CO2 and H2O chemisorption mechanism on different potassium-promoted sorbents for SEWGS processes

The sorption kinetics and capacities of CO2 and H2O were investigated for two different potassium-promoted hydrotalcite sorbents and potassium-promoted alumina. Thermogravimetric analysis (TGA) and packed-bed reactor (PBR) breakthrough experiments were performed using sequences of adsorption and desorption steps in different gas mixtures containing CO2 and H2O. Experiments were carried out at an operating temperature of 400 °C with different partial pressures ranging from 0.025 bar to 0.3 bar for CO2 and 0.1 to 0.3 bar for H2O respectively. It was found that a sorption mechanism with different adsorption sites, developed for one of the sorbents, also applies for the other sorbents where capacities are different and depending on the sorbent. From experimental results it was deduced that K2CO3 promotion is mainly responsible for a reactive CO2 adsorption site, which can only be regenerated with steam. The adsorption capacity for this site is enhanced for K2CO3 promoted alumina compared to K2CO3 promoted hydrotalcite. A second adsorption site for CO2, which can be regenerated with N2 is dominant on the K2CO3 promoted hydrotalcite with a high MgO content. This indicates that MgO is probably responsible for the formation of basic sites on the surface of the sorbent, which are relatively easily regenerated at the investigated experimental conditions. The results also show that the sorbent with the highest MgO loading has the highest cyclic working capacity under dry adsorption conditions, whereas the hydrotalcite-based adsorbent with a lower MgO content has the highest cyclic working capacity for CO2 at wet conditions and is therefore the preferred sorbent for sorption-enhanced water-gas shift applications.


Introduction
In the sorption-enhanced water-gas-shift (SEWGS) process, the water-gas-shift reaction is combined with in-situ CO 2 capture on a solid adsorbent. This process has been widely investigated due to its large potential for pre-combustion CO 2 capture. A relatively low energy penalty for CO 2 capture in comparison to alternative processes (e.g. Selexol) makes this technology very promising [1]. Therefore, over the past years much research on various aspects of SEWGS processes has been carried out [1][2][3]. Modeling of the CO 2 and H 2 O adsorption kinetics and determination of the cyclic working capacity of the adsorbent is a very important aspect for the process design (e.g. the number of adsorption columns, cycle design and process integration) and further process optimization [4][5][6]. Different adsorbents have been investigated in the past for their use in SEWGS [7][8][9][10]. Hydrotalcite (HTC) based adsorbents are very promising candidates because of their low cost, sufficiently high CO 2 cyclic working capacity, fast adsorption kinetics and good mechanical stability over many adsorption/desorption cycles and long time periods [11,12]. Potassium-promoted hydrotalcites with different Mg/Al ratios were investigated in the past and their cyclic working capacity was determined [13]. Usually a potassium carbonate promotion of around 20% is used (optimum loading in terms of cyclic working capacity), where the Mg/Al ratio was varied between 0.54 (30 wt% MgO compared to Al 2 O 3 ) and 2.9 (70 wt% MgO compared to Al 2 O 3 ). In a previous work the adsorption capacity and kinetics for CO 2 and H 2 O was studied for two potassium-promoted HTCs with a different Mg/Al ratio and a potassium-promoted alumina. It was found that a higher Mg content can improve the cyclic working capacity for CO 2. A slower continuous adsorption rate after the first initial fast adsorption rate was observed for all adsorbents and this becomes more prominent for a material with a higher Mg content. At a higher operating temperature the desorption rate of CO 2 increases, which increases the cyclic working capacity of HTC-based adsorbents, however, not for a potassium-promoted alumina adsorbent [13]. Various authors have reported that the presence of steam significantly affects the cyclic working capacity for CO 2 [8,9]. Different adsorption sites on a HTC-based adsorbent are most likely responsible for the strong influence of steam on the cyclic working capacity of CO 2 [14].
The existence of one adsorption site for each sorbate species CO 2 (B) and H 2 O (A) was elaborated recently with the help of thermogravimetric analysis (TGA) and packed bed reactor (PBR) breakthrough experiments for a potassium-promoted HTC with a Mg/Al ratio of 0.54. Part of the adsorbed CO 2 can only be regenerated with H 2 O from the adsorbent which was attributed to an exchange site (C). Therefore, regeneration with steam significantly increases the cyclic working capacity for CO 2 . A schematic overview of the different sorption sites can be found in Fig. 1. It was concluded from TGA experiments that the cyclic working capacity for CO 2 increased for all subsequent experiments after the adsorbent was exposed to a stream containing both CO 2 and H 2 O. This increase in adsorption capacity was attributed to another site D. In this paper we will investigate whether the observed behavior needs to be described with an additional adsorption site, or whether it can also be attributed to an increase in the capacity of the three other sites.
The influence of the material composition of the adsorbent on the capacity of the different adsorption sites is important for sorbent selection and depends on the selected process conditions. The investigation of different sorption sites and their behavior as function of material composition is crucial in order to design a good adsorbent or select the sorbent for SEWGS.
In this paper, it is investigated whether the developed adsorption mechanism for HTCs is generally applicable, even for different sorbents such as potassium-promoted alumina to further improve our understanding about the nature of the different adsorption sites. These results can significantly improve future sorbent development for different sorption processes where adsorption of CO 2 can enhance the performance of the process. Experiments with both TGA and PBR were carried out for a potassium-promoted HTC with a Mg/Al ratio of 2.95 and a potassium-promoted alumina.

Materials and methods
A potassium-promoted hydrotalcite-based adsorbent, with a Mg/Al ratio of 2.95 (KMG70) and a potassium-promoted alumina (KSORB) were used in this work and compared to a potassium-promoted hydrotalcite-based adsorbent with a Mg/Al ratio of 0.54 (KMG30). The compositions are summarized in Table 1. The commercial materials were pre-calcined by the manufacturer at 250°C and 450°C for 24 h [8].
KMG30 and KMG70 adsorbent pellets (4.7 × 4.7 mm) and KSORB (spheres with a diameter of 2 mm) were crushed to powders with the same particle size, which have been used in TGA-measurements. Sorption experiments were carried out in an in-house developed TGA setup which was already described elsewhere [14,15]. Prior to the adsorption experiments all materials were pretreated in the same way (heating the samples to 600°C in N 2 with a dwelling at 600°C for 120 min). For TGA measurements the cyclic working capacity (cwc) is defined according to Eq. (1), where the average between adsorption and desorption capacity with respect to the pretreated sample mass is calculated for each experiment.
This cwc was determined for each adsorption/desorption step using the mass change averaged over the last three consecutive experiments. Note that we use the mass-based cyclic working capacity, as the TGA does not provide additional information on the species absorbed/desorbed. For all experiments the measured weight change was corrected with a blank experiment carried out at the same condition without the sorbent sample. Experiments were carried out two times starting one time with H 2 O as a first adsorption step after the pretreatment and a second time starting with CO 2 . Details on the experimental setups and the procedures used in this work can be found in an earlier published work [14]. Packed bed reactor experiments were carried out using a small packed bed reactor with an inner diameter of 27 mm and 350 mm length (AISI 316 L). A detailed description on the experimental Setup was described elsewhere previously [15,16]. The reactor was filled with different sorbent particles with characteristics reported in Table 2.
A multipoint thermocouple (10 measuring points with a distance to each other of 20 mm) was installed to measure the axial temperature profile in the bed and to monitor the temperature fronts due to sorption.
Different molar fractions were used during the PBR and TGA experiments as summarized in Table 3. Changing the gas composition was performed by bypassing the reactor for 5 min and measuring the concentration (to check the feed gas composition). After this stabilization time, the feed was sent to the reactor from bottom to top while monitoring the outlet composition of the reactor for 30 min. Experiments were conducted at atmospheric pressure and different operating temperatures between 300 and 500°C. The mixing in the empty volume of the reactor and the tubing was determined using an inert (CH 4 in this case). The breakthrough of the Intert was used to determine the contribution of gaseous species to the breakthrough time for each material independently because different porosity and bed height will have an impact on the inert breakthrough time. The determined adsorption capacity during each adsorption/desorption step was corrected with the determined contribution of gas phase. An overview of all experiments including different partial pressures of sorbate species is listed in the Table 4.
For each experiment listed in Table 4 a sequence of adsorption and desorption steps have been performed as reported in Table 5. Cycles have been repeated 5 times for the TGA experiments. The cyclic working capacity for the TGA experiments was calculated based on the last three adsorption/desorption cycles. Every cycle for the PBR experiments was carried out twice. We will refer to the different experiments performed with 'n.m', where the first number refers to the  Table 4 (1-27) and the second number to the sub-experiment as listed in Table 5 (1-8).

Results and discussion
In the results and discussion section, we will show the presence of the earlier introduced different sorption sites based on TGA and PBR experiments. The different adsorption sites will be introduced consecutively, and their cyclic working capacity will be determined for the different sorbent materials. The importance of the different sorption sites for the different materials will be discussed with a simple model to predict the weight change observed in the TGA experiments. The total CO 2 and H 2 O cyclic working capacities will be determined for all sorbents at different operating conditions to find the best operating conditions.

TGA data
Arrows are used in the figures that show the normalized weight change as a function of time to indicate which sites are involved in the adsorption/desorption steps, where the mass at the start and end of each step is indicated by a dotted line. Red arrows represent the mass of CO 2 exchanged in one experimental step, whereas blue arrows represent the mass of H 2 O exchange. The length of the arrow represents the total mass change obtained in the experimental step and the contributions for the different sites are indicated by their letter. An increase in the adsorbate content is represented by a solid arrow, a decrease by a shaded arrow. An example for a representation of three different sites involved in an adsorption step can be found in Fig. 2.

PBR data
The areas in the figures (integration of analyzer signal over the time with respect to the baseline) are colored in the same way as for the TGA results (see Fig. 2). The signal for CO 2 and the corresponding areas are plotted in red, whereas the signal and areas for H 2 O are plotted in blue. Again, we distinguish between adsorption (solid area above the signal) and desorption (shaded area below the analyzer signal). Note that the integrated areas were corrected afterwards for gas phase contribution during the experiments. The graphs show the uncorrected analyzer signals.

Sorption of H 2 O (site A)
The three different sorbents were exposed in the TGA to a H 2 O/N 2 mixture with a partial pressure of 0.34 bar H 2 O (during the first part of experiment 6). The cyclic working capacity for H 2 O (site A) follows the following order: KMG30 > KMG70 > KSORB (see Fig. 3). The same trend was already reported in an earlier publication using a different experimental setup [13]. The absolute value of the cwc of site A for KMG30 is somewhat lower than reported before [16], most probably due to some aging of the material during these tests.
The packed bed reactor experiments confirm the higher cyclic working capacity of H 2 O for KMG70 compared to KSORB. Both the surface basicity and surface area affect the adsorption capacity of the sorbents, explaining the obtained results since a direct correlation between surface area and adsorption capacity of the materials for steam could not be found. Surface area of the different sorbents has been already published earlier [13]. The breakthrough curves for KSORB can be found in the Appendix. Note that the longer breakthrough times for KSORB are a result of the higher sorbent mass used in the experiments. Since it was the first time that the adsorbent was exposed to a sorbate not only adsorption site A is filled with H 2 O, but also site C (Fig. 3b, first adsorption), explaining the larger amount of H 2 O adsorbed in the first cycle (larger blue area in Fig. 3b). This was also observed in the TGA experiments, but in Fig. 3a only the last cycle is plotted to focus on the cyclic behavior of the different materials.

Sorption of CO 2 (site B)
The cyclic working capacity for CO 2 of site B has been determined based on results obtained by TGA during experiment 2 (see Fig. 4a),    Table 4 Experiments conducted for different sorbent materials in the TGA and PBR setups. For the experiments marked with * only cyclic sequences 2-5 were carried out (see also which are also confirmed with the results from the breakthrough curves in the packed bed reactor experiments (Fig. 4b).
The results show that the order in the cyclic working capacity for CO 2 is changed compared to H 2 O. The sorbent with the highest Mg content (KMG70) has the highest cyclic working capacity for site B. This trend was also found in a previous study and was explained with the higher surface basicity of the sorbent when a larger amount of MgO is present in the sorbent structure. It seems that MgO induces a second reaction, which could be explained e.g. by diffusion of carbonate species into the bulk of the material which is enhanced for a more basic sorbent. This explanation is supported by the observed difference in adsorption profile for KSORB (without MgO), where an adsorption equilibrium is established after 30 min half-cycle time, while this is not the case for KGM30 and KMG70. The findings reported here confirm earlier measurements [13].

Replacement of CO 2 by H 2 O (site C)
To determine the cyclic working capacity of the adsorption site which can be either occupied by H 2 O or CO 2 (site C) experiment 3 was carried out (feeding CO 2 and H 2 O consecutively). Fig. 5a (TGA) and b (PBR) show the results obtained for KSORB and Fig. 5c and d show the results for KMG70. A significant difference in the cyclic working capacity between the different sorbents was observed. The cyclic working capacity for KSORB (11.3 mg/g) is slightly smaller than for KMG30 (12.6 mg/g) but still comparable, whereas KMG70 shows a very low cyclic working capacity of 5.9 mg/g. When comparing the results obtained during the breakthrough experiments the ratio between the area of site B and site C should be compared rather than the breakthrough time or the total area due to the difference in sorbent mass in the experiments. Based on the PBR breakthrough results we can conclude that MgO cannot be responsible for this adsorption site, because the sorbent with the highest MgO content shows the lowest capacity for site C. The impregnation of the sorbents with K 2 CO 3 could be the reason for the higher capacity of this adsorption site. It has been proposed in the literature that indeed the interaction of K 2 CO 3 with Al is a crucial for the availability of adsorption sites on a potassium-promoted hydrotalcite [17]. These findings explain the rather high adsorption capacity of this site for KSORB compared to KGM70. Since KMG30 has the highest surface area [13] K 2 CO 3 can interact to a larger extent with the available Al on the surface resulting in the highest cyclic working capacity of site C for KMG30.
Based on the packed bed reactor experiments, the replacement ratio between CO 2 and H 2 O was determined. Note that this information cannot be obtained from the TGA experiments. The average cyclic working capacity for CO 2 of site C was determined based on the desorbed amount of CO 2 during step d of an experiment n.3 (were n represents the experiment number from Tables 4 and 3 refers to step 3 as listed in Table 5) and the adsorbed amount of CO 2 during step a of experiment n.3 minus the cyclic working capacity of site B determined based on step c of each experiment according to Eq. (2): (3)

Simultaneous feed of CO 2 and H 2 O
Experiment 4 was used to determine the cyclic working capacity of site C at equilibrium conditions when feeding both CO 2 and H 2 O simultaneously, indicated with C eq . The TGA and PBR results are shown in Fig. 6. Fig. 6a and c show that a higher cyclic working capacity for the site C eq is found for KSORB compared to KMG70. This is expected because the cyclic working capacity of site C eq should be directly correlated to the cyclic working capacity of site C. For KMG30 the highest cyclic working capacity for site C and C eq was found.

Adsorption capacity change for CO 2 and H 2 O (site D)
In an earlier work [16] a site D was introduced to describe the higher adsorption capacity of all sorbents when CO 2 and H 2 O are fed together during the first step of experiment 4. The increase in cyclic working capacity is more pronounced for KMG70 (6.3 mg/g) compared to KSORB (3.3 mg/g). However, it could also be considered that the increase in working capacity is mainly related to an increase in the cyclic working capacity of site B, which would also explain the higher capacity of this adsorption site for KMG70.
The results from experiment 5 can be described with the 4 sites mechanism introduced until this point. Fig. 7a and b show the results of KSORB for both TGA and PBR experiments, and similarly Fig. 7c and d show the results for KMG70. When comparing the ratios of the areas of site B and C in the PBR breakthrough experiments, it can be concluded that indeed the replacement effect (cyclic working capacity of site C) is  much smaller for KMG70 compared to KSORB. During step 2, the replacement of CO 2 is only barely visible for KMG70. Moreover, it was found that the sorbent was still adsorbing CO 2 , which can be explained with the slower adsorption kinetics of KMG70, which were shown before (Experiment 2).
For KSORB it was observed during experiment 5 that not all CO 2 can be desorbed from site B. This observation was obtained from both TGA and PBR experiments. If CO 2 would desorb completely from site B and H 2 O from site A during step three one would expect a weight decrease of about 16.5 mg/g. However, the obtained weight decrease is much lower (about 10 mg/g). PBR experiments confirm that indeed the amount of CO 2 desorbed during this step is smaller than during experiment 4. In the following step (step 4) some additional CO 2 is desorbed compared to the previous experiments. In Fig. 8 the area of the breakthrough curve for CO 2 is plotted for experiments 4, 5 and 6. It is evident, that the amount of CO 2 desorbed in experiments 5 and 6 is significantly lower compared to experiments 3 and 4, confirming the introduced hypothesis.
Also, the results obtained from experiment 6 confirm our proposed mechanism. The results from both the TGA and PBR experiments for both KGM70 and KSORB can be described. An important observation from the TGA experiments is the observed capacity change of adsorption site D. When comparing the total observed weight change during step 1 of experiment 5 (Fig. 8) and experiment 6 ( Fig. 9), we observed a lower weight change during experiment 6. This observation seems to be valid for all sorbents, but it is most apparent for KSORB. Deactivation of this adsorption site can be explained with the fact that the measured adsorption capacity is always dependent on the prior desorption step [14]. Therefore, if the desorption time is not sufficient to desorb the amount of CO 2 on site D, the capacity of this site will decrease. This would imply that the pseudo steady state between adsorption and desorption on site D was not yet reached or is not occurring at these operating conditions. To describe the cyclic working capacity during all experiments this adsorption site should deactivate with experimental time if the capacity of the other adsorption site is assumed constant.
During the experiment 7 (Fig. 10), the capacity of site D for KSORB is reduced to already 20% of its original value, indicating that this site  probably will not be needed to describe the steady state operation of a SEWGS process. For KMG70, the TGA experiments can be described more accurately using 70% of site D still participating in the adsorption and desorption steps. It appears that the deactivation of site D depends on the type of sorbent. It was found that the reported mechanism of KMG30 could also be improved by a slow deactivation of site D. Table 7 shows the model used for the different sorbents during the 8 different experiments and the different adsorption/desorption steps. It can be seen, that indeed the strongest deactivation of site D seems to be present for the potassium-promoted alumina (KSORB), where a higher MgO content results in a slower deactivation rate of this adsorption site. Most probably the slow continuous adsorption of CO 2 which was reported earlier for HTC based sorbents with a higher MgO content, can balance the effect of the deactivation of site D for KSORB. Therefore, the deactivation of site D is less important for sorbents with a higher MgO content. Still one can expect for experiments with more than 1000 cycles that probably for all sorbent materials this site will be completely deactivated and is therefore not needed to describe the cyclic working capacity of the sorbent. The experimentally determined cyclic weight change in the TGA experiments together with the difference with the predicted weight change by the developed mechanistic model can be found in Table 8 in the Appendix. The determined cyclic working capacities using the model according to Table 7 are plotted in Fig. 11a. The error bars represent the standard deviation between two different experimental sets which were carried out for all sorbents independently. The determination of the adsorption sites using the proposed model is quite accurate. Although a direct comparison between TGA and PBR results is not possible due to the difference in experimental conditions (partial pressure of the sorbate species, transient behavior of the PBR) the trend in adsorption capacities of the different adsorption sites can be compared. Error bars in the PBR experiment plot represent the standard deviation for determination of cyclic working capacities based on the adsorbed and desorbed amount within one cycle. Because adsorption and desorption of site C and site C eq goes along with adsorption and desorption of site A and B it's determination is more difficult. The adsorption capacities of the adsorption sites A, B and C for the different materials show indeed the same trend in terms of cyclic working capacity. These results support the identification of different adsorption sites on weight bases measured with TGA experiments. To our best knowledge this is the first time that different adsorption sites on various HTC based adsorbents has been identified using a direct combination of both measurement techniques. Site D has not been determined using packed bed reactor experiments since it is expected that the determination of this site cannot be done in an accurate way due to the axial concentration profiles associated with the transient behavior of the packed bed.   Step 3) and Experiment 6 (Step 2). Amount of CO 2 desorbed is significantly reduced in Experiments 5 and 6 compared to Experiment 4.

Importance of adsorption site D
To investigate the behavior of the adsorption site D compared to the other adsorption sites a TGA experiment with identical conditions (CO 2 and H 2 O partial pressure at 400°C) was performed with a longer step duration of 60 min instead of 30 min. It can be seen from the results shown in Fig. 12 that the adsorption capacity is increasing for all adsorption sites except for site D if the step time is increased compared to the experiment with a shorter step duration. It is known that a longer experimental time would lead to a lower loading of CO 2 and H 2 O on the sorbent, leading to an increase in measured cyclic working capacity. In a previous work it was reported that slower desorption kinetics are the reason for this change in capacity [14]. The fact, that the determined cyclic working capacity for site D is decreasing support our hypothesis that site D is a combination of a capacity increase of the adsorption sites B and C and mainly caused by the slow desorption rate. By using a longer step time, more CO 2 and H 2 O can be desorbed from the site A -C making the correction with site D less important. This observation, together with the deactivation of this site D needed to accurately describe the TGA experiments, illustrates that indeed site D is not required to describe long term experiments and steady state behavior of the different adsorbents.

Cyclic working capacities of different sorbents
For the application of the different adsorbents in a SEWGS process, the total cyclic working capacities of gaseous sorbate species at operating conditions play a key role in the selection of the best sorbent. In Fig. 13 one can find a summary of the cyclic working capacity of CO 2 and H 2 O for the all sorbents determined by TGA (Fig. 13a) and PBR (Fig. 13b) experiments. Depending on the process conditions, the total cyclic working capacity of CO 2 and H 2 O will change mainly because of the adsorption site C which is dependent on feed partial pressures of the sorbates in the gas phase. KMG30 has the highest cyclic working capacity for both CO 2 and H 2 O for a dry gas feed of CO 2 (Total CO 2 ) and for a wet gas feed (Total CO 2 capacity at a feed of P CO2 = 0.66 bar and P H2O = 0.34 bar). The high cyclic working capacity of the adsorption site C has also a major influence on the total cyclic working capacity for H 2 O on KMG30. Depending on the process conditions a lower adsorption capacity for steam can be beneficial in terms of energy efficiency. Reducing the required amount of steam is usually a crucial factor in order to reduce energy costs of a chemical process [18]. KMG70 has a high cyclic working capacity for CO 2 with a much lower adsorption capacity for H 2 O. PBR reactor experiments confirm the trend observed in the TGA experiments. However, because of the lower partial pressures of the feed gas the reported cyclic working capacities are much lower compared to the ones determined from the TGA experiments. Also the transient behavior of the packed bed reactor reveals an incomplete regeneration of the sorbent compared to the experiments performed in the TGA. Sorbate species desorbing from the sorbent material can re-adsorb on the other sorbent particles along the reactor during the regeneration step hindering a complete desorption. It can be seen in Fig. 13b that, under certain process conditions, for KMG70 (see Total CO 2 at EQ), the cyclic working capacity for CO 2 can be even higher than for the other sorbent materials. Considering the generally higher cyclic working capacity of this sorbent at dry regeneration conditions (only site B active), KGM70 can be an interesting material if the process conditions do not allow regeneration with steam. However, it has to be proven that this sorbent is mechanically stable throughout cyclic operation in long term tests under dry adsorption conditions, since it was found that high partial pressures of steam and CO 2 result in the formation of MgCO 3 leading to a loss in mechanical stability [11].

Influence of operating conditions on cyclic working capacity
To achieve a high cyclic working capacity, the influence of operating conditions (variations in operating temperature and operating partial pressure) were investigated with PBR experiments. An operating temperature higher than 400°C was not considered for KSORB because it was found that the presence of steam at high operating temperature leads to an irreversible decrease in the CO 2 cyclic working capacity due to decomposition of carbonate species [13]. It can be discerned from Fig. 14a and b that a decrease in operating temperature leads to an increase in the cyclic working capacity for H 2 O on site A and a decrease for CO 2 on site B. This is in agreement with earlier studies for KMG30, where the same trend was observed [16]. Indeed, it seems that this behavior is in general valid for potassium-promoted HTC's and potassium-promoted Al 2 O 3 . For KMG70, site C (for CO 2 ) shows the highest cyclic working capacity at 300°C, which is probably due to the higher amount of CO 2 being replaced by H 2 O. H 2 O in general shows a higher affinity to adsorb at lower operating temperatures. KSORB exhibited an opposite behavior. This can be explained by the increased desorption kinetics at higher operating temperatures. Since the cyclic working capacity of site C is much higher for KSORB compared to KMG70 and the total mass of sorbent was higher during the experiments the desorption rate becomes more important for this site than the replacement.
Experiments at different partial pressures for CO 2 and H 2 O confirm the opposite behavior of the two sorbents. In Fig. 14(c and d) the cyclic working capacities of the different adsorption sites are plotted for different partial pressures. A higher CO 2 partial pressure consistently leads to an increase in the cyclic working capacity for site B, where for site C the cyclic working capacity is decreased for KMG70, but increased for KSORB. Note that for KMG70 the small amount of CO 2 and H 2 O being exchanged by site C result in a small area for integration which makes the determination of the cyclic working capacity of this adsorption site less accurate (Fig. 15).

Conclusions
A detailed experimental study using TGA and PBR experiments with three different adsorbents for the application in a SEWGS process, showed that the mechanism for CO 2 and H 2 O adsorption on a potassium-promoted hydrotalcite reported in our earlier publication, is also valid for similar sorbents with a different material composition. Depending on the material composition, the cyclic working capacity of certain adsorption sites for CO 2 and H 2 O are changing. It was found that an increase in MgO content in the sorbent leads to an increase in the cyclic working capacity for CO 2 (called site B) for an adsorption site, which can be relatively easily regenerated with e.g. N 2 due to the increase of basic adsorption sites on the surface of the sorbent materials. The cyclic working capacity for CO 2 on a reactive adsorption site (called site C), which can only be regenerated using H 2 O is highest for potassium-promoted alumina and a HTC-based adsorbent with a low MgO content. This indicates that mainly K 2 CO 3 interactions with Alcenters would be responsible for this reactive adsorption site. The cyclic working capacity on an adsorption site for H 2 O (called site A) is slightly higher for potassium-promoted HTC compared to the potassium-promoted alumina. The activation of an additional adsorption site for CO 2 Table 7 Model description for the two sorbent materials used for the weight changes in Table 3.    if the sorbent is exposed to CO 2 and H 2 O together during one step (called site D) was confirmed for all sorbents. It was found that the additional capacity of this site is decreasing with time on stream. This can be described with a deactivation of this adsorption site. Therefore, in order to describe the steady state adsorption behavior this adsorption site does not need to be included. It was proven with PBR breakthrough experiments comparing the cyclic working capacity of the proposed adsorption sites, that indeed the proposed mechanism can describe the experimental results of both measurement techniques. For SEWGS KGM30 shows the highest cyclic working capacity for CO 2 and is therefore the preferred sorbent for the process. For adsorption process where dry regeneration of the sorbent is used, KGM70 could be a very promising alternative, if its mechanical stability is demonstrated during long-term cyclic experiments.

Acknowledgement
The research leading to these results has received support through the ADEM innovation lab program, project number TUE-P05.

Appendix A
This appendix contains the results from breakthrough experiments for H 2 O and CO 2 adsorption performed in a packed bed reactor filled with HTC based materials. Additionally, the weight change during different experiments are given together with the deviation with predictions by a developed model using different adsorption sites. The total deviation in mg/g predicted by the model for each experiment is provided in Table 8 Fig . 15. a) EXP1 KMG70 at 400°C with PH 2 O = 0.1 bar b) EXP2 KMG70 at 400°C with PCO 2 = 0.05 bar. Table 8 Summary of the weight change in mg/g observed during TGA experiments compared to predicted weight change by the model for KMG70 and KSORB. The values of the model represent the total deviation between the predicted weight change and weight change measured during the experiments.