3.1 Effects of respirator material and UV-C fluence
Coupons from two 3M N95 respirator models (8110s & 9210) were inoculated with MS2 and exposed to an array of UV-C 280 nm fluences ranging from 50 to 1,000 mJ cm− 2 to examine the relative effects of fluence and type of respirator material. The results of this experiment are shown in Fig. 3.
FFR model 9210
At 50, 100, and 500 mJ cm− 2 (Fig. 3), MS2 LRVs observed for the 9210 FFR differed significantly between L1 and L3 (p = 0.0009, 0.01, & 0.01, respectively). At fluences of 700 and 1,000 mJ cm− 2, no significant difference in MS2 reduction was observed between L1 and L3 (p = 0.9 & 0.07, respectively). This difference suggests that the outer-most layer (L1) requires a higher UV-C fluence for inactivation. For L1, MS2 LRV increases between fluences were significant up until 500 mJ cm− 2, after which no significant differences were observed. For L3, LRV increases were significant between 50 and 100 mJ cm− 2 (p = 0.02), after which no significant increases in LRV were observed. In Supplementary Figure S1, P. aeruginosa LRVs observed for 9210-L1 were similar to that observed with MS2. The similarity in kinetics between MS2 and P. aeruginosa was surprising, given that MS2 typically requires UV fluences which are orders of magnitude higher than what is required for a similar LRV of P. aeruginosa 28,29. Supplementary Figure S1 illustrates an observable difference in P. aeruginosa LRVs between 100 and 500 mJ cm− 2. Between 500 and 1,000 mJ cm− 2, the LRVs were virtually the same, both having two treated samples with non-detectable levels of P. aeruginosa.
FFR model 8110s/8210
A significant difference in LRV was observed between L1 and L3 of the 8110s FFR at 500 mJ cm− 2 (p = 0.02) but not at 100 and 1,000 mJ cm− 2 (p = 0.3 for both fluences). Further, there was no significant increase in LRV at fluences after 100 mJ cm− 2 (p > 0.05). The most drastic LRV differences were observed between the 9210 and 8110s respirator models. Significant differences (p < < 0.05) were observed between both respirator models for both arrangements and all paired fluences (100, 500, & 1,000 mJ cm− 2).
The disparities in LRV observed between the two respirator models are consistent with previous studies 5,30, who found that the effectiveness of various decontamination methods was model-dependent, given their differences in design, materials, and hydrophobicity. However, a more recent study31 did not find a statistical difference between hydrophobic versus hydrophilic respirator materials. In summary, the outer-most layer (L1) of the 9210 FFR requires higher UV-C fluences than the inner-most layer (L3) to reach the maximum observed MS2 LRV of just over 5. Additionally, UV is considerably more effective for inactivating MS2 on 9210 FFRs relative to 8110s FFRs and higher than 5 LRVs on MS2 are achievable at UV-C 280 nm fluences of 100 and 500 mJ cm− 2 for L3 and L1 of the 9210 FFR, respectively.
As shown in Fig. 3, both respirator models and the inoculation/exposure arrangement considerably affected UV efficacy. Although both FFR models are comparable in terms of certification by The National Institute for Occupational Safety and Health (NIOSH), they responded differently in terms of UV disinfection. LRVs above 1.5 were not observed with the 8110s/8210 FFR model, even at fluences of 1,000 mJ cm− 2. The 9210 FFR model resulted in a more pronounced disinfection curve, where LRVs between 5 and 6 were observed for both L1 and L3 starting at doses of 500 and 100 mJ cm− 2, respectively. Proposed stringent regulations of 3 and 6 LRVs 32 for Tier 3 and Tier 2 certification, respectively; however, these results suggest that UV disinfection technologies may behold specificity for FFR models, which is not contemplated in regulatory constructs. Future UV validation studies for FFR disinfection should place emphasis on the model of FFR investigated to understand disinfection efficacy and specificity further.
3.2 FFR elastic material and UV-C fluence
Strap segments from two FFR models were inoculated with MS2 and exposed to UV-C irradiation at fluences of 280-nm. The objectives were to i) understand the UV-C kinetics for MS2 on FFR straps and ii) determine if strap material plays a role in UV efficacy for treating FFR straps. Rubber-elastic (9210) and polyester-elastic (1860) straps were tested. The LRVs observed with the rubber elastic from the 9210 FFR increased steadily until 500 mJ cm− 2, where all samples were reduced to either below the LOQ or below detection. However, the LRVs observed with the polyester elastic from the 1860 FFR were considerably less. LRVs remained at or below 0.5 for fluences of 50 to 1,000 mJ cm− 2. When each side of the elastic was exposed to 700 mJ cm− 2 for a total of 1,400 mJ cm− 2, LRVs were higher than when one side was exposed to 1,000 mJ cm− 2; however, the difference was relatively small (0.17, p = 0.003).
Figure 5 shows that similarly to the 9210-respirator material, MS2 LRVs > 5 are observed at a fluence of 500 mJ cm− 2 for 9210 rubber-elastic straps. When treating the polyester-elastic 1860 respirator straps, it was impossible to achieve more than one LRV, even when exposed to UV-C fluences higher than 1,000 mJ cm− 2, which are generally sufficient under other conditions. Additionally, even when the 1860 polyester straps were exposed to UV-C fluences (700 mJ cm− 2 per side), LRVs were not substantially increased. SEM results (Fig. 6) show that the polyester strap is much more intricate at a microscopic level than the rubber strap from other FFR models. This intricate geometry is likely absorbing the inoculum and preventing UV-C radiation from penetrating the strap material, in contrast with rubber straps where the inoculum droplets stay on top of the material.
Including strap material in disinfection experiments is critical to determine if specific N95 FFRs models will be suitable candidates for a given disinfection strategy. The results presented in Fig. 5 show that the 1860 FFR model and any other model with polyester nature straps may not be suitable for UV-C disinfection, at least below an applied fluence of 1,400 mJ cm− 2.
3.3 Effects of bacterial loading
The relative effects of microbial concentrations on disinfection performance are not well understood, especially concerning FFR materials. For this reason, we investigated the relative effects of P. aeruginosa cell-density inoculated onto 9210 FFR coupons on LRVs (Fig. 6).
P. aeruginosa was recovered at similar concentrations in treated samples (p < 0.05 for all comparisons) across all cell-densities examined, resulting in an increased LRV as cell-densities increased (Fig. 6). Increased cell-densities that would result in more significant shielding were expected but were not observed in this result. The data in this study suggest that if cell-densities are too low, there is a diminishing return in efficacy with respect to LRVs. The implications of these results are essential for the design of standardized performance-testing protocols. Furthermore, a more direct comparison between studies would be possible if cell-densities were reported. For example, a recent study 31 achieved > 5 LRV with MS2, whereas other studies 13,33only achieved around 3 LRV with MS2 as well. None of these studies mentioned cell density; additionally, inoculation medium, FFR model used, and inoculation technique was also different. These disparity among studies may be impacted by cell-densities or other factors that were different, such as respirator type and inoculation location, as shown in Fig. 3.
A second experiment was conducted to address potential differences at higher cell-densities and a higher UV-C 280-nm fluence. As shown in Fig. 5, there may exist a critical cell-density threshold somewhere between 6.2 and 7.7 log CFU cm− 2, in which cell-densities exceeding such a threshold result in over-aggregation of bacterial cells, leading to reduced LRVs.
3.4 Materials characterization of three N95 FFR models
Figure 6 summarizes the material characterization of the 1860, 8210 and 9210 N95 respirators. Only the three primary and more easily separated layers were analyzed. Layer nomenclature is as depicted in Fig. 1A. The main polymers found in the layers of the respirator were Polypropylene (PP), Polyethylene Terephthalate (PET-P), Polydimethylsiloxane (PDMS), and Eltec P HP-603 Polypropylene.
PP is considered a thermoplastic polymer used in a wide range of applications. PP presents non-polar properties, which indicates a low interaction with water 34. In contrast, PET is a polar plastic, commonly found in plastic bottles 35. PET has an intrinsic viscosity and hygroscopic nature (retains water from its surroundings). Moreover, PDMS is a commercially-available silicon rubber 36 that is viscoelastic and hydrophobic (repels water).
The 1860 model layers were mainly comprised of PP (L1 and L2) and PET (L3), while the straps were composed of PET. Comparatively, the 9210 model mainly had PP (L1, L2 and L3) and PDMS (rubber) for the straps, while the 8210 model primarily consisted of PET (L1 and L3), Eltec P HP-603 Polypropylene (L2), and Bunatex K 71 (straps).
The respirator materials and their configuration within the layers of the respirators could explain the difference in disinfection found in the respirator microbial testing. Figure 3 and Fig. 4 show that the 9210 respirator achieved higher LRVs in both MS2 and P. aeruginosa tests. The 9210 and 1860 respirator models have a more plastic feel on L1 than the 8210 model, which has a softer and more fabric-like texture.
The 1860 respirator straps, which have a fabric-like texture, resulted in low LRVs compared to the 9210 model, even when applying 700 mJ cm− 2 per side. The low LRV value achieved on the 1860 strap may be attributed to the hygroscopic nature of PET, contrary to the 9210-respirator strap, which is mainly composed of hydrophobic PDMS.
The higher disinfection efficiency on the 9210 respirator strap, compared to the 1860-respirator strap, may be attributed to the hydrophobic nature of the strap. Hydrophobic layers ensure that the bulk of UV exposure occurs on the surface of respirators. In contrast, hygroscopic layers enhance the penetration of inoculum deeper into the textile, inhibiting the microorganisms from being adequately exposed to UV light. Figure 6 shows the overall structure for each of the tested respirator layers and depicts gaps in the material where droplets could reach. It is worth noting that since FFRs are single-use PPE items, the original design of the straps and choice of material was likely based on other desired features, such as comfort and durability. Incidentally, the pandemic has created a unique demand on the repurposing of FFRs (REF), and the strap material must be assessed in disinfection experiments as it appears to have a profound effect on UV disinfection efficiency.
The findings presented in this manuscript provide evidence that respirators with hygroscopic properties may not be suitable for UV disinfection alone, as their absorbent materials may attract and retain droplets where microorganisms are present. Other studies have also found that different FFR models respond differently to UV disinfection 13,31. However, comparison between FFR models and their material properties has not often been incorporated in previous studies, or their results have been inconclusive 37,38.
3.5 Implications of UV disinfection on surfaces
UV disinfection has been used primarily in the drinking water and wastewater industries since the maturation of the technology in the past decades 39. More recently, UV technology has been increasingly used for disinfection of surfaces in hospital settings 40–42. A recent study examined the impact of common hospital surfaces (plastic, stainless steel and copper) on UV disinfection efficiency 43; however, there is still a significant gap in UV surface disinfection knowledge. As mentioned by a recent study44, the applied UV fluence delivered onto a surface does not necessarily reflect the UV fluence received. Moreover, surface irregularities and crevices at the microscopic level could create shadowed areas where the UV light cannot penetrate. Furthermore, the porous multilayer structure of an N95 FFR requires roughly two orders of magnitude higher applied UV fluence for sufficient inactivation when compared to a smooth surface material.
The COVID-19 pandemic has created an urgent need and interest to disinfect a broader range of complex surfaces, including N95 FFRs, surgical masks and other forms of PPE. This scenario presents a challenge for the development of disinfection protocols as there are many factors that can influence the effectiveness of UV treatment, such as surface geometry, material type and FFR construction. Recent studies have successfully applied UV technology for the repurposing of PPE in clinical settings 38,45; moreover, systematic reviews on the topic have concluded that UV disinfection is a suitable technology for PPE repurposing 12,44,46−48. However, not all studies have considered the effect of different FFR layer materials on UV disinfection efficacy.
While the current literature agrees that UV disinfection is suitable for FFR repurposing, there has not been a consensus on the UV fluence required. However, an application of at least 1000 mJ cm− 2 is the most common value reported 13,30,45. Additionally, there is not unanimity on the required LRV to claim successful FFR disinfection, as these values have ranged from > 3 to > 6 LRV and involved different target microorganisms. Moreover, there is still debate whether the type of FFR material dictates the UV fluence required for disinfection, even though some studies have found evidence that the hydrophobicity/hydrophilicity of materials play a role in disinfection efficacy 13,30,49. In contrast, other study 31did not find a difference in disinfection performance between hydrophilic and hydrophobic materials used in FFR layers; however, the authors of the study did not provide a characterization of the materials.
The inconsistency of results between studies for UV disinfection of FFRs could be attributed to the exclusion of the impact of respirator materials on disinfection performance. This work indicates that not all materials used in the construction of FFRs respond equally to UV treatment. To the author’s knowledge, this is the first manuscript that analyzes FFR disinfection efficiency as a function of layer material and composition when using a UV-C light source at 280 nm. It is hypothesized that differences in disinfection efficiency will be similarly impacted by material type across the UV-C spectrum.
3.6 Respirator layer analysis
N95 FFRs are designed to repel droplets from the outer layers and electrostatically trap microorganisms within the respirator’s inner layers 50. However, the reuse of N95 FFRs may still pose a health hazard to users if pathogenic microorganisms are not adequately inactivated. An experiment was conducted to assess several combinations of inter-layer MS2 inoculation and UV-C exposure direction at a fluence of 500 mJ cm− 2. Figure 7 shows the differences in LRVs achieved.
MS2 recoveries for 9210 FFR layers varied layer to layer, which impacted the level of measurable LRVs in many cases. Arrangements 3 and 5 were the only arrangements that used the front section of the 9210 FFR, which included L2 (Fig. 1). Tests with all other arrangements were carried out with the top section of the 9210 FFR, which did not include L2.
The high LRVs for arrangements where embedded layers were inoculated were L3-a, depicted in Fig. 1, was decisively the layer that blocked the most UV-C light. This can be concluded by the fact that arrangements 4 and 5 were the only arrangements that did not result in LRVs greater than 4. These results suggest that if 9210 FFRs are exposed to UV-C 280 nm from both sides, LRVs above 4 may be expected at fluences of 500 mJ cm− 2. However, this may be a best-case scenario and does not account for areas on the respirator where additional blockage may occur, such as straps and nose pads.