3.1. VLE Experiments
The CO
2 absorption of a liquid crystalline sample containing MEA (60% L92/10% MEA) was measured, and the VLE data is shown in
Figure 1, were the partial pressure of CO
2 in the reactor (P
CO2) has been plotted as a function of loading (α). The final loading of the sample is 38.6 g CO
2/kg sample. The VLE data of 60% L92/10% MEA shows that the sample exhibits dual behavior: first chemical absorption takes place and then physical absorption occurs. After each of the first 11 injections, the pressure in the reactor drops to approximately 0 bar, indicating that all CO
2 has reacted with MEA. Hence, the loading of the solution increases, while the partial pressure of CO
2 in the reactor remains low. After loading 32 CO
2/kg sample, the partial pressure in the reactor starts to increase fast and linearly, indicating that physical absorption is occurring [
27]. In
Figure 1, together with experimental data, the vapor–liquid equilibria model for 10% MEA has been plotted at 30 °C [
33]. The behavior of the two systems (both containing 10% MEA) is very similar. This indicates that the absorption of CO
2 into the developed liquid crystal (g CO
2/kg sample) is limited by the amount of amine, and that all the amine in the liquid crystal is still active and can be used to absorb CO
2. However, due to the high viscosity of the unloaded and loaded liquid crystal (see
Section 3.3.1), the VLE experiments took much longer than in typical aqueous MEA solutions.
During the experiment a change in phase behavior of 60% L92/10% MEA was observed, where the physical properties of the sample changed. In order to investigate this phase transition, loading of the sample was performed in a round bottom flask.
3.2. Phase Behavior of 60% L92/10% MEA during CO2 Absorption
Pure CO
2 was bubbled through 60% L92/10% MEA in a round bottom flask in order to investigate the phase transition occurring during the loading process. Samples were acquired at various arbitrary times during the experiment and analyzed by TGA to check if there was water loss due to CO
2 bubbling. The CO
2-loading of the samples was quantified by titration. In this manuscript, the samples are named according to the CO
2-loading, α (
Table 1).
Results of the TGA measurements are shown in
Figure 2, and the initial (unloaded) sample was used as a control reference. It is evident from the plot that 40% of the weight is lost between 20 °C and 150 °C. This corresponds to loss of water and MEA. Subsequently, the curve reaches a plateau, followed by a decrease around 350 °C. This behavior corresponds to the polymeric portion of the sample, which starts to decompose above 350 °C. The TGA results confirm that the initial sample consisted of 39% solvents and 61% of polymer. The results showed that until the sample reached a loading of 27 g CO
2/kg sample, the composition did not change. The polymer concentration of the sample with loading 35 g CO
2/kg sample reached 71%, due to evaporation of water caused by bubbling of CO
2. Therefore, the composition of this sample was adjusted by adding Milli-Q water to the reaction mixture. The loading experiment was considered to be complete when the loading of the sample reached 38 g CO
2/kg sample. TGA results showed that the final sample composition consisted of 57% polymer and 43% solvent.
The CO
2-loading of the calorimeter sample at one bar is approximately 34 g CO
2/kg sample, whereas the final loading obtained in the round-bottom flask experiment is 38 g CO
2/kg sample. The 10% difference between these two values might be caused by slight variations in the composition of the samples, as well as the different techniques used to determine the loading of the samples. The loadings obtained for this liquid crystalline system at 1 bar, are slightly higher than the values for three different thermotropic liquid crystals reported by Chen et al. (MBBA, PCH5, and PCH8-CNS), where the highest loadings they measured were between 7–8 mg CO
2/g liquid crystal [
34].
SAXS spectra of 60% L92/10% MEA with loadings α = 0, 15, 27, and 38 g CO
2/kg sample were recorded at 15, 25, 35, and 45 °C. However, it should be noted that it was not possible to distinguish between hexagonal and reverse hexagonal phases. Therefore, the symbol used for hexagonal phases (H) represents both normal hexagonal and reverse hexagonal phases. The spectra measured at 35 and 45 °C showed that the microstructure of the samples was destroyed, and therefore these spectra are not presented. The results for unloaded 60% L92/10% MEA (α = 0 g CO
2/kg sample) were previously reported [
21]. The SAXS spectra displayed in
Figure 3 and
Figure 4, show how the microstructure of the sample changes as the CO
2-loading increases at 15 and 25 °C, respectively. In addition, the corresponding lattice parameters have been calculated, and are shown in
Table 2.
The SAXS spectrum recorded at 15 °C (
Figure 3) shows that the unloaded sample forms a lamellar phase (Lα) [
21]. At α = 15 g CO
2/kg sample, a broad intense peak forms, with some shoulders at higher q values. However, the relative peak positions do not follow the pattern of any liquid crystalline structure. The broad peak at α = 15 g CO
2/kg sample is slightly shifted to lower q values in comparison with the most intense peak at α = 0, implying that the distance between aggregates increases. When the loading reaches 27 g CO
2/kg sample, a new Bragg peak appears at low q values, and the most intense peak in this spectrum has a shoulder. The ratio between the three peaks follows the relationship established for hexagonal phases. In this case, the peaks are shifted towards higher q values, meaning that the aggregates are more compressed than the ones with loadings 0 and α = 15 g CO
2/kg sample. Similarly, the spectrum of the sample with α = 38 g CO
2/kg sample follows the same trends as the sample with α = 27 g CO
2/kg sample. In this case, there is a clearer distinction between the two most intense peaks that are merged. The relationship between the less intense peak at low q values and the two intense peaks is 1÷
÷2, which corresponds to a hexagonal phase.
At 25 °C (
Figure 4), when there is no CO
2 present, the sample has a defined microstructure consisting of coexisting lamellar and hexagonal phases [
21]. In the case of the other samples, the phase behavior is analogous at 15 and 25 °C. As the degree of CO
2-loading increases, the microstructure shifts towards hexagonal phase. Comparing the samples with loadings 0 and 15 g CO
2/kg sample, it can be seen that a broad intense peak appears, which seems to be the result of the combination of the most intense peaks of the lamellar and hexagonal phases present in the unloaded sample, and suggests the presence of amorphous aggregates. As discussed for 15 °C, the main peak of α = 15 g CO
2/kg sample is slightly shifted to lower q values with respect to the intense peaks of the unloaded sample, indicating the larger distance between aggregates. Moreover, there are shoulders in the spectrum α = 15 g CO
2/kg sample in the positions where the Bragg peaks of the unloaded sample appear, indicating that part of the structure is still remaining. At higher q values two shoulders appear which follow the relationship 1÷
÷2 with respect to the most intense peak, possibly indicating the presence of remaining hexagonal aggregates. The lattice parameters of this sample are not calculated due to its undefined microstructure.
The spectra obtained for loadings of 27 and 38 g CO
2/kg sample follow the same behavior as described for 15 °C. In both cases, a small Bragg peak starts to appear at low q values. The spectra of both samples show the presence of an intense peak with a shoulder, which seems to correspond to two merged peaks. The distinction between these peaks is clearer at 25 °C than at 15 °C, and it is more defined for the highest loading, suggesting that the aggregates are more ordered and that at 15 °C the transition towards hexagonal phase is still ongoing when the loading is 27 g CO
2/kg sample. At 25 °C, the relationship between the peaks at 27 and 38 g CO
2/kg sample follows the ratio 1÷
÷2, which corresponds to a hexagonal phase. However, the positions of the main peaks of α = 27 and 38 g CO
2/kg sample at 25 °C appear to be shifted with respect to the peak positions at 15 °C. From the lattice parameters calculated in
Table 2, it can be seen that at α = 27 g CO
2/kg sample the transition to hexagonal phase is completed, since the lattice parameters a of α = 27 and 38 g CO
2/kg sample are the same.
At 15 °C, the lattice parameters at α = 27 and α = 38 g CO
2/kg sample (
Table 2) decrease slightly with increasing degree of CO
2-loading, most likely due to the formation of more ordered aggregates at higher CO
2 loading. On the other hand, at 25 °C, the lattice parameters of these two samples are the same, as mentioned above. However, the values at 25 °C are lower than the results obtained at 15 °C, probably due to higher mobility of the molecules. Nevertheless, it should be noted that the microstructures formed when α = 27 and 38 g CO
2/kg sample are less defined than the ones of the unloaded sample, as it can be seen in
Figure 3 and
Figure 4. In general, loading with CO
2 leads to peak-broadening, indicating that the distances between aggregates are less well-defined. In addition, the tails that the CO
2-loaded samples present at low q values of the spectra suggest the presence of larger aggregates that could be of amorphous nature.
The change in the phase behavior of 60% L92/10% MEA is likely due to the increase in polarity of MEA during the loading process as a result of the formation of carbamate and ammonium ions by the reaction between MEA and CO
2 [
25]. This variation in the polarity increases the affinity of water (the most polar component in the system) for MEA, and at the same time decreases its affinity for the apolar polymer [
14,
15]. Increase in the solvent polarity favors the formation of normal structures over the reverse ones [
11,
12]. Therefore, one could infer that the hexagonal phases in the system are normal hexagonal phases, not reverse. In terms of packing parameter, the transition from lamellar to hexagonal phase can be explained by the presence of charged species embedded in the PEO blocks, which would increase the effective head group area, and therefore reduce the packing parameter.
The spectra of 60% L92/10% MEA with a loading of 38 g CO
2/kg sample and the sample loaded in the calorimeter are compared in
Figure 5. Both spectra follow the same pattern, and show the presence of hexagonal phases and larger aggregates. However, the peak positions are slightly shifted, which might be a result of slight differences in the sample composition. Therefore, the lattice parameter of the calorimeter-loaded sample is slightly larger than the value obtained for the sample with loading α = 38 g CO
2/kg sample (
Table 2). In general, when the polymer concentration decreases, the distance between aggregates (lattice parameters) increases [
12].
3.4. Thermal Degradation Experiments
To study thermal degradation of MEA in the system, six parallels consisting of 60% L92/10% MEA with loading 38 g CO
2/kg sample were placed into stainless steel cylinders and stored in a heating cabinet at 80 °C. Three parallels were stored for a week, and the remaining three samples were stored for 7 weeks [
30]. After 1 and 7 weeks, all parallels had separated into two phases, but the weight of the samples remained constant. The two phases of one of the parallels of each set of samples were carefully separated and analyzed by TGA to determine the composition of each phase. The results showed that the upper phases contained approximately 96% of polymer, whereas the lower phases contained below 3% of polymer (
Figure 9). Both phases were isotropic when examined by cross-polarized visual observation [
8].
Amine analysis was used to determine the proportion of amine groups remaining in the samples before and after the thermal degradation experiments. Results are shown in
Table 3. The sample with loading 38 g CO
2/kg sample before degradation, and two samples from the thermal degradation experiments (after 1 week of heating, and after 7 weeks of heating) were titrated. For the samples of the thermal degradation experiments, the top and bottom phases were titrated separately to determine the amine content in each phase. The results show that there are no amine groups in the top phase of any of the two samples, indicating that it mostly consists of polymer, which is in agreement with the TGA results. On the other hand, the bottom phases after 1 and 7 weeks of storage contain 3.6 and 3.8 mol amine/(g CO
2 + g sample), respectively, which is about three times more that the amine content of the un-degraded sample (α = 38 g CO
2/kg sample). The un-degraded sample consisted of one phase and contained polymer, water and MEA. However, after heating the six parallels of this sample to 80 °C, the bottom phases only contain water and MEA, and therefore, the MEA concentration is higher than in the reference sample. If the overall amine concentration of the phase separated samples is estimated by calculating the ratio between the volume of both phases, the amine concentration is approximately the same as in the un-degraded sample. Therefore, the results suggest that after storage at 80 °C for 1 and 7 weeks, there was no actual thermal degradation of MEA. The estimated amine concentration after 7 weeks increases (3%) with respect to the original sample, which is within experimental error.
Aqueous amine solutions are stable at 80 °C, and the results gained in this work indicate the L92 will not accelerate the MEA degradation, at least not at this low temperature [
38]. In addition, the degradation experiments have also shown that the samples lose stability upon heating and phase separate, remaining in a two-phase state after being cooled down to room temperature. In a real process, phase separation would be beneficial because it would allow regeneration of the amine solvent separately from the polymer. Moreover, the formation of two isotropic phases would also be advantageous due to the lower viscosity of isotropic samples compared with liquid crystals.