.epa.gov/pesticide-registration/list-n-disinfectants-use-against-sars-cov-2-covid-19.
21. L. Chu and A. Price, “Addressing COVID-19 face mask shortages,” 2020, https://stanfordmedicine.app.box.com/v/covid19-PPE-1-2.
22. M.-H. Woo, A. Grippin, D. Anwar, T. Smith, C.-Y. Wu, and J. D. Wander, “Effects of relative humidity and spraying medium on UV decontamination of filters loaded with viral aerosols,” Appl. Environ. Microbiol. 78, 5781–5787 (2012). [CrossRef]
23. R. Wallace, M. Ouellette, and J. Jean, “Effect of UV -C light or hydrogen peroxide wipes on the inactivation of methicillin-resistant staphylococcus aureus, clostridium difficile spores, and norovirus surrogate,” J. Appl. Microbiol. 127, 586–597 (2019). [CrossRef]
24. C.-C. Tseng and C.-S. Li, “Inactivation of viruses on surfaces by ultraviolet germicidal irradiation,” J. Occupational Environ. Hyg. 4, 400–405 (2007). [CrossRef]
