J. Planar Chromatogr. 36, 3-8 (2023). HPTLC of Gentiana extract on silica gel with ethyl acetate - methanol - water 4:1:1. High Dynamic Range (HDR) photography was performed by combining nine Low Dynamic Range exposures (the whole camera range in 1-EV steps). Principal component analysis was applied on the classic exposures and HDR image, proving that HDR image contained the highest amount of extracted information from the TLC plate.

J Chromatogr A 1666, 462863 (2022). Theoretical discussion on the factors determining the RF value of a given substance in a chromatographic system: A) the stationary phase (SP); B) the mobile phase (MP), the composition of which can be different from the solvent mixture prepared because of evaporation, saturation and liquid or gas adsorption effects over migration time; C) the difference of the free energies for the analyte transfer from SP to MP; D) external parameters like temperature and humidity. The universal HPTLC mixture (UHM) is a mixture of reference compounds that can be used for the system suitability test (SST) for the full RF range in all HPTLC experiments. Its composition is: thioxanthen-9-one (0.001 %), guanosine (0.05 %), phthalimide (0.2 %), 9-hydroxyfluorene, octrizole, paracetamol, sulisobenzone and thymidine (each 0.1 %), in methanol. The purpose was to study the potential of UHM to replace SST (described with specific markers in European Pharmacopoeia monographs) and to assess the quality of HPTLC results. TLC and HPTLC silica gel on different support (aluminium, glass) or with different granulometries and binders (classic, Durasil, Adamant), of the UHM, an acetonitrile extract of Abelmoschus manihot flowers (Malvaceae), a methanol extract of Sambucus canadensis flowers (Adoxaceae), and essential oils of Lavandula angustifolia, of Mentha × piperita (Lamiaceae) and of Myristica fragrans (Myristicaceae), as well as the following specific markers (standards): borneol, bornyl acetate, linalool, linalyl acetate (terpenoids), isoeugenol, isoeugenol acetate, chlorogenic acid (phenylpropanoids), gossypin (flavone), gossypetin-glucuronide, hyperoside (flavonol heterosides). Development (after 20 min plate conditioning with a saturated MgCl2 solution) with one of the following mobile phases: (MP1) toluene – ethyl acetate 19:1, especially for essential oils; (MP2) ethyl acetate – butanone – formic acid – water 5:3:1:1, especially for S. canadensis; (MP3) ethyl acetate – acetic acid – formic acid – water 100:11:11:26, especially for A. manihot. Documentation in UV 254 nm and 350 nm, and with white light (reflection + transmission), before and after derivatization. RF values were determined by scanning densitometry at 254 nm in absorption mode (for octrizole, at 366 nm in fluorescence mode with mercury lamp and optical filter K400 nm). For each HPTLC condition, intra-laboratory precision assay of UHM separation was performed (at least 5 analyses) with average RF values and 95 % prediction intervals, and calculating RF differences between pairs of UHM constituents and 95 % confidence intervals, which were max. +/-0.012 of the RF values for all UHM and markers. The sensitivity of UHM, and thus its usefulness as generic SST was demonstrated by repeating the HPTLC experiments with modifying by 10 % the quantity of one of the solvent each time. There were always significant changes in RF values of UHM components and/or in RF differences between pairs of UHM bands; it was often but no always the case with the official specific markers. UHM underwent also significant changes (although less than A. manihot extract) when several silica gel phases were compared under the same HPTLC conditions. This property is crucial to verify the right stationary phase before doing any RF correlations, and could make UHM a universal tool to identify discrepancies between different analyses. Finally, the use of UHM for a computer-supported evaluation of HPTLC results was discussed, either for zone identification and RF corrections (within confidence intervals), or for correlations of entire fingerprints as first step to implement machine learning algorithms.

J Chromatogr A, 1669, 462942 (2022). Samples were medroxyprogesterone acetate (MPA) as standards and commercial drug extracts, dissolved in dichloromethane. TLC on silica gel (preactivated by 30 min heating at 120 °C) with dichloromethane – ethyl acetate 10:1, followed by 30 min drying at 120 °C. Derivatization by spraying with sulfuric acid (50 % in ethanol). Visualization in a 3D-printed chamber designed especially for this purpose, blocking extraneous light and including a smartphone holder, a fluorescent lamp and an optical density step tablet. Pictures were taken with the smartphone digital camera, after spraying (6 background images) and after 10 min heating at 120 °C (6 foreground images). In the last case, MPA appeared as black spots (hRF 16–20). Using an image processing software program: (1) one averaged background image and one averaged foreground image were created by concatenation and were split into 3 colour channels; (2) the green colour channels were corrected to remove background noise, by subtraction of an averaged darkfield image (taken on blank plate without light) and by comparison ratio to an averaged blankfield image (taken on blank plate with light); (3) the pixel values of the MPA bands were converted to optical density values through the Robard’s function, by comparison to a reference image of a theoretical optical density step tablet; (4)  furthermore, the corrected background image was subtracted from the corrected (and denoised with a Gaussian Blur) foreground image; a triangle threshold algorithm was applied on the resulting image, and was converted to a mask (white spots on black background); (5) applying the binary mask to the original corrected images (obtained in (2)), the final integrated density values of MPA spots were obtained. This method was validated for linearity range (1.25–3.75 mg/mL), for precision, for reproducibility, for robustness, and for accuracy expressed as average recovery values (101 % overall mean) by comparison of TLC results with HPLC-DAD results.

J Chromatogr A 1638, 461830 (2021). The purpose was to find the first universal HPTLC mixture (UHM), a mixture of reference compounds that could be used for the system suitability test (SST) for the full RF range in all HPTLC experiments.
(Part 1) UHM composition: First, 56 organic molecules, detectable without derivatization, were tested on HPTLC silica gel with 20 different mobile phases (MP) belonging to different Snyder’s selectivity groups and with several polarity indices. Visualization under UV 254 nm and 366 nm. Densitometry scanning at 254 nm in absorption mode, and at 366 nm in a fluorescence mode (mercury lamp 366 nm, with wavelength filter <400 nm). For selected bands, spectra were recorded in absorbance-reflectance mode (wavelength range 190 – 450 nm, deuterium and tungsten lamp). This procedure allowed 8 molecules to be selected for their better spot resolution and for their specific RF values (at least 3 different values distributed throughout the full RF range for each MP). The final composition of UHM was: thioxanthen-9-one (0.001 %), guanosine (0.05 %), phthalimide (0.2 %), 9-hydroxyfluorene, octrizole, paracetamol, sulisobenzone and thymidine (each 0.1 %), in methanol.
(Part 2) UHM validation: Afterwards, UHM was submitted again to a panel of HPTLC assays with always two MP: (A) toluene – methanol – diethylamine 8:1:1; (B) ethyl acetate – formic acid – water 15:1:1; and for each MP, the means, standard deviation and 95 % confidence intervals of the RF values were calculated. (a) UHM was validated for intermediate intra-laboratory precision, as well as for inter-laboratory reproducibility, with ΔRF 0.045. (b) The capacity of UHM to detect small variations was demonstrated by significant changes in at least some RF values, when separation was deliberately performed at different levels of relative humidity (0 %, 33 %, 75 %, 100 %), or with smaller humidity variations (7 % compared to 0–5 %, and 49 % compared to 33 %), or when performing vs. omitting the 10min chamber pre-saturation, or when modifying the MP (+/-10% of one solvent at each time). These response characteristics (the opposite of robustness) made UHM a powerful tool for SST. (c) Finally, UHM stability was studied with UHM aliquots under several storage conditions (-78 °C, -20 °C, 4 °C, room temperature, 45 °C; or 40 °C with 75 % relative humidity) and durations (2 weeks or 2 months). The densitometric peak profiles at 254 nm were compared to those of the fresh compounds, qualitatively (RF value, UV spectrum) and quantitatively (peak area). UHM was stable at room temperature or below, for 2 months (at higher temperature, guanosine, phthalimide and paracetamol degraded).

J. Liq. Chromatogr. Relat. Technol. 44, 484-489 (2021). HPTLC of caffeic acid (1), rutin (2), naringin (3), and hesperidin (4) in samples of citrus fruits on polyamide plates with a microemulsion - formic acid 47:1. Microemulsion was prepared as follows: 2.7 g of SDS in 90 mL water, followed by adding 6.3 mL n-butanol and 1.0 mL n-heptane, and the mixture was shaken to produce a uniform and transparent O/W microemulsion. Detection by spraying with 1 % aluminum trichloride aqueous solution and photographed using a smartphone under UV light at 365 nm. Image retrieval techniques combined with support vector machine (SVM) directly classified the chromatograms based on migration path images.

J Chromatogr A, 1624, 461239 (2020). Samples were chemical standards of acetylcholinesterase (AChE) inhibitors (azamethiphos, caffeine, donepezil, galanthamine, methiocarb-sulfoxide, paraoxon-ethyl) and of neurotoxic compounds, as well as drinking or contaminated water samples enriched through solid phase extraction. HPTLC on spherical silica gel (pre-washed twice by 20 min immersion in isopropanol, heated 20 min at 120 °C before and after pre-washing with acetonitrile). First separation (preparative TLC) with automated multiple development (16 steps). Effect-directed analysis for AChE inhibitors by immersion (speed 5 cm/s, time 1 s) into enzyme solution, incubation 5 min at 37 °C and immersion into substrate solution (indoxyl acetate 2 % in methanol); visualization under UV 366 nm. Active zones from untreated layers were eluted through the oval head of a TLC-MS interface to a second plate for a second separation with a panel of other mobile phases. Bands of interest were eluted from the second layer with water through the oval elution head of the TLC-MS interface pump, into a RP18 liquid chromatography guard column, followed by a quadrupole time-of-flight mass spectrometer. Full scan mass spectra (m/z 100–1200) were recorded in negative and positive modes using electrospray ionization (and collision-induced dissociation for MS2). Among the water contaminants, lumichrome (riboflavin photolysis product), paraxanthine and linear alkylbenzene sulfonates were identified as AChE inhibitors.

J Chromatogr A, 1619, 460945 (2020). Samples were lamotrigin as standard, or extracted with an oil-in-water microemulsion (10 µL butyl acetate, 4 mL n-butanol, 925 mg sodium dodecyl sulphate, 8.6 mL water) either from patients’ raw plasma (for separation from blood proteins) after spiking, or from commercial tablets dissolved in methanol. TLC on silica gel with a water-in-oil microemulsion of 9 mL butyl acetate, 1 mL n-butanol, 250 mg sodium dodecyl sulphate, 250 µL water. Both optimal microemulsions were predicted using Taguchi orthogonal array and Plackett-Burman design. Evaluation in UV 254 nm, quantification from the digital picture using four image processing software programs. For lamotrigin (hRF 24), limits of quantification were 170 ng for pure drug and 10 ng for spiked plasma. Linearity (in range 20–200 ng/spot) was directly obtained for the calibration curve in spiked plasma; however, for pure drug, linearity was obtained only when using log values of the calculated densities (300–3000 ng/spot).

J Chromatogr A, 1619, 460945 (2020). Samples were lamotrigin as standard, or extracted with an oil-in-water microemulsion (10 µL butyl acetate, 4 mL n-butanol, 925 mg sodium dodecyl sulphate, 8.6 mL water) either from patients’ raw plasma (for separation from blood proteins) after spiking, or from commercial tablets dissolved in methanol. TLC on silica gel with a water-in-oil microemulsion of 9 mL butyl acetate, 1 mL n-butanol, 250 mg sodium dodecyl sulphate, 250 µL water. Both optimal microemulsions were predicted using Taguchi orthogonal array and Plackett-Burman design. Evaluation in UV 254 nm, quantification from the digital picture using four image processing software programs. For lamotrigin (hRF 24), limits of quantification were 170 ng for pure drug and 10 ng for spiked plasma. Linearity (in range 20–200 ng/spot) was directly obtained for the calibration curve in spiked plasma; however, for pure drug, linearity was obtained only when using log values of the calculated densities (300–3000 ng/spot).