Thermal decomposition of dolomite under CO2: insights from TGA and in situ XRD analysis

Thermal decomposition of dolomite under CO2: Insights from1 TGA and in-situ XRD analysis2 Jose Manuel Valverdea∗, Antonio Perejonb, Santiago Medinac, Luis A. Perez-Maquedad 3 aFaculty of Physics. University of Seville. Avenida Reina Mercedes s/n, 41012 Sevilla, Spain4 bFaculty of Chemistry. University of Seville. C/5 Profesor Garcia Gonzalez s/n, 41012 Sevilla, Spain6 cX-Ray Laboratory (CITIUS), University of Seville,7 Avenida Reina Mercedes, 4B. 41012 Sevilla, Spain8 dInstituto de Ciencia de Materiales de Sevilla (C.S.I.C.-Univ.9 Seville), Americo Vespucio 49, 41092 Sevilla, Spain10 ∗Corresponding Author: Fax: +34 95423 9434; Tel: +34 95455 0960; E-mail: [email protected]11 1 Abstract12 Thermal decomposition of dolomite in the presence of CO2in the calcination environment13 is investigated by means of in-situ X-ray diffraction (XRD) and thermogravimetric analy-14 sis (TGA). In-situ XRD results suggest that dolomite decomposes directly at a temperature15 around 700◦C into MgO and CaO. Immediate carbonation of the nascent CaO crystals leads16 to the formation of calcite as intermediate product of decomposition. Subsequently, decarbon-17 ation of this poorly crystalline calcite occurs when the reaction is thermodynamically favorable18 and sufficiently fast at a temperature depending on the CO2partial pressure in the calcina-19 tion atmosphere. Decarbonation of this dolomitic calcite occurs at a lower temperature than20 limestone decarbonation due to the relatively lower crystallinity of the former. Full decom-21 position of dolomite leads also to a relatively low crystalline CaO, which exhibits a relatively22 high reactivity as compared to limestone derived CaO. At CO2capture conditions in the23 Calcium-Looping (CaL) process, MgO grains remain inert yet favor the carbonation reactivity24 of dolomitic CaO specially in the solid-state diffusion controlled phase. The fundamental mech-25 anism that drives the crystallographic transformation of dolomite in the presence of CO2is26 thus responsible for its fast calcination kinetics and the high carbonation reactivity of dolomitic27 CaO, which makes natural dolomite a potentially advantageous alternative to limestone for28 CO2capture in the CaL technology as well as SO2in-situ removal in oxy-combustion fluidized29 bed reactors.30 2 I. INTRODUCTION31 Recent studies on the conservation of cultural heritage have revealed that the use of32 dolomitic lime (MgO·CaO) derived from calcination of dolomite (CaMg(CO3)2) as binder in33 mortars was a generalized practice in ancient masonry [1]. Dolomitic lime mortars had higher34 strength and resistance to deterioration than lime (CaO) mortars derived from limestone35 (CaCO3) calcination [1–3] if their preparation followed certain specific rules. Archeological36 records show that dolomite thermal decomposition in ancient ovens was carried out under37 high CO2concentration at relatively low temperatures, which highlights the extraordinary38 relevance of calcination conditions on the structural and physicochemical characteristics of39 the calcium and magnesium oxides produced during thermal decomposition. Nowadays,40 the thermal decomposition of dolomite under CO2remains at the root of a number of41 industrial processes aimed at using Ca and Mg oxides and carbonates as raw materials for42 the production of portland cement and Mg-based products, respectively [4].43 Natural limestone and dolomite have been used commercially in the energy sector indus-44 try since the 1970s for SO2capture in fluidized bed coal combustors [5, 6]. CaO resulting45 from in-situ thermal decomposition undergoes sulphation (CaO+SO2+1 2O2→CaSO4) at46 temperatures between 800 and 900◦C at a rate that depends on the partial pressure of CO2 47 in the calcination environment. During fluidized bed atmospheric combustion (with typ-48 ically 20% excess air), the partial pressure of the generated CO2is about 0.15 atm. On49 the other hand, in the close to commercial oxy-combustion technology for CO2capture, the50 concentration of CO2in the flue gas is enriched up to ∼95% by burning the fuel with a51 mixture of nearly pure oxygen and a CO2rich recycled flue gas at combustor temperatures52 typically between 850◦C and 950◦C. Thermal decomposition of limestone under high CO2 53 3 partial pressure (P≃1 atm) occurs sufficiently fast only at temperatures above 930◦C [7, 8].54 Thus, SO2retention in the oxy-combustion reactor would occur by the less efficient direct55 sulphation reaction (CaCO3+ SO2+1/2O2→CaSO4+ CO2) [9, 10]. Another promising56 CO2capture process, which has been identified as a cost-effective approach, is the recently57 emerged Ca-looping (CaL) technology based on the multicyclic calcination/carbonation of58 CaCO3using natural limestone as raw material [11–13]. In this post-combustion capture59 technology, the combustor effluent gas at atmospheric pressure with typically 15% vol con-60 centration of CO2is used for fluidization of CaO particles in a gas-solid reactor (carbonator)61 wherein CO2is retained by carbonation of CaO at optimum temperatures around 650◦C.62 The carbonated solids are driven into a second fluidized bed reactor (calciner) in which63 CaCO3is regenerated by calcination under CO2at high partial pressure. In order to obtain64 a stream of highly concentrated CO2from the calciner and to rise its temperature up to high65 enough values, fuel is burned in-situ in the calciner by oxy-combustion [14, 15]. Since natural66 limestone decomposition under CO2partial pressures close to P= 1 atm is extremely slow67 near to equilibrium conditions (T≃895◦C) [7, 8], the calciner temperature in practice has68 to be increased up to T≃930◦C in order to achieve a high calcination efficiency, which im-69 poses an important energy penalty to the technology [12, 13, 16]. Furthermore, the decay of70 CaO carbonation reactivity with the number of cycles must be compensated by periodically71 feeding the calciner with a makeup flow of fresh limestone while a fraction of the circulating72 sorbent is purged.73 Thermogravimetric analysis (TGA) indicates that the use of dolomite as CaO precursor in74 a calcination environment of high CO2partial pressure would allow reducing the calcination75 temperature as compared to limestone [17]. The use of dolomite would thus presumably76 improve the in-situ SO2capture performance in oxyfuel combustors as well as the calciner77 4 efficiency in the CaL technology. Moreover, the decay of carbonation reactivity of dolomitic78 CaO with the number of calcination/carbonation cycles at CaL conditions is mitigated [17–79 21]. The main purpose of the present work is to understand the fundamental mechanisms80 that drives this behavior. To this end, the calcination reaction of natural dolomite, as81 affected by the CO2partial pressure in the calciner atmosphere, will be investigated by82 means of in-situ X-ray diffraction (XRD) analysis coupled with TGA at similar conditions.83 II. MECHANISM OF DOLOMITE THERMAL DECOMPOSITION84 TGA and differential thermal analysis (DTA) have long demonstrated that the CO2 85 partial pressure in the calcination environment determines essentially the mechanism of86 dolomite thermal decomposition [4, 22–24]. At CO2partial pressures below P≃0.1 atm87 decomposition of dolomite into CaO and MgO is observed to occur apparently via a single88 stage reaction:89 CaMg(CO3)2→CaO + MgO + 2CO2(1) In-situ XRD analysis of calcination at low absolute pressures (<10−6bar) and low temper-90 atures (490 - 600◦C) [25] as well as in-situ TEM observations by exposure of dolomite to91 an e-beam under vacuum [26, 27] have revealed that pure CaO and MgO crystals nucleate92 from a metastable FCC solid solution Mg0.5Ca0.5O, which stems as a direct result of dolomite93 breakdown by a topotactic process. Transient Mg-rich CaO (CaϵMg1−ϵO) and Ca-rich MgO94 crystals (MgδCa1−δO, with ϵ∼δ∼0.1) are identified as resulting from the equimolar solid95 solution [25, 27], which is thermodynamically unstable because of the mismatch of cation96 sizes (the equilibrium solid solution contains at most 8 mol% Ca in MgO and 22 mol% Mg97 5 in CaO [26]). Pure CaO and MgO crystals form afterwards by diffusion of Ca2+ and Mg2+ 98 ions across the phase boundary between the mixed oxide crystallites.99 Decomposition under CO2at partial pressures higher than about 0.1 atm consists of a100 two-stage process. As temperature is increased, the first and simultaneously emerging prod-101 ucts of dolomite decomposition are calcite (CaCO3) and periclase (MgO) whereas CaCO3 102 decomposition takes place subsequently at a higher temperature to form CaO:103 CaMg(CO3)2→CaCO3+ MgO + CO2(2) CaCO3→CaO + CO2(3) Experimental TG and DTA results show that the so-called half-decomposition of dolomite104 (reaction 2) is not fundamentally affected by the CO2partial pressure Pwhereas decom-105 position of CaCO3(reaction 3) is shifted towards higher temperatures as the CO2partial106 pressure Pis increased [23, 24] according to the thermodynamic equilibrium of the CaCO3 107 calcination reversible reaction [8, 28]:108 P(atm) ≈4.083 ×107exp(−20474/Teq) (4) where Teq is the temperature for the reaction to be at equilibrium 3.109 The physicochemical mechanism that governs dolomite half-decomposition under the110 presence of CO2remains a controverted issue [2, 3, 22, 27, 29, 30]. Traditionally, it was111 believed that calcite and magnesite (MgCO3) microdomains were initially developed by the112 counter current diffusion of Mg2+ and Ca2+ ions [29]:113 CaMg(CO3)2→CaCO3+ MgCO3(5) 6 Half-decomposition would thus obey to the thermodynamical instability of MgCO3at rel-114 atively lower temperatures than CaCO3[18, 31]. However, the equilibrium temperature of115 magnesite decomposition under pure CO2at atmospheric pressure would be around 400◦C116 [18]. In contrast, half-decomposition is generally observed at higher temperatures (around117 700◦C) [32] in non-isothermal tests usually carried out at low heating rates (∼10◦C/min),118 which could be nevertheless explained from the existence of a certain activation energy for119 ion diffusion [22, 29]. In fact, half-decomposition is observed at comparable temperatures120 with that at which magnesite itself decomposes (T∼400 ◦C) for dolomite samples previously121 subjected to prolonged and vigorous grinding that would arguably promote the migration122 of Mg2+ and Ca2+ ions [29, 33].123 Other decomposition mechanisms proposed contemplate the formation of calcite/magnesite124 (CaCO3(1-n)MgCO3) [34] and calcite/MgO ((1-x)CaCO3(1-y)MgO) intermediates near to125 equilibrium as suggested from ex-situ XRD analysis [3]. An alternative view is that MgO126 and CaO crystals are directly generated during the breakdown of the dolomite structure,127 which is immediately followed by the direct carbonation of the nascent CaO crystals at tem-128 peratures below the thermal decomposition of calcite (Teq in Eq. 4) [4, 22–24]. Accordingly,129 a recent study based on ex-situ XRD/2D-XRD analysis and FESEM/TEM observations [27]130 suggests that dolomite decomposition in air follows a similar mechanistic path to that found131 under vacuum [27]. Yet, the occurrence of dolomite decomposition in air at higher tem-132 peratures (650-700◦C) than in vacuum would enhance ion diffusion [27]. Thus, Mg-calcite133 crystals would be initially formed by carbonation of the nascent MgδCa1−δO (δ≤0.1) phase134 with CO2present in the air and/or released upon decomposition [27]. As the temperature135 is further increased, Mg2+ ions diffuse out of the Mg-calcite structure, after which calcite136 would decompose at T > 750◦C to yield pure CaO and MgO as final products [27].137 7 The present study is focused on analyzing the structural evolution of dolomite under138 atmospheric pressure as it suffers thermal decomposition in the presence of CO2at partial139 pressures ranging between 0 and 1 atm. To this end, in-situ XRD analysis coupled with140 Rietveld refinement has been performed. Additional TG analysis of calcination under similar141 conditions was carried out, which has been useful also to accurately measure in-situ the142 carbonation reactivity of the CaO stemming as final product of decomposition. Further tests143 on limestone decomposition under the same conditions have been performed whose results144 highlight the distinct role of the presence of CO2on decomposition of both materials.145 III. MATERIALS AND METHODS146 In our study we have used a powdered natural dolomite from Bueres quarry (Asturias,147 Spain). Raw dolomite was sieved (opening size 45 µm) in order to avoid decrepitation148 phenomena during thermal decomposition, which may be significant for particles of size149 above ∼100 µm as reported elsewhere [31, 35]. The average particle size of the sieved150 powder is dp≃35 µm (volume weighted mean) as measured using a Malvern Mastersizer151 2000 instrument by laser diffractometry. Such small particle size allows us also neglecting152 intra-particle diffusion resistance effects on the reaction rate that would be relevant only for153 particles of size larger than 300 µm [36, 37]. The major mineral phase identified by XRD154 analysis at ambient temperature is CaMg(CO3)2(dolomite) with a minor CaCO3calcite155 phase. Phase quantification by Rietveld refinement yields a 94.4wt% of dolomite (SD=0.3%)156 the rest being calcite. For comparison, some experiments were carried out using a natural157 limestone of high purity (99.6wt% CaCO3) from Matagallar quarry (Pedrera, Spain) with158 average particle size of 9.5 µm.159 TGA tests were performed using a Q5000IR TG analyzer (TA Instruments) equipped with160 8 a high sensitivity balance (<0.1 µg) characterized by a minimum baseline dynamic drift (<10161 µg). Samples of small and fixed mass (10 mg) were tested in order to avoid undesired effects162 due to CO2diffusion resistance through the powder bulk, which would become relevant in163 this type of analysis for sample masses above ∼40 mg [38]. Heat transfer phenomena is164 also minimized by placing the sample inside a SiC enclosure heated by four symmetrically165 placed IR halogen lamps, which ensures consistent and uniform heating. Active water-166 cooling of the surrounding furnace body provides an efficient heat-sink and favors accurate167 temperature and heating rate control. The temperature is registered by a thermocouple168 positioned underneath and close to the sample. Quick heating of the gas up to the desired169 temperature is achieved by using a small gas flow rate (100 cm3min−1). At this small flow170 rate the gas velocity has no influence on the reaction rate [39]. In the tests reported in the171 present study, the sample is subjected to a N2/CO2controlled gas mixture at atmospheric172 pressure and the temperature is increased from ambient temperature at 10◦C/min up to the173 target calcination temperature, which is kept constant for 1h, after which the temperature174 is quickly decreased down to 650◦C (300◦C/min rate) and the sample is subjected to a gas175 mixture of 15%CO2/85%N2vol/vol (typical ratio of post-combustion flue gas) in order to176 test in-situ the carbonation reactivity of the CaO product that results immediately after177 calcination.178 In-situ XRD analysis has been carried out using a powder diffractometer (Bruker D8179 Advance) equipped with a high temperature chamber (Anton Paar XRK 900) and a fast180 response/high sensitivity detector (Bruker Vantec 1) with radial Soller slits, which allows181 us investigating the evolution of the solid crystal structure as calcination progresses. The182 reactor chamber is specifically designed to ensure its homogeneous filling with the reaction183 gas at the desired temperature and avoid temperature gradients across the sample in order184 9 amorphous or very poorly crystalline as not to yield visible diffraction peaks. It becomes323 also apparent from the breadths of Bragg reflection peaks that the coherent crystal length324 (crystallite size) of the final CaO and MgO products is larger when dolomite is calcined325 under CO2. The results of a quantitative analysis on this issue will be shown below.326 XR diffractograms obtained for dolomite and limestone samples calcined in pure CO2 327 immediately at the end of the temperature ramp (T= 925◦C) and after 1h at 925◦C are328 superposed in Fig. 7. The notably less sharpness of the reflection peaks for calcite and lime329 obtained from dolomite (Fig. 7a) reveals a lower degree of crystallinity for these products330 of decomposition. Another interesting feature illustrated in the inset of Fig. 7b is that331 the peaks of the CaO phase obtained from dolomite decomposition are slightly shifted with332 respect to the peaks obtained for CaO derived from decomposition of limestone. As will333 be seen, a Rietveld refinement analysis suggests that this relative displacement is possibly334 caused by the presence of Mg impurities in the structure of CaO derived from dolomite.335 A quantitative phase analysis of the registered XRD scans has been carried out by means336 of the Rietveld method [47] and using the software TOPAS 4.2 (Bruker) [48]. Zero error337 (2θ), sample displacement, absorption (1/cm), lattice parameters of the phases as well as338 site occupancy of Mg impurities in calcite and lime and Ca impurities in periclase were339 allowed to vary to provide the best fits to the experimental diffractograms. Dilation of the340 structure as affected by temperature is also considered in the analysis. The background was341 fitted by a fourth-order Chebychev polynomial. Lorentz and polarization geometric factors342 for the measurement configuration were used. For the fittings to be as accurate as possible343 the resulting value of the GOF (Goodness-of-fit) parameter should be above unity and as344 close as possible to it [47]. Furthermore, the residual factors (Rwp and RBragg) have to be345 small for the measurement configuration used [47]. In general, acceptable fitting indicators346 16 were obtained in our analysis (GOF ∼1-1.5, Rwp∼10, RBragg∼1).347 The evolution with temperature of the wt% for the CaMg(CO3)2and MgO phases during348 dolomite decomposition as depending on the CO2vol% in the calcination environment is349 shown in Fig. 8. As may be seen, the drop of CaMg(CO3)2wt% is accompanied by a increase350 of the wt% of MgO that stems as a direct product of half decomposition and remains inert to351 the presence of CO2. In accordance with the TGA results, dolomite decomposition is started352 in the absence of CO2at relatively lower temperatures whereas there is not a remarkable353 dependence of the decomposition temperature on the concentration of CO2present. Figure 9354 shows a comparison of the evolution with temperature of CaCO3and CaO wt% for dolomite355 calcination under pure CO2(Fig. 9a) and pure N2(Fig. 9b). Note that the small amount356 of CaCO3prior to half-decomposition (about 6%) is due to calcite originally present in357 the dolomite samples. Half-decomposition, as inferred from the emergence of MgO Bragg358 reflection peaks, is initiated at 700◦C for decomposition under CO2and at 650◦C in the case359 of decomposition under N2. As seen in Fig. 9a half-decomposition under CO2leads to a360 remarkable rise of the CaCO3wt% that results as a product of decomposition together with361 MgO. On the other hand, the products of half-decomposition under N2are MgO and CaO362 while the CaCO3initially present in the sample drops abruptly to zero as decarbonation of363 calcite at the temperature of decomposition becomes thermodynamically favorable and fast364 under N2.365 Interestingly, it is seen in Figs. 9a-9b that, after the initiation of decomposition, the366 CaCO3and CaO wt% for calcination under CO2and N2, respectively, follow compara-367 ble trends reaching both a similar maximum value (around 63%) at the same temperature368 (750◦C). This observation suggests that the mechanism that rules dolomite decomposition in369 both cases basically consists of the direct breakdown of the dolomite structure into MgO and370 17 CaO with the immediate recarbonation of the nascent CaO crystals when calcination is car-371 ried out under CO2. Thus, CaO would remain carbonated under CO2until the temperature372 reaches a sufficiently high value for decarbonation to be thermodynamically favorable and373 fast enough. As seen in Fig. 9a this occurs for temperatures above 925◦C when calcination374 is carried out under pure CO2. Remarkably, best Rietveld fittings for decomposition under375 CO2suggest that the product of the reaction is pure calcite without Mg impurities. This376 contrasts with previous ex-situ XR studies indicating the formation of Mg-calcite [27] or a377 solid solution of MgO in calcite [3]. It is also noticeable that, for decomposition under N2,378 MgO and CaO reflection peaks start to become visible at a temperature (650◦C) which is379 roughly 100◦C over the temperature at which TGA tests indicate the initiation of weight loss380 (Fig. 3a). This suggests that the nascent oxides are poorly crystalline or even amorphous381 as inferred from other works [27]. In contrast, for decomposition under CO2, MgO reflection382 peaks and the intensification of calcite peaks occur at a temperature (700◦C) close to that383 at which weight loss is started according to TGA results (Fig. 3a), which indicates that the384 nascent oxides are already in a well defined crystalline form if decomposition takes place385 under CO2.386 Taking into account the presence of CaCO3by a small amount (≃6%) initially in the387 dolomite powder, the theoretical wt% for the final CaO and MgO products after decompo-388 sition would be close to 39% and 61%, respectively if pure CaO/MgO phases are assumed.389 On the other hand, the Rietveld analysis reveals that the wt% of CaO and MgO in the390 fully decomposed samples after the 1h isotherm at 925◦C are in the ranges 37-38% and391 62-63%, respectively. This slight difference between the expected and obtained wt% can be392 explained from the presence of Mg impurities in the CaO structure. Figure 10 illustrates an393 example of the best Rietveld fits to the XRD pattern measured for dolomite decomposition394 18 in pure CO2(1h at 925◦C after the end of the temperature ramp) either assuming pure CaO395 and MgO phases (Fig. 10a) or allowing for the presence of Ca and Mg impurities in the396 MgO and CaO phases, respectively. A considerably better fit is obtained in the latter case397 (fitting indicators are given in the figure caption). Thus, the shift of the CaO peaks with398 respect to the peaks of pure CaO derived from limestone above observed (inset of Fig. 7b)399 can be accounted for by allowing a certain occupation of Mg atoms in the CaO structure.400 Arguably, the presence of Mg impurities in CaO could contribute to the enhancement of the401 solid-state diffusion controlled carbonation of dolomitic CaO inferred from the TGA tests.402 On the other hand, best Rietveld fits indicate that the MgO phase does not contain Ca403 impurities.404 B. Evolution of Crystallite Size405 Crystallite size for the main phases involved in dolomite decomposition (dolomite406 CaMg(CO3)2, calcite CaCO3, lime CaO and periclase MgO) has been calculated from407 the XR patterns by means of the Le Bail method [49] and using the software TOPAS 4.2408 (Bruker) [48]. Data are plotted in Fig. 11 as depending on the CO2vol% in the calcination409 atmosphere. Generally, the values obtained for the fitting indicators such as the GOF and410 residual factors (Rwp and RBragg) are indicative of reliable results (GOF∼1-1.5, Rwp≤8,411 RBragg≤0.5) [47]. Data reproducibility is demonstrated in Fig. 11a, where the evolution412 of crystallite size with temperature is plotted for independent calcination tests carried out413 under CO2and using diverse samples. Data on CaO and MgO crystallite size are consistent414 in order of magnitude with data reported elsewhere from ex-situ XR analysis [27, 50].415 An interesting advantage of the in-situ XR analysis over ex-situ studies is that it yields416 noninvasively the evolution of crystallite size during calcination. As may be seen in Fig. 11b,417 19 CaMg(CO3)2crystallite size remains roughly constant independently of the CO2partial418 pressure before decomposition. Only when the temperature approaches T≃700◦C, the419 crystallite size of CaMg(CO3)2declines as its reflection peaks disappear and calcite (or CaO420 under N2) and MgO reflection peaks start to emerge. As temperature is increased above421 700◦C, CaCO3crystallite size, which may be accurately measured in the interval between 700422 and 925◦C for decomposition under pure CO2, grows with temperature similarly to CaO and423 MgO crystallite size. This result is consistent with the argument that the direct products424 of dolomite decomposition are MgO and CaO, the latter being immediately carbonated425 when calcination is carried out under high CO2concentration. It can be noticed also that426 the size of CaO crystallites, which emerge at 925◦C for calcination under pure CO2(Fig.427 11c), is larger than that of the CaO crystallites that emerged at 700◦C for calcination at428 lower CO2concentrations. As demonstrated from dolomite calcination under vacuum by429 in-situ TEM [27], the nascent crystals would grow by oriented aggregation and sintering.430 Initially, attractive forces between the nanosized crystals due to the universal van der Waals431 interaction would lead to their aggregation. Since these attractive forces are enhanced by432 the presence of adsorbed CO2molecules [51] on the solid surfaces, crystal coarsening would433 be expectedly promoted when calcination is carried out under high CO2partial pressure.434 Accordingly, it can be seen in Fig. 11 that the rate of growth of CaO and MgO crystals435 with temperature is notably increased with the concentration of CO2in the calcination436 environment.437 Figure 12 shows data on the size of CaO and MgO crystallites calculated from the in-situ438 XRD patterns registered during the 1 h isotherm at 925◦C. As can be observed, the size439 of the crystallites is only slightly increased with time regardless of the CO2concentration440 in the calcination atmosphere. Thus, the final size of crystallites is mainly determined441 20 by their growth during the temperature ramp. Crystallite coarsening at 925◦C far from442 the equilibrium temperature would be mainly driven by sintering [41], which essentially443 depends on the Tamman temperature. The Tammann temperature of a ceramic material444 (Tt) marks the initiation of notable sintering by diffusion of chemical species across the445 crystalline lattice and corresponds to about half the melting temperature in K (Tt≃1170◦C446 for CaO and Tt≃1276◦C for MgO) [52, 53]. At lower temperatures, as in the present case,447 sintering would be governed by surface diffusion albeit lattice diffusion could be promoted448 by the presence of defects and impurities in the crystal structure [54]. Our results indicate449 that the major influence of CO2on the growth of CaO and MgO crystallites stemming from450 dolomite decomposition is attributable to the promotion of nanosized crystal aggregation in451 the initial nucleation phase. Note also that the rate of growth of crystallite size is similar452 for both CaO and MgO.453 The evolution of calcite and CaO crystallite size during calcination of dolomite under454 pure CO2is shown in Fig. 13(a-b) in comparison to data obtained for calcination of lime-455 stone under the same conditions. Note that the size of CaCO3crystallites in limestone456 before decomposition (L∼100 nm) is comparable to the size of dolomite crystallites be-457 fore decomposition (Fig. 11a). On the other hand, CaCO3crystallites that nucleate after458 half-decomposition of dolomite are notably smaller (L∼50 nm) and increase in size with459 temperature as discussed above. CaO appears first for decomposition of both materials at460 925◦C with a similar crystallite size (L∼50 nm). However, the crystallites of CaO derived461 from limestone experience a marked increase during the initial stages of calcination at 925◦C462 up to reach a value of around 100 nm whereas the growth of CaO crystallites for dolomite is463 hindered (Fig. 11b). The enhanced growth of CaO crystallites during limestone calcination464 would be arguably caused by aggregation of the nascent CaO nanocrystals due to surface465 21 attractive forces between them [41]. Presumably, these short ranged van der Waals forces466 would be mitigated in the case of dolomite (as compared to limestone) by the interposi-467 tion of MgO grains in between the CaO nanocrystals. It is also noticeable that the size of468 dolomite derived CaCO3crystallites is, at the end of the temperature ramp, L≃70 −80469 nm (Fig. 13a) which is considerably smaller than the size of CaCO3crystallites in the nat-470 ural limestone before decomposition (L≃100 −120 nm, Fig. 13a). This would explain471 why decarbonation of CaCO3in the second decomposition stage of dolomite occurs at lower472 temperatures than CaCO3decomposition in limestone as seen above from the TGA tests473 (Fig. 3). An analogous effect is seen when limestone is mechanically milled, which reduces474 the size of CaCO3crystallites [55]. Generally, a decrease of the crystallite size leads to a475 higher reactivity of the material, which reduces the nucleation period that usually precedes476 gas-solid reactions and accelerates decomposition [41]. Data on the size of CaO crystallites477 obtained from decomposition of both materials under N2are plotted in Fig. 13(c-d). As seen478 for decomposition under CO2, the crystallinity of CaO derived from dolomite is considerably479 reduced as compared to limestone derived CaO.480 VII. CRYSTAL STRUCTURE, SINTERING AND CAO REACTIVITY481 Previous works on limestone decomposition [8, 40, 56] demonstrate that the enhanced482 growth of CaO crystallite size when calcination is carried out at high temperatures and high483 CO2partial pressure leads to a marked decrease of the CaO surface area which would be484 available for reaction-controlled carbonation. Thus, the notably smaller size of CaO crystal-485 lites stemming from dolomite decomposition would explain their higher reactivity towards486 carbonation demonstrated by the TGA results above shown (Fig. 5) as compared to the487 carbonation reactivity of CaO obtained from limestone. Figure 14a shows that the final488 22 size of CaO and MgO crystallites derived from dolomite decomposition is correlated to the489 BET surface area Smeasured for the calcined samples according to a power law L∝Sλ 490 with a similar exponent λ∼ −0.5 for both CaO and MgO. This result questions the gen-491 erally accepted role of MgO as a thermally stable support to mitigate CaO sintering during492 multiple calcination/carbonation cycles of dolomite [53]. Dolomite samples subjected to493 calcination/carbonation cycles exhibit sintering of the CaO grains and decay of CaO car-494 bonation reactivity as the CaCO3/CaO transformation is repeated in each cycle. As CaO495 grains, MgO grains suffer significant sintering during decomposition as observed in the SEM496 pictures of the calcined samples (Fig. 1). Yet, MgO grains are inert during subsequent497 calcination/carbonation cycles and their size remain essentially unchanged [17]. It may be498 thus argued that, according to our results, most of CaO sintering would occur along multiple499 calcination/carbonation cycles by enhanced aggregation during the CaCO3/CaO transfor-500 mation, which would be enhanced by the presence of CO2in the calcination environment501 at high partial pressure. The main effect of the inert MgO grains would be thus to reduce502 surface forces between the nascent CaO crystals, which would mitigate subsequent CaO503 sintering in each calcination stage. As seen in Fig. 14b, CaO conversion in the kinetically504 controlled phase Xrscales proportionally to the BET surface area for CaO derived from505 dolomite calcination under diverse CO2partial pressures, which leads to a very low particle506 porosity and small surface area (S < 10 m2/g). On the other hand, conversion of CaO507 derived from the sample calcined under N2and with a relatively higher porosity, is lower508 than expected from this linear law (Xr≃0.56 for S≃28 m2/g). In this case, a further509 limiting factor for fast carbonation would be the diffusion of CO2through the pores of the510 particles [57].511 23 VIII. CONCLUSIONS512 A main conclusion of the present work is that the fundamental mechanism of dolomite513 decomposition under CO2determines essentially the carbonation reactivity of the generated514 CaO. Our results indicate that dolomite suffers a direct breakdown into MgO ad CaO at515 temperatures around 700◦C while the nascent CaO crystals are immediately carbonated.516 CaCO3remains stable until a temperature is reached for decarbonation to be thermody-517 namically favorable and sufficiently fast, which depends on the CO2partial pressure in the518 calcination environment. CaCO3crystallites derived from this carbonation have a small size519 as compared to CaCO3crystallites in limestone, which explains their higher reactivity to-520 wards decarbonation. Thus, the CaCO3/CaO transformation occurs at lower temperatures521 for dolomite. This behavior may be useful for applications requiring the generation of CaO522 by calcination under CO2such as the Calcium-Looping technology for CO2capture or in-situ523 SO2removal in oxy-combustion reactors. The use of dolomite instead of limestone in the524 CaL technology as CaO precursor would allow reducing the calcination temperature thus525 minimizing the energy penalty. Moreover, the CaO resulting from dolomite decomposition526 show a substantially higher reactivity towards carbonation as compared to limestone derived527 CaO. CaO reactivity in the reaction controlled phase is directly correlated to CaO crystal-528 lite size. A further important observation in our work is that carbonation reactivity in the529 solid-state diffusion controlled phase is remarkably enhanced for CaO derived from dolomite530 in the presence of CO2. The use of dolomite would therefore allow increasing significantly531 the carbonation efficiency by prolonging the residence time of the solids in the carbonator532 reactor.533 24 IX. ACKNOWLEDGEMENTS534 This work was supported by the Spanish Government Agency Ministerio de Economia y535 Competitividad (contracts CTQ2014-52763-C2-1-R and CTQ2014-52763-C2-2-R) and An-536 dalusian Regional Government Junta de Andalucia (contract FQM-5735). We gratefully537 acknowledge the X-ray service of the Innovation, Technology and Research Center of the538 University of Seville (CITIUS).539 [1] Seeley, N. J. Journal of Architectural Conservation 2000, 6, 21 - 29.540 [2] Beruto, D.; Vecchiattini, R.; Giordani, M. Thermochimica Acta 2003, 404, 25–33.541 [3] Beruto, D.; Vecchiattini, R.; Giordani, M. Thermochimica Acta 2003, 405, 183 - 194.542 [4] Caceres, P.; Attiogbe, E. Minerals Engineering 1997, 10, 1165 - 1176.543 [5] O’Neill, E.; Keairns, D.; Kittle, W. Thermochimica Acta 1976, 14, 209 - 220.544 [6] Ulerich, N.; O’Neill, E.; Keairns, D. Thermochimica Acta 1978, 26, 269 - 282.545 [7] Hyatt, E. P.; Cutler, I. B.; Wadsworth, M. E. Journal of the American Ceramic Society546 1958, 41, 70–74.547 [8] Valverde, J. M.; Sanchez-Jimenez, P. E.; Perez-Maqueda, L. A. The Journal of Physical548 Chemistry C 2015, 119, 1623-1641.549 [9] Garca-Labiano, F.; Rufas, A.; de Diego, L. F.; de las Obras-Loscertales, M.; Gayn, P.;550 Abad, A.; Adanez, J. Fuel 2011, 90, 3100 - 3108.551 [10] de Diego, L.; Rufas, A.; Garca-Labiano, F.; de las Obras-Loscertales, M.; Abad, A.; Gayn, P.;552 Adanez, J. Fuel 2013, 114, 106 - 113.553 25 4 8 12 16 Deriv. Weight (%/min) 650 700 750 800 850 900 Temperature (°C) 50 60 70 80 90 100 110 Weight (%) 65 75 85 95 105 Time (min) Universal V4.7A TA Instruments Limestone (CO2) Dolomite (CO2) Limestone (N2) Dolomite (N2) Limestone (N 2 ) Dolomite (N 2 ) 2 4 6 8 Deriv. Weight (%/min) 450 500 550 600 650 700 750 800 Temperature (°C) 50 60 70 80 90 100 110 Weight (%) 45 55 65 75 85 Time (min) Universal V4.7A TA Instruments a) b) FIG. 3: Thermograms obtained for limestone and dolomite decomposition during calcination under pure N2(a) and CO2(b) at atmospheric pressure. 32 5 10 Deriv. Weight (%/min) 650 700 750 800 850 900 Temperature (°C) 50 60 70 80 90 100 110 Weight (%) 60 65 70 75 80 85 90 95 100 Time (min) Universal V4.7A TA Instruments 5 10 15 Deriv. Weight (%/min) 650 700 750 800 850 900 Temperature (°C) 50 60 70 80 90 100 110 Weight (%) 60 65 70 75 80 85 90 95 100 Time (min) Universal V4.7A TA Instruments 5 10 15 20 Deriv. Weight (%/min) 650 700 750 800 850 900 Temperature (°C) 50 60 70 80 90 100 110 Weight (%) 60 65 70 75 80 85 90 95 100 Time (min) Universal V4.7A TA Instruments FIG. 4: Thermograms obtained for dolomite and limestone decomposition during calcination at atmospheric pressure under CO2/N2gas mixtures at diverse CO2concentrations as indicated. 33 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.5 1 1.5 2 2.5 3 3.50 0.5 1 1.5 2 2.5 3 3.5 CaO from dolomite calcination in 25%CO2 / 75%N 50%CO2 / 50%N 75%CO2 / 25%N 100%CO2 100%N2 a) 0 0.1 0.2 0.3 0 0.5 1 X t (min) FIG. 5: Time evolution of CaO conversion during carbonation (650◦C, 15%CO2/85%N2vol/vol) of calcined dolomite (a) and limestone (b) samples carried out by in-situ TGA immediately after calcination under diverse CO2/N2mixtures as indicated. For comparison, results for CaO derived from dolomite calcined in N2(dashed line) are shown in the main graph of b) and for dolomite calcined under CO2in the inset. The deviation from the dotted straight line in a) determines the boundary between carbonation regimes controlled by reaction at the surface of the solids and by solid-state diffusion. 34 CaO 220 CaO 111 CaO 200 MgO 200 400ºC 925ºC 925º C 1h D 10ºC/min T=25ºC 0 5000 10000 15000 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 2Theta (Coupled TwoTheta/Theta) WL=1.54060 CaMg(CO3)2104 CaCO3 104 0 7000 14000 21000 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 2Theta (Coupled TwoTheta/Theta) WL=1.54060 a) Dolomite calcination under CO2 b) Dolomite calcination under N2 CaCO3 104 CaO 111 CaMg(CO3)2104 CaO 200 MgO 200 CaO 220 400ºC 925ºC 925º C 1h D 10ºC/min T=25ºC FIG. 6: XRD scans obtained during in-situ calcination of dolomite under pure CO2(a) and N2(b) atmospheres. Main Bragg peaks of dolomite (CaMg(CO3)2), calcite (CaCO3), lime (CaO), and periclase (MgO) are indicated. The shadowed scans correspond to the observed onset of first and second decomposition stages under CO2(a) and direct decomposition under N2 (b). 35 0 1000 2000 3000 4000 5000 Counts 21 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 2Theta º 22 0 1000 2000 3000 4000 5000 7000 Counts 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 2Theta º MgO 200 CaO 220 CaO 200 MgO 111 CaO 111 a) b) After 1h at 925ºC 01000 2000 3000 Counts 53 53.1 53.2 53.3 53.4 53.5 53.6 53.7 53.8 53.9 2Theta º FIG. 7: XRD scans obtained during in-situ calcination of limestone and dolomite under pure CO2. a) Just at the end of the temperature ramp (10◦C/min) at T= 925◦C. b) After 1h at 925◦C from the end of the temperature ramp. The inset in b) is a zoom of the CaO 220 peak. Bragg reflection peaks of calcite (CaCO3), lime (CaO), and periclase (MgO) are indicated. 36 0 10 20 30 40 50 60 70 80 90 100 400 500 600 700 800 900 100%CO2 75% CO2 50% CO2 25% CO2 100% N2 FIG. 8: Evolution with temperature of the wt% for dolomite (a) and MgO (b) during in-situ XRD calcination calculated by means of a Rietveld quantitative analysis for calcination tests under atmospheres of diverse CO2concentrations as indicated. 37 1 10 100 400 500 600 700 800 900 CaCO3 1 10 100 400 500 600 700 800 900 CaCO3 CaO FIG. 9: Evolution with temperature of the wt% for CaCO3and CaO during in-situ XRD calcination. Calculated by Rietveld quantitative analysis for calcination tests under carried out under pure CO2(a) and N2(b). The vertical dashed line indicates the temperature at which the presence of MgO is first detected as indicative of the initiation of half-decomposition. 38 a) b) CaO 111 CaO 200 MgO 200 CaO 220 CaO 111 CaO 200 MgO 200 CaO 220 CaO 220 CaO 220 FIG. 10: Experimental diffractogram (blue) obtained for calcined dolomite in pure CO2(after 1h at 925◦C from the end of the temperature ramp) and Rietveld best fits (red) for pure CaO and MgO phases (a) and allowing for the presence of Mg and Ca impurities in the oxides (b). The bottom grey line shows the deviation between the best fits and experimental diffractogram. Rietveld fitting parameters for pure CaO and MgO phases in a) are: 39.6wt% MgO (SD=0.39), 60.40wt% CaO (SD=0.39), GOF=1.98, Rwp=14.27, RBragg(MgO)=1.376, RBragg(CaO)=6.115. Rietveld fitting parameters allowing for Ca and Mg occupations in CaO and MgO, respectively in b) are: 37.62wt% Mg1−xCaxO with x= 0 (SD=0.023) and 62.380wt% Ca1−yMgyO with y= 0.138 (SD=0.014), GOF=1.47, Rwp=10.62, RBragg(MgO)=0.432, RBragg(CaO)=4.133. Right side figures show a zoom of the CaO 220 Bragg reflection peak. 39 FIG. 11: CaO crystallite size L(calculated from Le Bail analysis of in-situ XRD diffractograms) of dolomite, calcite, CaO and MgO during decomposition of dolomite as a function of temperature (increased at 10◦C/min). Calcination tests are carried out at atmospheric pressure under diverse CO2/N2mixtures as indicated. Figure a) shows data from independent tests of dolomite calcination under pure CO2demostrating reproducibility. 40 0 10 20 30 40 50 60 70 0 20 40 60 D t (min) a) CaO, T=925ºC 0 10 20 30 40 50 60 70 b) MgO, T=925ºC 0 20 40 60 D t (min) 75%CO2 / 25%N2 50%CO2 / 50%N2 25%CO2 / 75%N2 100% N2 100%CO2 L (nm)L (nm) FIG. 12: Crystallite size L(calculated from Le Bail analysis of in-situ XRD diffractograms) of CaO (a) and MgO (b) during the isotherm at 925◦C at the end of the temperature ramp. Calcination tests are carried out at atmospheric pressure under diverse CO2/N2gas mixtures as indicated. 41