Maximilian
Krödel
,
Alexander
Oing
,
Jan
Negele
,
Annelies
Landuyt
,
Agnieszka
Kierzkowska
,
Alexander H.
Bork
,
Felix
Donat
and
Christoph R.
Müller
*
Department of Mechanical and Process Engineering, Laboratory of Energy Science and Engineering, ETH Zurich, Leonhardstrasse 21, 8092, Zurich, Switzerland. E-mail: muelchri@ethz.ch
First published on 10th October 2022
Improving the cyclic CO2 uptake stability of CaO-based solid sorbents can provide a means to lower CO2 capture costs. Here, we develop nanostructured yolk(CaO)–shell(ZrO2) sorbents with a high cyclic CO2 uptake stability which outperform benchmark CaO nanoparticles after 20 cycles (0.17 gCO2 gSorbent−1) by more than 250% (0.61 gCO2 gSorbent−1), even under harsh calcination conditions (i.e. 80 vol% CO2 at 900 °C). By comparing the yolk–shell sorbents to core–shell sorbents, i.e. structures with an intimate contact between the stabilizing phase and CaO, we are able to identify the main mechanisms behind the stabilization of the CO2 uptake. While a yolk–shell architecture stabilizes the morphology of single CaO nanoparticles over repeated cycling and minimizes the contact between the yolk and shell materials, core–shell architectures lead to the formation of a thick CaZrO3-shell around CaO particles, which limits CO2 transport to unreacted CaO. Hence, yolk–shell architectures effectively delay CaZrO3 formation which in turn increases the theoretically possible CO2 uptake since CaZrO3 is CO2-capture-inert. In addition, we observe that yolk–shell architectures also improved the carbonation kinetics in both the kinetic- and diffusion-controlled regimes leading to a significantly higher cyclic CO2 uptake for yolk–shell-type sorbents.
CaO-based sorbents are non-toxic and cost-effective (estimated capture cost of <30 USD per tCO2) compared to the current benchmark technology for CO2 capture, i.e. amine scrubbing (estimated capture costs of 40 to 55 USD per tCO2).7–9 However, CaO-based sorbents are prone to rapid deactivation, i.e. a reduction of their cyclic CO2 uptake capacity from a theoretical maximum of 0.78 gCO2 gCaO−1 to typically <0.2 gCO2 gCaO−1, largely due to sintering at the high operating temperatures (around 650–900 °C).10–12 Various approaches to mitigate sintering-induced deactivation have been reported, including (I) the addition of high Tammann-temperature stabilizers such as MgO, Al2O3 or ZrO2,13–18 (II) the (re-)structuring of the morphology of sorbents,14,19e.g. using steam treatment,20 or (III) the manufacture of materials with a high surface area and pore volume.19
The stabilization of CaO-based sorbents using metal oxides such as MgO, Al2O3 and ZrO2 has been widely applied and currently there are very few new sorbent formulations that do not contain either such structural stabilizers, promoters (e.g. alkali metal salts like Na2CO3) or other performance-enhancing precursors (e.g. Ca-precursors containing alcohol groups).5,14,16,17,21–25 The best reported materials exhibit cyclic CO2 uptakes of up to 0.65 gCO2 gSorbent−1 after 10 carbonation–calcination cycles, which is six times higher compared to the benchmark limestone.26 Here, the high Tammann-temperature stabilizer MgO is believed to act as a physical barrier (spacer) between the CaO grains, reducing in turn their sintering (or the sintering of the CaCO3 formed).5,21,27,28 Hence, the size and distribution of these structural stabilizers within the CaO matrix must play an important role in mitigating sintering. It is important to note that the homogeneity of the distribution of the stabilizer might dynamically change with cycling (and indeed possibly even within one carbonation–calcination cycle). For example, Kim et al.15 observed that the stabilizer Ca3Al2O6 migrates to the surface of CaO particles during cycling, reducing in turn the stabilizer's effectiveness in reducing sintering. Therefore, one of the proposed design criteria for effective CaO-based sorbents is a low mobility of the stabilizing phase to avoid the loss of its stabilizing functionality. In addition, many investigated stabilizers form mixed phases with CaO at CaL conditions (650–900 °C) that are inactive for CO2 uptake, e.g. Ca3Al2O3 or CaZrO3.11,19,29,30 The formation of mixed phases that do not absorb CO2 results in a reduced amount of reactive CaO, reducing the theoretically achievable CO2 uptake of the sorbent. Hence, the synthesis of structures in which the interaction between CaO and the stabilizer is minimized to limit the formation of CO2-capture-inactive mixed oxides is desirable.
Like many gas–solid reactions, the carbonation reaction of CaO proceeds in two principal reaction regimes: a kinetically-controlled, rapid CO2 uptake regime that is followed by a much slower, diffusion-controlled CO2 uptake regime.31–33 While the CO2 uptake during the kinetically-controlled regime is largely controlled by the available pore volume and surface area, the apparent rate of reaction in the diffusion-controlled reaction regime is affected by morphological and structural parameters, e.g. the effective diffusivity of CO2 in the material.21 In previous works, the transition between the two reaction regimes has been associated with the formation of a product layer of CaCO3 on top of CaO, which causes the blockage of pores such that a certain fraction of the pore volume and surface area of the sorbent becomes inaccessible for CO2 molecules.34–36 For a two-dimensional surface of a non-porous material it has been hypothesized that once the product layer reaches a critical thickness (estimated to be around 30–50 nm by various works34,37,38), the reaction transitions from kinetic control to diffusion control.39 To maximize the CO2 uptake in the kinetically-controlled regime, the nanostructuring of CaO (i.e. particle size < 60–100 nm) is highly favorable. Previous works concerned with CaO (or CaCO3) nanoparticles of size < 100 nm investigated uncoated nanoparticles or nanoparticles with coatings that have an intimate contact with the CaO nanoparticle core (core–shell structures), resulting in the formation of CO2-capture-inactive mixed phases and hence comparatively low CO2 uptake.40,41 For example, core–shell structures with CaCO3 cores and shells of ZrO2 or SiO2 have been reported to yield materials with a significantly stabilized cyclic CO2 uptake compared to bare CaCO3.29,42,43 However, these materials deactivated over prolonged cyclic operation via the formation of mixed oxides with calcium that both reduce the amount of CaO available for reaction and also affect negatively the carbonation kinetics. Instead, the structure of an optimized model CaO-based sorbent should exhibit the following attributes: (I) the structure should be composed of CaO nanoparticles of size < 100 nm to reduce the effect of product layer diffusion limitation.5,34,37 (II) It should utilize a stabilizer to avoid physical contact (and hence sintering) between the individual CaO nanoparticles. Applying a stabilizer shell around the CaO nanoparticles would yield this physical separation, but the direct contact between the stabilizer and CaO might lead to the formation of CO2-capture-inactive mixed phases. To prevent such direct contact, a sacrificial layer can be placed between the CaO nanoparticle and the stabilizer coating, creating upon its removal a space between the CaO nanoparticle and the stabilizing coating. (III) The stabilizer shell needs to be porous to allow for a fast transport of CO2 to the surface of CaO. Overall, these requirements would demand a yolk–shell-type structure. Hence in this work, we report on the synthesis of such model, yolk–shell-structured sorbents to probe whether the aforementioned requirements concerning sorbent design are indeed correct and to understand better the deactivation and stabilization mechanisms of nanostructured sorbents. We note that such sorbents are very likely not to be directly usable in an industrial CO2 sorption process in which mechanical stability of sorbent particles may be one of many additional constraints. However, the insights gained from this study will help to understand better the functioning of stabilizers, aiding the design of practically applicable, yet still superior sorbent systems.
To this end, we use CaCO3 nanoparticles with an average particle size < 100 nm that are stabilized by a porous shell of ZrO2 in a yolk(CaO)–shell(ZrO2) architecture. The presence of a yolk–shell-type structure is shown to ensure a high sintering resistance and high rates of transport of CO2 to CaO, while the formation of the CO2-capture-inactive CaZrO3 phase is avoided. The design approach is depicted in Fig. 1.
A FEI Magellan 400 FEG high resolution scanning electron microscope (SEM) was used for the textural and morphological analysis of the synthesized sorbents. The instrument was operated at an acceleration voltage of 10 kV and a beam current of 25 pA. To minimize charging effects during imaging, the samples were sputtered with a 4 nm thick film of Pt/Pd (80:20) prior to SEM analysis.
(1) |
In the second step of the synthesis, the Ca@C particles were coated with a layer of Zr(OH)4 utilizing a wet-chemistry-based approach (see Materials section for details).45 A homogeneous coating with Zr(OH)4 (see below) was confirmed via EDX-TEM, see Fig. 2d. The resulting yolk–shell structure after calcination (synthetic air, 900 °C) is shown in Fig. 2e and will be discussed in more detail further below. Throughout this work, the nomenclature for yolk–shell sorbents will be Ca@xZr, where x is the experimentally determined amount of ZrO2 (in wt%) in the sorbent. As a reference material, CaCO3 NPs (Ca-NP) were also directly coated with zirconia using the same coating procedure but without addition of the carbonaceous shell, yielding core–shell sorbents. Throughout the manuscript, the core–shell sorbents will be referred to as Ca–xZr.
The structure of the as-synthesized (i.e. before the first calcination) sorbents was assessed by XRD and Raman spectroscopy. The XRD patterns of the reference material and the zirconia-stabilized yolk–shell and core–shell-type sorbents are shown in Fig. 2f. The diffractogram of Ca-NP (as received) showed only the characteristic diffraction peaks of CaCO3 (calcite polymorph). Also all of the yolk–shell and core–shell sorbents showed the diffraction peaks due to CaCO3. An amorphous halo with a maximum around approximately 2θ = 30°, relating to amorphous Zr(OH)4, was observed in the sorbents prior to their calcination.46 The ZrO2 content in the sorbents was estimated using Rietveld analysis of XRD data acquired of sorbents that have undergone 20 cycles and are listed in Table 1 (see Fig. S2 and Table S1† for details on the Rietveld refinement). The Raman spectra of the as-synthesized sorbents are shown in Fig. 2g. All yolk–shell sorbents (prior to calcination) showed features due to the D-band (1350 cm−1) and G-band (1574 cm−1) of carbon confirming the presence of a carbonaceous coating. These sorbents also show a weak band related to amorphous Zr(OH)4 between 590 and 670 cm−1, confirming the Zr phase observed in XRD.47 Additional features related to CaCO3 (158 cm−1, 284 cm−1, 713 cm−1, 1087 cm−1) were observed as well.
Type | Sorbent | ZrO2 content [wt%] |
---|---|---|
Core–shell | Ca–15Zr | 15 |
Core–shell | Ca–26Zr | 26 |
Yolk–shell | Ca@16Zr | 16 |
Yolk–shell | Ca@12Zr | 12 |
Yolk–shell | Ca@55Zr | 55 |
To assure that the carbonaceous template was removed completely during calcination in synthetic air, we studied the decomposition and removal of the template by in situ Raman spectroscopy. Fig. S3† shows the Raman spectra of Ca@C during heat treatment in synthetic air (up to 550 °C). We observe the disappearance of hydroxyl groups ν(OH), at approximately 2800 cm−1, in the temperature range 50 to 100 °C, corresponding to mass release (I) as observed in the TGA measurement. The modes relating to the G-band and D-band of carbon decrease in intensity between 100 and 500 °C. At 550 °C, no features related to the D- and G-band were observed, confirming the complete removal of the carbonaceous template at 550 °C. The characteristic mode for CO32− (∼1087 cm−1) in CaCO3 is present at all temperatures (i.e. up to 550 °C). EDX-TEM analysis of the particles before and after calcination confirms the carbon removal; the respective images are shown in Fig. S3b and c.†
The structure of the as-prepared (i.e. calcined), zirconia-stabilized sorbents was assessed by XRD (Fig. 3b). The calcined core–shell particles show the characteristic diffraction peaks of both CaO (Fmm space group, cubic) and CaZrO3 (Pcmn space group, orthorhombic). Unexpectedly, the intensity of the peaks related to CaZrO3 increased from Ca–15Zr to Ca–26Zr, indicating the formation of larger amounts of the mixed phase with increasing ZrO2 contents. Using Rietveld analysis (see Fig. S2 and Table S1†), we estimated the fraction of the CaZrO3 mixed phase in the core–shell sorbents Ca–15Zr and Ca–26Zr as 15 and 37 wt%, respectively. The XRD patterns of the yolk–shell sorbents Ca@16Zr and Ca@12Zr showed peaks due to CaO and very low intensity peaks possibly due to CaZrO3, which we were not able to quantify using Rietveld refinement (due to their small intensity). This suggests, considering the detection limit of the XRD setup, that in such sorbents the weight fraction of CaZrO3 is <1 wt%. In addition, no peaks due to ZrO2 were observed. Hence, in the yolk–shell sorbents Ca@12Zr and Ca@16Zr Zr is very likely present in an amorphous phase. However, for Ca@55Zr, tetragonal zirconia (t-ZrO2) was observed in addition to both CaO and CaZrO3. Rietveld refinement showed that this sorbent contained about 11 wt% CaZrO3 and 52 wt% t-ZrO2. In conclusion, compared to the core–shell particles, CaZrO3 formation was reduced in the yolk–shell-type sorbents (at similar ZrO2 contents), in particular when comparing Ca@16Zr and Ca–15Zr.
The morphologies of the calcined yolk–shell sorbents and core–shell sorbents were imaged using (EDX-)TEM, see Fig. 3c and d. The yolk–shell-structured sorbents Ca@12Zr (Fig. 3c) and Ca@16Zr (Fig. S4a†) exhibit individual CaO nanoparticles encapsulated by a Zr-containing shell with an occasionally dendritic appearance. The void space between the yolk and the shell, characteristic for the yolk–shell architecture, is clearly visible for Ca@12Zr. Ca@55Zr (Fig. S4b†) showed both larger agglomerates coated with a Zr-containing shell as well as some yolk–shell structures similar to the ones observed for Ca@16Zr and Ca@12Zr. For the larger agglomerates, the distribution of Ca and Zr, based on EDX-TEM, overlapped in intensity, indicating the formation of large CaZrO3 particles or CaZrO3-coated CaO particles, in line with our XRD observations (Fig. 3b). For high Zr contents (Ca@55Zr), TEM also shows the presence of ZrO2 particles, which are, taking into account the XRD results, related to t-ZrO2. The core–shell sorbent Ca–26Zr (Fig. 3d) exhibited a Zr-containing coating covering the whole surface of the CaO particles. Based on our XRD analysis, this coating is composed of a thick layer of CaZrO3. On the other hand, the calcined core–shell sorbent Ca–15Zr (Fig. S4c†) showed an inhomogeneous Zr-containing coating on the CaO particles, indicating that the added amount of ZrO2 was insufficient to completely coat the CaO nanoparticles with a Zr-containing phase. Based on our XRD analysis, we also expect the Zr-containing coating of Ca–15Zr to be CaZrO3.
The reference sorbent Ca-NP (without Zr) has a high CO2 uptake of 0.63 gCO2 gSorbent−1 in the first cycle. However, the CO2 uptake decreases rapidly with cycle number, reaching 0.21 gCO2 gSorbent−1 (decrease of 66% compared to the 1st cycle) and 0.17 gCO2 gSorbent−1 (decrease of 73% compared to the 1st cycle) after 10 and 20 cycles, respectively. The CO2 uptake of the yolk–shell-structured sorbents varied substantially with the content of Zr. Compared to Ca-NP (0.17 gCO2 gSorbent−1), Ca@16Zr and Ca@12Zr show significantly higher CO2 uptakes after 20 cycles, reaching, respectively, 0.38 gCO2 gSorbent−1 (maximum theoretical CO2 uptake of 0.65 gCO2 gSorbent−1) and 0.61 gCO2 gSorbent−1 (maximum theoretical CO2 uptake of 0.69 gCO2 gSorbent−1). In particular, the CO2 uptake performance of Ca@12Zr is promising, as it showed a relatively stable CO2 uptake over 20 cycles with only a small decrease from 0.66 gCO2 gSorbent−1 in the 1st cycle to 0.61 gCO2 gSorbent−1 (decrease of 8%) in the 20th cycle. For the sorbent with the highest zirconia content investigated (Ca@55Zr), very low CO2 uptakes of, respectively, 0.17 gCO2 gSorbent−1 and 0.04 gCO2 gSorbent−1 in the 1st and 20th cycle were observed. The low CO2 uptake of Ca-NP@55Zr was due to the large fraction of CO2-capture-inactive t-ZrO2 and CaZrO3 in this material as evidenced by XRD measurements (Fig. 3b). Turning to the core–shell structured sorbents, the most promising core–shell sorbent Ca–26Zr showed a significantly lower CO2 uptake in the 20th cycle (0.29 gCO2 gSorbent−1) compared to Ca@12Zr (0.61 gCO2 gSorbent−1). Further, the core–shell sorbent Ca–15Zr had a CO2 uptake of 0.27 gCO2 gSorbent−1 after 20 cycles, which is significantly lower than the CO2 uptake of its yolk–shell analogue with a similar ZrO2 content, i.e. Ca@16Zr (0.38 gCO2 gSorbent−1). Hence, a yolk–shell architecture yielded significantly better performing sorbents compared to the core–shell architecture (and the Ca-NP reference). This observation will be rationalized in the following sections.
First, we analyzed the effect of the yolk–shell and core–shell architectures on the kinetically-controlled regime. Based on the normalized CO2 uptake rate reaching a plateau, see Fig. 4d, we defined the end of the kinetically-controlled regime at t = 150 s for all sorbents. Within the kinetically-controlled regime in the 1st cycle, the yolk–shell sorbents reached significantly higher CO2 uptakes of 0.56 gCO2 gSorbent−1 (Ca@12Zr) and 0.51 gCO2 gSorbent−1 (Ca@16Zr) compared to the core–shell sorbents (0.3 gCO2 gSorbent−1 for Ca–15Zr and 0.28 gCO2 gSorbent−1 for Ca–26Zr) and Ca-NP (0.35 gCO2 gSorbent−1). Also, the maximum normalized CO2 uptake rate (Fig. 4d) achieved for the yolk–shell sorbents was significantly higher than for Ca–15Zr, Ca–26Zr and Ca-NP. In the 10th cycle, the CO2 uptake at the end of the kinetically-controlled regime was comparable for Ca@12Zr and Ca–26Zr, whereas Ca@16Zr, Ca–15Zr and Ca-NP showed a lower CO2 uptake. The maximum achieved CO2 uptake rate was also slightly higher for the yolk–shell sorbents than for the core–shell sorbents and the Ca-NP reference. In the kinetically-controlled regime in which the CO2 mass transfer resistance to reach the reaction surface is negligible, the reaction proceeds as a surface reaction. Hence, the CO2 uptake rate (dmCO2/dt) is proportional to the reaction surface area S0 and the reaction rate ks:
(2) |
As the reaction itself (i.e. the carbonation of CaO) has an intrinsic reaction rate ks,33,50,51 a higher maximum CO2 uptake rate implies a higher reaction surface area of CaO. Hence, we can conclude that a yolk–shell architecture improved the accessibility of CaO for reaction with CO2, i.e. provided sorbents with a higher reaction surface area compared to the core–shell sorbents and the Ca-NP reference. N2 physisorption measurements to further confirm this conclusion could not be conducted due to the small amount (<20 mg) of sorbent retained from a single synthesis.
The analysis of the kinetics in the diffusion-controlled regime, i.e. t > 150 s, will be discussed next. All sorbents exhibited a comparable CO2 uptake in the diffusion-controlled regime in the 1st cycle, see Fig. 4b. In the 10th cycle (Fig. 4c), the yolk–shell sorbents achieved a large part of their total CO2 uptake after 20 minutes of carbonation in the diffusion-controlled regime with an average normalized CO2 uptake rate (for 150 s < t < 1200 s) for Ca@12Zr and Ca@16Zr of 2.5 and 1.5 × 10−4 s−1, respectively. The CO2 uptake of the core–shell sorbents in this regime was significantly smaller, reaching an average CO2 uptake rate of 1.04 × 10−4 s−1 and 6.73 × 10−5 s−1 for Ca–15Zr and Ca–26Zr, respectively. The Ca-NP reference exhibited a CO2 uptake rate of 8.0 × 10−5 s−1 and its overall CO2 uptake in the 10th cycle was comparably low. The average CO2 uptake rates in the 20th cycle were consistent with those observed in the 10th cycle and scaled in the following order: Ca@12Zr (2.2 × 10−4 s−1) > Ca@16Zr (1.3 × 10−4 s−1) > Ca–15Zr (9.3 × 10−5 s−1) > Ca–26Zr (7.3 × 10−5 s−1) ≈ Ca-NP (6.7 × 10−5 s−1). Hence, for yolk–shell-structured materials also the CO2 uptake rates in the diffusion-controlled regime were improved compared to the core–shell sorbents and the Ca-NP reference. In particular, the yolk–shell-structured material Ca@16Zr compared favorably with the core–shell sorbent Ca–15Zr which possesses a very similar Zr content. According to the random pore model (RPM), which is widely used to describe the carbonation reaction of CaO, a higher CO2 uptake rate in the diffusion-controlled regime is indicative of a higher effective diffusivity of CO2.52 Hence, there is a strong indication that the yolk–shell structure improves the effective diffusivity of CO2 compared to core–shell architectures. It is worth noting that the accessibility and availability of CaO for reaction with CO2 may be affected by several factors, including the morphology of the sorbent as well as the structural properties, which will be discussed below.
After 10 cycles, the yolk–shell-structured sorbents Ca@12Zr (Fig. 5a) and Ca@16Zr (Fig. S5a†) exhibit Zr-containing shells with a skeletal appearance that encapsulate individual CaO nanoparticles. Compared to the unreacted sorbents, the shells have to some degree lost their shape and integrity, in particular for Ca@16Zr (see Fig. S4a† for comparison). The yolk–shell structure of the as-prepared sorbents is lost to some degree over cycling, resulting in shell structures with higher porosity. Nonetheless, SEM micrographs of Ca@12Zr (Fig. S5b†) confirm that after cycling the sorbent is still composed of individual CaO nanoparticles that are surrounded by a protective shell of dendritic appearance. Based on EDX-TEM analysis, the shells of the yolk–shell sorbents are Zr-containing. After 20 cycles (see Fig. S5d–h†), large agglomerates of yolk–shell particles (in this case Ca@12Zr) are observed. The morphology of the individual particles is therefore difficult to assess. Some of the larger dendritic structures observed for the cycled yolk–shell sorbents consist of smaller crystallites as shown in Fig. 5c. The high-resolution TEM micrograph of these smaller crystallites reveal plane distances that resemble the spacings of the [0 0 2] and [0 1 1] planes of CaZrO3.
On the other hand, electron microscopy-based analysis of the core–shell sorbent Ca–26Zr after 10 cycles shows large CaO agglomerates surrounded by a Zr-containing layer (Fig. 5b). These particles also exhibited Zr-containing dendritic structures covering the surface of the CaO particles. EDX-TEM shows that the Zr coating is in intimate contact with the CaO surface, indicative of the formation of the Ca–Zr mixed phase. In addition, Ca–15Zr (Fig. S5c†) showed enlarged (i.e. sintered) CaO particles with dendritic coating structures at the surface. Indeed, cycled Ca–15Zr does not feature a homogeneous coating around CaO particles anymore. To conclude, a yolk–shell-structured architecture protected the single CaO nanoparticles from sintering more effectively than a core–shell-structured morphology. We expect the ZrO2- (or CaZrO3-) shell to act as a physical barrier against sintering between single CaO nanoparticles, reducing in turn sintering,6,29,53 whereby the yolk–shell architecture seems to be more effective in reducing the interfacial area between the individual CaO nanoparticles. Further, the high porosity of the yolk–shell-structured morphology allowed for a fast transport of CO2 to the surface of the CaO particles. In contrast, for core–shell-type particles the intimate contact between zirconia and CaO led to the formation of a thick Zr-containing shell (likely CaZrO3, see below) that reduced the rate of CO2 uptake by creating diffusional resistance for CO2. The formation of CaZrO3 over cycling will be discussed below.
The XRD patterns of zirconia-stabilized yolk–shell and core–shell sorbents after 10 and 20 cycles are shown in Fig. 5d and e, respectively. Using Rietveld analysis (Table S1†), we estimate the fraction of CaZrO3 in the different materials and define the parameter α, i.e. the molar ratio of CaZrO3 to the total molar amount of CaO in the as-synthesized sorbent (see Table 2):
(3) |
Sorbent | Ratio as-prepared | Ratio after 10 cycles | Ratio after 20 cycles |
---|---|---|---|
Ca–15Zr | 0.07 | 0.08 | 0.08 |
Ca–26Zr | 0.21 | 0.20 | 0.24 |
Ca@16Zr | <0.01 | 0.03 | 0.09 |
Ca@12Zr | <0.01 | 0.04 | 0.07 |
In the core–shell sorbent Ca–15Zr the ratio α increased slightly from 0.07 (as-prepared) to 0.08 after 20 cycles. This small increase is however within the accuracy of the experiment. Therefore, we conclude that in Ca–15Zr most, if not all, of the CaZrO3 phase already formed after the initial calcination step. For Ca–26Zr, α was 0.21 (as-prepared), 0.20 (10 cycles) and 0.24 (20 cycles). Hence, also for Ca–26Zr most of the Zr is in the form of CaZrO3 after the initial calcination step. The intimate contact between Ca and Zr-containing phases in the core–shell-based architectures leads to the rapid formation of a CO2-capture-inactive CaZrO3 layer around CaO particles (as also evidenced by TEM analysis). Hence, the deactivation of the core–shell sorbents over repeated carbonation–calcination cycles is not driven by the additional formation of CaZrO3. Instead, it is very likely that deactivation is due to the sintering of the individual CaO particles and an increase in the density (and thickness) of the CaZrO3 layer itself (as a result of sintering). This reduces the size of the active surface area and decreases the effective diffusivity of CO2.
The XRD patterns of the yolk–shell sorbents after 10 and 20 cycles (Fig. 5d and e) show the presence of CaZrO3. The values of α after 10 cycles are 0.03 and 0.04 for Ca@16Zr and Ca@12Zr, respectively. Hence, for yolk–shell-structured sorbents the fraction of CaZrO3 increased significantly with cycle number (the as-prepared materials have α < 0.01). After 20 cycles, α increased further to 0.09 for Ca@16Zr and 0.07 for Ca@12Zr, indicating a continuing formation of CaZrO3 during cycling. In addition, a small peak due to t-ZrO2 (approximately 1 wt%) was observed for Ca@16Zr (20 cycles). This continuing formation of CO2-capture-inactive CaZrO3 for the yolk–shell-structured materials, in particular for Ca@12Zr, is one explanation for their decreasing CO2 uptake with cycle number (∼40% decrease from the 1st to 20th cycle for Ca@16Zr), yet sintering of CaO nanoparticles must also contribute to the decay in the CO2 uptake for Ca@16Zr. For Ca@12Zr on the other hand, the formation of CaZrO3 would result in a theoretical loss of approximately 6 wt% of the initially available CaO. At the same time, Ca@12Zr shows a drop in the cyclic CO2 uptake from the 1st to the 20th cycle of ∼8%, i.e. for Ca@12Zr the loss of reactive CaO due to CaZrO3 formation accounts to ∼75% of the decay in CO2 uptake. Hence, the yolk–shell-structured architecture of Ca@12Zr largely prevented sintering of the CaO nanoparticles, making the loss of reactive CaO through the formation of CaZrO3 the dominating deactivation mechanism. The formation of CaZrO3 is likely a result of the partial loss of the yolk–shell-type structure during cycling, as evidenced in our TEM analysis (Fig. 5a). Hence, the sacrificial template can reduce the amount of CaZrO3 formation during the first cycles, but cannot prevent entirely the contact between CaO and ZrO2.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nr04492g |
This journal is © The Royal Society of Chemistry 2022 |