Hardware and software solution for planar chromatography

Abstract

Pushed by the citizens and governments, the demand for quality control is increasing in numerous industries. In analytical chemistry especially, innovative apparatus and data analysis solutions are constantly needed to face those challenges.Using the open-source RepRap 3D printing environment, thin layers of silica gel suitable for planar chromatography were printed. A low-cost 3D printer was modified by replacing the plastic extruder by a slurry doser for production of stationary phases. The final apparatus opened new possibilities for tailored plates, both in terms of layer shape and composition. The development of this device was greatly facilitated by the minimal modifications needed to repurpose the 3D printing electronics and software. The same strategy was applied to produce an apparatus for office chromatography. In a single miniaturized device, several steps of the planar chromatography pipeline were performed. Liquids were printed via a thermal inkjet cartridge, enabling dropon-demand of 150 pL with a resolution of 96 dpi. The device was controlled by dedicated software hosted on a Raspberry Pi and available on the local network. Sample application was quantitative with correlation coefficients superior to 0.999. Mobile phase printing for the separation of dyes and parabens was also conducted via inkjet. The Raspberry Pi camera enabled an easy implementation of the documentation step. LEDs were used for the illumination and opened new possibilities in terms of selectivity. The final apparatus was compact, modular and affordable.In modern analytical chemistry, where there is hardware to produce data, software is needed to analyze it. Using the R programming language and in particular the shiny package, several web applications were developed for data analysis.To fulfill the need for free, dedicated and fully featured solutions for quantitative evaluation of videodensitograms, quanTLC was created. For an intuitive user experience, the user interface was kept minimalistic, enabling a fast and reproducible analysis. Multivariate analysis of planar chromatography data is gaining interest among researchers and industry. However, there is no all-in-one solution to perform such analysis. The rTLC software was developed for this purpose. All necessary steps were implemented, i.e. data extraction, preprocessing, variable selection, unsupervised and supervised statistics.Before chromatogram images are evaluated, preprocessing is often necessary. The use of unsupervised learning with artificial neural network was investigated. Inhomogeneous background and noise were removed with this technique. In addition, the new features learned by the network show an improvement in resolution while keeping the quantitative aspect.Due to the tremendous amount of information produced in high resolution mass spectrometry, data interpretation can be time-consuming. If several data analysis software are available for this task, none of them is focused to the hyphenation of mass spectrometry with planar chromatography. A tailored software, called eicCluster, was created for this technique. After bucketing of the m/z values, the powerful t-Distributed Stochastic Neighbor Embedding algorithm was used to cluster the extracted ion chronograms. The data dimension was reduced to a 2D map where isotopes and fragments from the same molecules were clustered together. The user could then explore the dataset via interactive visualization tools and draw conclusions otherwise hidden.