(474e) Decoding Optical Responses of Contact-Printed Arrays of Liquid Crystals Using Machine Learning: Detection of Aqueous Amphiphiles with Enhanced Sensitivity and Selectivity | AIChE

(474e) Decoding Optical Responses of Contact-Printed Arrays of Liquid Crystals Using Machine Learning: Detection of Aqueous Amphiphiles with Enhanced Sensitivity and Selectivity

Authors 

Qin, S. - Presenter, University of Wisconsin-Madison
Van Lehn, R., University of Wisconsin-Madison
Wang, F., University of Wisconsin-Madison
Acevedo-Velez, C., University of Puerto Rico At Mayaguez
Lynn, D., University of Wisconsin-Madison
Zavala, V., University of Wisconsin-Madison
Surfactants are amphiphilic molecules that contain hydrophilic head groups and hydrophobic tail groups [1]. Given this unique structure, surfactants and other amphiphilic molecules are used extensively in household products, industrial processes, and biological applications, and are also common environmental contaminants [2-5]; as such, methods that can detect, sense, or quantify them are of great practical relevance.
State-of-the-art methods for detecting natural and synthetic amphiphiles include mass spectrometry and high-performance liquid chromatography [4, 6]. While these methods provide excellent sensitivity and selectivity, they often require complex sample preparation procedures, costly laboratory infrastructure, and highly trained personnel.

To address these limitations, new approaches have been developed in recent studies, including many optical, electrochemical, and mechanical-based techniques [4, 7]. Among these new approaches, aqueous emulsions of thermotropic liquid crystals (LCs) [8-13] have gained growing attention because they can exhibit distinctive optical responses in the presence of surfactants. When coupled with machine learning techniques such as convolutional neural networks (CNNs), LC-in-water emulsions have emerged as sensitive, rapid, and inexpensive sensors or reporters of environmental amphiphiles [14-18]. However, many existing CNN-based LC sensing methods require the use of expensive instrumentation (e.g., flow cytometry [14]) or computational resources (e.g., videos [18]) for quantitative characterization, owing to variations in optical responses among individual LC droplets with different sizes in an LC-in-water emulsion.

Here, we report an LC-based surfactant sensing platform that takes a step toward addressing several of the abovementioned issues and can predict concentrations and types of model surfactants in aqueous solutions. Our approach uses microcontact-printed arrays of micrometer-scale droplets of thermotropic LCs and hierarchical CNNs to automatically extract and decode rich information about topological defects and color patterns available in optical micrographs of LC droplets to classify and quantify adsorbed surfactants. The microcontact-printed LC droplets are surface-immobilized and uniformly sized, reducing variations in optical responses. The proposed hierarchical CNN contains two levels of 2D CNNs that are trained on different data representations. The selection of the data representation for each level relies on the detection focus. The first level is intended to differentiate defect patterns, and therefore grayscale images, which highlight the contrast between light and shade and contain sufficient information for defect pattern classification. The second level, on the other hand, is forced to learn patterns that are less distinguishable and thus require color information (e.g., RGB). Overall, we show that the combination of microcontact-printed LC arrays and machine learning provides a convenient and robust platform that could prove useful for developing high-throughput sensors for on-site testing of environmentally or biologically relevant amphiphiles.

References

[1] Rosen, M. J., & Kunjappu, J. T. (2012). Surfactants and interfacial phenomena. John Wiley & Sons.
[2] Gallou, F., Isley, N. A., Ganic, A., Onken, U., & Parmentier, M. (2015). Surfactant technology applied toward an active pharmaceutical ingredient: more than a simple green chemistry advance. Green Chemistry, 18(1), 14-19.
[3] Torchilin, V. P. (2001). Structure and design of polymeric surfactant-based drug delivery systems. Journal of controlled release, 73(2-3), 137-172.
[4] Kubicek-Sutherland, J. Z., Vu, D. M., Mendez, H. M., Jakhar, S., & Mukundan, H. (2017). Detection of lipid and amphiphilic biomarkers for disease diagnostics. Biosensors, 7(3), 25.
[5] Arora, J., Ranjan, A., Chauhan, A., Biswas, R., Rajput, V. D., Sushkova, S., ... & Jindal, T. (2022). Surfactant pollution, an emerging threat to ecosystem: Approaches for effective bacterial degradation. Journal of Applied Microbiology, 133(3), 1229-1244.
[6] Behrens, B., Helmer, P. O., Tiso, T., Blank, L. M., & Hayen, H. (2016). Rhamnolipid biosurfactant analysis using online turbulent flow chromatography-liquid chromatography-tandem mass spectrometry. Journal of Chromatography A, 1465, 90-97.
[7] Chen, Y. T., & Yang, J. T. (2015). Detection of an amphiphilic biosample in a paper microchannel based on length. Biomedical Microdevices, 17, 1-8.
[8] Lin, I. H., Miller, D. S., Bertics, P. J., Murphy, C. J., de Pablo, J. J., & Abbott, N. L. (2011). Endotoxin-induced structural transformations in liquid crystalline droplets. Science, 332(6035), 1297-1300.
[9] Carlton, R. J., Hunter, J. T., Miller, D. S., Abbasi, R., Mushenheim, P. C., Tan, L. N., & Abbott, N. L. (2013). Chemical and biological sensing using liquid crystals. Liquid crystals reviews, 1(1), 29-51.
[10] Agarwal, H., Nyffeler, K. E., Manna, U., Blackwell, H. E., & Lynn, D. M. (2021). Liquid Crystal-Infused Porous Polymer Surfaces: A “Slippery” Soft Material Platform for the Naked-Eye Detection and Discrimination of Amphiphilic Species. ACS applied materials & interfaces, 13(28), 33652-33663.
[11] Guo, X., Manna, U., Abbott, N. L., & Lynn, D. M. (2015). Covalent immobilization of caged liquid crystal microdroplets on surfaces. ACS Applied Materials & Interfaces, 7(48), 26892-26903.
[12] Manna, U., Zayas‐Gonzalez, Y. M., Carlton, R. J., Caruso, F., Abbott, N. L., & Lynn, D. M. (2013). Liquid crystal chemical sensors that cells can wear. Angewandte Chemie International Edition, 52(52), 14011-14015.
[13] Ortiz, B. J., Boursier, M. E., Barrett, K. L., Manson, D. E., Amador-Noguez, D., Abbott, N. L., ... & Lynn, D. M. (2020). Liquid crystal emulsions that intercept and report on bacterial quorum sensing. ACS applied materials & interfaces, 12(26), 29056-29065.
[14] Jiang, S., Noh, J., Park, C., Smith, A. D., Abbott, N. L., & Zavala, V. M. (2021). Using machine learning and liquid crystal droplets to identify and quantify endotoxins from different bacterial species. Analyst, 146(4), 1224-1233.
[15] 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.
[16] Bao, N., Jiang, S., Smith, A., Schauer, J. J., Mavrikakis, M., Van Lehn, R. C., ... & Abbott, N. L. (2022). Sensing Gas Mixtures by Analyzing the Spatiotemporal Optical Responses of Liquid Crystals Using 3D Convolutional Neural Networks. ACS sensors, 7(9), 2545-2555.
[17] Smith, A. D., Abbott, N., & Zavala, V. M. (2020). Convolutional network analysis of optical micrographs for liquid crystal sensors. The Journal of Physical Chemistry C, 124(28), 15152-15161.
[18] Frazão, J., Palma, S. I., Costa, H. M., Alves, C., Roque, A. C., & Silveira, M. (2021). Optical gas sensing with liquid crystal droplets and convolutional neural networks. Sensors, 21(8), 2854.