A critical review on the role of leakages in the facemask protection against SARS‐CoV‐2 infection with consideration of vaccination and virus variants

Abstract The protection provided by facemasks has been extensively investigated since the beginning of the SARS‐CoV‐2 outbreak, focusing mostly on the filtration efficiency of filter media for filtering face pieces (FFP), surgical masks, and cloth masks. However, faceseal leakage is a major contributor to the number of potentially infectious airborne droplets entering the respiratory system of a susceptible individual. The identification of leaking spots and the quantification of leaking flows are crucial to estimate the protection provided by facemasks. This study presents a critical review on the measurement and calculation of facemask leakages and a quantitative analysis of their role in the risk of SARS‐CoV‐2 infection. It shows that the pairing between the mask dimensions and the wearer's face is essential to improve protection efficiency, especially for FFP2 masks, and summarizes the most common leaking spots at the interface between the mask and the wearer's face. Leakage is a crucial factor in the calculation of the protection provided by facemasks and outweighs the filtration performances. The fit factors measured among mask users were summarized for different types of face protection. The reviewed data were integrated into a computational model to compare the mitigation impact of facemasks with vaccination with consideration of new variants of SARS‐CoV‐2. Combining a high adoption rate of facemasks and a high vaccination rate is crucial to efficiently control the spread of highly infectious variants.


| INTRODUC TI ON
Facemasks gained an explosive news coverage in early 2020 at the beginning of the COVID-19 pandemic as a measure to mitigate the spread of the infection, and since the middle of 2020, mask recommendations and mandates have been regularly updated and modulated to adapt to the successive waves of infections. 1,2 Mask mandates were reintroduced at the end of 2021 in a number of countries amid the emergence of the latest SARS-CoV-2 Omicron variant [3][4][5] with early reports suggesting a potentially higher infectivity compared with previous variants and a reduction of the protection provided by vaccines. [6][7][8] Consequently, governments strongly advised or imposed the use of the highly efficient FFP2 (filtering face piece 2) masks in various indoor and outdoor settings. 9,10 The role of mask wearing in slowing the spread of  has been heavily investigated. Chu et al. 11 published a metaanalysis to assess the efficiency of social distancing and mask wearing, concluding that medical or surgical masks might result in a large reduction in virus infection, with N95 masks leading to a larger reduction than surgical masks. The findings on the efficiency of masks have been supported by Brooks and Butler 12 focusing on specific indoor settings (a hair salon, a warship).
From a community perspective, a high rate of mask wearing has been found to significantly reduce the reproduction number. 13 However, Bartoszko et al. 14 and Haller et al. 15 pointed out the lack of evidence to conclude that N95 masks provide a higher level of protection than medical masks.
An accurate estimation of the protection provided by facemasks is challenging as it requires not only information on the filtration efficiency as a function of the particle sizes, but also knowledge of the fit between the mask and the wearer's face. Faceseal leakage plays a significant role in estimating the inhalation of infected respiratory droplets leading to the risk of SARS-CoV-2 infection. Leakage depends on numerous parameters such as the geometry of the mask, its resistance to the inhaled or exhaled airflow, the skin's roughness, or the relative size of the mask and the wearer's head.

| REG UL ATI ON S ON THE LE VEL OF LE A K AG E
The majority of masks commonly used to reduce the transmission of SARS-CoV-2 can be classified into three categories: high-efficiency face protection also called respiratory protective devices (e.g., filtering face piece 2, also known as FFP2, and N95) intended for respiratory protection; medical and surgical masks designed for source control with generally lower protection efficiencies and looser fit; community masks, defined by the Swiss National COVID-19 Science Task Force 16 as non-professional masks designed to protect the public from infection through source control. These three types of face protections are regulated by various standards to ensure a minimum filtration efficiency and/or a sufficient fit quality. The respiratory protection and source control abilities of other devices, including face shields and face coverings such as clothes or scarves, are not regulated and their use is generally not encouraged. 17 High-efficiency protections are defined by both a minimum filtration efficiency and a maximum total inward leakage. The European standard EN 149:2001 + A1:2009 describes the requirements for FFP, divided into FFP1, FFP2, and FFP3 masks. The total inward leakage (given as the ratio of the particles concentration measured in the volume enclosed by the mask over the ambient particles concentration) is calculated from the particles penetrating through the filtering part of the mask and the particles entering through the imperfectly sealed interface between the mask and the wearer's head. As these masks are primarily intended for respiratory protection, only the inward leakage is regulated. The values for filter penetration and total Practical implications • The literature review provides an overview of the major   regulations on facemasks, the measurement and simulation methods developed to locate and quantify air and particles leakage, and their integration within largerscale epidemiological models.
• The presented computational model is a tool to estimate the impact of various parameters on the infection risk and helps inform decisions on mask mandates and future regulations on facemasks.
• The model focuses on the impact of the leaking and filtration properties on the infection risk and can be further adapted to consider other airborne viruses and SARS-CoV-2 variants.
• Future developments on facemasks need to focus on improving the fit between the mask and the wearer's head.
inward leakage are given in Table 1. The measurement of the total inward leakage is done on volunteers performing different exercises (walking, moving the head, speaking). The EN 149 standard allows the presence of a one-way valve designed to reduce the pressure drop at exhalation. The use of masks equipped with such a device is however highly discouraged by health authorities 17,18 in the context of the SARS-CoV-2 outbreak as they do not filter the released airflow, and therefore, do not protect others from emitted respiratory particles. The NIOSH-42 CFR Part 84 is the equivalent of the EN 149 in the United States and sets the requirements for protection equipment such as N95 or N99 masks.
Medical (or surgical) masks are regulated by the EN 14683 + AC:2019 standard. They are designed for source control, thus preventing the spread of droplets emitted by the wearer. They are classified into two groups, Type I and Type II, based on their filtration efficiency as given in Table 1 (a third group, Type IIR, differs from Type II by the maximum allowed breathing resistance). As these masks regulate the exhaled airflow, no requirement for the inward leakage is defined in the standard. Comparable standards for medical masks include ASTM F2100-21 defining the requirements for medical face masks labeled Level 1, Level 2, and Level 3. The ASTM F2100-21 standard requires not only a minimum filtration efficiency at 3 μm like the EN 14683 standard, but also a minimum filtration at 0.1 μm.
Community masks are regulated by the CWA 17553:2020 standard at the European level, which was developed in 2020 following the outbreak of COVID-19. They are primarily designed to minimize the projections of respiratory droplets (source control), but they also provide a certain degree of respiratory protection. Community masks following the CWA 17553:2020 standard are divided into two groups based on their filtration efficiency as shown in Table 1. The CWA 17553 standard highlights the importance of the fit on the wearer's face by including size requirements based on the average face morphology of the European population (adults and children).
It also defines the area covered by the mask with an emphasis on the nose, cheeks, and chin where most leakages occur. Inhalation and exhalation valves are prohibited. The ASTM F3502-21 is a similar standard adopted during the outbreak of COVID-19 to regulate barrier face coverings in order to ensure a sufficient protection from exhaled droplets and aerosol, but also reduce the level of aerosol inhaled by the wearer.

| THE FIT FAC TOR A S AN IND I C ATOR OF THE PROTEC TI ON PROVIDED BY MA S K S
The Occupational Safety and Health Administration (OSHA) defines the fit factor as a "quantitative estimate of the fit of a particular respirator to a specific individual" which is calculated as the ratio of the aerosol concentration in the environment to the concentration in the volume enclosed by the mask and the wearer's head during a series of exercises involving movements of the head. 19 The fit factor is measured during a fit test designed to ensure that the inward penetration of particles is below the prescribed limits in operational conditions. This metric shows a high inter-and intra-user variability as it is not only dependent on the type of mask, but also on the skills and carefulness of the wearer and on the agreement between the sizes of both the mask and the wearer's head.
Originally developed to guarantee an optimal protection in a professional context, the fit factor (also called protection factor when it is not calculated within a standardized fit test) and the total inward leakage have been widely adopted to compare the protection levels of different types of masks either on volunteers [20][21][22] or on manikins. [23][24][25] Manikins help reducing the variability resulting from movements of the head and to a certain extend from the skills of the volunteers in adjusting the mask. They allow a rough control of the fit configuration (by means of a fully or a partially sealed interface, 24 or via the introduction of artificial leaks). 25,26 Measurements have shown that the largest fraction of the penetrating aerosol enters via faceseal leakage rather than through the filtering part. [26][27][28] The fit factor is thus a better indicator of the level of protection than the sole measurement of the filtration efficiency.  The EN 149 standard gives two values for the total inward leakage: the first value is the limit that should not be exceeded for 46 out of 50 individuals exercise results (five exercises for each of the 10 tested individuals) and the second value is the maximum arithmetic mean for 8 out of the 10 tested individuals.

| Measurement of the protection efficiency of masks used for source control
Measurements of the aerosol penetration was initially developed to estimate the level of respiratory protection. Assessing the protection efficiency in a source control application is more challenging as the emitted aerosol has to be distinguished from the ambient aerosol by, for example, radioactive 29,30 or fluorescent 31 markers. The aerosol exposure can be calculated as the ratio of the radioactivity deposited on the filter of the receiving manikin over the radioactivity emitted by the source manikin, with the use of soft manikins to simulate a realistic fit. 29,30 Alternatively, the sampling apparatus can be placed directly in front of the mouth to reduce the mixing with ambient aerosol and avoid using harmful tracers. 32 Measurements can also be performed in a closed volume with a stable ambient particles concentration, 33 but such a setup can impact the results as a small volume might not allow a realistic spread of the emitted particles (further discussion can be found in Appendix S1). Source control was generally found to be significantly more efficient than respiratory protection at reducing the exposure to aerosol. 29,30,34,35 However, this relation was less pronounced and even inverted in particular settings: in small volume enclosures, 33 or when the susceptible individual was placed next to or behind the source. 35 A computational model has been developed to compare the efficiencies of mask usage on the emitter and the receiver with various levels of leakage. 36 The authors did not find significant differences in the efficiency of source control versus respiratory protection. The study considered a uniform distribution of particles in the interaction volume which is valid for long-range interactions but does not consider short-range interactions. The main benefit of source control-a reduced velocity of the exhaled airflow and the carried particles-has therefore not been included in the model, which reduced the calculated efficiency of source control compared with respiratory protection.

| Impact of head movements
The protection efficiency is degraded by head movements (bending over, talking, moving the head side-to-side and up-to-down, grimacing) as facial muscles modify the contact surface between the mask and the wearer, 27,37-41 thus creating additional leaking spots. The degradation of the protection is strongly dependent on the quality of the initial fit and the compatibility between the mask and the wearer's head. Head movements have been found to cause a lower degradation of the protection efficiency of well-fitting masks (i.e., N95 or FFP) compared with masks providing a lower fit quality (i.e., surgical masks). 39,41 A computational framework developed to model and further investigate this aspect 42,43 is presented in the modeling section.

| Influence of the expiratory activity
The type of expiratory activity is also likely to impact the fit: speaking has been found to degrade the respiratory protection efficiency, 37,38,40,41 but the impact has been partially attributed to additional particles generated by the emitter 37 and to a measurement artifact resulting from a limited inhalation time and a longer exhalation time compared with breathing. 38 The source control efficiency of a surgical mask has been found to be higher for speaking than for breathing. 32 Coughing and sneezing are likely to impact the source control efficiency. On one hand, they are both violent expiratory activities and cause the airflow-and the carried particles-to be expulsed from the mouth or nose over a short period of time. This leads to high flowrates and to an increase of the pressure in the space between the mask and the wearer's head which is likely to modify the balance between the airstreams flowing through the gaps and the filtering part of the mask. On the other hand, sneezing and coughing generate larger particles 44,45 which are expulsed at a higher velocity than speaking or breathing, potentially leading to a higher fraction of the particles impacting the facemasks as they cannot follow the leaking airflow. Such particles are more likely to be filtered as the masks' filtration efficiencies increase for particle sizes above their most penetrating particle size. The capture efficiencies (not considering leaks) of N95 and surgical masks have been found to be higher upon coughing versus tidal breathing. 30 Investigations on the outward protection upon exhaling and coughing have not highlighted significant differences between the two activities. 46 The testing setup has been conceived to gather and measure particles from all around the emitter, including particles exiting from sideward and backward leaks. A comparison of the inward and outward protection efficiency as a function of the relative position of two manikins 35 (front-to-front, front-to-back, and side-to-side) coughing and breathing has shown a significant advantage of source control over respiratory protection in the front-to-front and front-toback orientations upon coughing, while the relation was inverted in the side-to-side measurements. The comparative advantage of source control appeared to be reduced for front-to-front and back-to-back for breathing, and both mitigation measures showed similar impact in the side-to-side configuration. A mask on the emitter therefore efficiently stops the forward motion of a cough jet but redirects a higher fraction of the particles to the leakage compared with breathing. Significant differences in the spread of emitted particles between coughing and sneezing have been measured on the side and the back of a masked source, 47 with a sneeze leading to a larger spread of the aerosol around the emitter. The forward movement of the particles was efficiently contained, with sneezing leading to a noticeably higher spread than coughing. Measurements on a manikin featuring a pulsatile flow simulator 48 have led to the conclusion that a succession of expiratory pulses (e.g., during a series of consecutive coughs) degrades the fitting of facemasks (higher leaking airflow) more than single isolated pulses.

| Summary of the fit factors provided by different types of masks
Fit factors measured on trained and non-trained users were summarized and organized into the three groups previously mentioned.
The results are given in Figure 1 and highlight the diversity in fit qualities likely to be found within a population of users with various levels of training in mask usage. The data presented in Figure 1 is based on measurements of the total inward penetration (i.e., including the penetration through the filter and the inward leakage). The displayed fit factor refers to the ratio between the concentration outside of the mask and the concentration penetrating inside of the mask. We use the terms fit factor and protection factor to refer to the total penetration through the filter and the leaks: the fit factor is reserved to the quantification of a mask's penetration during a standardized fit test. Measurements of the fit factor on 14 experienced individuals (working in a Biosafety Level 3 laboratory) 49  with an average passing rate of 16% (range 0%-76%). Low success rates obtained with a blend of FFP masks 51 (half of the tested masks had a passing rate <10%) have been attributed to a mismatch between the facial dimensions of the wearers and the masks' sizes.
Self-assessment of mask fit by way of fit checks (feeling the leaking flow around the mask) was not correlated with the measured fit factors, as the N95 masks were highly sensitive to small leaks 52 that could not be detected by the wearers. On the contrary, a significant improvement of the fit factor has been measured when users were allowed to adjust their N95 masks after performing a seal check. 53 It is worth noting that a seal check is a necessary step to ensure that a certified mask (e.g., N95 or FFP) provides the intended level of protection from airborne pollutants. However, a seal check is unlikely to be performed by the general public wearing FFP or N95 masks, due to a lack of information and/or training.

| Schlieren optical technique
Schlieren optical technique is a powerful method to visualize the exhaled airflow. It is based on the differences in optical refraction index between the warm exhaled air and the colder surrounding made visible by a relatively simple optical setup as shown in Figure 49 were set to 200 if measured >200. Both the fit factor and the protection factor refer to the total inward leakage (penetration through the filter and the leaks); however, the fit factor is a reserved term to quantify the performances of a mask during a standardized fit test. 54

| Light scattering
Light-scattering methods allow a direct observation of the particles' trajectories. Light sensors can be implemented to measure the intensity of the scattered light and derive a semi-quantitative analysis. 65 This method requires the utilization of tracers mostly with manikins (e.g., nebulized NaCl solution, 66 artificial fog 67 ) but also with volunteers (e.g., with smoke from e-cigarettes 68 ). The exhaled airflow can also be visualized in a room filled with tracers prior to inhalation or exhalation. 69 The size and position of the area covered by the laser is a critical factor as a fraction of the emitted flow might be outside of the targeted area. Particularly, the fast and narrow jet generated from an unmasked breath might appear dimmer than the slower plume generated with a mask. 66 An example of experimental setup to track particles by light scattering is given in Figure 2B.

| Thermal imaging
A thermal camera measures the changes in the temperature of the skin at the interface with the mask caused by the inward flow (temperature decrease from colder air entering the mask) and outward flow (temperature increase from warmer air flowing out). 76 The ability to detect inward leakage constitutes a significant advantage of this technique over the methods described earlier which are only able to detect outward leaks. The method has been used to validate results from CFD simulations predicting the positions of leaking spots with N95 masks. 77 However, thermal cameras have a limited resolution in leakage detection and cannot replace fit testing. They are limited to the detection of massive leaks. 78

| Particles tracing
Fluorescent tracers can reveal the trajectories of particles and their deposition patterns on masks at inhalation 80,81 and exhalation. 31 Fluorescent particles sprayed on surgical masks have been used to investigate the deposition patterns and estimate surface contamination. 82 Fluorescent particles have also been used to assess the impact of gender, mask brand and exercise (movements of the body and the head) on the location and shape of faceseal gaps. 83

| Measurement of airborne particles concentrations
Direction and magnitude of leaking flows can be inferred from the measurement of the particles' concentration around the source, and to compare masks and protection devices 47,84 in their ability to contain the forward flow and limit leakage. Source control with surgical masks was found to reduce the release of particles by 70% for speaking and 90% for coughing. 85 The measurements of the concentration around the emitter do not provide detailed information about the airflow or a precise identification of the leaking area at the mask/face interface but permit the identification of the areas where the emitted particles accumulate. The nomenclature of the leaking spots is taken from Viola et al. 60 with data taken from Viola et al. 60 and Tang et al. 58 The airflow through the filter media (front flow) appeared larger for N95/FFP2 masks in the Schlieren imaging Tang et al., 58 while laser observations showed a lower scattering intensity in the front flow compared with peripheral leaking flows, indicating a lower particles concentration as a result of filtration.

| Synthesis of the most common leaking spots
FFP2/N95 masks given in Figure 2C and medical/community masks given in Figure 2D. For example, a CFD simulation to describe the spread of contaminated droplets in a ventilated room considered gaps between 1 mm on the sides and 6 mm around the nose. 90 The simulation has shown a drop of the mean diameter of the released particles with mask, as large particles were well filtered by the filter media and tended to stick to the mask instead of following the leaking flow.
The fate of the leaking particles will be discussed in the following paragraphs with the help of various models and simulations. Another CFD simulation focused on the fate of particles (stick, penetrate, or follow the leaking flow) interacting with surgical and cloth masks, 92 showing that a high fraction of smaller particles (d < 20 μm) were able to follow the leaking flow, while larger particles rather stuck to the mask as they could not follow the curvature of the leaking flow. Results were similar to the previously mentioned model 91 : less particles were able to leak from cloth masks, which was compensated by a higher penetration through the filter media.

| Analytical and numerical models
Analytical and numerical models have been applied to calculate the leaking fraction as a function of various parameters, such as the resistance generated by facemasks against the inhaled or ex- curves gathered from the literature. 109 The total inward penetration has been derived from the standards for respiratory protective devices and set to a uniform distribution between 83% and 91%. The inward penetration for surgical masks was based on the available literature and set to a uniform distribution between 25% and 80%. In an indoor scenario simulator developed to investigate the impact of several parameters (room size, ventilation rate, type of mask, exhaling activity) on the infection risk, 110 FFP masks have been assumed to pass the fit test and the inward leakage has therefore been taken from the EN 149 standard. An overall protection efficiency of 80% has been applied to all masks at exhalation, with the exception of masks with a valve set to an outward protection of 5%. Neither the inward nor the outward protection has been assumed to be dependent on the particle diameter. A similar quantification of the leaks 111 assumed a protection efficiency of 30% for masks used as respiratory protection, and 60% in source control. A high-efficiency mask (95% filtration) has also been included. In an estimation of the mask efficiency to reduce the horizontal spread of droplets, Wang et al. 112 have represented the leaking flow as a fraction of the particles not being filtered by the mask, which has been based on experimental values available in the literature.
Other models 113,114 have only dealt with the filtration efficiency of the mask, assuming a perfect fit on the wearer's face. As the leakages account for the largest fraction of particles entering (respiratory protection) or released (source control) via the mask, such a hypothesis is likely to lead to an over-estimation of their protection.

| Integration into compartmental epidemiological models
Compartmental epidemiological models dynamically assign a given

| E S TIMATI ON OF THE IMPAC T OF MA S K LE AK AG E ON THE INFEC TI ON RIS K WITH CON S IDER ATI ON OF VACCINATI ON AND VIRUS VARIANTS
The synthesized information on mask leakage was applied to evaluate the infection risk in different scenarios including the influence of vaccination and variants of SARS-CoV-2. Three types of masks were modeled: a FFP2 mask, a medical mask, and a community mask each with realistic levels of leakage taken from the compiled data presented in Figure 1. The filtration efficiency curves of the masks are given in Appendix S1. Three levels of leakage (low, intermediate, and high) were considered: the low leakage level corresponded to the 5th percentile of the values given in Figure 1 (i.e., 5% of the population was expected to have a fit factor lower than this value), the high level to the 95th percentile, and the intermediate level to the median value. The resulting leaking fractions for each mask are indicated in Table 2. The leakage was given as the fraction of the total exhaled (mask on the emitter) or inhaled (mask on the receiver) air stream that flows through the gaps between the mask and the wearer's head. We considered that the leaking flow escaped around the nose in an upward vertical direction (crown leak according to the nomenclature presented in Figure 2) as this configuration posed a higher risk than back or side leaks, assuming the receiver in front of the emitter. The three scenarios feature the same numerical values for the inward and the outward leakage. The outward leakage is applied to the emitter and the inward leakage is applied to the susceptible receiver.
The infection risk was calculated in three scenarios describing realistic interactions between an infected emitter and a receiver.
The Indoor setting was considered to be the reference as both individuals were interacting (speaking and breathing) for 15 min and separated by 1 m, which corresponded to the guidelines for contact tracing given by the World Health Organization. 118 Table 2 is shown in Figure 3C for masks used as source control, respiratory protection, and both. Source control had advantages over respiratory protection as it slowed down the particles and the airflow penetrating the mask and diverted the leaking flow away from the receiver. Therefore, most scenarios demonstrated a significantly lower infection risks for source control over respiratory protection. However, the Office scenario constituted one exception with a slightly lower infection risk through respiratory protection compared with source control (−10% for a FFP2 in the low-fit scenario).
A detailed discussion is available in Appendix S1.
The variation of the relative infection risk as a function of the usage of FFP2, medical and community masks with the three leakage scenarios, and as a function of the vaccination rate (without a mask) is given in Figure 3D. The mask usage indicated the fraction of the population wearing masks and determined the probability that the emitter, the receiver, or both wore a mask in the given scenario. In a similar way, the vaccination rate indicated the probability that the emitter, the receiver, or both were vaccinated. The effect of vaccine was modeled with lower infectivity and viral charge concentration (details in Appendix S1 The values indicate the flow through the leaking spots as a fraction of the total flow. The same values were adopted for inward and outward leaks. The low level corresponded to the 5th percentile of the fit factors presented in Figure 1, the intermediate level corresponded to the median value, and the high level corresponded to the 95th percentile.
The combined mitigation impact of vaccination and masks is given in Figure 3E, with the assumption that 100% of the infected the average risk as a function of the leaking fraction for FFP2, medical, and community masks worn by the receiver in the indoor scenario is given in (B); the average infection risk in the different scenarios with masks on the emitter, on the receiver, or on both is given in (C) considering the low-level fit scenario; the adoption of facemasks with the three realistic levels of leakage described in Table 2 was compared with the vaccination rate and shown in (D). The calculations considered a random encounter with an infected emitter and a receiver, each one having a probability equal to the mask usage rate to wear a mask, or equal to the vaccination rate to be vaccinated. The reference for the calculation of the relative infection risk was the average risk in the Indoor scenario without mitigation measures. The combined mitigation impact of vaccination (70% vaccination rate) and masks, with the assumption that 100% of the infected individuals carried the variant (Delta variant, additional data in Appendix S1) and that both individuals wore masks, is given in (E

| LIMITATI ON S OF THE DE VELOPED MODEL
The model presented in this work was used to compare the impact of mitigation measures (masks and vaccination) on the infection risk in different scenarios, as well as to consider variants with an increased infectivity. However, the model had several limitations.
The size distribution of the emitted droplets had a significant impact on the infection risk and on the evaluation of the protection provided by facemasks, as shown by a comparable model considering a slightly different emission size distribution. 124 As the amount of viral charges in a particle was considered to be mostly dependent on We also based our calculations on estimated fit factors for the masks, intended to reflect the fitting qualities of FFP2 and medical masks worn by non-trained users. However, as we highlighted in the review part, the fit factor is likely to show a high variability and is influenced by numerous parameters such as facial features or dynamic phenomena like coughing or sneezing. Finally, we considered in our scenarios that the inward leakage was equal to the outward leakage, as most of the data available in the literature to quantify the level of leakage focuses on inward leakage. This may lead to inaccuracy, as the outward leakage is likely to be higher than the inward leakage due to the higher pressure of the air enclosed between the mask and the wearer's face at exhalation, which might create additional leaking spots comparing to the lower pressure scenario at inhalation.

| CON CLUS ION
In the present work, we reviewed various aspects of facemask leak- All authors approved the final version of the manuscript.

ACK N OWLED G M ENTS
The work was partially supported by Innosuisse project 46668.1 IP-ENG "ReMask: Strategies for innovations for Swiss masks needed in pandemic situations".

CO N FLI C T O F I NTE R E S T
The authors report there are no competing interests to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.