Measurement and analysis of pore size variability in microporous membrane chips acquired from Aquamarijn

The purpose of this study was to characterize pore sizes on Aquamarijn (Aquamarijn Membranes B.V., the Netherlands) membrane chips. Membranes provided to us by Aquamarijn have a design specification of 450 nm pore size across 14 windows in the chip. Chips were sputter coated with gold particles at 20 mA for ~60s after which SEM imaging was used to analyze pores and measure pore-size, including degree of variability across chips and membrane windows with a margin of 400-500 nm being the permissible pore-size range.

SEM imaging was used to capture several images, one in a random spot on each window at 20,000X magnification. 5-7 random pores were selected in each image in each window and diameter measurement (a single reading for each pore) was taken using the ROI feature in the imaging program (Method 1; Fig.1.A.). The pores appeared to be more rough than smooth (Fig.1.A.).  A previous measurement method (Method 2) involved capturing images of single pores at 140,000X magnification and then taken two measurements across two diametric ends of the pore (Fig.1.B.). This method analyzed one pore per window per chip. Using a single Aquamarijn chip, a comparison in pore size measurement was made using both methods, which resulted in similar measurements (Fig.1.C.). Overall measurements combining readings from both methods showed that the single Aquamarijn chip had an average mean pore size of 452.20 ± 12.55 nm (Fig.1.D.), while having statistically significant variability from window to window (p<0.0001; Fig.1.E.) based on a total of 112 measurements.

Figure 1. Aquamarijn chip pore size measurement using two methods. A. Representative SEM image of pore size measurements made using method 1, and B. method 2; C. Comparison of pore size measurements between method 1 (n = 84) and method 2 (n = 28) ; D. Overall pore size measurements combining both method 1 and 2 data points (n = 112); E. Comparison of pore size measurements among the 14 windows of the Aquamarijn chip representing window to window variability (n = 8 measurements for each window). Ordinary one-way ANOVA shows significant differences among means (p < 0.0001). 

With the arrival of two more chips (treated as another lot) from Aquamarijn, a more robust measurement of pore size, and window-to-window, as well as lot-to-lot variability was made using Method 1. A representative image of the array of pores was obtained, in which pores once again appeared to be more rough than smooth across the three chips among 2 lots (Fig. 2.A.). The overall window to window variability across 3 chips in 2 lots was statistically significant (*p < 0.05) while all measurements fell comfortably within the 400-500nm permissible pore size (Fig. 2.B.). The overall measurement of pore sizes obtained, factoring in all three chips, was 448.59 ± 13.99 nm, and a statistically significant degree of variability in pore sizes were observed between the two Aquamarijn lots imaged (*p < 0.05; Fig. 2.C-D.).

Figure 2. Comparison of pore sizes among lots and manufacturers. A. Representative SEM image of Aquamarijn 450nm pore size chip taken at 20,000X magnification with pore size measurements; B. Representative graph with measurements of overall window to window variability across 3 Aquamarijn chips (N = 3 chips among 2 lots, n= 20-21 measurements for each group). Ordinary one-way ANOVA shows significant differences among means (*p < 0.05); C. Overall pore size measurement across 3 Aquamarijn chips (N = 3 chips among 2 lots; n = 257 total measurements); D. Graphic comparison of pore size measurements between two Aquamarijn lots (N = 1-2 chips per lot, n = 84-173 measurements per lot). Unpaired t-test shows significant differences between means (*p < 0.05). Dotted lines indicate permissible pore size range; E. Graphical comparison of pore size measurements among 4 SiMPore lots (N = 3 chips per lot, n = 80-90 measurements per lot). Ordinary one-way ANOVA shows significant differences among means (****p < 0.0001). Dotted lines indicate permissible pore size range; F. Graphical comparison between Aquamarijn and SiMPore lots (N = 3-12 chips per group, n = 257-350 measurements per group); G. Graphical comparison showing percentage deviation in pore size from specifications (Aquamarijn – 450 nm, SiMPore – 500 nm; N = 3-12 chips per group, n = 257-350 measurements per group). Unpaired t-test with Welch’s correction shows significant difference between means (****p < 0.0001); H. Graphical comparison of lot-to-lot variability among Aquamarijn lots (N = 1-2 chips per lot, n = 84-173). Unpaired t-test shows significant difference between means (*p < 0.05); I. Graphical comparison of lot-to-lot variability among SiMPore lots (N = 3 chips per lot, n = 80-90). Ordinary one-way ANOVA shows significant difference among means (****p < 0.0001).

When factoring in how different these chips are compared with SiMPore chips, pore size measurements among different SiMPore lots were made previously resulting in an overall (all 12 chips among 4 lots) pore size measurement of 449.95 nm ± 61.32 nm. The comparison among these lots indicated a stronger likeliness of significant pore size variability (****p < 0.0001; Fig. 2.E.). Furthermore, when considering the difference in pore size specifications of the SiMPore (500 nm) versus the Aquamarijn (450 nm) chips, an overall comparison between the two (among all lots/chips per group analyzed) showed no statistically significant variability in means possibly due to high variability in pore size measurements between data sets (Fig. 2.F.). However, a strong likeliness of a significant difference in % deviation from respective specifications (****p < 0.0001; Fig.2.G.) was observed; deviation calculated as follows:

Where, SP is specified pore size, and MP is measured pore size. Additionally, lot-to-lot deviations from specifications were statistically significant in the case of Aquamarijn chips (* p< 0.05) but showed a stronger statistical likeliness for significant deviation among SiMPore lots (**** p < 0.0001) (Fig.2.H-I.).

In conclusion, while there was a statistically significant degree of variability in pore size measurements among Aquamarijn lots as well as among membrane window-to-window reported, it was clear that the chips were overall more or less consistent with the fabrication specifications in pore size when compared with SiMPore chip lots imaged. Some errors in measurement were to be expected, as the measurements are being made visually by identifying boundaries of the pores in slightly higher contrast images. However, this error might be made up in part by an extensive number of measurements made per pore, per membrane window, per chip, and potentially, moving forward, by increasing sample size, i.e., the total number of chips analyzed. Overall, the Aquamarijn microporous chips pose as good candidates for further studies requiring close to accurate membrane pore sizes.