Vapor–liquid equilibrium curves of aqueous 30 wt % MEA and NMP + 30 wt % MEA both at 120 °C

The liquid stream coming out of the reboiler is the lean solvent, which is later recirculated back to the absorber. In the reboiler, this lean solvent is in equilibrium with a vapor-containing solvent and CO₂ at a certain pressure p_R. This pressure will then limit the operability of the reboiler. In other words, the vapor–liquid equilibrium behavior of the solvent will always tie together the composition of the lean solvent and the pressure of the reboiler.

Wanderley et al. (10) have obtained VLE curves for a variety of water-lean solvents at 120 °C. For the moment, we shall focus our attention on solvents with low volatility such as N-methyl-2-pyrrolidone, as has been demonstrated in {C}Section 4.2{C} that these are the ones most likely to provide lower reboiler duties. The VLE behavior of NMP + 30 wt % MEA has been compared to that of aqueous 30 wt % MEA at 120 °C in Figure 13.

Supposing one wants to produce a lean solvent with α = 0.2 mol CO₂·mol MEA^–1, the operational pressure of the reboiler with aqueous 30 wt % MEA at 120 °C is p_R = 195 kPa. This is a feasible scenario. For NMP + 30 wt % MEA, however, the operational pressure is p_R = 95 kPa, a subatmospheric condition.

This imposes a series of limitations for water-lean solvents with low volatility. If one wants to actually recover the solvent with the same lean loading as that of aqueous MEA, they will need to resort to different strategies. One possible strategy would be elevating the temperature of the reboiler above 120 °C. However, what prevents the operation of the reboiler at higher temperatures is the degradation of the solvent, and it has been previously assessed by our group (58) that primary amines in water-lean solvents are more vulnerable to thermal degradation (and possibly oxidative degradation) than when in aqueous solutions. Another alternative is employing an inert stripping gas, something already suggested by Rivas and Prausnitz in 1979, (7) or at least a nonreactive cosolvent with high volatility such as suggested by Frimpong et al. (59) But then again, adding a volatile liquid to the mixture seems to defeat the purpose of choosing a water-lean solvent with low volatility to begin with. A stripping gas might then be a feasible solution, with the caveat that the CO₂ henceforth produced at the top of the desorber will most likely not be pure enough for storage.

For recovering high-purity CO₂ ready for storage or utilization when employing a low-volatility nonaqueous solvent, we can finally propose two alternatives with very clear drawbacks. The first is to recover the solvent by operating the desorber under subatmospheric conditions, i.e., by vacuum stripping, with an obvious penalty in electrical power. This seems to be the approach adopted by some researchers working with ionic liquids (60) (many proposals for CO₂ regeneration after capture by ionic liquids can be operated at 100 kPa, but only because absorption itself is carried out at very high pressures so that the full potential of a pressure-swing can be attained (61−63)). The second is to recover a lean solvent with higher loading than that of aqueous amines.

If one fixes the pressure of the reboiler at p_R = 200 kPa and T_R = 120 °C, it will imply a lean loading of α = 0.31 mol CO₂·mol MEA^–1 for NMP + 30 wt % MEA (as against α = 0.22 mol CO₂·mol MEA^–1 for aqueous 30 wt % MEA), as seen on Figure 13. This is a significant increase. As mentioned in Section 2.7, due to the vapor–liquid equilibrium shift in water-lean formulations, these solvents are already operating with lower rich loadings in the absorber. If the lean loading must be higher as well, this means that the cyclic capacity of water-lean solvents is lower than that of aqueous amines. The consequence is that more liquid is required to capture the same amount of CO₂, which not only jeopardizes any benefit acquired in terms of mass transfer coefficients as discussed in Section 2 but also will increase the reboiler duties calculated in Section 4.2 under the assumption of comparable cyclic capacities Δq.

The present analysis has been carried out for the case of NMP + 30 wt % MEA. A similar analysis could be made with other solvents of very low volatility, such as MEG or THFA, both presented by Wanderley et al. (10) For solvents with lower volatility than water but higher than NMP, such as the 2-methoxyethanol (a.k.a. methyl cellosolve) studied by Guo et al., (41) one could possibly find a compromise between lower vaporization heat and the capacity of regenerating the solvent at higher pressures. Conversely, the addition of some water to a water-lean solvent may moderately increase its mass transfer coefficient and provide just enough steam in the reboiler so that the lean amine can be recovered at atmospheric pressures without excessive solvent volatilization. This seems to be the approach adopted by Semenova and Leites (13) for their water-lean solvents evaluated in pilot plant conditions and also by the semiaqueous solvents developed by ION Engineering (64) among others.