Modelling and prediction of retention in high-performance liquid chromatography by using neural networks

Summary Multi-layer feed-forward neural networks trained with an error back-propagation algorithm have been used to model retention behaviour of liquid chromatography as a function of the composition of the mobile phases. Conventional hydro-organic and micellar mobile phases were considered. Accurate retention modelling and prediction have been achieved using mobile phases defined by two, three and four parameters. With micellar mobile phases, the parameters involved included the concentrations of surfactant and organic modifier, pH and temperature. It is shown that neural networks provide a competitive tool to model varied inherent nonlinear relationships of retention behaviour with respect to the mobile phase parameters. The soft models defined by the weights of the networks are capable of accommodating all types of linear and nonlinear relationships, neural networks being specially useful when the relationships between retention behaviour and the mobile phase parameters are unknown. However, to train neural networks more experimental points than with hard-modelling methods are required, hence the use of the networks is recommended only for those cases where adequate theoretical or empirical models do not exist.     

Calculation of amino acid hydrophilicity indices for retention of peptides on amide, diol and silica columns in normal-phase liquid chromatography

Summary A set of hydrophilicity parameters in a normal-phase liquid chromatography of peptides is presented in order to clarify the contribution of individual amino acid residues to peptide retention and to predict retention times. The retention of 100 peptides was studied using normal-phase liquid chromatography on amide, diol and silica columns. An acetonitrile-water mixed solution containing 0.2% trifluoroacetic acid +0.2% triethylamine was used as the mobile phase in a linear gradient elution system. The contribution of each residue upon retention was calculated by linear multiple regression analysis. This paper described the contribution values as “hydrophilicity retention coefficients”. Using these hydrophilicity retention coefficients, retention times could be predicted for peptides of known amino acid content and sequence. A set of hydrophilicity retention coefficients on each column was successfully explained by contributions to the degree of retention.     

Estimation of the degree of self-association of water using a novel retention model

Summary An attempt has been made to gain a semi-quantitative insight into the self-association of water molecules through hydrogen bonds. This was only possible with the use of a new solute retention model for the chromatographic systems by considering intermolecular interactions between the constituents of binary mobile phases. Four different sizes of the average associative aqueous multimer were assumed. By comparing measured and calculated retention values, the existence of associated aqueous multimers consisting of 100 aqueous monomer units is postulated as an average multimer structure.     

Convergence of retention for small solutes in reversed-phase liquid chromatography

Summary It is shown theoretically that when the concentration of organic solvent in the mobile phase increases, or solute size decreases, log k values of small solutes in reversed-phase liquid chromatography (RPLC) will tend to have a minimum value called the convergence point. A theoretical model for evaluating the convergent coordinates of small solutes is presented by using a stoichiometric displacement model for retention (SMDR). The physical meaning of the coordinates of each kind of convergence are also elucidated. The convergence points have either two-dimensional coordinates with a common ordinate (the logarithm of the phase ratio of the column, log ) or threedimensional corrdinates with two common axes: — log and the logarithm of the molar concentration of the pure displacing agent in mobile phase, log a_D. The other axis relates to the nature of the solutes, such as carbon number of a homolog, van der Waal's surface area, hydrophobic fragment constant etc. for the latter and those and/or concentration axis for the former. The model was tested with published data and found to give a good fit.     

Retention of ionizable compounds on HPLC. Modelling retention for neutral and ionizable compounds by linear solvation energy relationships

Summary A global LSER model that relates HPLC retention to mobile phase composition and pH is tested for a varied group of solutes, both neutral and ionizable, in a polymeric column and methanol-water mobile phases. It is compared to the local LSER model developed only for a given mobile phase, i.e., a fixed organic modifier content, and to the global LSER model set only for neutral solutes. The global LSER model for neutral and ionizable solutes requires a few supplementary parameters over the other models tested, but it accounts for retention under any experimental conditions for a given column and methanol-water mobile phases, describing properly the interactions established in the HPLC system (hydrophobicity, hydrogen-bond acidity and basicity, dipolarity/polarizability…). This paper is number 13 of a series with the same general title: “Retention of Ionizable Compounds on HPLC” published in various journals.     

Dye-coated stationary-phases: A retention model for anions in ion-interaction chromatography

Summary The separation of inorganic anions (NO₃ ⁻, NO₂ ⁻, Cl⁻, Br⁻, I⁻, SO₄ ²⁻, S₂O₃ ²⁻) by ion-interaction chromatography mediated with a specific dye has been investigated. Chromatography was performed on a LiChrospher RP-18 colum dynamically coated with crystal violet, using acetonitrile-water buffered with phthalate as the mobile phase. The presence of the dye in the eluent enabled indirect spectrophotometric detection of the analytes, which have no significant UV absorption. Retention data were collected for the different anions by varying the composition of the mobile phase according to a full factorial experimental design. A theoretical model for the retention of singly- and doubly-charged analytes, on the basis of the two main processes of ion-exchange and ion-pair formation, has been proposed and validated with the experimental data.     

Two- and three-parameter equations for representation of retention data in reversed-phase liquid chromatography

Two-parameter equations that describe the dependence of ln kappa upon psi, where kappa is the retention factor and psi the volume fraction of the organic modifier in the mobile phase, are examined in what concerns the underlying approximations and their performance to fit experimental data obtained from reversed-phase liquid chromatography. Using 293 experimental systems, it was found that the performance of these equations to describe ln kappa versus psi data is rather low, since the percentage of the systems that can be described satisfactorily ranges from 40 to 60% depending on the fitting equation. This percentage may be raised to 75%, if the discreteness effect is properly taken into account. A further improvement to 90% of the systems studied can be achieved only by the use of three-parameter equations, which may arise by refinements of the rough approximations of the two-parameter equations. Although the refinements do not lead always to better equations, we developed a new three-parameter expression of In kappa that works more satisfactorily, since it combines simplicity, linearity of its adjustable parameters and the highest applicability.     

Optimisation and validation of an automated solid phase extraction technique coupled on-line to enzyme-based biosensor detection for the determination of phenolic compounds in surface water samples

Summary A fully integrated screening system for phenolic compounds was developed incorporating on-line solid phase extraction, fractionation and biosensor detection. Two different types of biosensors, solid graphite and carbon paste electrodes incorporating the enzyme tyrosinase, were compared and used in the screening system. Interfacing of the solid phase extraction and fractionation with the biosensor detection was given special attention since the biosensors were not compatible with the organic modifier used for desorption of phenols from the solid phase extraction step. The system was validated with conventional analytical techniques. Surface water samples from the Ebro river were spiked with 1,10, and 25g L^–1 of catechol, phenol,p-cresol, respectively. Three out of seven samples were spiked and the correct samples were identified, containing phenols equivalent to the spiked concentrations.     

Optimisation and validation of an automated solid phase extraction technique coupled on-line to enzyme-based biosensor detection for the determination of phenolic compounds in surface water samples

Summary A fully integrated screening system for phenolic compounds was developed incorporating on-line solid phase extraction, fractionation and biosensor detection. Two different types of biosensors, solid graphite and carbon paste electrodes incorporating the enzyme tyrosinase, were compared and used in the screening system. Interfacing of the solid phase extraction and fractionation with the biosensor detection was given special attention since the biosensors were not compatible with the organic modifier used for desorption of phenols from the solid phase extraction step. The system was validated with conventional analytical techniques. Surface water samples from the Ebro river were spiked with 1,10, and 25 μg L⁻¹ of catechol, phenol,p-cresol, respectively. Three out of seven samples were spiked and the correct samples were identified, containing phenols equivalent to the spiked concentrations.     

The on-line determination of the enantiomeric purity of enzymatically produced chiral compounds

Summary A prototype of a coupled-column chromatographic system has been developed to determine the enantiomeric purity of enzymatically produced chiral compounds. In this system, an immobilized enzyme HPLC reactor based upon -chymotrypsin (ACHT-IMER) was linked through a switching valve to a chromatographic column containing a crown ether-based chiral stationary phase (CR-CSP). Although ACHT is normally stereospecific for the L-enantiomorphs of aromatic amino acids, L-tryptophan (L-TRP), L-tyrosine and L-phenylalanine, we were able to achieve the hydrolysis of D-TRP methyl and ethyl esters on the ACHT-IMER. The percent of hydrolysis of the D-TRP esters could be manipulated by varying the resident time in the ACHT-IMER. In a series of experiments, D,L-TRP methyl or ethyl esters were injected into the ACHT-IMER, the flow stopped and after resuming the flow, the eluent which contained the hydrolyzed aminoacids, D- and L-TRP, was switched onto the CR-CSP where the enantiomeric composition of the TRP peak was determined. This configuration is a model for other IMER/CSP coupled-column systems which can be used for analytical and preparative applications.     

Supported liquid membrane work-up in combination with liquid chromatography and electrochemical detection for the determination of chlorinated phenols in natural water samples

Summary A sample preparation method has been developed for the determination of chlorinated phenols in water. The method is based on a supported liquid membrane extraction system connected on-line to liquid chromatography with electrochemical detection. The supported liquid membrane technique utilizes a porous PTFE membrane. The membrane is impregnated with an organic solvent which forms a barrier between two aqueous phases and enables selective extraction. The technique can easily be coupled in a flow system. In this investigation five chlorinated phenols (1–5 chlorine atoms) were extracted from natural water samples. Extraction for 30 minutes resulted in detection limits of approximately 25 ng L^–1.     

An improved method for the determination of S-Carboxymethyl-L-cysteine in biological fluids with precolumn derivatization and on-line clean-up

Summary In order to follow levels of S-Carboxymethyl-L-cysteine in biological fluids for a period as long as three half-lives after drug administration during pharmacokinetic studies, an improved method for its determination had to be developed. Like the previous one, this method uses a protein precipitation step followed by an O-Phthalaldehyde derivatization step and then an HPLC on-line clean-up. This latter was obtained by means of a switching valve system, including a Nucleosil CN 5 m (3 cm × 4.6 mm i.d.) precolumn and a Spherisorb ODS 5 m (15 cm×4.6 mm i.d.) analytical column. The sensitivity limit was improved to 0.1 g/ml in plasma samples and 0.2 g/ml in urine samples.This method was applied in studies comparing single (0.75 g) and repeated (0.75 g tid) oral administration of the drug to 30 elderly patients and 20 healthy volunteers. Results showed that the half-life was 40% longer in elderly patients than in healthy volunteers, and that area under the plasma concentration versus time curve (AUC) values in elderly patients were twice those obtained with young subjects.     

An integrated coupled column liquid chromatography system for the determination of leukotriene metabolites in biological samples

Summary A fully integrated chromatographic system was developed for the determination of leukotrienes in biological samples using photodiode-array detection (PDAD), which eliminates time consuming manual sample handling steps. A special solid phase extraction, (SPE) methodology for leucotriene metabolite stability was developed which increased the recoveries and eliminates the contamination risk of biological samples. The inherent instability (autooxidation) of many of the leukotriene mediators, and the adsorption effects onto exposed surfaces in vials and in the chromatographic system were found to be very important parameters to control in order to circumvent high loss of sample analytes. By binding the cell supernatants to the functionalities of the SPE support stabilised these mediators. Cell culture samples were eluted through a disposable C₁₈ SPE column. The SPE columns were allowed to thaw and deposited in an automated sample handling unit (ASPEC XL). Desorption of the analytes was followed by a second on-line SPE step, to eliminate remaining interfering matrix compounds. Typical recoveries when stored at −70°C were in-between 55–97% except for (LTE₄) which was found to be around 40% after 72 days of storage. Seven reversed-phase packings were studied. Selectivity factors, as well as the separation efficiencies, were found to differ for the various C₁₈ bonded silica stationary phases. This integrated on-line column liquid chromatographic system was applied to the determination of leukotriene B₄, leukotriene C₄, leukotriene D₄, leukotriene and E₄ in human cell extracts using prostaglandin B₂ as the internal standard. More than 1500 biological samples were analysed. Some validation data are presented for unattended operations.     

Accelerated solvent extraction then liquid chromatography coupled with mass spectrometry for determination of 4-t-octylphenol, 4-nonylphenols, and bisphenol A in fish liver

Summary Analysis of extracts from fish liver containing alkylphenol contaminants can be hindered by the presence of co-extracted fats and proteins that interfere with chromatographic analysis. In this study accelerated solvent extraction (ASE), Florisil clean-up, then combined liquid chromatography—mass spectrometry (LC-MS) with an electrospray (ESI) interface have been used to optimize an analytical procedure for the analysis of octylphenol, nonylphenol, and bisphenol A in fish liver. After comparison of the efficiency of ASE with conventional Soxhlet extraction the developed procedure was applied to the analysis of liver samples. Calibration plots of the relationship between concentration and the ratio of the responses to the analyte and to the internal standard, 4-n-nonylphenol, were determined by linear regression analysis over the concentration range 0.05 to 10 ppm and resulted in good fits (r ²>0.994). Recoveries (evaluated for each liver sample as the ratio between response to the surrogate compound, 4-n-nonylphenol, and that to the internal standard, 4-n-heptylphenol, relative to the same ratio for a reference standard solution) were 53±20%. Under the experimental conditions used in this work the limits of detection (LOD), calculated by use of a signal-to-noise ratio of 3∶1, were 5 ng g⁻¹ for 4-t-octylphenol, 15 ng g⁻¹ for bisphenol A, and 20 ng g⁻¹ for nonylphenol. The method can be satisfactorily applied to screening analysis of octyl-and nonylphenols and bisphenol A in biological samples such as fish liver.     

Solute retention in the stationary phase of a liquid-solid chromatographic system

Summary Utilizing the UNIFAC group model of activity coefficients the retention behaviour of a solute in the stationary phase of a liquid-solid chromatographic system is studied. By comparison of experimentally observed capacity ratios and calculated activity coefficients of solutes in the mobile phase, varying the concentration of a polar moderator, it is shown that the calculated activity coefficients in the stationary phase fit very well the equation formally identical with the Langmuir function. Comparison of activity coefficients in the mobile and the stationary phase proves equivalence between the solvent interaction and the competition theory.     

Improvement of on-line solid-phase extraction for determining phenolic compounds in water

Summary Modifying the most common design for the on-line coupling of a precolumn to reversed phase LC with diode array detection has resulted in reduction of the broadening of the peaks which results when the compounds of interest are strongly retained by a highly hydrophobic sorbent. The modification consists of the desorption of the analytes trapped on the precolumn solely by the organic solvent used to modify the solvent strength of the mobile phase. Results obtained using this design were compared with those obtained with the conventional design, with C₁₈ and PLRP-S precolumns. The performance of the system was also tested with a highly cross-linked styrene-divinylbenzene copolymer (ENVI-chrom P) precolumn for the determination of phenolic compounds in real samples. The advantages and disadvantages are discussed. Ion-pair solid phase extraction is used in order to increase the breakthrough volumes of more polar compounds, mainly phenol. The use of the new design enables phenolic compounds to be determined at the low μg L⁻¹ level with limits of detection ranging between 0.1 and 2 μg L⁻¹ in tap water when a 10 mL sample was analyzed.     

On the mechanism of solute retention in RP-HPLC systems with B+AB1+AB2 type eluents

Summary Intermolecular interactions between the solute and the moieties constituting the mobile phase significantly contribute to the overall retention pattern of a given solute in a given chromatographic system. In this paper retention of solute is discussed in the case of the B+AB₁+AB₂ type mobile phase, which, in a quasithermodynamic way, can be divided into seven individual moieties. One evaluates the influence of each moiety on solute retention, and refers these regularities to the polarity of the solutes.     

Correlation of the retention data of polyaromatic hydrocarbons obtained on various stationary phases used in normal- and reversed-phase liquid chromatography

Summary The correlation between the retention data of polyaromatic hydrocarbons (PAH) obtained in normal-and reversed-phase liquid chromatography is investigated in order to determine the dominant factors controlling the retention. It is clear that the separation of PAHs on various chemically-bonded packing materials in normal- and/or reversed-phase modes is primarily controlled by the molecular structure and shape. The -electron interaction between the solute and the stationary phase also contributes to the retention, although pure silica shows a somewhat different behavior.     

Influence of solvent effects on retention for a porous polymer sorbent in reversed phase liquid chromatography

Summary The influence of acetonitrile, methanol and isopropanol as retention selectivily modifiers in reversed phase liquid chromatography on a poly(styrene-divinylbenzene) macroporous polymer sorbent (PLRP-S) is evaluated using the solvation parameter model. Retention results from a combination of adsorption and partitioning and is influenced by the equilibrium absorption of organic solvent by the polymer from the mobile phase. The sorption of solutes is dominated by the ease of cavity formation in or on the solvated sorbent, with a small contribution from lone pair-lone pair electron interactions. All polar interactions, such as dipole-type and hydrogenbond formation, are more favorable in the mobile phase and reduce retention. Changes in the uptake of organic solvent from the mobile phase affect kinetic properties of the column such as band broadening and porosity as well as retention. The PLRP-S solvated sorbent is suitable for solid-phase extraction and is more retentive than typical silica-based, bonded phase sorbents for extraction from water. As a surrogate system for estimating solute lipophilicity and biological activity through retention-property correlations it provides a poor fit for hydrogen-bond acid solutes and is too dipolar/polarizable to fit some models.