How Does K₂CO₃ Promote the CO₂ Uptake of MgO?

Abstract

Alkaline earth metal oxides are economically attractive, earth-abundant sorbents for CO₂ capture. In particular, MgO is a promising CO₂ sorbent for applications in the intermediate temperature range (200–400 °C). However, MgO requires the addition of promoters such as alkali metal nitrates or carbonates to overcome the slow kinetics of CO₂ sorption. It has been observed experimentally that the addition of K₂CO₃ increases the CO₂ uptake of MgO, yet its role remains poorly understood; this is a critical knowledge gap for the design of more efficient sorbents. In this work, using a combination of in situ XRD and Raman spectroscopy, we gain insight into the CO₂ uptake mechanism of K₂CO₃-promoted MgO, which proceeds in two steps. An initial rapid CO₂ uptake (within 1 min) occurs via the formation of a potassium-rich amorphous carbonate phase (K₂CO₃·xMgCO₃·yH₂O). This is followed by a slower CO₂ capture step, in which the intermediate K-rich phase transforms into a magnesium-rich amorphous carbonate, along with the formation of small amounts of crystalline (hydrated and anhydrous) carbonates such as K₂CO₃, baylissite (K₂Mg­(CO₃)₂·4H₂O), and nesquehonite (MgCO₃·3H₂O). In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) analysis reveals that surface hydroxides, which remain on the MgO surface even after a thermal treatment at 720 °C in N₂, facilitate the formation of hydrated carbonates at 315 °C in dry CO₂. However, the presence of steam during carbonation lowers the CO₂ uptake and inhibits the second CO₂ uptake step, most likely by stabilizing the K-rich amorphous intermediate. Cyclic CO₂ uptake and regeneration experiments reveal the crucial role of a high K₂CO₃ dispersion within the sorbent for maintaining good cyclic stability.

Introduction

The challenge of reducing atmospheric CO₂ emissions to limit the extent of global warming has driven significant interest in developing efficient and cost-effective materials for CO₂ capture from large anthropogenic point sources. − In this regard, alkaline earth metal oxides such as MgO are being studied as promising alternatives to the benchmark aqueous amine technology. − This class of sorbents has the potential to be less energy-intensive and avoid the release of toxic gases arising from the oxidative degradation of amines. − MgO-based CO₂ sorbents have emerged as promising candidates for industrial-scale CO₂ capture due to their high theoretical CO₂ uptake capacity of 1.09 g_CO2 g_MgO ^–1, low cost, and ability to capture and release CO₂ in the intermediate temperature range (200–450 °C). This temperature range is well-suited for precombustion CO₂ capture processes. ,, The associated cyclic CO₂ capture process of MgO-based CO₂ sorbents is based on the reversible reaction of MgO with CO₂:

$$\mathrm{MgO}(\mathrm{s})+{\mathrm{CO}}_2(\mathrm{g})\rightleftharpoons {\mathrm{MgCO}}_3(\mathrm{s}){\mathrm{\Delta }\mathrm{H}}_{298\mathrm{K}}=-116.9{\mathrm{kJ}\mathrm{mol}}^{-1}$$ 1

However, bare MgO exhibits slow CO₂ uptake kinetics, resulting in CO₂ uptake capacities in the range of 0.02–0.04 g_CO2 g_MgO ^–1, well below the theoretical maximum. , This limitation has driven research into various strategies to enhance the CO₂ uptake performance of MgO, including the use of promoters. ,,

Alkali metal salts, including nitrates (e.g., LiNO₃, NaNO₃ and KNO₃) and carbonates (e.g., Na₂CO₃ and K₂CO₃), are among the most effective promoters for MgO-based CO₂ sorbents. ,,− For example, MgO promoted with a mixture of (Li, K)­NO₃ and (Na, K)₂CO₃, achieves a CO₂ uptake of 0.84 g g_sorbent ^–1 (corresponding to a MgO conversion of 95%) after 4 h of carbonation at 325 °C. Despite their efficacy, the mechanisms by which alkali metal salts enhance the CO₂ uptake of MgO-based sorbents remain poorly understood and most of the recent mechanistic studies have focused on NaNO₃, which is molten under typical carbonation conditions. In contrast, relatively few studies have investigated the mechanisms by which alkali metal carbonates promote the CO₂ uptake of MgO, and most of these works have focused on mixtures of alkali metal nitrates and carbonates, such as NaNO₃–Na₂CO₃. , In the presence of a molten alkali metal nitrate, carbonate promoters are generally understood to act as nucleation seeds for MgCO₃ formation. ,,− Moreover, many carbonates such as Na₂CO₃ and CaCO₃ readily react with MgO and CO₂ to form double carbonates. Compared to alkali metal nitrates, alkali metal carbonates are thermally more stable, less corrosive, and remain in the solid state under typical operating conditions, which makes them more suitable for usage in large-scale reactors such as packed or fluidized beds. ,

Among the alkali metal carbonates studied in the absence of nitrates, the promotion of MgO with K₂CO₃ has yielded the highest CO₂ uptakes. However, there is a very limited body of work on K₂CO₃-promoted MgO for CO₂ capture, and such studies have been conducted under diverse operating conditions, including differences in temperature, gas atmosphere and the amount of K₂CO₃ used for promotion (Table ). − Overall, the CO₂ uptake capacities of K₂CO₃-promoted MgO are in the range of 0.09–0.18 g g_sorbent ^–1,, well above the typical values of unpromoted MgO (0.02–0.04 g g_MgO ^–1). The increase in the CO₂ uptake of MgO upon its promotion with K₂CO₃ is typically explained by the reaction of K₂CO₃ and MgO with CO₂, leading to the formation of a double carbonate: ,−

$${\mathrm{K}}_2{\mathrm{CO}}_3(\mathrm{s})+\mathrm{MgO}(\mathrm{s})+{\mathrm{CO}}_2(\mathrm{g})\rightleftharpoons {\mathrm{K}}_2\mathrm{Mg}({\mathrm{CO}}_3)(\mathrm{s}){\mathrm{\Delta }\mathrm{H}}_{298\mathrm{K}}=-113.01{\mathrm{kJ}\mathrm{mol}}^{-1}$$ 2

1. Overview of K₂CO₃-Promoted MgO-Based CO₂ Sorbents Reported in the Literature.

Material	CO2 uptake (gCO2/gsorbent)	Operating conditions (all experiments performed at 1 bar)	Reference
MgO-30mol% K2CO3	0.033	10 vol% H2O; 10 vol% CO2 at 200 °C (carbonation time not specified)
MgO-95mol% K2CO3	0.085	1 bar CO2; 350 °C for 2 h
MgO-15mol% K2CO3	0.119	1% CO2, 9% H2O (1 bar) and 60 °C for 2 h	,
MgO-15mol% K2CO3	0.179	1% CO2, 11% H2O and 50 °C for 3 h
MgO-10mol% K2CO3	0.144	1 bar CO2 at 330 °C for 6 h

However, the CO₂ uptakes observed for K₂CO₃-promoted MgO (Table ) exceed the expected uptake according to eq both at low and at higher temperatures, indicating that the CO₂ uptake by MgO itself was promoted in the presence of K₂CO₃ (or the double carbonate formed). − Yet, as no detailed characterization of the final product has been reported thus far it is unclear how the (additional) CO₂ is captured by MgO, e.g. due to the formation of surface carbonates as in the absence of K₂CO₃, or due to bulk MgCO₃ formation, promoted by K₂CO₃ (or K₂Mg­(CO₃)₂). Furthermore, the reported CO₂ uptake of K₂CO₃-promoted MgO depends strongly on the amount of K₂CO₃ added. Indeed, the CO₂ uptake of a nearly stoichiometric MgO-K₂CO₃ sorbent (0.085 g g_sorbent ^–1 in CO₂ at 350 °C) is lower than that of a sorbent containing only 10 mol % K₂CO₃ (0.144 g g_sorbent ^–1 in CO₂ at 330 °C). The lower CO₂ uptake in the presence of a higher loading of K₂CO₃ could be due to a smaller interfacial area between MgO and K₂CO₃, or a reduction in the surface area of the sorbent when a large amount of K₂CO₃ is added. , It is also worth noting that the presence of water vapor strongly affects the CO₂ sorption capacity of K₂CO₃-promoted MgO. At low temperatures (<140 °C), water vapor was found to be essential for CO₂ uptake to occur, whereas at higher temperatures (200–320 °C), steam partially inhibits CO₂ uptake.

Given the lack of systematic studies on the effect of K₂CO₃ on the CO₂ uptake performance of MgO, this work aims to elucidate how the addition of K₂CO₃ promotes the CO₂ uptake of MgO in the intermediate temperature range. To this end, we investigated the evolution of both amorphous and crystalline carbonates during the carbonation of K₂CO₃-promoted MgO at 315 °C in CO₂ using a combination of in situ X-ray diffraction (XRD), in situ Raman spectroscopy and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). We observed an initial, rapid formation (<1 min) of a K-rich amorphous carbonate (K₂CO₃·xMgCO₃·yH₂O) through the reaction of K₂CO₃ with MgO and CO₂. After approximately 20 min of carbonation, the K-rich amorphous carbonate further reacted with MgO and CO₂, resulting in the formation of a Mg-rich amorphous carbonate, explaining the capture of additional CO₂. Simultaneously, small amounts of baylissite (K₂Mg­(CO₃)₂·4H₂O), nesquehonite (MgCO₃·3H₂O) and K₂CO₃ slowly crystallized. Furthermore, in situ DRIFTS revealed that the water involved in the formation of the hydrated carbonates originates from H-bonded OH-groups, which remain on the surface of the sorbent even after thermal treatment at 720 °C and recombine to water during carbonation. Interestingly, the addition of steam during carbonation inhibits the second CO₂ capture stage, likely by stabilizing the amorphous K-rich intermediate. Finally, we assessed the potential of the sorbent for application in a cyclic CO₂ uptake process.

Results and Discussion

CO₂ Uptake Performance of K₂CO₃-Promoted MgO Sorbents

A series of K₂CO₃-promoted MgO sorbents was prepared by ball-milling MgO with varying amounts of K₂CO₃ (0, 2.5, 5, and 7.5 mol %). The sorbents are referred to as MgO-xK, where x denotes the nominal K₂CO₃ loading. Inductively coupled plasma optical emission spectroscopy (ICP-OES) and SEM-EDX analyses (Table S1) confirmed that the molar ratios of Mg:K in the sorbents were consistent with the targeted values. XRD analyses of the as-prepared sorbents (Figure S1) show only Bragg peaks due to MgO for sorbents containing 0, 2.5, and 5 mol % K₂CO₃, while also peaks due to potassium carbonate sesquihydrate (K₂CO₃·1.5H₂O) were observed in MgO-7.5K. The absence of any XRD peaks corresponding to potassium-containing phases and the presence of a diffuse scattering halo in MgO-2.5K and MgO-5K indicates that potassium was present in an amorphous form. To identify the nature of the amorphous K-containing phase(s), the as-prepared sorbents were interrogated by Raman spectroscopy (Figure S2). In all of the as-prepared K₂CO₃-promoted sorbents (MgO-2.5K, MgO-5K, and MgO-7.5K), a broad peak at 1082 cm^–1 was observed, a peak that is assigned to baylissite, i.e. a hydrated double carbonate of magnesium and potassium (K₂Mg­(CO₃)₂·4H₂O, see Figure S3). Baylissite readily formed during exposure of the samples to air via the reaction of K₂CO₃-promoted MgO with water and CO₂ from the atmosphere. , Furthermore, a relatively sharp peak at 1061 cm^–1, corresponding to the characteristic ν₁ peak of K₂CO₃, was present in MgO-5K and MgO-7.5K. TEM-EDX (Figure a-c) and SEM-EDX (Figure S5) mapping of the as-prepared sorbents revealed that potassium was well-distributed in MgO-2.5K and MgO-5K, although potassium was found to be partially agglomerated in MgO-7.5K. N₂ physisorption experiments (Figure S6) indicate that the surface area of the different sorbents decreased with increasing K₂CO₃ content, resulting in surface areas of 129, 52, 25, and 21 m² g^–1 for MgO-0K, MgO-2.5K, MgO-5 and MgO-7.5K, respectively. The reduction in surface area in the presence of K₂CO₃ can be attributed to K₂CO₃ covering the MgO surface, thereby blocking a fraction of the pores of MgO.

1.

STEM-EDX maps of as-synthesized (a) MgO-2.5K, (b) MgO-5K, and (c) MgO-7.5K. (d) CO₂ uptake of MgO, MgO-2.5K, MgO-5K, and MgO-7.5K during carbonation at 315 °C in CO₂ in a TGA with (e) a magnification of the red dashed area in (d). (f) Stacked XRD patterns of MgO, MgO-2.5K, MgO-5K, and MgO-7.5K obtained after 10 h of carbonation at 315 °C in CO₂ in overlay with the reference patterns of nesquehonite (ICSD-91710), baylissite (ICSD-200006), β-K₂CO₃ (ICSD-662), γ-K₂CO₃ (ICSD-10191), K₂Mg­(CO₃)₂ (ICSD-31295), and MgO (ICSD-52026). (g) Enlarged view of the red dashed area in (f). (h) Visualization of the unit cells of γ-K₂CO₃, β-K₂CO₃, MgCO₃·3H₂O (nesquehonite), K₂Mg­(CO₃)₂·4H₂O (baylissite), and K₂Mg­(CO₃)₂.

The CO₂ uptake of the K₂CO₃-promoted MgO-based sorbents at 315 °C in CO₂ was determined using a thermogravimetric analyzer (TGA) (Figure d,e). Prior to carbonation, the sorbents were pretreated at 450 °C in N₂ for 30 min to remove adsorbed CO₂ and H₂O. In the absence of K₂CO₃, the CO₂ uptake of MgO was very low, viz. 0.013 g_CO2 g_sorbent ^–1 (weight increase by 1.3 wt.%) after 10 h of carbonation (Figure d). The presence of K₂CO₃ enhanced substantially the CO₂ uptake, yielding CO₂ uptakes of 4.4 wt % (0.044 g_CO2 g_sorbent ^–1) for MgO-2.5K, 11.4 wt % (0.114 g_CO2 g_sorbent ^–1) for MgO-5K and 10.2 wt % (0.102 g_CO2 g_sorbent ^–1) for MgO-7.5K. All K₂CO₃-promoted sorbents showed a fast CO₂ uptake at the onset of the carbonation reaction with similar CO₂ uptake kinetics (Figure e). Previous studies have attributed this very rapid, initial CO₂ uptake to the formation of K₂Mg­(CO₃)₂ according to eq . ,,, The amount of CO₂ captured in this first, rapid CO₂ uptake step (reaction stage I) correlated indeed with the K₂CO₃ content in the sorbent (when also considering that a small amount of CO₂ is captured in the form of surface carbonates on MgO), see Table S2.

Assuming that CO₂ is captured through the formation of the double carbonate (K₂Mg­(CO₃)₂) during reaction stage I, the results in Table S2 show that MgO-5K reached full conversion of K₂CO₃. In contrast, a fraction of K₂CO₃ remained unreacted in MgO-7.5K (full conversion would yield an uptake of 0.069 g_CO2 g_sorbent ^–1, but only 0.057 g_CO2 g_sorbent ^–1 was observed). This difference is most likely due to a poorer dispersion of K₂CO₃ in MgO-7.5K. Interestingly, in MgO-2.5K the CO₂ uptake (0.035 g_CO2 g_sorbent ^–1) exceeded the amount expected for the full conversion of K₂CO₃ into K₂Mg­(CO₃)₂ (0.026 g_CO2 g_sorbent ^–1), indicating that the uptake cannot be explained by K₂Mg­(CO₃)₂ formation alone, but instead points to the formation of an addtional phase. Interestingly, for MgO-5K and MgO-7.5K, a second, slower CO₂ capture stage was observed. A similar two-step CO₂ uptake behavior was also reported by Kwak et al. for MgO promoted with 10 mol % K₂CO₃ at 330 °C in CO₂. However, the authors could not identify the formed phases based on their ex situ XRD or solid state NMR analyses.

To identify the phases formed during the carbonation of K₂CO₃-promoted MgO, XRD analysis of the carbonated sorbents was performed (Figure 1 f,g). For MgO-0K, no crystalline carbonate phases were detected, indicating that the small amount of CO₂ captured was due to the formation of surface carbonates on MgO, consistent with previous studies and confirmed by DRIFTS experiments (Figure S16). ,, The XRD pattern of MgO-2.5K shows no sharp reflections; instead, the presence of diffuse X-ray scattering was observed as highlighted in Figure g. The XRD patterns for MgO-5K and MgO-7.5K revealed several small peaks in addition to a broad amorphous background, indicating the presence of several crystalline carbonate phases, including MgCO₃·3H₂O (nesquehonite), K₂Mg­(CO₃)₂·4H₂O (baylissite), K₂Mg­(CO₃)₂ and K₂CO₃ (β- and γ-type), in addition to one or multiple amorphous phases. The unit cells of the crystalline carbonates present in the carbonated sorbents are visualized in Figure 1 h and more details on the crystal structures are provided in Table S3. Interestingly, only a small amount of K₂Mg­(CO₃)₂ was detected by XRD in MgO-5K and MgO-7.5K. Additional K₂Mg­(CO₃)₂ may be present in an amorphous form, may have transformed into baylissite by reacting with water upon exposure to air during the XRD measurements, or a different K₂CO₃-containing phase may have formed.

To obtain more structural insight into the nature of the amorphous phase(s) formed during carbonation, MgO-2.5K and MgO-5K were probed by Raman spectroscopy after 10 h of carbonation (Figures S8 and S9, respectively). The Raman spectrum of MgO-2.5K reveals a broad peak centered in the range 1040–1130 cm^–1, while MgO-5K exhibited a broad peak in the range 1040–1140 cm^–1. By applying Gaussian deconvolution to the Raman spectra, we identified that the broad peak in both MgO-5K and MgO-2.5K can be deconvoluted into three carbonate phases with their peaks located at 1061 cm^–1, 1082 cm^–1 and 1103 cm^–1 whereby MgO-5K contains a significantly larger fraction of the carbonate phase that has a characteristic peak at 1103 cm^–1. Measurements of several reference carbonates (Figure S7) revealed that the peaks at 1061 cm^–1 and 1082 cm^–1 can be assigned to K₂CO₃ and baylissite (K₂Mg­(CO₃)₂·4H₂O), respectively. The peak at 1103 cm^–1 is most likely due to nesquehonite (MgCO₃·3H₂O), although the peak was slightly shifted compared to the reference position (1101 cm^–1). This shift could be explained by the partial dehydration of the nesquehonite-type structure during carbonation at 315 °C, as nesquehonite dehydrates at temperatures above 270 °C. , Indeed, Hales et al. observed that when heating nesquehonite to 400 °C in air, the Raman signal of the main carbonate peak gradually shifted from 1098 cm^–1 to 1105 cm^–1, associated with the formation of an amorphous magnesium carbonate phase as a result of nesquehonite dehydration. , Hence, Raman spectroscopy combined with XRD analysis indicates the presence of amorphous carbonates in MgO-2.5K and MgO-5K that have been exposed for 10 h to carbonation conditions, whereby the amorphous carbonate phase exhibits a variety of local motifs that are similar to those found in baylissite (K₂Mg­(CO₃)₂·4H₂O), K₂CO₃ and nesquehonite (MgCO₃·3H₂O).

In Situ XRD and Raman Analysis of MgO-5K

To investigate the evolution of the different carbonates of the most active sorbent, i.e. MgO-5K during carbonation at 315 °C in CO₂, we conducted in situ XRD and in situ Raman spectroscopy experiments (Figure ). In the in situ XRD experiment (Figure a), MgO-5K was first pretreated at 450 °C in N₂ for 1 h to remove adsorbed H₂O and CO₂, and to decompose hydrated phases such as baylissite (K₂Mg­(CO₃)₂·4H₂O) and K₂CO₃·1.5H₂O (Figure S10). After such pretreatment, the sorbent was cooled down to 315 °C in N₂. The XRD pattern collected at 315 °C in N₂, i.e. just before carbonation, revealed only Bragg peaks due to MgO and α-K₂CO₃. α-K₂CO₃ (hexagonal) is the high-temperature polymorph of K₂CO₃, which is stable at temperatures above 420 °C. , Upon cooling down the sorbent to 315 °C, a phase transition to β-K₂CO₃ (monoclinic) would typically be expected. However, only a minor fraction of α-K₂CO₃ converted into the β-phase, possibly due to kinetic limitations, and most of the K₂CO₃ remained as α-K₂CO₃ at the onset of the carbonation (Figure S11). To initiate carbonation, the gas atmosphere was changed from pure N₂ to an 80% CO₂/20% N₂ mixture. During the first 20 min of carbonation, the Bragg peaks corresponding to α-K₂CO₃ disappeared, and a broad, diffuse X-ray scattering background appeared, indicating the formation of an amorphous phase. After ca. 1 h of carbonation, a mixture of different crystalline carbonate phases started to appear, including baylissite (K₂Mg­(CO₃)₂·4H₂O), nesquehonite (MgCO₃·3H₂O), β-K₂CO₃, γ-K₂CO₃, and a very small amount of K₂Mg­(CO₃)₂. The intensity of these phases gradually increased with increasing carbonation time, revealing a slow crystallization process. Note, however, that the amorphous signal remained present, but slightly reduced in intensity.

2.

(a) In situ XRD data acquired during the carbonation of MgO-5K at 315 °C in CO₂, overlaid with reference diffractograms of nesquehonite (MgCO₃·3H₂O), baylissite (K₂Mg­(CO₃)₂·4H₂O), K₂Mg­(CO₃)₂, β-K₂CO₃, γ-K₂CO₃, and MgO. The dashed lines indicate the position of the background signal. The slight mismatch between the reference patterns and the experimental data arises from thermal lattices expansion at 315 °C. (b) In situ Raman spectroscopy data in the symmetric stretching region of carbonates, recorded after 0, 0.5, 1, 2, and 5 h of carbonation of MgO-5K at 315 °C in CO₂. Deconvolution using Gaussian functions reveals peaks corresponding to K₂CO₃, baylissite, and nesquehonite. (c) Area under the v₁ Raman peaks of K₂CO₃, baylissite, and nesquehonite as a function of carbonation time, derived from the deconvolution results in (b). (d–f) Quantitative analyses of the in situ XRD data shown in (a): (d) MgO conversion, (e) the evolution of the scattering signal due to amorphous phase(s), and (f) the evolution of the crystalline carbonates as a function of the carbonation time.

To complement the in situ XRD results and gain more insight into the evolution of the amorphous phase, we performed in situ Raman spectroscopy (Figure b (carbonate region) and Figure S12 (full spectral range)). As in the in situ XRD experiment, also in the in situ Raman spectroscopy experiments the sorbent MgO-5K was pretreated at 450 °C in N₂ for 30 min, subsequently cooled down to 315 °C in N₂ and exposed to CO₂ at 315 °C for 5 h. At the start of the carbonation reaction, a single Raman band located at 1057 cm^–1 was observed, corresponding to the ν₁ symmetric stretching mode of K₂CO₃, confirming the absence of other carbonate phases. This v₁ symmetric stretching mode of K₂CO₃ was shifted to a lower wavenumber compared to its room-temperature reference due to the thermal expansion of the K₂CO₃ lattice. During the first hour of carbonation, this carbonate band gradually shifted from 1057 cm^–1 to 1077 cm^–1. Gaussian peak fitting analysis of this band after 30 min and 1 h of carbonation revealed that this signal was a convolution of two contributions, viz. 1080 cm^–1 (baylissite (K₂Mg­(CO₃)₂·4H₂O)) and 1057 cm^–1 (K₂CO₃), with the amount of the baylissite-type structure increasing over time. As carbonation proceeded, the carbonate peak shifted further to higher wavenumbers, reaching 1104 cm^–1 while maintaining a shoulder at ca. 1057 cm^–1. Gaussian fitting of the carbonate peak after 2 and 5 h of carbonation suggests that the peaks contained now three components: 1057 cm^–1 (K₂CO₃), 1080 cm^–1 (baylissite (K₂Mg­(CO₃)₂·4H₂O)) and 1104 cm^–1 (nesquehonite (MgCO₃·3H₂O)), whereby the nesquehonite-type structure is the main component. In reaction stage I (after 30 min and 1 h), peak broadening of the ν₁ carbonate band was observed, consistent with the formation of disordered, amorphous phases, as confirmed by in situ XRD analysis (Figure a). In reaction stage II (after 2 and 5 h), the ν₁ carbonate band narrowed, reflecting a partial crystallization of the carbonates. It is worth noting that the in situ Raman results are in line with ex situ Raman measurements conducted on different areas of MgO-5K after 30 min and 10 h of carbonation (Figure S9). However, after 30 min of carbonation the ex situ sample no longer shows unreacted K₂CO₃. This observation could be explained by a higher MgO conversion in the TGA compared to the Raman reaction cell, or due to a further reaction upon exposure to air, since baylissite can also form at room temperature by reaction with atmospheric CO₂.

By tracking the intensity of the Raman bands during carbonation (Figure c), we find that in reaction stage I, the amount of K₂CO₃ decreased while there was a simultaneous emergence of a baylissite-type structure (a mixed K–Mg carbonate), indicating that the baylissite-type structure formed through the reaction of K₂CO₃ with MgO and CO₂. During reaction stage II, a nesquehonite-type structure (i.e., a Mg-rich carbonate) appeared and the amount of it gradually increased. Note that in this stage, the amount of K₂CO₃ remained stable, while the amount of the baylissite-type structure decreased. This suggests that the baylissite-type structure evolved into a nesquehonite-type structure by incorporating additional Mg and CO₂. At the end of the carbonation reaction (i.e., after 5 h), the material consisted predominantly of nesquehonite-type domains.

A quantitative analysis of the in situ XRD data provided further insight into the carbonation mechanism (see methods for details). Specifically, the MgO conversion (Figure d), the evolution of the amorphous signal (Figure e), and the evolution of the amount of the various crystalline phases (Figure f) were determined from the in situ XRD data. The conversion of MgO in MgO-5K during reaction stage I was determined as 12.0 ± 2.0% and increased to 16.0 ± 1.5% after 5 h of carbonation (Figure d). Note that the conversion of MgO after 10 h of carbonation, calculated based on the TGA results, was only 12% (determined using Eq. S3). The slightly higher conversion of MgO in the in situ XRD is likely due to differences in the gas–solid contact pattern. The evolution of the amorphous signal as a function of the carbonation time (Figure e) indicates that a large amount of the amorphous phase rapidly formed in reaction stage I, followed by a slight decrease in the total amount of the amorphous phase in reaction stage II. Tracking the evolution of the amount of crystalline carbonates (β-K₂CO₃, baylissite (K₂Mg­(CO₃)₂·4H₂O) and nesquehonite (MgCO₃·3H₂O)) (Figure f), reveals that the different carbonates appeared simultaneously and that their appearance matches well with the onset of the decrease of the amount of the amorphous phase. Quantification of the in situ XRD data also indicates that the main fraction of the carbonates formed exists in an amorphous state with only a small fraction being in the form of crystalline carbonates.

To analyze the effect of promoter loading on the observed phase transformations, we conducted an in situ XRD measurement during the carbonation of MgO-7.5K (Figure S13). The phase transformations closely resembled those observed in MgO-5K. However, in reaction stage I, a fraction of the K₂CO₃ remained unreacted and did not become amorphous, consistent with the TGA results (Figure e and Table S2). Furthermore, after 5 h of carbonation, MgO-7.5K contained significantly more K₂Mg­(CO₃)₂ (Figure S14), which can be attributed to the higher K₂CO₃ loading, in combination with a comparable surface hydroxide concentration to MgO-5K (the surface areas of MgO-5K (25 m² g^–1) and MgO-7.5K (21 m² g^–1) are similar, see Figure S6). Hence, the overall water content in the amorphous double carbonate formed in reaction stage I is lower in MgO-7.5K compared to MgO-5K. The slower reaction kinetics observed in reaction stage II for MgO-7.5K compared to MgO-5K, suggest that the lower water content in the amorphous phase affects the further CO₂ uptake.

The Role of Surface Hydroxides in the CO₂ Uptake Mechanism

To investigate the origin of the water required to form hydrated carbonates (baylissite (K₂Mg­(CO₃)₂·4H₂O) and nesquehonite (MgCO₃·3H₂O)) during the carbonation of MgO-5K at 315 °C in CO₂, in situ DRIFTS experiments were conducted (Figure ). Specifically, DRIFTS spectra of MgO-5K were collected after pretreatment at 450 °C in N₂ for 30 min and after carbonation at 315 °C in CO₂ for 1 h, Figure a (hydroxide region) and Figure S16b (carbonate region). The DRIFTS spectrum collected at 315 °C in N₂ (following the pretreatment and prior to carbonation) (Figure a), showed a sharp peak at 3740 cm^–1 corresponding to isolated −OH groups, and a broad band in the 3700–3200 cm^–1 region, which contains several features at 3620, 3513, and 3374 cm^–1. The broad band can be assigned either to adsorbed water or multicoordinated H-bonded −OH groups, which can recombine to form water at temperatures below 400 °C. − After 1 h of carbonation, the content of surface hydroxides on MgO-5K remained largely unchanged, indicating that no water was captured nor released during carbonation, as also confirmed by TGA-MS experiments (Figure S15). Interestingly, the small amount of isolated −OH groups present on the surface of MgO in MgO-5K after pretreatment almost completely disappeared after carbonation, and only the broad band in the 3700- 3200 cm^–1 region, corresponding to adsorbed water or crystal water, remained. The hydroxyl groups in the 3700–3200 cm^–1 region can be assigned to the vibrational modes of −OH groups and H₂O molecules in the crystal structure of baylissite and nesquehonite. Note that after pretreatment, the type of surface hydroxides in unpromoted MgO (MgO-0K, Figure S17) was similar to that in MgO-5K. However, there was a larger fraction of isolated −OH groups (peak at 3740 cm^–1) in MgO-0K. Furthermore, unlike in MgO-5K, such isolated −OH groups remained on the surface of MgO after the carbonation of MgO-0K.

3.

DRIFTS spectra of the hydroxide region of MgO-5K after pretreatment in N₂ for 30 min at (a) 450 °C, (b) 600 °C, and (c) 720 °C. The darker curves represent the spectra obtained at 315 °C in N₂ after pretreatment, while the lighter curves correspond to the spectra obtained after an additional carbonation step at 315 °C in CO₂ for 1 h. (d) CO₂ uptake curves obtained in a TGA after pretreatment at 450 °C (yellow), 600 °C (black), and 750 °C (green).

To examine the effect of the pretreatment temperature on the amount of surface hydroxides in MgO-5K, in situ DRIFTS experiments were performed for pretreatment temperatures of 600 °C (Figure b) and 720 °C (Figure c). The DRIFTS spectrum of MgO-5K after pretreatment at 600 °C in N₂ for 30 min (denoted as MgO-5K-pre600) showed a lower intensity of the broad −OH band compared to the spectrum after pretreatment at 450 °C, confirming that a pretreatment at a higher temperature removed surface hydroxides more effectively. However, a significant amount of isolated −OH groups and H-bonded −OH groups remained. In contrast, pretreatment at 720 °C (MgO-5K-pre720) resulted in the near-complete removal of the isolated −OH groups and a substantial reduction in the amount of H-bonded −OH groups. Interestingly, during the carbonation of both MgO-5K-pre600 and MgO-5K-pre720, the peak due to isolated −OH groups almost completely disappeared. However, the intensity of the broad −OH band increased slightly, revealing that the sorbents adsorbed small quantities of residual water present in the gas stream or the surrounding environment of the in situ DRIFTS cell. These findings indicate that H-bonded −OH groups on the surface of K₂CO₃-promoted MgO cannot be avoided under typical CO₂ capture conditions, even when applying high temperature pretreatments.

To better understand the role of water/surface hydroxides on the CO₂ uptake and the underlying carbonation mechanism of K₂CO₃-promoted MgO, the CO₂ uptake curves of MgO-5K pretreated at 450 °C, 600 or 750 °C were compared (Figure d). All sorbents exhibited a two-step CO₂ uptake, but the CO₂ uptake in both reaction stages decreased with increasing pretreatment temperature, viz. 0.099, 0.087, and 0.065 g_CO2 g_sorbent ^–1 for pretreatment temperatures of 450 °C, 600 and 750 °C, respectively. As we observe the two-step CO₂ uptake behavior for all sorbents, there is no indication of a change in the carbonation mechanism for lower surface hydroxide coverages (viz. higher pretreatment temperatures). Furthermore, the reduction in CO₂ uptake with increasing pretreatment temperature (a 34% reduction at 750 °C compared to 450 °C) does not scale with the much larger decrease in the quantity of surface hydroxides (a ca. 76% reduction at 720 °C compared to 450 °C, based on the integrated area in the 3200–3850 cm^–1 region). Instead, the lower CO₂ uptakes when using higher pretreatment temperatures (i.e., 600 and 750 °C) are most likely due to sintering. Indeed, the average crystallite size of MgO in MgO-5K after carbonation increased with increasing pretreatment temperature, viz. 20 nm (450 °C), 24 nm (600 °C) and 34 nm (750 °C) (Figure S18), indicative of a higher degree of sintering with increasing temperature. Overall, our results suggest that variations in the amount of water/surface hydroxides do not alter the carbonation mechanism. Instead, the formation of crystalline hydrated phases such as nesquehonite and baylissite occurs due to the unavoidable presence of surface hydroxides under typical carbonation conditions.

Finally, the role of water in the carbonation mechanism was investigated by carbonating MgO-5K in humid CO₂ (Figure S19). Compared to dry CO₂, carbonation under humid conditions resulted in a 32% lower CO₂ uptake, with only one CO₂ uptake step being observed. Interestingly, when switching from humid to dry CO₂, the CO₂ uptake recovers rapidly, and the second CO₂ uptake step appears. To probe these effects, DRIFTS spectra were obtained during carbonation in humid CO₂. After 1 h of carbonation in humid CO₂, the spectrum (Figure S20) closely resembles that obtained after carbonation in dry CO₂ (Figure a and Figure S16b), i.e. there is no evidence of bicarbonates, indicating that humidity does not affect the nature of the adsorbed carbonates. Upon switching from humid to dry CO₂, a slight decrease in the amount of adsorbed water (3200–3700 cm^–1) was observed, with no detectable changes in the surface carbonates. This suggests that only bulk carbonates, to which DRIFTS is less sensitive, were affected. Furthermore, based on TGA results, only a small additional CO₂ uptake was observed after switching to dry CO₂ and holding for one hour (Figure S19). Finally, the carbonation kinetics may differ in the DRIFTS setup.

To further rationalize these findings, density functional theory (DFT) calculations were performed. CO₂ adsorption energies on baylissite (K₂Mg­(CO₃)₂·4H₂O) and partially dehydrated and fully dehydrated baylissite surfaces were calculated as −0.46 eV and −0.57 eV, respectively (Table S4 and Figure S21), indicating that CO₂ adsorption is energetically more favorable onto a fully dehydrated baylissite surface as compared to a partially dehydrated one. These results suggest that humidity affects the CO₂ uptake of K₂CO₃-promoted MgO, potentially due to stabilization of the baylissite-type amorphous phase and/or a competition between CO₂ and H₂O for the available sorption sites. This observation aligns with the well-established role of water in mineralization pathways, such as the stabilization of metastable amorphous intermediates. ,

Carbonation Mechanism

Combining insights from the analyses presented above, we propose a mechanism for the carbonation of K₂CO₃-promoted MgO (schematically shown in Figure ). In reaction stage I, a K-rich amorphous phase forms rapidly, which consists predominantly of baylissite-type (K–Mg carbonate) domains as well as some K₂CO₃-type domains (unreacted K₂CO₃). More specifically, the amorphous K–Mg carbonate forms through the reaction of K₂CO₃ with MgO, CO₂, and residual water that is adsorbed on the surface of MgO via the following equation:

$${\mathrm{K}}_2{\mathrm{CO}}_3+\mathrm{x}\mathrm{MgO}+{\mathrm{x}\mathrm{CO}}_2+{\mathrm{y}\mathrm{H}}_2\mathrm{O}\rightleftharpoons {\mathrm{K}}_2{\mathrm{CO}}_3·{\mathrm{x}\mathrm{MgCO}}_3·{\mathrm{yH}}_2\mathrm{O}$$ 3

4.

Schematic representation of the evolution of the different crystalline and amorphous phases during the carbonation of K₂CO₃-promoted MgO.

In reaction stage I, x ≤ 1 as the amorphous carbonate is rich in K₂CO₃.

The rapid kinetics of double carbonate formation via eq can be rationalized by considering the binding strength of CO₂ in baylissite (K₂Mg­(CO₃)₂·4H₂O). DFT calculations of MgCO₃, baylissite and partially dehydrated baylissite yield CO₂binding strengths of −1.50 eV, −4.7 eV and −1.91 eV, respectively (Table S4). These values indicate that the binding strength of CO₂ is much higher in baylissite (and partially dehydrated baylissite) than in MgCO₃.

In reaction stage II, the K-rich amorphous carbonate further reacts with additional MgO and CO₂, forming a Mg-rich amorphous carbonate (x > 1 in eq ), that mostly contains nesquehonite-type domains. Concurrently, small amounts of crystalline carbonates - including K₂CO₃, nesquehonite (MgCO₃·3H₂O) and baylissite (K₂Mg­(CO₃)₂·4H₂O) - appear.

The promotional mechanism differs from that reported for Na₂CO₃ and CaCO₃, in which crystalline double carbonates (NaMg­(CO₃)₂ and CaMg­(CO₃)₂) form rapidly and likely act as nucleation seeds for MgCO₃ in the presence of molten nitrates. ,, In the case of K₂CO₃, a hydrated amorphous double carbonate forms that, however, does not act as a nucleation seed for MgCO₃ formation. Instead, we hypothesize that this amorphous phase promotes carbonation in a manner analogous to molten alkali metal nitrates such as NaNO₃, i.e. by dissolving MgO and CO₂ and thereby facilitating the formation of a Mg-rich amorphous carbonate – even in the absence of a molten alkali metal nitrate salt.

Cyclic CO₂ Uptake Performance of MgO-5K

To evaluate the cyclic CO₂ uptake and release performance of K₂CO₃-promoted MgO, MgO-5K was pretreated at 450 °C in N₂ for 30 min, carbonated for 5 h at 315 °C in CO₂, and regenerated at 450 °C in N₂ for 15 min in a TGA. The carbonation-regeneration cycle was repeated five times. During the first five carbonation-regeneration cycles (Figure a), the CO₂ uptake decreased continuously from 0.114 g_CO2 g_sorbent ^–1 to 0.058 g_CO2 g_sorbent ^–1 (i.e., a reduction by 49%). While there was a slight decrease in the CO₂ uptake in reaction stage I; most of the decrease in the CO₂ uptake occurred in reaction stage II due to an extended induction period before its onset. HAADF-STEM images combined with elemental mapping of carbonated MgO-5K after the first (Figure b,c) and the fifth (Figure d,e) cycle revealed that while potassium was initially distributed uniformly, it partially agglomerated after five cycles. The increasingly heterogeneous distribution of potassium, and consequently of the K-rich intermediate with cycle number resulted in a decrease in the interface area with MgO. However, a high interfacial surface area between the K-rich amorphous carbonate and MgO is crucial to allow for a high additional CO₂ uptake in reaction stage II. The XRD pattern of the carbonated sorbent in the fifth cycle (Figure S22) confirms that after 5 h of carbonation, no crystalline carbonates were detected. Instead, only an amorphous halo, most likely due to the baylissite-type phase, was present, consistent with TGA data indicating that reaction stage II had just started. These results suggest that while MgO-5K is suitable for cyclic CO₂ capture from a process perspective, the segregation of K₂CO₃ significantly reduces its CO₂ uptake capacity with cycle number. Therefore, future research should focus on maintaining a uniform K₂CO₃ distribution within the sorbent, e.g. by nanostructuring of the sorbent or by the addition of inert stabilizers, such as Al₂O₃, SiO₂ or ZrO₂. Such stabilizers have been studied extensively for CaO-based CO₂ sorbents where they are shown to reduce sorbent sintering by acting as physical barriers between adjacent sorbent grains, thereby stabilizing the pore network of the sorbent. −

5.

(a) Cyclic CO₂ uptake of MgO-5K over five consecutive carbonation–regeneration cycles recorded in a TGA (carbonation: 5 h at 315 °C in CO₂; regeneration: 15 min at 450 °C in N₂). (b) Plot of the temperature profile used during the TGA experiment in (a). HAADF-STEM images and elemental mapping (Mg and K) of carbonated MgO-5K (c, d) after the first cycle and (e, f) after five carbonation-regeneration cycles.

Conclusions

MgO-based sorbents are cost-efficient materials for CO₂ capture, yet their slow kinetics require the use of promoters, typically alkali metal salts. Here, we elucidated the mechanism by which K₂CO₃ promotes the CO₂ uptake of MgO and identified the carbonate phases that form during carbonation at 315 °C in CO₂, using in situ XRD and in situ Raman spectroscopy. Initially, there was a rapid (<1 min) formation of a K-rich amorphous carbonate (K₂CO₃·xMgCO₃·yH₂O); in a second CO₂ uptake step this intermediate phase further reacted with MgO and CO₂, resulting in the formation of a Mg-rich amorphous carbonate. Furthermore, small amounts of hydrated crystalline carbonates, including nesquehonite (MgCO₃·3H₂O) and baylissite (K₂Mg­(CO₃)₂·4H₂O), formed. In situ DRIFTS revealed that the water required for the formation of the hydrated carbonate phases was present in the form of surface hydroxyl groups at the onset of the carbonation reaction, even after a thermal pretreatment at 720 °C in N₂. Introducing steam during carbonation, resulted in a lower CO₂ uptake and inhibited the second CO₂ uptake step, most likely by stabilizing the amorphous baylissite-type intermediate. It was also demonstrated that the sorbent can be used for cyclic CO₂ capture operation, although minimizing K₂CO₃ segregation will be crucial to maintain a satisfying long-term cyclic performance. Overall, this study identifies a K-rich amorphous carbonate as the key promoter for CO₂ uptake in K₂CO₃-promoted MgO and provides valuable insights to guide the design of more effective, nitrate-free MgO-based sorbents for CO₂ capture. Furthermore, the phases identified here are likely to play also an important role in MgO-based sorbents that are promoted with a mixture of Na and K nitrates and carbonates, which are currently the best performing MgO-based CO₂ sorbents reported in the literature.

Experimental Section

Chemicals

All chemicals were used as received: Magnesium hydroxide (Mg­(OH)₂, SLR, Fisher Chemical), potassium carbonate (K₂CO₃, extra pure, Fisher Chemical), sodium carbonate (Na₂CO₃, 99.5%, Acros Organics), magnesium nitrate hexahydrate (Mg­(NO₃)₂·6H₂O, Thermo Scientific), potassium nitrate (KNO₃, ≥ 99.0%, Merck), hydromagnesite (Mg₅(CO₃)₄(OH)₂·4H₂O, extra pure, Acros Organics), magnesium chloride hexahydrate (MgCl₂·6H₂O, ≥ 99.0%, Sigma-Aldrich) and ethylene glycol (>99%, Sigma-Aldrich).

Materials Preparation

K₂CO₃-Promoted MgO Sorbents

MgO was prepared by calcining Mg­(OH)₂ for 5 h at 500 °C in air. For the preparation of MgO-5K₂CO₃ (MgO promoted with 5 mol % K₂CO₃), 1 g of MgO, 0.17 g of K₂CO₃ and 10 mL isopropanol were loaded into a Pulverisette 7 planetary micro mill (Fritsch) at 150 rpm for 20 h (60 cycles of 10 min, followed by a 10 min break after each cycle) using 1 mm ZrO₂ balls in a Si₃N₄ cylinder. The ball-milled material was collected and dried in an oven at 100 °C for 2 h. The same method was used to prepare MgO-2.5K₂CO₃, MgO-7.5K₂CO₃ and the unpromoted MgO reference by adjusting the amount of K₂CO₃. Finally, the sorbents were calcined at 500 °C for 4 h.

Reference Materials

Baylissite (K₂Mg­(CO₃)₂·4H₂O)

was synthesized using an adapted method from literature. Specifically, 1 mL of a magnesium nitrate solution was added dropwise to a concentrated K₂CO₃ solution (3 g K₂CO₃ in 10 mL H₂O) at 90 °C under constant stirring. After 10 min, the stirring was stopped, and the temperature was lowered to 60 °C. The mixture was maintained at 60 °C for 72 h to allow crystallization. The resulting mixture was then filtered (without washing as washing would have dissolved the crystals) and dried at room temperature. XRD, Raman and SEM-EDX analysis (Figures S3 and S4), revealed that the synthesized baylissite was almost phase-pure with only a small K₂CO₃ impurity.

K₂Mg­(CO₃)₂

was prepared by mixing as-synthesized baylissite with 20 mol % KNO₃ upon grinding. The material was subsequently treated in a TGA at 340 °C for 3 h in CO₂ (80 mL/min and a heating rate of 50 °C/min) to remove the crystal water, resulting in KNO₃-promoted K₂Mg­(CO₃)₂ (see Figure S23 for more details).

MgCO₃

was synthesized from hydromagnesite via a wet chemistry method, using a suspension of hydromagnesite in a 95 wt % ethylene glycol solution in water. This suspension was refluxed at 150 °C under a continuous flow of CO₂. The phase purity was verified by XRD (Figure S24).

Nesquehonite (MgCO₃·3H₂O)

was synthesized using a method adopted from the literature. Solutions of 0.5 M MgCl₂ and 0.5 M Na₂CO₃ were mixed at a volume ratio of 1:1 and stirred at 1000 rpm for 2 h at room temperature. The resulting white suspension was filtered out, washed with distilled water, collected and dried overnight in air. The phase purity was verified by XRD (Figure S25a).

Nesquehonite-Type Amorphous Magnesium Carbonate

was prepared by heating nesquehonite to 315 °C in a TGA under a N₂ atmosphere at a heating rate of 5 °C min^–1, resulting in a weight loss of 36 wt %. This value aligns closely with the theoretical weight loss for the dehydration of MgCO₃·3H₂O (39 wt %), indicating the nearly complete removal of crystal water (Figure S25b). The amorphous nature of the resulting magnesium carbonate was confirmed by the absence of crystalline peaks in the XRD pattern (Figure S25a).

Characterization Methods

Raman Spectroscopy

Raman spectra were collected with a Thermo Scientific DXR2 Raman spectrometer equipped with a 532 nm laser using a spot size of 1.8 μm. The spectra were acquired in the range of 100–3,500 cm^–1 with a spectral resolution of 0.964 cm^–1. For the ex situ measurements, the Raman spectra represent the average obtained from measurements taken in five different regions of each sample. For the in situ measurements, 20 mg sample was loaded into a Linkam CCR1000 in situ cell. The sample was subjected to a pretreatment for 30 min at 450 °C in N₂, followed by carbonation for 5 h at 315 °C in CO₂. The heating and cooling rates were set to 30 °C min^–1 and a gas flow rate of 30 mL min^–1 was used. The Raman band intensity was obtained from the in situ Raman data by integrating the area under the ν₁ bands for K₂CO₃, baylissite and nesquehonite. The areas were obtained through deconvolution using Gaussian functions (Figure b) as a function of the carbonation time. Note that the deconvolution results are inherently semiquantitative: while they allowed us to track the temporal changes in the area under each individual Raman band, they do not enable a reliable quantification of the relative amounts of the different phases. Additionally, the Raman bands reflect distinct local structural motifs within the amorphous phase rather than specific crystalline phases. These motifs are analogous to those observed in the reference carbonates used for the deconvolution.

CO₂ Uptake Measurements

TGA experiments were carried out in a Mettler Toledo TGA/DSC 3+. In a typical analysis, 15 mg of sample (or 50 mg when a mass spectrometer was connected for product gas analysis) was loaded into a 70 μL alumina crucible. For a typical carbonation experiment, the flow of reactive gas (N₂ for pretreatment and regeneration and CO₂ for carbonation) was set to 80 mL min^–1 and a heating rate of 50 °C min^–1 was used. To introduce moisture (humidity) into the system, the reactive gas was sent through a gas washing bottle at room temperature to achieve a relative humidity of close to 100% at room temperature (which corresponds to a relative humidity of ca. 4.5% at 315 °C). For selected experiments, the gas composition at the outlet of the TGA was analyzed with a mass spectrometer (MKS Cirrus TM 3-XD), focusing on the signal for the following masses (m/z): 18 (H₂O), 28 (N₂), 32 (O₂) and 44 (CO₂).

X-ray Diffraction

XRD patterns were collected using a PANalytical Empyrean X-ray powder diffractometer (45 kV and 40 mA using Cu K_α radiation) equipped with a Bragg–Brentano HD mirror. Ex situ scans were collected in the 2θ range of 8–70 ° (step size 0.033 ° and time per step 3.2 s). For the in situ measurements, 20 mg sample was placed onto a quartz disc and loaded into an Anton Paar XRK 900 reaction chamber. XRD patterns were collected in the 2θ range of 8–50° (step size of 0.12 ° and time per step of 3.4 s), while the sample was subjected to pretreatment (at 450 °C in N₂) and carbonation (at 315 °C in 80 vol% CO₂/N₂) conditions using a gas flow rate of 200 mL min^–1 and a heating rate of 20 °C min^–1.

Phase Quantifications Using the in Situ XRD Data

The MgO conversion was calculated as (A_MgO(200)(t = 0) – A_MgO(200)(t)) × A_MgO(200)(0)⁻¹ × 100%, where A_MgO(200)(t) is the area under the (200) peak of MgO (2θ = 42.9°) at a given time t. It should be noted that this calculation considers crystalline phases only. To estimate the amount of the amorphous phase as a function of carbonation time, the background of the XRD pattern at carbonation t = 0 (determined using a linear interpolation) was subtracted from all subsequent patterns collected during carbonation. The amorphous fraction was then calculated by fitting the remaining (adjusted) background using a linear interpolation (spline) and integrating the area under the interpolation. The evolution of the crystalline phases was determined by integrating the area under the most intense, nonoverlapping Bragg peak for each phase as a function of carbonation time: α-K₂CO₃ (102) (2θ = 31.1°), β-K₂CO₃ (130) (2θ = 31.4°), baylissite (120) (2θ = 29.7°) and nesquehonite (200) (2θ = 23.1°). Note that the results for both the amorphous and the crystalline phases are semiquantitative. While they provide insight into the temporal evolution of the amount of each phase, they do not allow for a direct comparison between the different phases.

Electron Microscopy

Transmission electron microscopy (TEM) and scanning TEM combined with energy dispersive X-ray spectroscopy (STEM-EDX) measurements were acquired with a FEI Talos F200X electron microscope operated at 200 kV and equipped with a Super-X EDS system. SEM-EDX measurements were obtained using a Quanta 200F SEM microscope operated at 10 kV. Potassium loadings based on the SEM-EDX results were determined by averaging the quantification results from 3 different regions of the sample.

N₂ Physisorption

BET surface areas of the materials were determined from a N₂ physisorption isotherm recorded at 77 K on an Anton Paar Nova 800 apparatus. The samples were degassed at 350 °C under vacuum (10^–3 mbar) for 3 h prior to measurement.

ICP-OES

The relative amounts of Mg and K in the sorbents were determined using an Agilent 5100 VDV ICP-OES instrument. Samples were dissolved in aqua regia prior to the measurements and the measurements were performed in triplicate for reproducibility.

In Situ DRIFTS

In situ DRIFTS experiments were performed using a Nicolet 6700 FTIR spectrometer (resolution 4 cm^–1) coupled with a Harrick Praying Mantis high temperature reaction chamber with ZnSe windows. The sample holder was filled with quartz wool and ca. 10 mg of ground sample was placed on top and pressed flat. In a typical experiment, the sample was heated to the pretreatment temperature (e.g., 450 °C, 600 or 720 °C) in N₂ and kept at this temperature for 30 min, followed by cooling down to 315 °C. At 315 °C, the sample was exposed to CO₂ for 2 h. The heating rate was set to 10 °C min^–1 and the gas flow rate was 30 mL min^–1. The FTIR signal of KBr powder treated under similar conditions was used as a spectral background.

DFT Calculations

DFT calculations were performed using the Projector Augmented Wave (PAW) method , as implemented in the Vienna Ab initio Simulation Package (VASP). , The calculations were completed with a plane-wave cutoff energy of 600 eV and Gamma k-points. The electronic self-consistent calculation converged to 1 × 10^–6 eV and ionic relaxation steps were performed using the conjugate-gradient method (IBRION = 2) and continued until the total force on each atom was below a tolerance of 0.01 eV/Å. The generalized gradient approximation (GGA) was used for the exchange correlation functionals as parametrized by Perdew–Burke–Ernzerhof (PBE). The bulk and (100) surface calculations, including MgO, MgCO₃ and baylissite, were conducted to study the adsorption/binding of CO₂ in/on different systems, both bulk and surfaces with the presence and absence of water molecules. The binding of CO₂ (E_b‑CO2) with the substrate was calculated using the following equation.

$${\mathrm{E}}_{\mathrm{b}‐\mathrm{CO}2}=\mathrm{E}[\mathrm{substrate}+{\mathrm{CO}}_2]-\mathrm{E}[\mathrm{substrate}]-\mathrm{E}[{\mathrm{CO}}_2]$$ 4

where E [substrate + CO₂], E­[substrate] and E­[CO₂]­are the electronic energies of the adsorbed system, the clean bulk/surface, and CO₂, respectively.

The Supporting Information is available free of charge at . The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c00838.

• Additional description of methods, XRD patterns, and Raman spectra of the reference materials, calculation of MgO conversion, and supplementary results including ICP-OES, XRD, Raman, TGA­(-MS), DRIFTS, SEM-EDX, and N₂ physisorption data (PDF)

CRediT: Annelies Landuyt conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing-original draft; Felix Donat conceptualization, investigation, supervision, validation, writing-review and editing Paula M. Abdala conceptualization, supervision, validation, writing-review and editing; Christoph R. Müller conceptualization, funding acquisition, project administration, resources, supervision, validation, writing-review and editing.

The authors declare no competing financial interest.