This work focused on determining the dissociation constants (pKa) for eight amines, namely, 3-(Diethylamino) propylamine, 1,3-Diaminopentane, 3-Butoxypropylamine, 2-(Methylamino) ethanol, Bis(2-methoxyethyl) Amine, α-Methylbenzylamine, 2-Aminoheptane, and 3-Amino-1-phenylbutane at temperatures ranging from 293.15 K to 323.15 K. The protonated order of two polyamines, 3-(Diethylamino) propylamine and 1, 3-Diaminopentane, were determined using computational chemistry methods. The dissociation constants at the standard temperature of 298.15 K were estimated using group functional models (paper-pencil) and computational methods using software such as COSMO-RS and Gaussian. In addition, the pKas at various temperatures were calculated using computational methods for two different thermodynamic cycle. A simple artificial neural network (ANN) method was also employed to reduce the calculation time as well as improve the accuracy. Instead of using the experimental property data, these could be generated using Aspen Plus or CosmothermX. The simulated model provided a very good fit to the pKa values.

In this study, the dissociation constants of the eight amines, namely, N-(2-aminoethyl)-1,3-propanediamine, 2-methylpentamethylene diamine, N, n-dimethyldipropylene-triamine, 3,3’-Diamino-n-methyldipropylamine, Bis[2-(n, n-dimethylamino) ethyl]ether, 2-[2-(Dimethylamino) ethoxy] Ethanol, 2-(dibutylamino) Ethanol and N-propylethanolamine were determined from 298.15 K to 313.15 K. Using the van’t Hoff equation, thermodynamic properties such as the standard state changes of enthalpy, entropy and Gibbs free energy were calculated. Using computational chemistry calculations, the amino group protonated first was predicted. Furthermore, computer free group contribution methods such as the original Perrin-Dempsey-Serjeant (PDS), the modified PDS and the Qian-Sun-Sun-Gao (QSSG) model were used to estimate the dissociation constants of the studied amines. In these methods, the QSSG provided the most accurate results as the database used in this method was updated with additional the functional groups as well as information about group positions. Finally, an artificial neural network was used to predict the pKa values.