Particle Count Statistics Applied to the Penetration of a Filter Challenged with Nanoparticles

Abstract

Statistical confidence in a single measure of filter penetration (P) is dependent on the low number of particle counts made downstream of the filter. This paper discusses methods for determining an upper confidence limit (UCL) for a single measure of penetration. The magnitude of the UCL was then compared to the P value, UCL ≤ 2P, as a penetration acceptance criterion (PAC). This statistical method was applied to penetration trials involving an N95 filtering facepiece respirator challenged with sodium chloride and four engineered nanoparticles: titanium dioxide, iron oxide, silicon dioxide and single-walled carbon nanotubes. Ten trials were performed for each particle type with the aim of determining the most penetrating particle size (MPPS) and the maximum penetration, Pmax. The PAC was applied to the size channel containing the MPPS. With those P values that met the PAC for a given set of trials, an average Pmax and MPPS was computed together with corresponding standard deviations. Because the size distribution of the silicon dioxide aerosol was shifted towards larger particles relative to the MPPS, none of the ten trials satisfied the PAC for that aerosol. The remaining four particle types resulted in at least 4 trials meeting the criterion. MPPS values ranged from 35 – 53 nm with average Pmax values varying from 4.0% for titanium dioxide to 7.0% for iron oxide. The use of the penetration acceptance criterion is suggested for determining the reliability of penetration measurements obtained to determine filter Pmax and MPPS.

Comparing aerosol refractive indices retrieved from full distribution and size- and mass-selected measurements

“…To test the sensitivity of the Set Average FD-RIR to measured N near 0 m −3 (i.e., low N at larger D m where count significance is questionable), the average N at each D m was treated as a Poisson distribution [96, 97]. The CPC counts c particles per unit volume V giving λ poi = c/V and the probability distribution λ Poi V such that P(N)=e−λPoiV(λPoiV)cc! Therefore, the average N and approximate uncertainty take the form [96] N±TN where T is the left-tailed t -statistic for a given cumulative percentile and two degrees of freedom; at 1 σ and 2 σ , P = 84.1 % and T = 1.32 and P = 97.7 % and T = 4.50, respectively. At D m with N that were not statistically significant (i.e., N−T√N≤0), N was set to 0, and m recalculated; retrieved m are shown in the Poisson − 1 σ and Poisson − 2 σ columns of Table 3 with uncertainties tabulated in Table S7 and S8 and plotted in Figures 6 and S26 through S32.…”

Filtration Efficiency, Breathability, and Reusability of Improvised Materials for Face Masks

“…where C is the observed number of counts, is the confidence interval, and χ % is the Chi-squared function. 45 The challenge particles for these experiments are naturally occurring particles in indoor ambient air. Although ambient air provides less control and standardization over the input particle size and composition compared to latex or NaCl challenge particles, 19,22 the filtration of naturally occurring particles should be a representative placeholder for respiratory particles and is widely used for quantitative mask fitting (e.g., TSI PortaCount).…”