(123c) Modelling the Phase Behaviour of the CO2+H2O+Amine Mixtures Using Transferable Parameters with SAFT-VR | AIChE

(123c) Modelling the Phase Behaviour of the CO2+H2O+Amine Mixtures Using Transferable Parameters with SAFT-VR

Authors 

Galindo, A. - Presenter, Imperial College London
Adjiman, C. S. - Presenter, Imperial College London,Center for Process Systems Engineering
Jackson, G. - Presenter, Imperial College London


The reduction in CO2 emissions from anthropogenic sources has become a topic of widespread interest in recent years. As the power generation sector is by far the largest stationary-point-source of CO2, being responsible for approximately 35% of total global CO2 emissions1 this issue has particular relevance for the energy sector. The current method of choice for large-scale CO2 capture is amine-based chemisorption; typically in packed columns, with the solvent of choice being a primary alkanolamine: monoethanolamine (MEA). Despite the widespread use if this technique, there is an ongoing debate as to how best to model these systems, owing to the complex reactions and phase equilibrium they exhibit. To capture the interactions that occur in these systems, we use the statistical associating fluid theory (SAFT)2. This is a molecular approach, specifically suited to hydrogen-bonding, chain-like fluids. In this contribution we use the SAFT approach for potentials of variable range (SAFT-VR3) to calculate the fluid phase behaviour of amine + H2O + CO2 mixtures. The molecules are modelled as homonuclear chains of bonded square-well segments of variable range, and a number of short-ranged off-centre attractive square-well sites are used to mediate the anisotropic effects due to association in the fluids.

We model MEA as a pair of tangentially bonded homonuclear spherical segments. Six distinct association sites are required to mediate the hydrogen bonding interactions exhibited by this molecule. In discriminating between potential models of MEA, we consider both symmetric models - where no distinction is made between the functional groups on MEA ? and asymmetric models of MEA ? where we explicitly consider the multifunctional nature of MEA. We find that despite giving an adequate description of the MEA + H2O phase behaviour a purely symmetric model of MEA is not suitable for describing either the phase behaviour MEA + CO2 binary mixture or that of the MEA + CO2 + H2O mixture. Further, as the chemistry of this system is well known, we note that it is not possible to preserve the stoichiometry of this mixture with a purely symmetric model of MEA. In order to reduce the number of adjustable parameters required to describe MEA, we exploit the molecular nature of SAFT-VR by transferring parameters from SAFT-VR models of alkanols and alkylamines developed in this work. We select our asymmetric model of MEA, by consideration of its ability to represent vapour-liquid equilibrium (VLE) data as well as enthalpy of vaporisation and interfacial surface tension data. Models for H2O and CO2 are taken from previous work4,5 with two effective "reaction" sites being incorporated in the CO2 model to mediate the chemical interaction between the MEA and CO2. In modelling binary mixtures of MEA+H2O, parameters are transferred from binary mixtures of ethanol+ H2O and ethylamine+ H2O. Unlike interaction parameters to describe the MEA+ CO2 mixture are determined by comparison with ternary experimental data. Our final calculations for the ternary mixture predict the phase behaviour of with an absolute average deviation in the mole fraction of CO2 in the liquid phase 0.01. The excellent accuracy of the asymmetric MEA model and of the MEA+H2O+CO2 mixture model over a wide range of conditions makes the proposed approach suitable for incorporation in process modelling.

Building on this success, we also investigate alternative solvents. There is a well recognised requirement for an improvement on the MEA based capture methods6, with emphasis on increasing CO2 loading capacity and absorption rate. Recent work7 has shown that alkylamines could be ideal candidates to address both of these requirements. However, before these solvents can be deployed in practice, a detailed knowledge of their phase behaviour in aqueous CO2 mixtures is required. The unlike interaction parameters required to describe aqueous alkylamine mixtures are obtained by comparison to experimental VLE data. Due to the lack of data for the RNH2 + CO2 mixture, we transfer binary interaction parameters obtained from NH3+ CO2 mixture. Our predictions confirm that ternary mixtures of RNH2+H2O+CO2, for alkyl chains with more than 2 carbon atoms, exhibit liquid-liquid immiscibility, the extent of which increases with alkyl chain length. The consequence of this phenomenon is that the aqueous phase separates into a water-rich and amine-rich region. This phase split results in one phase of essentially pure water, and a concentrated amine-CO2 phase, containing only a small amount of water. This provides a valuable basis for experimentation and preliminary assessment of alkylamines as solvent alternatives.

1. Steeneveldt, R. et al., ChERD, 84(A9): 739-763, 2006

2.Chapman, W.G. et al., Ind. Eng. Chem. Res., 29, 1709-1721, 1990

3.Gil-Villegas, A. et al., Chem. Phys. 106 (10), 1997

4.Clark, G. N. I. et al., Mol. Phys.,04(22-24):3561-3581, 2006

5.Galindo, A. et al., J. Phys. Chem. B., 106: 4503-4515, 2002

6.Wolsky, A. M., Daniels, E. J. and Jody, B. J., Environ. Prog., 13, 214, 1994

7.Singh, P., Niederer, J. P. M. and Versteeg, G. F. , Int. J. of Greenhouse Gas Control, 1, 2007, 5 - 10