A "Reference Series" Method for Prediction of Properties of Long-Chain Substances

      Long chain substances pose special challenges for property prediction, as many of their properties cannot be measured because of thermal instability. Nikitin et al. (2001), for example, noted that critical temperatures (Tc) and pressures (Pc) of n-alkanes have been measured only up to hexatriacontane (C₃₆H₇₄) and for 1-alkanols up to 1-docosanol (C₂₂H₄₅OH). The critical constants of the heavier members of the homologous series can only be predicted.

      Current methods used to predict physical and thermodynamic properties can be classified into group contribution methods (GC, e.g., Marrero and Gani, 2001), asymptotic behavior correlations (ABCs, e.g., Marano and Holder, 1997, Nikitin et al., 2005)) and various quantitative-structure-property relationships (QSPRs, e.g., Brauner et al, 2008). All of these methods use available experimental data for low carbon number (n_C) compounds in order to obtain either the "group contribution" values or the QSPR parameter values. The so-obtained group contributions or QSPRs are used for prediction of properties of long chain members of homologous series by extrapolation. The ABCs are non-linear correlations in terms of n_C , which use in addition to the experimental property data also an estimation of the property value at the limit n_C → ∞, y^∞.  Kontogeorgis and Tassios, 1997 compared several GC methods and methods that converge to a finite y^∞value, for predicting T_C and P_C of  heavy alkanes. They concluded that only methods that converge to finite y^∞ values yield reliable predictions for T_C and P_C of heavy alkanes.

      We (Paster et al., 2011) have recently developed a technique for applying linear QSPRs to prediction of properties of long chain substances in homologous series. Using this method, molecular descriptors collinear with a particular property are identified based on available experimental data. From among these, the descriptors whose asymptotic behavior is similar to the property behavior are eventually used for prediction.  For the cases studied in that work the QSPRs developed represented the available experimental data satisfactorily and converge to theoretically accepted values for n_C → ∞. It has been also found that the limiting property values for different homologous series are very close in value to each other, as the effect of the particular functional groups (e.g., –CH3, –COOH, –CO) diminishes with increasing n_C, where the role of the –CH2– chain becomes the dominant one. 

      The use of the method of Paster et al.(2011) can be challenging, as in order to predict the property for a particular (target) compound its 3D (or 2D) molecular structure (MOL) file must be available, together with a program for calculating the required molecular descriptor values. In the present work a new method is presented where the method of Paster et al. (2011) is applied only to one "reference" homologous series, thus molecular descriptors need to be calculated only for members of this series. Typically the n-alkane series, for which the largest amount of experimental data is available, is used as the reference series. To predict properties for other ("target") series the following two general characteristics of homologous series are utilized: 1.The relationship between the property y at a particular n_c of members of the "reference" series with the same property of the "target" series can be approximated locally by a straight line (see Peterson, 2010, for example) and 2. The property value for the "reference" and "target" series should approach the same (theoretically accepted) value for n_C → ∞.

Using these characteristics a nonlinear equation with three parameters that converges to y^∞ of the "reference" series (at the limit n_C → ∞) is fitted to the available property data of the "target" series vs. the property data of the "reference" series. This nonlinear equation enables predicting the property value for members of the "target" series using only n_C and the property value of the members of the "reference" series with the same n_C.

The proposed method has been applied to several "target" series (1-alkenes, aldehydes, 1-alcohols, n-aliphatic acids) for several properties, with very encouraging results. Detailed results of these studies and discussions will be provided in the extended abstract and the presentation.

 

References

1.       Brauner, N.; Cholakov, G. St.; Kahrs, O.; Stateva, R. P.; Shacham, M. Linear QSPRs for Predicting Pure Compound Properties in Homologous Series. AIChE J. 2008, 54(4), 978-990.

2.       Kontogeorgis G. M.; Tassios D.P.; Critical constants and acentric factors for long-chain alkanes suitable for corresponding states applications. A critical review. Chemical Engineering Journal. 1997;66:35-49.

3.       Marano, J.J.; Holder, G.D. General Equations for Correlating the Thermo-physical Properties of n-Paraffins, n-Olefins and other Homologous Series. 2. Asymptotic Behavior Correlations for PVT Properties. Ind. Eng. Chem. Res. 1997A, 36, 1895.

4.       Marrero, J.; Gani, R. Group-contribution based estimation of pure component properties. Fluid Phase Equilibria. 2001, 183.

5.       Nikitin, E.D.; Pavlov, P.A.; Popov, A.P. Critical temperatures and pressures of some alkanoic acids (C2 to C22) using the pulse-heating method. Fluid Phase Equilibria. 2001;189:151-161.

6.       Nikitin, E.D.; Popov, A.P.; Bogatishcheva, N.S. Critical properties of long-chain substances from the hypothesis of functional self-similarity. Fluid Phase Equilibria. 2005;235:1-6.

7.       Paster, I.; Shacham, M.; Brauner, N. Adjustable QSPRs For Prediction of Properties of Long-Chain Substances. AIChE J, 2011; 57(2); 423–433.

8.       Peterson, B. K., Relationships between the Properties of Families of Materials, Ind. Eng. Chem. Res., 2010, 49(7), 3492-3495