Image Analysis of Structured Surfaces for Quantitative Topographical Characterization

Abstract

<p>In the fields of functional materials, interfacial chemistry, and microscale devices, surface structuring provides an opportunity to engineer materials with unique tunable properties such as wettability, anti-fouling, crack propagation, and specific surface area. Often, the resulting properties are related to the feature sizes of the structured surfaces and therefore, it is necessary to accurately quantify these topographies. This work presents a step-by-step description of a method for the quantification of the size of periodic structures using 2D discrete Fourier Transform analysis coupled with data filtering techniques to optimize feature size extraction and reduce user bias and error. The method is validated using artificial images of periodic patterns as well as scanning electron microscopy images of gold films that are structured on different substrates. While image Fourier Transform has been used previously and is a built-in feature in some commercial and open-source image analysis software, this work details image pre-processing and feature extraction steps, and how to best apply them, which has not been described in detail elsewhere. This method can analyze engineered or natural periodic topographies (e.g., wrinkles) to enable the design of patterned materials for applications including photovoltaics, biosensors, tissue engineering, flexible electronics, and thin film metrology.</p>

Direct Comparison of Three Buckling-Based Methods to Measure the Elastic Modulus of Nanobiocomposite Thin Films

“…Images of buckled films taken at different magnitudes were cropped to 900 × 900 pixels and adjusted for contrast and brightness to improve feature detection, followed by Canny edge detection to detect wrinkle edges, using ImageJ (ImageJ 1.52a, Wayne Rasband, National Institutes of Health). Fourier analysis and associated data filtering and curve fitting were performed using an algorithm developed in-house in MATLAB (MATLAB R2014b, MathWorks) . This algorithm was optimized in previous work to accurately extract characteristic wrinkle wavelengths of films buckled using thermal shrinking and SIEBIMM and eliminate user bias.…”

A Direct Comparison of Three Buckling-Based Methods to Measure the Elastic Modulus of Nanobiocomposite Thin Films

“…The elastic moduli of CNC-PEI films calculated from biaxial/uniaxial thermal shrinking and SIEBIMM were directly compared. A Fourier analysis routine was used to obtain accurate wrinkle sizes from microscopy images 44 which is an improvement over the weighted average of peaks used in previous Fourier analyses for identifying characteristic wavelengths of wrinkled films, 2 or manual measurements of selected wrinkles. 15 More specifically, the routine was optimized for high throughput, unbiased, and accurate measurement of periodic surface features.…”

“…Characteristic wrinkle wavelengths could be reliably extracted using the Fourier analysis routine, even when the structures were not perfectly periodic across the entire sample, or when there were discontinuities in the periodicity. 44 It should be noted that although broadening occurs in the spatial frequency peak of the PSD plots due to the dispersity in the size of the periodic features, a characteristic wrinkle wavelength can still be identified. This peak broadening effect increases from SIEBIMM, to uniaxial shrinking, to biaxial shrinking as the wrinkles become more randomly oriented.…”

“…These requirements justify the use of buckled gold calibration films and the recently developed Fourier analysis algorithm with curve fitting. 44 In the case of SIEBIMM, the elastic modulus is impacted the most by changes in the wrinkle wavelength, followed by the elastic modulus of the PDMS substrate (𝐸 * ), and the film thickness (Figure 7A). To ensure more reproducible measurements for SIEBIMM, a large number (at least 9) of repeat measurements of the wrinkle wavelength should be obtained since there can be variability in wrinkle size across the sample.…”

“…Fourier analysis and associated data filtering and curve fitting were performed using an algorithm developed in-house in MATLAB (MATLAB R2014b, MathWorks). 44 This algorithm was optimized in previous work to accurately extract wrinkle wavelengths of films buckled using thermal shrinking and SIEBIMM and eliminate user bias.…”