Solubility models for amino acids and antibiotics

Abstract

An approach is presented for modeling the effects of temperature and pH on the solubility of amino acids in water. The amino acids studied are alanine, amino-butyric acid, glycine, hydroxy-proline, proline, serine, threonine, and valine. The data employed are activity coefficients and solubilities of the amino acids in neutral water and the dissociation constants for the various amino-acid ionization reactions. Activity coefficients are correlated with the modified UNIFAC group contribution model and with new glycine and proline groups being introduced.A similar approach is presented for modeling the solubilities of certain antibiotics in mixed nonaqueous solvents. Fusion temperature data and solubilities of the antibiotics in pure organic solvents are used to deduce modified UNIFAC energy interaction parameters between new large antibiotic groups and the standard alkane (CH,), alcohol (OH), aromat (ACH), and ester (CCOO) groups. IntroductionIn a typical case, the cost involved in bioseparations may be as high as ninety percent of the total cost of manufacturing. Unfortunately, this area has not been given much research attention and bioseparation units are most often designed empirically rather than on the basis of rational information.The prediction of activity coefficients for amino acids has had only recent attention. Orella and Kirwan (1987), in their unpublished work, have made an attempt to predict solubilities of several amino acids (glycine, alanine, phenylalanine, valine, leucine, and iso-leucine). They considered the activitycoefficient to be made up of terms due to chemical interactions (from UNIFAC) and due to electrostatic interactions (from the extended Kirkwood theory).Nass (1988) has correlated amino acid activity coefficient and solubility data. She assumed that the activity coefficient is a product of terms due to chemical reaction equilibrium and due to physical interaction. Wilson's equation was used for the physical activity coefficient, with Bondi's volumes (Bondi, 1968) inserted for the pure-component liquid volume ratios. Activity coefficients for alanine, serine, and threonine in water, and solubilities of phenylalanine, tyrosine, and diiodotyrosine in water have been correlated. The correlations are in good agreement with the experimental data. The number of parameters regressed varied from three to ten. In this work, an attempt has been made to build a predictive model for the activity coefficients of amino acids. Amino acids in the study are alanine, amino-butyric acid, glycine, hydroxyproline, proline, serine, threonine, and valine. These are eight of the twenty most common amino acids. The effect of pH and temperature on amino acid solubility are considered.We have also explored the possibility of constructing a predictive model for the solubility in mixed solvents of six antibioticsanisomycin, carbomycin A, chloramphenicol, chloramphenicol palmitate, griseofulvin, and hygromycin A. The selection of these antibiotics is based on their commercial importance and the availability of so...

Solubility of Amino Acids: A Group-Contribution Model Involving Phase and Chemical Equilibria

“…Nevertheless, they are about the same order of magnitude as the values obtained by Gupta and Heidemann (1990). For the prediction of the activity coefficients of valine, aminobutyric acid, and aminovaleric acid an incorrect slope was calculated for all the cases, withy* decreasing with increasing molality, while the data show the opposite trend.…”

“…They are all of good quality. Table 5 summarizes some of the information of both correlation and prediction, and a comparison is given with the results of Chen et al (1989) and Gupta and Heidemann (1990). The root mean square deviation (rmsd) values for correlation are quite satisfactory, and in the majority of the cases the deviations are even smaller than those calculated by Chen et al (1989), who only correlated one amino acid or one peptide a t a time.…”

“…More recently, Gupta and Heidemann (1990) tried to describe the activity coefficients of amino acids in water considering only short-range interactions, using the modifiedUNIFAC model (Larsen et al, 1987). Their definition * Author to whom all correspondence should be addressed.…”

L‐serine obtained by methanol addition in batch crystallization

“…The chemical term was modeled by chemical theory, while physical interactions were accounted for by a form of the Wilson equation. Another approach has been proposed by Gupta and Heidemann (1990). The use of the modified UNIFAC group contribution model (Larsen et al, 1987) is recommended to evaluate activity coefficients.…”

Prediction of pH in complex aqueous mixtures using a group‐contribution method

“…Activity coefficients for alanine, serine, and threonine in water and solubilities of phenylalanine, tyrosine, and diiodotyrosine in water were correlated. Gupta and Heidemann (1990) extended this work to antibiotics, using the UNIFAC group-contribution method to calculate activity coefficients. They assumed no ionization of amino acids, and the solution is assumed to behave ideally for the expression of equilibrium constants and pH estimation.…”