In the absence of standardized methods and research data for determining the VFE of masks, the present study focused on the evaluation of four different experimental setups that can be used for this purpose. In parallel, BFE was determined on the same mask samples according to the standardized method EN 14683:2019+AC:2019 and its modified version. This allowed a direct comparison and assessment of the standardized and non-standardized method, which further facilitated evaluation of the quality of VFE tests as well as the suitability of MS2 as a model virus for the VFE test in various experimental setups.
Evaluation of the experimental setups
Initially, 5 or 6 different experimental setups (including both BFE and VFE) were tested on two mask samples, A and B, both made of three-layer polypropylene, to compare and assess each of the experimental setup. Three to five subsamples of each mask sample were tested in 3-4 independent repetitions, each time giving similar FE results (Supplementary Tables S1-S6), indicating high reproducibility of the developed test systems (coefficient of variation, CV, ≤ 1%) (Table 2). This was also confirmed by the generation of droplets of similar average diameter of 3.1 μm ± 0.3 μm in the experiments with Andersen sampler along with the maintenance of stable bacterial and viral concentrations in the PCs of the same experimental setup (Supplementary Tables S1-S6).
Table 2. Reproducibility of bacterial (BFE) or viral (VFE) filtration efficiency determined with different experimental setups (BFE: setups I and II; VFE: setups III – VI). The coefficient of variation (%) was calculated from all subsamples of each mask sample in the same experimental setup (A – F).
Mask sample
|
Experimental setup
|
Coefficient of variation (%)
|
A
|
I
|
0.16
|
II
|
0.07
|
III
|
0.26
|
IV
|
0.15
|
V
|
0.17
|
VI
|
0.14
|
B
|
I
|
1.00
|
II
|
0.94
|
III
|
0.48
|
IV
|
0.09
|
V
|
0.61
|
C
|
I
|
2.14
|
IV
|
0.73
|
D
|
I
|
5.34
|
V
|
3.15
|
E
|
I
|
0.06
|
V
|
0.002
|
F
|
I
|
2.11
|
V
|
1.63
|
Interestingly, the results show that the average BFE and VFE are slightly lower when using Andersen sampler than when collecting bacteria and viruses in impingers (Table 3). This could be linked to the initial bacterial and viral concentrations, which were lower in experimental setups with Andersen sampler and up to 3.07 x 103 CFU or PFU (Supplementary Tables S1-S6). It is very important to use the correct initial concentration in the Andersen sampler, as the plates can become saturated with microorganisms, obstructing accurate determination of FE. Hence, the statistical correction, i.e., the positive hole correction23, is applied to determine CFU and PFU, which anticipates that more than 1 bacteria or virus can pass through each hole of an individual stage (representative plates for BFE and VFE tests in Andersen sampler are shown in Supplementary Figure S1). The determined concentration is thus estimated and can differ from the actual concentration determined by classical growing and counting of CFU and PFU. Since the VFE values obtained with the Andersen sampler are on average lower than the VFE values in the impingers, the experimental setup with the Andersen sampler presents a safer choice, as when it comes to protective equipment, it is better to underestimate the FE and test the “worst case” filtration efficiency than to overestimate it25. Moreover, working with this sampler allows to determine the average droplet size and to work with airflows corresponding to respiration. In addition, a single Andersen sampler is sufficient for all subsamples, unlike impingers, which require the use of a new glass cup for each subsample. In addition, the plates are transferred from the Andersen sampler directly to the incubator without the need for processing as with impingers.
On the other hand, if it is necessary to work with higher initial concentrations of viruses or if the samples require additional processing and testing for other properties, then impingers are a way to go. However, the type of impinger and pump must be selected based on the desired airflow and considering the practicality of the experimental setup. In our opinion (from the experimental setups that used impingers) the combination in experimental setup V worked the best. A commercial laboratory practice also supports our conclusion, as they use impinger only for determination of VFE with increased challenge, i.e., when the initial viral concentration is higher than 3.3 x 103 PFU/test, while otherwise use Andersen sampler17. They, however, do it in combination with phix 174 as a model virus. Only a few other groups have worked on determining VFE in a similar setup. They either worked with MS219 or phix 17420, which they sampled using the Andersen sampler. In addition, a completely different experimental setup was also developed for determination of VFE, using the mannequin head with an aerosol source simulator and a SARS-CoV- 2 pseudovirus as a model virus18.
Table 3. Average bacterial (BFE) and viral filtration efficiency (VFE)
Experimental setup
|
I
|
II
|
III
|
IV
|
V
|
VI
|
Typea
|
Mask sample
|
BFE (%)b
|
VFE (%)b
|
|
A
|
99.8
|
99.9
|
99.4
|
99.9
|
99.8
|
99.8
|
II
|
B
|
96
|
98
|
98
|
99.3
|
99
|
-
|
I or II
|
C
|
91
|
-
|
-
|
-
|
92
|
-
|
NA
|
D
|
79
|
-
|
-
|
-
|
87
|
-
|
NA
|
E
|
99.9
|
-
|
-
|
-
|
99.999
|
-
|
II
|
F
|
91
|
-
|
-
|
97
|
-
|
-
|
NA
|
BFE and VFE values were calculated from all subsamples of the same experimental setup for an individual mask sample; aAccording to EN 14683:2019+AC:2019, Type I and II are determined. This is not applicable (NA) to reusable cloth masks. bWhen the BFE and VFE values were between 99% and 100%, more decimal places were included to show the exact filtration efficiency of the mask.
Several other factors are known to affect the FE of masks, including airflow20. Higher airflow decreases FE, likely due to the shorter time available for droplets to diffuse or interact with the electrostatically charged fibers26. In the experimental setups with the Andersen sampler (I and III) and the homemade impinger (II and IV), the airflow velocity was similar (28.3 L/min vs. 31.2 L/min), while much lower velocities were measured with type II impinger (V and VI) (10.3 and ~6.1 L/min, respectively). Since the FE results of the same mask sample were similar, regardless of airflow, they indicate that air velocity is not that crucial in VFE tests. In addition to mask samples A and B, four other mask samples were used. This enabled us to evaluate if different experimental setups for VFE testing were suitable for testing of masks of different quality and FE. It also helped in the final determination of whether MS2 is an appropriate viral model for VFE testing. Finally, the inclusion of these mask samples allowed us to evaluate how efficient masks are in general and how much they can mitigate the spread of respiratory viruses when worn properly. As was the case for mask samples A and B, the results obtained for the other masks also indicate the robustness, reliability, and repeatability of the experimental procedures developed, as confirmed by the FE results, generation of droplets of similar average diameter along with the maintenance of stable bacterial and viral concentrations in the positive controls of the same experimental setup (Supplementary Tables S1,S2,S4,S5). In addition, the CVs for mask samples C-F were also low and, as expected, higher for the masks with the lower FE (Table 2). In addition to reliability, the results obtained with the 6 different mask samples also demonstrate that MS2 is a suitable viral model for VFE testing for different experimental setups and that VFE testing can be performed for masks made of different materials and with different FEs.
Despite some observed differences, the average VFE values of the same mask sample are quite similar regardless of the experimental setup and are comparable to BFE results (Figure 3, Table 3). This is not surprising since the same nebulizer was used in all experiments, producing droplets and aerosols of the same size, with diameters large enough to contain either bacteria (S. aureus has a diameter of 0.5-1 μm) or viruses (MS2 has a diameter of about 27 nm). A similar observation was also made by Rengasamy et al., 201720, when they compared BFE and VFE values for the same masks.
We have shown that all experimental setups tested for the determination of VFE can be sufficient and that the decision of which experimental setup to choose depends on several factors, as described above. Therefore, this comprehensive study can serve as a great foundation for implementing VFE testing into existing standards for mask testing.
Determination of filtration efficiency of masks
As expected, mask sample A, classified as a Type II surgical mask (EN 14683:2019+AC:2019), resulted in BFE and VFE above 99% in all experimental setups (Table 3, Figure 3). A similarly high FE was also obtained for mask sample B, with some observed differences between the experimental setups, which could classify the final product, i.e., surgical masks made from this material as Type I or II. Knowing that the characteristics of the material can vary and depend on several process parameters27, the determined differences in FE could be due to the fact that three separate layers of polypropylene were manually assembled for testing, whereas for the final product, all three materials are pressed together. Mask sample E also had a very high FE value ≥ 99.9%, which was expected considering that this mask sample was an FFP2 respirator, although a different standard is normally used to determine the FE of respirators. We also tested BFE and VFE of three reusable, cloth mask samples, C, D, and F. Mask samples C and F, both made of two layers of cotton, had FE of >90%. Mask sample D contained polyester and polypropylene in addition to cotton and had lower BFE and VFE of 79 and 87%, respectively. It is known that cotton can ensure high FE, but it depends on the number of layers and thread count, as they assure physical filtration. On the other hand, materials like polyester have moderate electrostatic discharge, which is better for filtration of smaller aerosols (<300 nm)16. Since we produced droplets with an average size of 3 µm, we do not know if FE of these masks would be better for smaller particles only and whether mask sample D would be superior in filtering such particles compared to the other two cloth masks.
Similar observations on the FEs of different masks were made by other groups, which found high FEs (either BFE, VFE, or particle filtration efficiency, PFE) in surgical masks and respirators, while cloth masks had different FEs, which in some cases were quite high, above 90%18–20,28. Caution should be exercised in interpreting these results, because the fit of surgical masks is usually not taken into account when FE is determined, and therefore the protection provided by the masks does not necessarily correspond to the measured FE. For example, the BFE systems described in the standards and the VFE systems based on them usually test the FE of one area of the mask (in our case it was 8 x 8 cm for respirators and 10 x 10 cm for other masks) that is tightly clamped so that the air carrying bacteria/viruses can only pass through the material. In reality, various masks, especially cloth masks, do not always fit tightly against the wearer's face and these openings can serve as a transmission route for various pathogens. Therefore, in addition to FE, another important property of the mask is fit, because the same mask can protect differently depending on the fit; the better the fit, the better the protection29. Thus, it is not surprising that respirators (which undergo total inward leakage testing) have been found to be most effective in containing the spread of SARS-CoV-230.
Face masks, when worn properly, are a very important part of the contingency plan to prevent the spread of respiratory pathogens such as SARS-CoV-2. This has been confirmed by various studies that have shown that surgical masks, respirators, cloth masks, or even just fabrics used to make various textiles16, have high FEs. The type of mask to choose for protection depends on several factors, such as frequency of contact with infected persons, length of time spent in poorly ventilated enclosed spaces, and the state of the immune system.