J Chrom Sci, bmad055 (2022). Standards of antiglycemic drugs were metformin hydrochloride (S1, a biguanide), glibenclamide (S2 = glyburide, a sulfonylurea), pioglitazone hydrochloride (S3, a thiazolidinedione), repaglinide (S4, a glinide). Samples were methanolic solutions of commercial tablets of S1 with each of the other molecules. The following method was developed by a software-assisted AQbD approach (analytical quality by design): (1) Several TLC separations were tried with toluene together with other solvents and with acidic or basic modifiers, with also variations of 24 method or instrumental parameters. (2) Principal component analysis (PCA) was performed in order to identify two principal components (PCs) responsible for 98 % of the observed variations: namely, resolution and tailing factor. Three critical method parameters (CMPs) had a statistically significant impact on the PCs: mobile phase (MP) composition, ammonium acetate concentration in MP, and saturation time. (3) To optimize these CMPs, the Box–Behnken design was implemented in 15 software-proposed experiments; the impacts of the 3 CMPs on the 2 PCs were evaluated by ANOVA, multiple regression analysis, and 2D and 3D contour plots. (4) The optimal CMPs ranges were determined by defining a MODR (method operable design region) on the superposed contour plots, and one TLC condition was selected as analytical control point.
TLC on silica gel pre-washed with 10 mL methanol, dried and activated 10 min at 100° C. Separation with toluene – ethyl acetate – methanolic solution of 4 % ammonium acetate 7:7:6 after 15 min pre-saturation with 35 % relative humidity. Absorption emasurement at UV 254 nm. The hRF values were 13 for S1, 72 for S2, 82 for S3, 38 for S4. LOQ were 263, 387, 73 and 35 ng/zone, respectively. Linearity range was 25–75 µg/zone for S1, 100–300 ng/zone for S2 and S4, 750–2250 ng/zone for S3. Intermediate precision was below 2 %. For accuracy tests, recovery rates were between 97.6–101.4 %.

J Chrom Sci, bmad045 (2022). Standards were azilsartan medoxomil (AZL) and cilnidipine (CLN). Samples were acetonitrile solutions of commercial tablets of AZL and CLN, and purified human blood plasma as biological fluid spiked with AZL and CLN. The following method was developed by a software-assisted AQbD approach (analytical quality by design): (1) Taguchi orthogonal array design was implemented in 8 screening experiments in order to identify the 3 critical method variables (CMVs), which were: volume ratio of toluene – ethyl acetate, volume of methanol and saturation time. These CMVs had statistically significant impact (one-way ANOVA and Pareto charts) on the 3 critical analytical attributes (CAAs, they were: resolution between AZL and CLN and their hRF values). (2) To optimize these CMVs, the Box–Behnken design was implemented in 15 software-proposed experiments; the impacts of the 3 CMVs on the 3 CAAs were evaluated by ANOVA, multiple regression analysis, and 2D and 3D contour plots; the response surface analysis allowed the software to find a mathematical (quadratic or linear) equation for each CAA, based on the CMVs values. (3) The optimal CMVs ranges were determined by defining an analytical design space (ADS) on the superposed contour plots, and one TLC condition was selected as analytical control point.
TLC on silica gel pre-washed with 10 mL methanol, dried and activated 15 min at 110° C. Separation with toluene – ethyl acetate – methanol 13:3:4 after 15 min pre-saturation with 35 % relative humidity. Absorption measurement at UV 254 nm. The hRF values were 49–51 for AZL and 70–71 for LRT. Linearity range was 400–2000 ng/zone for AZL and 100–500 ng/zone for CLN. Intermediate precision was below 1.6 % (n=3). LOQ were 121 ng/zone for AZL and 34 ng/zone for CLN. Recovery rates were 99.3–99.7 % for AZL and 98.1–99.5 % for CLN. Recovery rates from spiked plasma were 83.3 % for both molecules.

J. Planar Chromatogr. 3, 144-148 (1990). Study of RM versus solvent composition relationships with 16 binary eluents composed of four diluents (heptane, chloroform, dichloromethane, trichloroethylene) and four modifiers (ethyl acetate, MEK, ether, diisopropyl ether) for 57 aromatic solutes (azaarenes, anilines, phenols, phenolic acids and their derivatives). Linear RM versus logCmod plots were obtained in almost all of the above cases, permitting storage of the data in computer memory.

J. Planar Chromatogr. 4, 442-445 (1991). Presentation of a method for the computer-assisted selection of multicomponent solvent systems for the separation of multicomponent solvent systems for the separation of a synthetic product of unknown composition by HPTLC. The method is based on recognition of the order of the spots by comparison of spot area percent, followed by design of the solvent mixture by a statistical optimization method. Excellent agreement has been obtained between predicted data and experimental results.

J. Planar Chromatogr. 8, 205-215 (1995). Investigation of adsorption liquid chromatography with alcohol - hydrocarbon binary eluents. A selection of the interpretative methods have been considered, and their practical usefulness compared when implemented with the commonly accepted Soczewinski equation and with the alternative nonlinear relationship. On the basis of a comparison of experimental and the computed results, conclusions are drawn about the efficiency of the individual interpretative methods and of the two „theoretical“ approaches.

J. Planar Chromatogr. 12, 13-25 (1999). Use of the solvation parameter model to derive system maps for methanol, 2-propanol, 2,2,2-trifluoroethanol, acetonitrile, N,N-dimethylformamide, and acetone binary mobile phase components in water on RP-18. Interpretation of the system constants provides an explanation of solvent selectivity in terms of fundamental intermolecular interactions and changes in the phase ratio. Excellent agreement between predicted and experimental Rf values which highlights the strong potential for this new method of computer-aided method development in reversed phase chromatography.

J. Planar Chromatogr. 3, 264-268 (1990). Comparison of various diluent + modifier - silica systems for 57 solutes by graphical methods; i.e. RM vs. eluent spectra plotted on the basis of equieluotropic series of RM1 vs. RM2 correlations. The spread of points was largest when the binary eluents differed in both diluent and modifier. The differentiated selectivity characteristics of the diluent - modifier binaries demonstrate good prospects for computer-aided selection of an optimal system for TLC analysis of a given set of compounds from the data base.

J. Chromatogr. 593, 329-337 (1992). Study of the retention behavior of 14 closely related coumarins in normal-phase TLC and HPLC with the aim of testing the suitability of TLC as a pre assay of HPLC when the optimization of the mobile phase has been carried out according to the PRISMA system. Measurement of the retention in TLC and HPLC at 37 and 13 selective points, respectively. Two and three dimensional evaluations of capacity factor k' against selectivity points showed similar retention behavior for the coumarins in TLC and HPLC. Discussion of the dependency of k' on the change in solvent strength.