(285j) Color As a Source of Information in Liquid Crystal Sensors | AIChE

(285j) Color As a Source of Information in Liquid Crystal Sensors


Jiang, S. - Presenter, University of Wisconsin-Madison
Zavala, V. M., University of Wisconsin-Madison
Understanding the physical alignment of liquid crystals (LCs) near an interface is crucial for the design of electro-optical devices (e.g., liquid crystal displays [1]), nanoscale materials (e.g., semiconducting nanorods [2, 3]), and sensors [4, 5]. In this work, we are specifically interested in LC-solid interfaces, which provide a flexible platform for the development of sensors to detect air contaminants [6]. The working principle of such sensors relies on the alignment of LC molecules at interfaces, which can be controlled via tailored surface chemistries (e.g., metals and metal oxides) [6]. LC alignment and the evolution of such alignment can be visualized using orthoscopic polarized light microscopy [7]. This technique probes an LC sample with a beam of near-parallel rays of polarized light; LC alignment at the interface generates optical fields with different features (e.g., color and spatial textures) [4, 6, 8]. It has been recently shown that the features of these optical fields can be decoded by machine learning techniques to detect different types of air contaminants and to measure their concentration [13]. These studies also found that color plays a key role in distinguishing contaminant type and concentration.

In this work, we investigate color analysis techniques to transform macroscopic optical fields to physical parameters of interest such as the zenithal (tilt) and azimuthal (twist) angles of the LCs at the interface. This information is used to obtain insights into the physical origins of information in color fields. Alignment studies typically use a Berek compensator, which can quantify the optical retardance caused by birefringence (the difference between direction-dependent refractive indices) between the top and bottom interfaces of the sensor. Miller et. al. showed that LC alignment at the interface can be approximated using a Berek compensator [7]. However, a Berek compensator cannot measure optical retardance in a scalable manner. Specifically, it can not measure the evolution of the optical retardance for multiple samples simultaneously. A Michel-Levy chart is typically used to correlate color with optical retardance [10]. Sørensen et. al. formulated an equation to construct Michel-Levy charts quantitively [11] but they did not provide a direct equation to map color to LC alignment. Moreover, standard Michel-Levy charts do not consider the effect of different light sources. For instance, an approximation of optical retardance from samples under halogen light bulbs will not be accurate based on the standard Michel-Levy Chart.

In this work, we present a mathematical framework to map color in optical fields to LC alignments at the surface (and viceversa). The framework takes into consideration various aspects of relevance in LC-based sensor design (e.g., film thickness and light sources). We use our framework to construct a generalized Michel-Levy chart that captures optical retardance under different light sources. These capabilities allow us to analyze the effect of different light sources (e.g., halogen light vs. LED) and color space representations (e.g., RGB and LAB) on machine learning tasks. Specifically, we construct a 3D convolutional neural network to predict chemical species and concentrations from a temporal evolution of LC optical fields with varied light sources and color representations. This framework provides insights on how to co-design instrumentation (light sources) and the LC interface (film thickness) to extract maximum information and provides insights on the physical origins of information in LC optical responses. We use attribution methods (e.g., saliency maps [12]) to get insights of the governing dynamics of LC alignments.

[1] Drzaic, Paul S. Liquid crystal dispersions. Vol. 1. World Scientific, 1995.

[2] Karanikolos, Georgios N., et al. "Templated synthesis of ZnSe nanostructures using lyotropic liquid crystals." Nanotechnology 16.10 (2005): 2372.

[3] Yang, Chung-Sung, David D. Awschalom, and Galen D. Stucky. "Growth of CdS nanorods in nonionic amphiphilic triblock copolymer systems." Chemistry of Materials 14.3 (2002): 1277-1284.

[4] Bai, Yiqun, and Nicholas L. Abbott. "Recent advances in colloidal and interfacial phenomena involving liquid crystals." Langmuir 27.10 (2011): 5719-5738.

[5] Lowe, Aaron M., and Nicholas L. Abbott. "Liquid crystalline materials for biological applications." Chemistry of Materials 24.5 (2012): 746-758.

[6] Yu, Huaizhe, et al. "Amplification of Elementary Surface Reaction Steps on Transition Metal Surfaces using Liquid Crystals: Dissociative Adsorption and Dehydrogenation." Journal of the American Chemical Society 141.40 (2019): 16003-16013.

[7] Miller, Daniel S., et al. "Introduction to optical methods for characterizing liquid crystals at interfaces." Langmuir 29.10 (2013): 3154-3169.

[8] Lockwood, Nathan A., and Nicholas L. Abbott. "Self-assembly of surfactants and phospholipids at interfaces between aqueous phases and thermotropic liquid crystals." Current Opinion in Colloid & Interface Science 10.3-4 (2005): 111-120.

[9] Collings, Peter J. Liquid crystals: nature's delicate phase of matter. Princeton University Press, 2002.

[10] Spring, K. R., M. J. Parry-Hill, and M. W. Davidson. "Michel-Levy Birefringence Chart." Olympus Microscopy Resource Center (2010).

[11] Sørensen, Bjørn Eske. "A revised Michel-Lévy interference colour chart based on first-principles calculations." European Journal of Mineralogy 25.1 (2013): 5-10.

[12] Simonyan, Karen, Andrea Vedaldi, and Andrew Zisserman. "Deep inside convolutional networks: Visualising image classification models and saliency maps." arXiv preprint arXiv:1312.6034 (2013).

[13] Cao, Y., Yu, H., Abbott, N. L., & Zavala, V. M. (2018). Machine Learning Algorithms for Liquid Crystal-Based Sensors. ACS sensors, 3(11), 2237-2245.