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

Summary Two analytical methods have been developed for the determination in water of 18 priority phenolics listed in US EPA method 604 and on EEC list 76/464. A solidphase extraction system using eight different sorbents packed in a precolumn was coupled on-line with a liquid chromatograph with UV detection. The ensuing method uses 50–100 mL of ground water; its performance was compared with that of an off-line method using Empore extraction disks and 1 L water samples. Phenol recoveries varied from <20 to 100% for concentrations in the range 0.1–10 g/L at an acid pH. The presence of the phenols in water was confirmed by using thermospray LC-mass spectrometry in the negative ion mode. The stability of the phenols in water was studied at a 10 g/l level in ground and estuarine water at acid pH (2.5–3) and at 4°C for 1 month. The system was validated by various interlaboratory exercises with samples containing 2,4,6-trichlorophenol and pentachlorophenol at concentrations from 0.1 to 0.5 g/L.     

Modelling of retention behaviour of solutes in micellar liquid chromatography

In micellar liquid chromatography (MLC), the resolution for a given multi-component mixture can be optimized by changing several variables, such as the concentrations of surfactant and organic modifier, the pH and temperature. However, this advantage can only be fully exploited with the development of mathematical models that describe the retention and the separation mechanisms. Several reports have appeared recently on the possibilities of accurately predicting the solute retention in MLC. Although the retention and selectivity may strongly change with varying concentrations of surfactant, organic modifier and/or pH, the observed changes are very regular, and are well described by simple models. This characteristic enables a successful prediction of retention times and compensates the negative effect of the broad and tailed chromatographic peaks obtained for some solutes when micellar eluents are used. An overview of the models proposed in the literature to describe the retention behaviour in pure micellar eluents and micellar eluents containing an organic modifier, at a fixed pH or at varying pH, is given. The equations derived permit the evaluation of the strength of micelle-solute and stationary phase-solute interactions. The prediction of the retention based on molecular properties and the use of neural networks, together with the factors affecting the prediction capability of the models (linearization of the equations, dead time, critical micellar concentration, ionic strength and temperature) are commented on. The strategies used for the optimization of resolution are also given.     

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.     

Heteroscedasticity of retention factor and adequate modeling in micellar liquid chromatography

The two concepts of micelle formation (pseudo-phase and mass-action) could be the basis of retention models in micellar liquid chromatography (MLC). The separation of 4-hydroxybenzoic acid esters and seven polyaromatic hydrocarbons were performed to study the repeatability of retention factor in MLC. The full two factor experimental design was used for studying the dependence of retention factor variance on mobile phase composition (sodium dodecylsulfate, 1-butanol). The experimentally observed heteroscedasticity and perturbations after linearization were taken into account by using statistical weights obtained on the basis of errors propagation law and the modeling of retention by non-weighted and weighted least squares method was performed. The mechanistical retention models based on pseudo-phase and mass-action concepts of micelle formation were compared by fitting quality and prediction capability and high robustness of bilogarithmic dependence was observed. The significance of retention factor heteroscedasticity for retention hydrophobicity relationships was shown.     

Application of artificial neural networks for gradient elution retention modelling in ion chromatography

Gradient elution in ion chromatography (IC) offers several advantages: total analysis time can be significantly reduced, overall resolution of a mixture can be increased, peak shape can be improved (less tailing) and effective sensitivity can be increased (because there is little variation in peak shape). More importantly, it provides the maximum resolution per time unit. The aim of this work was the development of a suitable artificial neural network (ANN) gradient elution retention model that can be used in a variety of applications for method development and retention modelling of inorganic anions in IC. Multilayer perceptron ANNs were used to model the retention behaviour of fluoride, chloride, nitrite, sulphate, bromide, nitrate and phosphate in relation to the starting time of gradient elution and the slope of the linear gradient elution curve. The advantage of the developed model is the application of an optimized two-phase training algorithm that enables the researcher to make use of the advantages of first- and second-order training algorithms in one training procedure. This results in better predictive ability, with less time required for the calculations. The number of hidden layer neurons and experimental data points used for the training set were optimized in terms of obtaining a precise and accurate retention model with respect to minimization of unnecessary experimentation and time needed for the calculation procedures. This study shows that developed, ANNs are the method of first choice for retention modelling of inorganic anions in IC.     

Factors influencing retention and resolution of substituted alkylbenzenes in reversed-phase liquid chromatography

Summary The retention data of isomeric xylenes, ethyltoluenes and diethylbenzenes, and of mesitylene, benzene, toluene and ethylbenzene were obtained on a reversedphase column using methanol-water and ethanol-water mobile phases at four different temperatures. This database was used to relate the dependence of solute retention and resolution on the polarity of the mobile phase, solute dipole moment, and column temperature. The additivity of the free energy of the transfer of solute molecules or solute segments from the stationary phase to the mobile phase, was examined for the isomeric compounds. For this, the logarithm of the net retention volume was substituted for the free energy. Deviations from the additivity of free energies indicate that the separation of isomeric substituted alkylbenzenes is governed by their differential interactions with both the polar mobile phase and nonpolar stationary phase. Among the disubstituted alkylbenzenes,ortho-isomers favor the mobile phase more andpara-isomers tend to prefer the stationary phase more. Themeta-isomers are found to follow the additivity rule closely. These trends are amplified as the polarity of the mobile phase is increased indicating that these isomers are resolved better in water-rich mobile phases. These findings are substantiated by analogous results from gas-liquid chromatographic retention data, estimation of dipole moment effects, and examination of the entropic and enthalpic contributions to the net retention volume.Dedicated to Professor Leslie S. Ettre on the occasion of his 70th birthday.     

Development of an inorganic cations retention model in ion chromatography by means of artificial neural networks with different two-phase training algorithms

This paper describes development of artificial neural network (ANN) retention model, which can be used for method development in variety of ion chromatographic applications. By using developed retention model it is possible both to improve performance characteristic of developed method and to speed up new method development by reducing unnecessary experimentation. Multilayered feed forward neural network has been used to model retention behaviour of void peak, lithium, sodium, ammonium, potassium, magnesium, calcium, strontium and barium in relation with the eluent flow rate and concentration of methasulphonic acid (MSA) in eluent. The probability of finding the global minimum and fast convergence at the same time were enhanced by applying a two-phase training procedure. The developed two-phase training procedure consists of both first and second order training. Several training algorithms were applied and compared, namely: back propagation (BP), delta-bar-delta, quick propagation, conjugate gradient, quasi Newton and Levenberg-Marquardt. It is shown that the optimized two-phase training procedure enables fast convergence and avoids problems arisen from the fact that every new weight initialization can be regarded as a new starting position and yield irreproducible neural network if only second order training is applied. Activation function, number of hidden layer neurons and number of experimental data points used for training set were optimized in order to insure good predictive ability with respect to speeding up retention modelling procedure by reducing unnecessary experimental work. The predictive ability of optimized neural networks retention model was tested by using several statistical tests. This study shows that developed artificial neural network are very accurate and fast retention modelling tool applied to model varied inherent non-linear relationship of retention behaviour with respect to mobile phase parameters.     

Determination of ephedra alkaloids by high-performance liquid chromatography

Summary Two simple methods were developed for the simultaneous determination of six alkaloids (ephedrine, pseudoephedrine, norephedrine, norpseudoephedrine, methylephedrine and methyl-pseudoephedrine) inEphedrae Herba by high-performance liquid chromatography. The first method was carried out by using a Cosmosil 5C₁₈-MS column with a gradient solvent system consisting of a phosphate buffer and acetonitrile, and detection at 210 nm. The contents of alkaloids in non-pretreated ephedra herb extracts could be determined easily in 50 min. Alternatively, the alkaloids could be determined within 35 minutes by using a Cosmosil 5C₁₈-MS column with an isocratic solvent system of a sodium dodecyl sulfate-acetonitrile solution. The two methods are compared and discussed.     

Retention behavior of dipeptide isomers in reversed-phase high-performance liquid chromatography

Summary The effects of eluent pH and organic modifier concentration on the capacity factor (k) and selectivity of dipeptide isomers were investigated. It has been observed that the variation in the logarithm of the capacity factor of the dipeptide isomers is linearly dependent on the organic modifier concentration (C_b), however, the selectivity is almost independent of it. Both capacity factor and selectivity were seriously affected by the pH of the eluent. Both the capacity factor and the intercept of the ln k vs. C_b plot increased with increasing van der Waals volume of the non-polar amino acid subunit of the dipeptides.     

Hexadecyltrimethylammonium bromide/ethanol/water systems as mobie phase in reverse phase high performance liquid chromatography. Study of metal-diethyldithiocarbamate complexes retention

Summary This paper presents the, chromatographic retention of Co(II), Ni(II) and Cu(II) as diethyldithiocarbamate complexes in the presence of hexadecyltrimethylammonium bromide/ethanol/water systems, as mobile phase, by reversed phase high performance liquid chromatography. The presence of an organic modifier reduces the retention times and improves the efficiency. In order to evaluate the interaction between the metal complexes and the mixed micellar system the values of solute binding constants are calculated in, two ways: a) Arunyanart and Cline-Love's treatment and b) multiple regression analysis taking account of the ethanol percentages.     

Determination of some banned stimulants in sports by micellar liquid chromatography

Summary A procedure has been developed for the determination, in <12 min, of several stimulants (amphetamine, ephedrine, methoxyphenamine, phenylephrine and phenylpropanolamine) in spiked urine samples after direct injection, using a hybrid micellar mobile phase of 0.15 M sodium dodecyl sulfate and 3% pentanol at pH 7, on a C₁₈ column with UV detection. Recoveries were 94–102% and limits of detection 4.5 ng·mL⁻¹ for methoxyphenamine and 0.39 μg·mL⁻¹ for amphetamine, similar to those obtained for aqueous solutions. Linearity reached 0.99 and intermediate precision was <8.4 and 5.3, for the two different concentrations tested.     

Specific retention behaviour of metal (III) tetraphenylporphyrins in non-aqueous,reversed high-performance liquid chromatography

Summary The Co(II), Ni(II), Fe(III) and V(IV) complexes of tetraphenylporphine (TPP) can be eluted at short retention times from a LiChrosorb RP-18 column with pure ethanol. However, both Mn(III) and Co(III) complexes of metal TPP chloride type are so strongly retained on the column that they cannot be eluted. While the retention of other metal teraphenylporphine complexes was not affected, that of the metal(III) complexes of the TPP chloride type especially MnTPPCl and CoTPPCl, decreases dramatically with an increase in the concentration of NH₄Cl added into the mobile phase; a linear relationship between logk' and log[NH₄Cl], with the slope of about–1, has been observed for these two metal(III) complexes in the NH₄Cl concentration range from 2.5×10^–4 to 1.3×10^–2 mol/l. Thus, the specific control of the retention of the metal(III) complexes is enabled by conditioning the NH₄Cl content of the mobile phase, and the chromatographic separation is demonstrated.     

Correlation between retention and the C18 silica surface coverage in reversed-phase liquid chromatography

Summary Members of three homologous series and five non-homologous solutes with various functional groups were chromatographed on a series of well-characterized C₁₈ reversed stationary phases with a range of methanol-water mobile phases. Measured capacity factors of solutes were correlated with the concentration of C₁₈ ligands on the stationary phase. A linear relationship with the slope depending on a solute molecular structure and the volume fraction of methanol in the mobile phase was obtained. A method for the evaluation of phase ratio is also proposed. Presented at the 17th International Symposium on Chromatography, September 25–30, 1988, Vienna, Austria.     

On the chromatography mechanism of reversed-phase liquid chromatography

Summary An equation is derived which can describe how the retention of solutes is influenced by the composition of the mobile phase in reversed-phase liquid chromatography, the retention of solutes in alkyl bonded stationary phase regarded as the complexation between solute molecule and the active sites on the surface of the stationary phase. When the stationary phase is not fully saturated by the organic modifier, the activity of the active sites, the activity coefficient of the adsorbed solute as well as the activity coefficient of the solute in the mobile phase depend on the composition of the mobile phase. However, when the stationary phase is fully saturated, the composition of the mobile phase mainly influences the activity coefficient of the solute in the mobile phase. In addition, the selectivity of retention is discussed in terms of the derived equation.     

Analysis of ofloxacin in plasma samples by high-performance liquid chromatography

Summary A sensitive HPLC method has been developed for determination of ofloxacin (OFL) in biological fluids. Sample preparation was performed by adding phosphate buffer (pH 7.4, 0.1m) then extraction with trichloromethane. OFL and the internal standard, sarafloxacin (SAR), were separated on a reversed-phase column with aqueous phosphate solution-acetonitrile, 80∶20, as mobile phase. The fluorescence of the column effluent was monitored at λ_ex 338 and λ_em 425 nm. The retention times were 2.66 and 4.24 min for OFL and SAR, respectively, and the detection and quantitation limits were 8 and 15 ng mL⁻¹, respectively. Plots of response against ofloxacin concentration were linear in the range 8 to 2000 ng mL⁻¹. Recovery was 92.9% for OFL.     

Microcomputer-assisted retention prediction system in reversed-phase HPLC

Summary The microcomputer-assisted retention prediction system in C₁₈ reversed-phase HPLC is described. The system is based on the use of the hydrophobic parameter and the correlation factor of alkylbenzenes and polycyclic aromatic hydrocarbons (PAHs) which are highly correlated to their retention in reversed-phase HPLC. The system is evaluated by comparing the retention data between measured and predicted values. One of the typical examples applied for PAH analysis in the extracts of diesel particulate matter shows the high potential of the system investigated.