Microstructure analysis and image-based modelling of face masks for COVID-19 virus protection

The COVID-19 pandemic prompted widespread mask mandates, but the scientific rationale linking mask material microstructure to filtration efficacy and breathability remained unclear. Prior debates over mask effectiveness were compounded by limited 3D microstructural data on commonly used masks. This study addresses the research question: How do the 3D micro- and nano-fibrous structures of reusable, surgical, and N95 face masks govern droplet filtration and air permeability? The purpose is to quantitatively relate microstructural parameters (e.g., pore size distribution, fibre length/diameter, specific surface area, porosity) to mesoscale performance (permeability and thermal signatures during breathing), thereby informing improved mask design and public health guidance.

Filtration performance depends on microstructure (fibre diameter, thickness, porosity), surface charge, and environmental conditions (air velocity, aerosol size, temperature, humidity). Prior work has largely emphasized performance testing over structural characterization. Fischer et al. introduced a low-cost optical droplet filtering assessment; Aydin et al. used high-speed imaging and showed multiple fabric layers improve droplet blocking. However, these studies provided limited microstructural insight. Earlier investigations were mostly 2D (e.g., SEM) and cannot capture interior connectivity. Lee et al. recently used X-ray tomography to analyze N95 filter layers with NaCl particles, but comprehensive, comparative 3D microstructural studies across mask types were lacking. Conventional pore size measurements such as mercury intrusion porosimetry can be inaccurate for air-filled porous media and rely on model assumptions (e.g., Washburn equation). Non-destructive X-ray tomography with pore network extraction offers a direct, spatially resolved alternative to quantify pore connectivity and size distribution.

Specimens: Three widely used commercial masks were studied: a single-layer reusable mask (53% polyester, 45% cotton, 2% elastane), a three-layer polypropylene surgical mask (outer nonwoven, middle melt-blown filter, inner soft layer), and a multi-layer polypropylene N95 mask (without valve) with charged nano-fabrics. Samples were trimmed (~3 mm wide) for mounting without stretch. Imaging: High-resolution lab-based X-ray micro-CT (ZEISS Xradia 620 Versa) with 40 kV tungsten source; 20X and 40X optics; camera binned to achieve 350 nm voxel size with ~700×700 µm² or ~350×350 µm² FOV. Phase contrast was optimized via source-sample-detector distances; 501–1601 projections acquired. Reconstruction used standard and iterative (ZEISS OptiRecon) algorithms. Data processing: Reconstructed volumes were filtered (unsharp mask) and segmented into fibre and air via thresholding and watershed in Avizo 2019.4. Pore Network Modeling (maximal ball algorithm) extracted pores and throats from binarized porosity to compute pore size distributions and connectivity. Skeletonization (XFibre: cylinder correlation, trace correlation line) traced fibre centerlines for fibre length/diameter statistics, with custom templates per mask morphology. Artefacts near boundaries were noted. Image-based simulation: Absolute permeability along Z (normal to mask plane) was computed in Avizo XLab Hydro (Stokes flow). Identical sub-volumes (161×161×35 µm³; 460×460×100 voxels) were extracted for all masks to compare. Assumptions: steady flow, no volume change, constant environment (moisture), flow rate 5.6×10^11 µm³ s^-1, representative cough droplet velocity 20×10^6 µm s^-1, outlet pressure 0.01 Pa, air viscosity at 15°C = 1.8×10^-5 Pa·s. Complementary characterization: SEM (Zeiss EVO MA10) imaged selected layers (reusable, surgical Layer #2, N95 Layer #b). IR thermography (FLIR One Pro LT, 8.7 Hz) recorded facial thermal maps during inhalation/exhalation while wearing each mask (temperature range 20.2–33.6 °C) as a proxy for air and moisture transmission.

- 3D microstructure: Reusable mask shows a single layer of coarse fibres with large inter-fibre gaps (tens to >100 µm). Surgical mask exhibits a three-layer sandwich; the middle melt-blown layer (Layer #2, ~100 µm thick) has higher fibre density with micro- and nano-fibres. N95 includes a shield-like outer Layer #a (~80 µm) over a filter Layer #b (~250 µm, typical 200–400 µm), rich in nanoscale fibres; N95 nano-fibres are finer than surgical. - Phase fractions and specific surface area (SSA): Key layer porosities: reusable 82%, surgical Layer #2 69%, N95 Layer #b 86%; N95 Layer #a has 44% porosity and ~1% embedded particles were observed. Fibre volume fractions are generally <30%. SSA trend: N95 highest (Sarea ≈ 0.64 µm²/µm³) > surgical (≈ 0.40) > reusable (≈ 0.27). - Pore network modeling: Reusable mask has wide pore diameters (20–100 µm) with about 10% of pores >90 µm; surgical shows fewer large pores (~2.5% >90 µm) with Layer #2 having smaller pores (≈15–30 µm) than outer layers; N95 has no pores >65 µm in FOV and ~50% of pores <30 µm with homogeneous distribution. Average pore diameters: reusable ~47 µm, surgical ~33 µm, N95 ~30 µm. Table 2 ranges: min/max pore diameters (µm): reusable 20/98; surgical 10/92; N95 8/64. - Fibre statistics: Number of fibres traced: reusable 78; surgical 151; N95 296. Average fibre lengths: reusable ~219 µm; surgical ~143 µm; N95 ~126 µm. Max lengths: 751 µm (reusable), 463 µm (surgical), 454 µm (N95). Average fibre diameters: reusable ~13.8 µm; N95 ~2.2 µm (surgical intermediate, qualitatively between these). - Permeability (Z-axis): Reusable 0.37 µm²; surgical 0.34 µm²; N95 0.26 µm². Streamline simulations show reusable has most homogeneous, fastest outlet flow; N95 shows most heterogeneous, slowed flow, consistent with higher flow resistance. - IR thermography during breathing: Facial thermal maps show higher temperatures (greater air/moisture transmission, comfort) for reusable and surgical masks; N95 exhibits the coldest facial regions, indicating lowest moisture transmission and highest filtration barrier. - Structure-performance linkage: N95’s small pores and high SSA from dense nano-fibres enhance interception/impaction/diffusion and electrostatic capture (especially for <200 nm droplets/particles), yielding superior filtration; reusable mask prioritizes breathability/comfort with larger pores and coarser fibres. - Design recommendations: To improve N95 balance, reduce Layer #a thickness by ~5–20 µm to enhance breathability; increase Layer #b thickness by ~50–150 µm and incorporate more homogeneous nanoscale fibres to further elevate capture efficiency while maintaining high porosity.

The study quantitatively links 3D microstructural features to mask performance. Masks with smaller, more uniformly distributed pores and higher specific surface area (notably N95) impede droplet-laden airflow and reduce moisture transmission, supporting superior filtration. The pore-size distributions relative to respiratory droplet sizes (≈40 µm to 1 mm) show substantial overlap for the reusable mask (peak pores ~45 µm) with droplet distributions, while N95 pores (10–65 µm, peak ~30 µm) overlap less with droplet peaks (~100 µm), reducing transmission likelihood. N95’s multilayer architecture provides staged interception: Layer #a reduces momentum and blocks larger droplets but decreases breathability due to lower porosity (44%); Layer #b’s dense nano-fibres (high SSA ~0.64 µm²/µm³, porosity 86%) capture smaller droplets/particles via impaction, interception, diffusion, and electrostatic forces. Image-based permeability results (reusable > surgical > N95) and IR thermography (N95 lowest facial temperature) corroborate the trade-off between breathability and filtration. The integrated tomography-thermography framework provides mesoscale validation of microstructure-derived predictions and offers evidence to inform mask design and public health recommendations.

This work provides a comprehensive, comparative 3D microstructural analysis of reusable, surgical, and N95 face masks using X-ray micro-CT, pore network modeling, skeletonization, image-based permeability simulations, and IR thermography. It establishes clear structure–performance relationships: N95 masks, with the finest fibres, highest specific surface area, and smallest pore sizes, deliver the best droplet filtration but lower breathability; reusable masks offer the best breathability/comfort with larger pores and coarser fibres. The authors propose design modifications to next-generation N95s—thinning the outer shield layer (Layer #a) by ~5–20 µm to improve breathability, thickening the filter layer (Layer #b) by ~50–150 µm, and adding more homogeneous nano-fibres to enhance capture—aiming for improved filtration without undue breathing resistance. All datasets are released openly to support further research. Future work could expand to more mask brands and materials, include dynamic moisture/charge effects, and validate with standardized aerosol penetration and fit tests to generalize findings.

- Sample scope: Only three commercial masks were examined; results may not generalize across all brands or materials. - Field of view constraints: To achieve highest resolution for the N95 filter layer, only two layers were captured due to limited FOV, potentially missing full-stack interactions. - Sub-volume analysis: Permeability and statistics were computed on cropped sub-volumes (161×161×35 µm³), which may not capture macroscale heterogeneity. - Skeletonization artefacts: Fibre tracing can introduce artefacts near boundaries; shortest fibres (<50 µm) were excluded, potentially biasing length distributions. - Modeling assumptions: Image-based flow simulations assumed steady Stokes flow, constant environment (no humidity/morphology changes), fixed flow rate/viscosity, and set outlet pressure; real breathing and droplet dynamics may differ. - Layer averaging: Overall porosities for multi-layer masks were approximated by averaging layer values, which simplifies through-thickness gradients. - IR thermography: Thermal maps provide indirect proxies of air/moisture transmission and can be influenced by fit/leakage and environmental conditions; not a standardized filtration metric.