Face mask performance related to potentially infectious aerosol particles, breathing mode and facial leakage

Abstract

During the COVID 19 pandemic, wearing certified Respiratory Protective Devices (RPDs) provided important means of protection against direct and indirect infections caused by virus-laden aerosols. Assessing the RPD performance associated with infection prevention in standardised certification tests, however, faces drawbacks, such as the representativeness of the test aerosols used, the protection of third parties during exhalation or the effect of facial leaks. To address these drawbacks, we designed a novel test bench to measure RPD performance, namely the number based total efficiency, size-segregated fractional filtration efficiency and net pressure loss, for 11 types of certified surgical masks and Filtering Face Pieces dependent on breathing mode and facial fit. To be representative for the context of potentially infectious particles, we use a test aerosol based on artificial saliva that is in its size distribution similar to exhaled aerosols. In inhalation mode excluding facial leaks, all investigated samples deposit by count more than 85% of artificial saliva particles, which suggests a high efficiency of certified RPD filter media related to these particles. In exhalation mode most RPDs tend to have similar efficiencies but lower pressure losses. This deviation tends to be significant primarily for the RPDs with thin filter layers like surgical masks or Filtering Face Pieces containing nanofibers and may depend on the RPDs shape. Both the filtration efficiency and pressure loss are strongly inter-dependent and significantly lower when RPDs are naturally fitted including facial leaks, leading to a wide efficiency range of approximately 30–85%. The results indicate a much greater influence of the facial fit than the filter material itself. Furthermore, RPDs tend be more effective in self-protection than in third-party protection, which is inversely correlated to pressure loss. Comparing different types of RPDs, the pressure loss partially differs at similar filtration efficiencies, which points out the influence of the material and the filter area on pressure loss.

1. Introduction

Pathogen dissemination through aerosol particles emitted by the respiratory system best explains several super-spreading events during the COVID-19 pandemic (Katelaris et al., 2021; Kutter et al., 2021; Lu et al., 2020; Zhang et al., 2020) and is therefore in the focus of SARS-CoV-2 transmission. Particles that are formed and expelled through the respiratory system, for example when talking, coughing or breathing, may differ in size and number based on several factors such as the individual's physiology, health condition or activity (Archer et al., 2022; Morawska et al., 2009; Schwarz et al., 2010). In SARS-CoV-2 infected persons, these particles may act as vehicles for pathogens (Gutmann et al., 2022; Ma et al., 2021) and thus are determinant for the definition of protective measures. Present studies suggest that the mode of the exhaled particle size distribution most likely is in the order of 0.1–0.5 μm (Scheuch, 2020) allowing for the particles to stay airborne over several hours in indoor environments. Contrary to breathing, talking or coughing produces larger particles from the submicronic and small micrometre range to particles larger than 50 μm (Alsved et al., 2020; Asadi et al., 2019). Exposure to these respiratory-emitted particles leads to two possible routes of infection. On the one hand, particles may be transported directly from an infected person to a susceptible host (direct route of infection), whereby the probability of larger particles reaching a susceptible host decreases with the distance due to the particles' settling velocity. Airborne transmission, on the other hand, only occurs indoors, where the smaller fraction of respiratory-emitted particles may accumulate in the indoor air with increasing durations of stay, numbers of persons present and their activity. Since particle transport is still possible after an infected person has left the room, airborne transmission may also be referred to as an indirect infection route (Brlek et al., 2020; Cai et al., 2020).

To reduce the risk of both direct and indirect infections, infection control measures such as ventilation, air purifying technologies or the wearing of Respiratory Protective Devices (RPDs) were discussed and introduced during the COVID-19 pandemic in many areas of public life, such as schools, kindergartens, offices, public buildings, hospitals or the transportation sector. While ventilation and air purifying technologies may affect mostly the indirect route of infection (Nardell, 2021), the wearing of RPDs counteracts both direct and indirect infections by reducing the number of inhaled as well as exhaled particles and thus potentially provides an effective means of protecting oneself (self-protection) and others (third-party protection) (Asadi et al., 2020). Since certified RPDs, in particular surgical masks (DIN EN 14683:2019-10) and filtering face piece respirators such as FFP (DIN EN 149:2009-08), N95 (NIOSH approved 42 CFR 84) or KN95 (GB 2626–2006) are subjected to standardised test procedures, requirements for the separation performance are defined. Filtering Face Pieces according to DIN EN 149:2009-08 are categorized into three classes, with class FFP2 requiring a mass-based total efficiency of at least 94% and the total inward leakage not exceeding 11%. Test procedures use aerosols containing submicronic solid-phase sodium chloride or liquid-phase paraffin oil particles with a broad range allowed for the geometric standard deviation (Zoller et al., 2021) that partly overlap the size range of potentially infectious aerosols (Penner et al., 2022). As the test procedure originates from occupational health and safety, the focus is on self-protection against occupational pollutants, with third-party protection not being considered. Surgical masks according to DIN EN 14683:2019-10, on the other hand, are designated to protect others from infectious droplets during medical procedures. In certification, the number-based total filtration efficiency of the filter medium is determined by the use of infectious particles from a bacterial suspension with a median diameter of 3 μm that are one order of magnitude larger than exhaled virus-laden aerosol particles from the respiratory tract. As with all filtration processes, however, the efficiency of the separation mechanisms is highly dependent on the particle size and particle characteristics (Hinds and Zhu, 2022; Lee and Liu, 1982). To effectively remove respiratory particles from both the inhaled and exhaled air, RPD filter media need to be highly efficient with respect to particles in the relevant size range and with similar properties to infectious particles such as shape, charge and density. Furthermore, the overall efficiency is dependent on the face-to-mask seal, whereby leakage flows can cause unfiltered breathing air to be inhaled or exhaled that bypasses the filter medium (Koh et al., 2021; Pan et al., 2021). As a result, the performance related to potentially infectious particles considering the nature and size of exhaled aerosol particles and also the filtration performance associated with facial leakages in both self-protection and third-party protection may be a drawback of certification procedures for evaluating the RPD performance in the COVID-19 pandemic context.

Several studies with a focus on RPD performance related to infection protection have already been conducted. Studies involving submicronic particle collectives to determine total filtration efficiencies of certified RPDs show that the certified filter media are highly effective even when considered on a number basis. Rengasamy et al. (2014) reported penetration rates of less than 1% for sealed respirators and less than 10% for surgical masks at a flow rate of 40 l/min using a NaCl aerosol. Bagheri et al. (2021) suggested similar penetration rates for FFP2 masks with dolomite dust, which are all below 6%, but have found a higher variance in the penetration rates of different surgical masks. This includes several masks with penetration rates below 12%, as well as up to 75%. Other work (Bałazy et al., 2006; Grinshpun et al., 2009; Zangmeister et al., 2020) similarly shows that penetrations are distributed over a wider area in surgical masks than in respirators. When looking at fractional efficiencies, the most penetrating particle size (MPPS) varies for certified RPDs and is typically in the order of 30–300 nm, with the upper bound being more relevant for surgical masks (Bagheri et al., 2021; Bałazy et al., 2006; Grinshpun et al., 2009; Zangmeister et al., 2020). RPDs that have not been sealed in the test procedure show that facial leakage sharply decreases the overall efficiency. Grinshpun et al. (2009) found that the total inward leakage is particle size dependent from 7 to 20 times greater than the penetration through the filter medium for respirators and size independent from 4.8 to 5.8 times greater for surgical masks. Various studies point to facial leakages lead to similar filtration efficiencies for both respirators and surgical masks, independent of the initial efficiency of the filter medium (Grinshpun et al., 2009; Li et al., 2006; Rengasamy et al., 2014). When looking at the total outward leakage, which is relevant for the effectivity in third-party protection, however, only a few studies were conducted. Koh et al. (2021) and Pan et al. (2021) indicate that both inward and outward leakages are similar in respirators. For surgical masks, however, they indicate that the outward leakage exceeds the inward leakage.

Despite the clear evidence that the filter media used in certified RPDs is efficient for submicronic particles, to date little is known about how the filtration performance is modulated by facial leaks on both the self-protection and third-party protection. Questions on how the filtration performance is influenced by the real use case in the context of infection prevention remain unanswered. For example, how is the filtration efficiency affected when using a test aerosol representative for exhaled aerosols? Does the certification or characteristics of the RPD influence the filtration performance in the real use case considering facial leaks? How is the pressure loss, as a measure for the breathing resistance, effected by facial leaks or the flow direction? To answer these questions, the aim of this work is to determine performance parameters, namely the fractional filtration efficiency, number based total efficiency and net pressure loss, for different RPD classes and characteristics under conditions representative for infection prevention in the COVID-19 pandemic context. This includes the

• −set-up of a test bench to determine the performance under representative test conditions.
• −selection of a suitable fluid for particle generation related to respiratory-emitted aerosols and the evaluation of the test aerosol by comparison to human exhaled particles.
• −determination of RPD performance as a result of fractional filtration efficiency, representative number based total efficiency and net pressure loss.
• −comparison of performance parameters dependent on flow direction in terms of self- and third-party protection, as well as dependent on facial fit considering facial leakages.

First, in Sec. 2, the basic transport mechanisms of filtration and the equations of the performance parameters are presented. Then, in Sec. 3, the experimental setup, the test procedure and the materials used are described. In Sec. 4, the results are presented and discussed. This includes the validation of the test aerosol with human exhaled particles, as well as the screening of different RPDs. Finally, in Sec. 5, a conclusion of this work is drawn and an outlook on future work is given.

2. Theory

Aerosol particles may be removed from the gas phase by porous media when they reach the inner surface of a filter through various transport mechanisms, namely Brownian diffusion, direct interception, inertial impaction and electrostatic attraction (see Fig. 1 ).

Fig. 1.

Schematic illustration of the transport mechanisms in depth filtration.

Transport mechanisms are strongly dependent on particle size and flow velocity. Brownian diffusion is the dominating mechanism for small particle sizes and low flow velocities, whereby the particle motion is governed by a superordinated chaotic movement. Thus, particles do not follow the streamlines exactly and may randomly hit filter fibres. Brownian diffusion may be significant for the separation of the smaller particle fraction in the size of the infectious SARS-CoV-2. Particles that follow exactly the streamlines may be removed by direct interception, if the streamline passes within the particle radius on a filter fibre. Inertial impaction is mainly dominant on larger particles that, due to their inertia, are deflected of their streamline by its redirection around a fibre. With mask leakage, also the total leakage flow can be accelerated and redirected at the mask-to-face seal, potentially allowing inertial impaction to effect the deposition of larger particles in the case of unsealed RPDs (Hinds and Kraske, 1987). The interaction of the three transport mechanisms typically results in a most penetrating particle size (MPPS), which represents the least effectively separated particle size and is thus a characteristic of the respective filter medium. In air filtration, the MPPS is empirically around 0.3 μm and therefore in the size range of exhaled aerosol particles. In order to increase the removal probability of small particles, filter media, such as media based on nanofibres, aim to increase the efficiency at the MPPS through small pore sizes. However, materials based on synthetic melt blown fibres are most commonly used in RPDs. Meltblown fibres are electrostatically charged due to their manufacturing process and therefore able to attract particles of the opposite charge by electrostatic attraction. This may be advantageous in increasing the efficiency at the MPPS without reducing permeability and thus, increasing pressure loss.

To evaluate filtration performance dependent on particle size as well as to determine the MPPS, the fractional filtration efficiency is an elementary parameter. The fractional filtration efficiency is defined according to Eq. 1

$$E_{C_n}\left(x_i\right)=\frac{{dC}_{n,upstream}\left(x_i\right)-{dC}_{n,downstream}\left(x_i\right)}{{dC}_{n,upstream}\left(x_i\right)}$$ (1)

and represents the measurable difference in particle concentration of discretised particle size intervals in the raw gas (upstream of the filter) and clean gas (downstream of the filter) related to the raw gas particle concentration.

When only considering total particle concentrations, the total filtration efficiency is obtained according to Eq. (2).

$$E_{C_n}=\frac{C_{n,upstream}-C_{n,downstream}}{C_{n,upstream}}$$ (2)

Compared to the fractional filtration efficiency, the total filtration efficiency depends on the particle size distribution and the metric with which particle concentrations are measured (Zoller et al., 2021). Therefore, test aerosols with a size distribution similar to that of potentially infectious aerosol particles are a pre-requisite to evaluate the protective effect of RPDs. With mass concentrations, larger particles have a higher relative importance for the overall efficiency than with number concentrations. However, since particularly the small particles are relevant in the context of disease transmission, we only use efficiencies based on number concentrations.

To evaluate the wearing comfort of RPDs, the second performance parameter is the pressure loss, which is given in porous media by the Darcy equation if the flow is creeping, i.e. if the Reynolds number is smaller than 1:

$$\frac{\Delta p}{H}=\frac{\eta \bullet {\bar{v}}_f}{B}$$ (3)

The pressure loss $\Delta p$ related to the layer thickness $H$ depends on the dynamic viscosity $\eta$, the filter velocity ${\bar{v}}_f$ and the permeability$B$, which is a material constant depending on the fiber diameter and porosity. The net differential pressure ${\Delta p}_{net}$ of RPDs is determined similar to the procedure described in DIN EN 13274-3:2002-03

$${\Delta p}_{net}={\Delta p}_F-{\Delta p}_H$$ (4)

where ${\Delta p}_F$ is the measured differential pressure with the RPD mounted to a test head. ${\Delta p}_H$ takes into account pipe friction losses, changes in cross-section and diversions due to the measuring apparatus, which is determined from a second measurement.

3. Material and methods

In this section, the materials and methods used for testing RPDs are described. First, the mask test bench is presented in detail with focus on both the experimental set-up and the test procedure. Subsequently, the materials used and the considered RPDs are described.

3.1. Mask test bench

Fig. 2 first illustrates the experimental test set-up to determine filtration-specific performance parameters of RPDs. Essential components are an aerosol generator (1), a test head (2), a volume flow-controlled fan (3) and measuring devices for the fractional particle number concentration (4a) and the differential pressure (4 b).

Fig. 2.

Experimental test set-up for the determination of mask performance parameters (fractional filtration efficiency and net pressure loss).

The primary function of the aerosol generator (AGK 2000, Palas GmbH) (1) is to produce test aerosol particles from a feeding liquid by the use of a two-substance nozzle and compressed air in order to mimic infectious particles from the respiratory system. Test aerosol particles are injected at the beginning of the test bench tubing and thereby diluted with ambient air to obtain a dry aerosol in measurable concentration. The total volume flow is controlled and generated by a radial fan mounted on the suction side. For flow control, an ultrasonic flowmeter is used to contactless measure the pressure- and temperature-compensated volumetric flow without influencing the flow profile and thus interfering particle sampling. The different RPDs under consideration are mounted to an additively manufactured head within a measuring cell. The measuring cell allows the test head to be mounted in such a way that it can be either flowed through from the outside to the inside (third-party protection) or vice versa from the inside to the outside (self-protection). In order to be representative towards facial leaks the dimensions of this test head are similar to ISO/TS 16976-2:2015-04 and represent an average Central European head size. Differential pressure is measured by the use of static ring pressure taps upstream and downstream the measuring cell with two differential pressure sensors in the range of 250 Pa and 1250 Pa, respectively. Particle concentrations are measured in both the raw and clean gas using an optical particle counter (Promo 3000, Palas GmbH). Therefore, an intrument-specific sampling volume flow of 5 l/min is taken isokinetically upstream and downstream the measuring cell. The particle concentration is determined by scattered light using an optical particle sensor with a measuring range of 10⁶ P/cm³ (WELAS 2070).

3.2. Test procedure

A total of four different test scenarios are considered. First, RPDs are attached to the test head by their existing head or ear loops. This intends to mimic the natural fit with leakage flows through the mask-to-face seal may influence the RPD performance. Second, to exclude facial leakage, RPDs are firmly attached to the test head by the use of a sealing compound. This provides the performance of RPDs if they would perfectly fit to a wearer's face, which partly is a comparable configuration to standardised certification procedures. Both modes of attachment are looked at separately for two flow directions, inhalation and exhalation, by varying the position of the test head in the measuring cell. As a result, the mask performance for perfect and imperfect fitting RPDs can be ascertained distinctive for the flow directions of inhalation and exhalation in self-protection and third-party protection. Here, third-party protection only represents the efficiency of particle removal in the expiratory volume flow, while self-protection represents particle reduction in the inspiratory volume flow.

To prepare a measurement, first the test head is fitted with an RPD and then installed in the measuring cell according to the considered test scenario. A constant volume flow of 95 l/min is then applied to represent most unfavourable conditions. Thus, 95 l/min is an unrealistically high flow rate for breathing, it is also used in DIN EN 149:2009-08 DIN EN 149 as an inspiratory flow rate and intends to mimic the peak condition during sinusoidal breathing at 30 l/min according to DIN EN 13274–3. As a result, the determined mask performance is representative only for the peak condition of the breathing cycle. After preparation, tests are carried out under room air conditions (p = 10⁵ Pa, T = 25 °C, φ = 30–45%). Absolute pressure and temperature are measured online and used for volume flow compensation with regard to small fluctuations in ambient conditions. After equilibration, the pressure loss of the unloaded RPD is measured over a time interval of 30 s. The net pressure loss is then determined according to Eq. (4), subtracting the reference pressure loss of the test head and measuring cell in this configuration. After the pressure loss measurement is finished, test aerosol is injected into the tesdslbt tubing. In the first 300 s, the raw gas concentration is determined. Here, the loading time of 300 s is necessary to equilibrate the particle concentration in the measuring cell. The total number concentration in the raw gas is approx. 50,000 P/l, thus background particle concentration of 20 P/l is three orders of magnitude lower and is therefore neglected. After a steady-state particle concentration has been established, the clean gas concentration is determined during the next 300 s. To determine the fractional efficiency according to Eq. (1), the last 60 s of the raw gas measurement and the first 60 s of the clean gas measurement are used.

3.3. Artificial saliva

Aiming on a representative test aerosol to respiratory-emitted particles, a saliva substitute solution (apomix® Speichelersatzlösung SR) is used as a feeding liquid for aerosol generation. Saliva substitutes are often used to moisten the oral mucosa in patients with xerostomia and therefore intended to imitate certain properties of human saliva, such as viscosity (Łysik et al., 2019). In saliva substitutes, the viscosity is mainly influenced by either the additive carboxymethylcellulose (CMC) or mucin (Foglio-Bonda et al., 2022). Other components include electrolytes such as sodium chloride, potassium chloride, calcium chloride and magnesium chloride. Moreover, water, sorbitol and substances that serve as pH buffers and for preservation are contained.

3.4. RPDs

Certified surgical masks and filtering face pieces were selected for a screening in four different test scenarios, as described. The selected RPDs are listed in Fig. 3 and are categorized into five groups based on in their shape and characteristics. Four fish-shaped masks were considered, with two based on meltblown filter media (FFP2_3; FFP3_1) and two based on nanofibres (FFP2_1*; FFP2_2*). Further, duckbill-shaped (FFP2_4; FFP2_5) and classical axe-shaped filtering face pieces (FFP2_6; FFP2_7) all based on meltblown filter media were selected. In addition, two medical masks (SM_1; SM_2) as well as a reusable fabric mask with a nanofilter insert (FFP2_8* (R)) were screened.

Fig. 3.

Selected meltblown based and nanofibre* based RPDs for performance screening.

4. Results and discussion

Prior to the actual measurements, the particle size distribution of the test aerosol was investigated and compared to exhaled aerosols (Sec. 4.1) in order to evaluate its representativeness for respiratory-emitted aerosols. Thereafter, the actual RPD screening was done on five new mask samples in each configuration with the test bench and test procedure described in Section 3. Performance parameters, namely the net pressure loss, the number based total efficiency and the fractional filtration efficiency were determined, aiming at a differentiated distinction between flow direction (third-party/self-protection) and fitting (including/excluding facial leakage) (Sec. 4.2).

4.1. Test aerosol

Test aerosols for determining the filtration performance of RPDs can be generated from various liquids such as those used in certification, for example sodium chloride solutions and liquid paraffin oil (DIN EN 13274-7:2019-09,, DIN EN 149:2009-08,), or biogenic solutions containing viable bacteria (DIN EN 14683:2019-10). Unlike the norms, the focus of the RPD screening is to determine the mask performance based on a representative test aerosol that mimics respiratory emitted particles. Representative in this context means that the characteristic of droplets and the size distribution are similar between exhaled and technically generated particles. Therefore, we use a saliva substitute solution as a feeding liquid for technical aerosol generation (see Sec. 3.3). To evaluate representativeness, the particle size distribution of the technically generated aerosol from artificial saliva is compared to an exhaled aerosol optically measured in Penner et al. (2022) and compared in Fig. 4 . Since the total number of exhaled particles is several orders of magnitude lower than of technically generated aerosols, a different optical sensor with a lower measuring range was used for the exhalation measurement. To allow for comparison, the particle number concentration of each particle size interval (dC_n) is normalized to the total particle number concentration (C_n) as well as the logarithmic bin size (Δlog (x_i)) of the optical particle counter.

Fig. 4.

Comparison of the particle size distributions of the technically generated aerosol from saliva substitute solution and human exhaled aerosols of 13 subjects normalized to the total particle concentration and logarithmic bin size (Penner et al., 2022).

The size distribution of exhalation measurements is presented as the mean of 21 measurements of 13 test persons and compared to a single measurement of the saliva substitute solution that is technically dispersed with the aerosol generator. The results show that both aerosols contain particles in a similar size range that are mainly smaller than 2 μm. However, in the technically generated aerosol, the mode of the size distribution is close to the metrological boundary of the optical particle counter in the range of 0.2 μm. Here, counting errors may occur, which suggests that the actual concentration at the mode may be even higher. The mode for exhaled particles, on the other hand, is in the range of 0.4 μm and thus the exhaled size distribution contains relatively larger particles. Thus, the technical generation principle is based on a two-substance nozzle with larger particles partly being removed by a cyclone, the used aerosol generator does not mimic the generation mechanism of particles in the human lungs, which may explain the slight differences in both distributions. Another reason may be the test conditions for both set-ups, with the exhaled particle size distribution determined undiluted at an air humidity of approx. 90% due to the low particle concentration. Although the time required for exhaled particles to evaporate is very short due to their small size and the associated high surface tension (Gregson et al., 2022; Walker et al., 2021), incomplete evaporation cannot be ruled out in this set-up due to the high humidity. Technical aerosol, on the other hand, is diluted to a total flow of 95 l/min, which may result in a faster evaporation of the water content and thus to a smaller particle size. On the whole, the overall differences are minor; moreover, saliva substitute solution represents a more comparable fluid in terms of its composition and properties in the context of infection protection and is therefore used for the RPD screening.

4.2. Screening of RPDs in new condition

RPDs act as particle sinks for the inhaled and exhaled air and thus potentially provide an effective means of protecting oneself and others from direct and indirect infections. The filtration performance, however, may differ in self-protection and third-party protection for both perfect and natural fitted RPDs that partly allow for unfiltered breathing air to pass at the mask-to-face seal. The RPD screening aims to provide the performance related to this dependency on flow direction and facial leakage by the use of a representative test aerosol (Sec. 4.1) and a newly conceived test bench (Sec. 3.1). Therefore, 11 surgical masks and Filtering Face Pieces (Sec. 3.3) are tested at a steady-state volume flow of 95 l/min, which aims to represent the peak volume flow occurring during sinusoidal breathing at 30 l/min. For each type of RPD, the fractional filtration efficiency, number based total efficiency and the net pressure loss (Sec. 2) are determined using five new RPD samples. Fig. 5 illustrates the averaged fractional filtration efficiencies.

Fig. 5.

Fractional separation efficiencies of selected RPDs at 95 l/min using artificial saliva in sealed installation excluding leakage, as well as in natural installation including leakage, differentiated in self-protection and third-party protection. Each fractional filtration efficiency curve is the mean of five measurements on five new RPD samples.

The diagrams aligned vertically differ in whether RPDs were sealed or naturally fitted to the test head. When the RPDs were sealed, thus were “perfectly fitted”, 7 out of 11 masks exceed an efficiency of 95% at each particle size, which indicates a good filtration performance related to aerosol particles from saliva substitute. Surgical mask SM_1 is similar efficient compared to meltblown based Filtering Face Pieces, while the second surgical mask (SM_2) shows a lower efficiency that is still above 85% at the MPPS. RPDs containing nanofibres (FFP2_1*, FFP2_2*, FFP2_8*(R)) appear to have lower efficiencies of approx. 75% (disposable) and 85% (reusable) at the MPPS. For sealed nanofibre-based RPDs, however, the fractional filtration efficiency curves deviate significantly in self-protection and third-party protection, thus the differences cannot be explained solely by a lower efficiency of nanofibre-based filter media but may also be the result of a more complicated sealing of these materials to the test head with the sealing compound used.

When RPDs were naturally fitted and facial leakage is expected to occur, the fractional efficiency curve of each RPD type is significantly lower than in its sealed installation variant. This confirms the expectation in general. Moreover, it can be observed that the efficiency curves deviate over a wider range of approx. 20%–90%, which suggests a significant influence of facial leakage on filtration performance dependent on the RPD fit. To take a closer look, diagrams aligned horizontally differ only in inhalation and exhalation mode, which is intended to represent self-protection and third-party protection. Naturally fitted RPDs in self-protection, generally, tend to have a higher efficiency than their equivalent in third-party protection. This difference is most pronounced in the case of surgical masks, nanofibre-based RPDs and two of the meltblown-based FFP masks, with these masks depositing partly twice as the amount in self-protection as in third-party protection. The RPD models FFP2_3, FFP2_7 and FFP3_1, however, deviate in their filtration performance in both modes only slightly. As a result, this indicates that the efficiency of an RPD may strongly differ between inhalation and exhalation dependent on its properties to minimize facial leakage, which is discussed in the context of Fig. 7 in more detail.

Fig. 7.

Illustration of the relative change of the performance parameters in self-protection in relation to third-party protection. A relative value of 100% would mean that twice the value was measured in self-protection as in third-party protection.

With the sealed installation, there are fewer differences between the two flow directions compared to naturally attached RPDs. Deviating filtration efficiency curves can be seen in the surgical masks and nanofibre-based RPDs, which, in addition to the directionality of the facial leakage, also indicates a directionality of the filter material on filtration performance. Since these mask materials are generally thinner and less rigid, they may be more easily drawn to the test head in the inhalation mode, thus reducing the effective filter area and increasing the specific load. As described in Section 2, the transport mechanisms of particles to the inner surface of the filter material are dependent on the flow velocity, which would well explain the observed differences here. Assuming further that the distance between an RPD and a wearers face is very small, so that the time required for complete evaporation of the water content of the particles during exhalation is insufficient, larger particle sizes could be relevant for third-party protection. In this case, RPDs with increasing efficiency over particle size, such as the SM_1 and SM_2 surgical masks and the FFP2_1* and FFP2_2* nanofibre-based masks, would be more efficient in a real application.

Pressure loss, as the second key performance parameter, is an indicator of breathing resistance and thus crucial for the wearing comfort. RPDs with a low pressure loss impair breathing less and are thus desirable especially for vulnerable individuals with pre-existing conditions of the respiratory system or low tidal volumes. As with the filtration efficiency, also the pressure loss may depend on facial leakage and the direction of flow for different mask characteristics. In order to view both performance parameters side-by-side, the number-based total filtration efficiencies determined from the fractional efficiencies are illustrated above the net pressure loss in Fig. 6 . Each point is the mean of five measurements on five new RPD samples, with the error bars representing the standard deviation. The total filtration efficiency in sealed installation shows for most RPDs again that the requirement of DIN EN 149 for a lower penetration than 6% is fulfilled, if the total efficiency is determined on a number basis and with a representative test aerosol of saliva substitute solution, both in third-party and self-protection. Naturally fitted masks, on the other hand, vary in the range of 30% and 85%, thus the efficiency is significantly decreased due to facial leakage.

Fig. 6.

Comparison of the number-based total separation efficiency with the net pressure loss @ 95 l/min. Open symbols represent measuring points in self-protection, filled symbols represent the third-party protection. Each point is the mean of five measurements on five new mask samples, with the error bars representing the standard deviation of the fivefold determination.

A comparison of the different RPDs in sealed installation shows that the pressure loss varies over a wide range, with the surgical masks at the lower bound of approximately 30 Pa–90 Pa. FFP masks, for example FFP2_5 and FFP2_7, tend to highly differ in pressure loss although the filtration efficiency is similar. In general, these observed differences can simply be explained by different effective filter areas, material thicknesses and permeabilities. When comparing naturally fitted RPDs, on the contrary, the pressure loss is strongly reduced, resulting from the effect of facial leakage. RPDs with a sharp decrease in pressure loss, compared to its sealed fit, also show a sharp decrease in total filtration efficiency, suggesting that both performance parameters are affected in a mutually dependent manner. Nevertheless, when comparing different RPD types in the natural fit, such as FFP2_3 and FFP2_7, for example, then similar efficiencies but different pressure losses can be observed. Despite the strong influence of facial leakage, this demonstrates the still existing dependency on filter area and filter material. By looking at the dependency of pressure loss on flow direction in Fig. 6, with open symbols representing self-protection and filled symbols representing third-party protection, RPDs in the inhalation mode exhibit the highest pressure losses in both sealed and non-sealed installation. This suggests a greater resistance when inhaling than when exhaling. Comparing the surgical mask SM_2 and the Filtering Face Piece FFP3_1 in both breathing modes, it is also evident that this difference in pressure loss but also filtration efficiency between inhalation and exhalation is significantly greater with the surgical mask.

In order to take a closer look at how breathing mode-based differences result for different mask types, Fig. 7 aims at a relative comparison. Here, the performance parameters of self-protection are related to those of third-party protection, subdivided into the mask groups described in Section 3.

As already seen on the basis of the fractional filtration efficiencies in sealed installation, the filtration efficiency is similar in both breathing modes when avoiding facial leakage. However, axe-shaped FFP and surgical masks have a higher pressure loss during inhalation, which may be due to a deformation of the mask caused by the direction of flow. In axe-shaped masks, both halves of the mask may contact each other due to the negative pressure during inhalation, while medical masks may touch the test head due to their flexible material. Both would lead to a reduction in filter area, increasing the flow velocity at constant volume flow, which in turn leads to a higher pressure loss based on Equation (3). As discussed above, relatively higher filtration efficiencies but also pressure losses are determined in self-protection in the natural fit including facial leaks. Subdivided into the different RPD shapes, this difference is found to be most pronounced for surgical masks. Fish-shaped FFP masks show the smallest differences, while duckbill-shaped masks and axe-shaped masks are in between. A closer look at the measured pressure loss of fish-shaped RPDs shows that the pressure loss increase in self-protection related to third-party protection is also differently pronounced within this subgroup. FFP2_1 and FFP2_2 with nanofibre materials show a higher relative pressure loss increase on inhalation than FFP2_3 and FFP3_1 based on meltblown filter media, highlighting the still existing influence of the mask material. Mask shape and filter material most likely affect the extent to which RPDs are being drawn to the test head on inhalation due to the negative pressure. The reduction of leakage areas may be advantageous for self-protection, but the minimization of leakage areas also increases the pressure loss.

5. Conclusions

This work focused on a screening of certified surgical and FFP masks used in the COVID-19 pandemic context with respect to respiratory-emitted particles. To this end, we presented a novel experimental set-up that allows the determination of mask performance parameters, namely the fractional filtration efficiency and the net pressure loss, as a function of the flow direction (self and third-party protection) and of the facial fit (sealed and natural fit) using a test aerosol based on artificial saliva. The particle size distributions of exhaled breath and the test aerosol were compared in exhalation mode. Measurements show that they are in a similar size range up to 0.4 μm with most particles smaller than 2 μm. The results of the mask screening in sealed fitting show that both the FFP and surgical masks examined feature a high filtration efficiency with regard to artificial saliva and, with a few exceptions, would meet total number-based efficiencies of 94% related to the requirements in DIN EN 149 using artificial saliva. The filtration efficiencies of the sealed fit are similar in both flow directions, but higher pressure losses are found in self-protection. One reason for this might be a reduction in filter area during inhalation, which presumably results from the drawing of masks with less stiff and thinner materials against the test head or, especially in the case of axe-shaped masks, might be the result of two mask surfaces being merged. In natural fitting, facial leakage significantly decreases both the filtration efficiency and pressure loss for each mask model tested. Here, the total efficiencies between different masks are in the order of 30%–85%, whereas the pressure loss appears to decrease in an inter-dependent manner with filtration efficiency. As a result, we conclude that the mask performance is more influenced by the mask fit and sealing material qualities than the filtration-specific properties of the filter media. As far as the flow direction is concerned, the filtration efficiency and pressure loss tend to be lower in third-party protection than in self-protection. This can similarly be caused by a drawing to the test head during inhalation, which might reduce the size of leakage areas between test face and RPD. Here, the relative change of the performance parameters may be influenced by thin and less stiff materials that favour such drawing to the test head, but may also be influenced by different RPD shapes (fish-, duckbill-, axe-shape).

One can conclude from our study that, considering naturally fitted masks, in addition to filtration-specific material properties, the flow direction, the dimensional stability, the mask shape as well as the sealing material properties influence the RPD performance significantly. However, these properties may be influenced also through humid and particle-laden breath, which is why future work needs to focus on the influence of wearing time on RPD performance. In addition, the influence of RPD shape indicates potential for optimisation, especially for the development of well-separating masks with reduced pressure losses that are suitable for infection prevention even for high-risk patients with restricted tidal volume.

Data availability statement

Data are available from the corresponding author upon reasonable request.

Declaration of competing interest

The authors declare no conflict of interest.