Layered Double Oxides Supported on Graphene Oxide for CO2

Jun 9, 2015 - Layered Double Oxides Supported on Graphene Oxide for CO2 Adsorption: Effect of Support and Residual Sodium. Diana Iruretagoyena† ... ...
1 downloads 7 Views 3MB Size
Article pubs.acs.org/IECR

Layered Double Oxides Supported on Graphene Oxide for CO2 Adsorption: Effect of Support and Residual Sodium Diana Iruretagoyena,† Milo S. P. Shaffer,‡ and David Chadwick*,† †

Department of Chemical Engineering, ‡Department of Chemistry, Imperial College London, South Kensington Campus, London, SW7 2AZ, U.K. S Supporting Information *

ABSTRACT: The use of graphene oxide (GO) as support has been shown previously to improve the CO2 adsorption performance of layered double oxides (LDOs). In this contribution, the separate promoting effect of GO has been distinguished clearly from the influence of alkali species remaining from the layered double hydroxide coprecipitation. A range of sodium-free LDO and LDO/GO hybrids with relatively low GO loadings have been prepared and characterized. These have been compared to adsorbents with sodium residues deliberately left by minimum washing of the precipitates. The incorporation of GO was found to enhance considerably the thermal stability of the LDO and to reduce the loss of site heterogeneity during temperature swing cycling. The impact of the gradual loss of surface heterogeneity of the materials on the CO2 adsorption equilibrium is shown to be described by the Toth model. CO2-TPDs did not reveal any significant modification in the nature of the sites relevant for adsorption at 573 K induced by the presence of GO in the range of loadings studied. Sodium ions incorporated by leaving residual sodium from the synthesis greatly enhanced the adsorption capacity of the unsupported and supported LDOs. The improved thermal stability achieved by use of GO in the LDO/GO hybrids was not affected significantly by the presence of residual sodium. Compared to other LDO supports, GO hybrids were found generally to exhibit higher adsorption capacities per total volume of adsorbent after multiple adsorption−desorption cycles.

1. INTRODUCTION The impact of climate change and the depletion of cheap fuel sources have encouraged researchers in industry and academia to explore technologies that improve energy efficiency and render environmentally friendlier processes. In this context, sorption-enhanced hydrogen production appears to be an attractive strategy to increase H2 yields while overall CO2 emissions are reduced. In this sorption-assisted process, reforming and/or water gas shift reactions take place in the presence of a solid adsorbent that selectively removes CO2 as it is formed, thereby decreasing the thermodynamic constraints. The sorption-enhanced reactions are normally conducted between 573 and 773 K. In this temperature range, physisorbents for CO2 such as zeolites and activated carbons have relatively low capacities or are not tolerant to competing species such as water. On the other hand, chemisorbents such as supported amines, calcium oxide, and lithium zirconates are either unstable, exhibit slow adsorption kinetics, or are difficult to regenerate.1−3 Consequently, research efforts are currently being devoted to developing suitable adsorbents for sorptionenhanced hydrogen production.4−6 Layered double hydroxides (LDHs), also known as hydrotalcite-like compounds, appear as very attractive CO 2 adsorbents to be used in sorption-assisted H2 production.7−9 LDHs are two-dimensional nanostructured materials composed of M2+(OH)2 layers in which a fraction of divalent cations, octahedrally coordinated by hydroxyls (typically Mg2+), are substituted by trivalent cations (typically Al 3+ ). After calcination, LDHs decompose into layered double oxides (LDOs), which are nearly amorphous materials with a distribution of basic sites able to adsorb CO2 in the temperature © 2015 American Chemical Society

range required for sorption-enhanced hydrogen production.10,11 In general, LDH derivatives show fast adsorption and desorption kinetics and need relatively low temperatures to be regenerated. In addition, they show adequate performance when species such as H2O and H2S are coadsorbed, which makes them promising candidates for processes such as the sorption-enhanced water gas shift reaction.12,13,9,14 Despite the advantages of layered double oxides compared to other potential chemisorbents, LDOs show relatively low adsorption capacities for CO2 and their multicycle stability needs to be further improved. As a consequence, several strategies have been explored to overcome these limitations, including substitution of framework metals15−17 or promotion with alkali metals to increase their capacity18−20 as well as the addition of water in the feed to enhance their capacity and regenerability.21,22 Additionally, dispersion of the LDO particles over high surface area materials has been shown to increase the intrinsic capacity of LDOs (i.e., per mass of LDO)23−26 and their thermal stability.26 Different supports for LDOs have been reported such as zeolites,23 carbon nanofibers (CNF),25 alumina,24 and multiwalled carbon nanotubes (MWCNTs).26 However, the enhancement in the adsorption capacity of LDOs is observed only with very high loadings of support, which, in practice, would result in large sorption units. Aiming to improve the CO2 adsorption performance of layered double oxides by dispersing them in a morphologically compatible material, we reported the use of graphene oxide Received: March 31, 2015 Accepted: June 9, 2015 Published: June 9, 2015 6781

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research

333 K for 12 h under vigorous stirring. Subsequently, the precipitate was filtered using 0.2 μm nylon membranes and thoroughly washed with DI water (0.5 L per 1 g of solid) at 333 K. Samples were dried for 24 h at 393 K in a recirculating oven. Additionally, a sample was prepared as described above but was filtered with minimum washing (∼100 mL of DI water). This sample, which deliberately contains residual sodium, is denoted LDHMW. 2.2. Synthesis of LDH/GO Hybrids. Graphene oxide (Nanoinnova Technologies) was dispersed in a solution containing 1.2 M Na2CO3 and 4.7 M NaOH, from which the LDH was coprecipitated. The amount of GO added reflected the desired loading of GO in the hybrid. The resulting suspension was aged, filtered, thoroughly washed, and dried as described above for the unsupported LDH. Five different LDH/GO hybrids were prepared with nominal weight ratios 99:1 (LDH99), 20:1 (LDH20), 7:1 (LDH7), 3:1 (LDH3), and 2:1 (LDH2), using, respectively, 10, 50, 143, 333, and 500 mg of graphene oxide at constant LDH precursor concentration. A minimum washed LDH/GO hybrid with a ratio 20:1 (LDH20‑MW) was also synthesized. 2.3. Activation of LDHs and LDH/GO Hybrids. Prior to CO2 adsorption measurements, the materials were activated by calcination in situ, flowing 20 mL min−1 of N2 at 673 K for 4 h to produce the corresponding unsupported LDO and the hybrids LDO99, LDO20, LDO7, LDO3, and LDO2. The activation was performed in situ in the TGA to avoid irreversible loss in capacity due to CO2 captured from exposure to the atmosphere. For characterization purposes, the assynthesized and alkali containing samples were activated ex situ, flowing 100 mL min−1 of N2 at 673 K for 4 h using a quartz cylinder (i.d. = 4 cm, L = 60 cm) placed in a horizontal furnace (Lenton). It has been widely reported that a calcination temperature of 673 K produces LDH derivatives with an adequate balance between surface area and basic sites, which maximizes their CO 2 capacities and favors reversible adsorption.30,22 2.4. Characterization. N2 physisorption measurements at 77 K were conducted using a Micrometrics Tristar 3000 instrument. The samples were dried at 393 K under N2 for at least 12 h. The surface area was determined by the BET method and the pore volume was measured at P/P0 = 0.991. The average pore diameter was calculated from the desorption isotherm using the BJH method. Powder X-ray diffraction (XRD) was carried out using a PAnalytical X’Pert Pro diffractometer (Cu Kα radiation) in reflection mode at room temperature. The elemental composition of the adsorbents was measured by inductively coupled plasma-optical emission spectroscopy (ICP-OES) in a PerkinElmer Optima 2000DV apparatus. Thermogravimetric analysis (TGA), performed in a TAQ500 instrument, was used to determine the actual loading of GO in the hybrids. Approximately 5 mg of sample was dried at 393 K under N2 for 20 min and then heated from 393 to 1273 K at 10 K min−1 in 60 mL min−1 of air. Transmission electron microscopy (TEM) images were obtained using a JEOL 2000FX microscope. STEM-EDS images were collected in a JEM-2100F instrument. Temperature-programmed desorption (TPD) of CO2 was carried out using a quartz microflow-column system operated at atmospheric pressure. The precalcined sample (20 mg) was preconditioned by heating from room temperature to 673 at 10 K min−1, flowing 45 mL min−1 of Ar, and held for 1 h. After cooling to 313 K, the sample was exposed to a 20% CO2/Ar

(GO) as support for LDOs using GO in a wide range of loadings (5−90 wt %).27 It was shown that the multicycle stability of unsupported LDOs was improved significantly by adding moderate amounts of GO, which acts as a support or spacer. The LDO/GO hybrids also showed superior adsorption capacity compared to that of pure LDO. However, the independent effects of GO and residual sodium ions remaining from the preparation of the hybrids were not identified and therefore the origin of the enhancement in performance needs to be clarified. Subsequently, we reported CO2 adsorption isotherms at 573 K of the unsupported LDO and hybrids containing a relatively low amount of GO (5 and 25 wt %). The presence of GO at these loadings was found to maintain the heterogeneity of the adsorption sites after thermal cycling but not to increase significantly the CO2 adsorption capacity. On the other hand, higher adsorption capacities in the first exposure to CO2 were obtained with a LDO/GO hybrid containing 2 wt % of sodium residues from the synthesis. However, until now, the influence of sodium residues on the multicycle stability of the LDO/GO hybrids has not been studied. In this contribution, we distinguish clearly between the effects of graphene oxide and sodium residues on the CO2 adsorption capacity and on the thermal stability of unsupported and GO supported LDOs. For this purpose, we present a detailed study of CO2 adsorption by GO-supported LDOs that are prepared to be essentially sodium-free hybrids, in contrast to the previous work.27 A comparison is also made to minimum-washed materials, where sodium was deliberately left remaining from the synthesis. A range of GO loadings from 1 to 33 nominal wt % have been studied because these low GO contents result in higher capacities per total volume of adsorbent, which, in principle, would yield smaller adsorption units. Over this range of GO loadings, the sodium-free hybrids are shown to give no significant increase in adsorption capacity compared to that of pure LDO. However, the presence of GO is shown to improve the stability of the adsorbents (over more extended thermal cycling than that in our previous work27,28). The influence of GO on the nature of the adsorption sites has been investigated by temperature-programmed desorption of CO2 (CO2-TPD) and by the adsorption isotherms at 573 K. The isotherms for thermally cycled sodium-free LDO/GO materials reveal a transition from a heterogeneous surface toward a more uniform distribution of adsorption sites upon cycling, which is slowed by the presence of GO. This is shown to be well-described by the three-parameter isotherm of Toth.29 The effects of sodium retained in minimum-washed LDO/GO hybrids on the distribution and density of basic sites (determined by CO2-TPD), on the CO2 capacity and adsorption isotherms, and on the evolution of properties under thermal cycling are reported in detail. The present study provides insight on the roles of GO and sodium residues on the enhanced adsorption performance of LDO/GO hybrids.

2. EXPERIMENTAL SECTION 2.1. Synthesis of Mg-Al LDHs. Unsupported LDHs, Mg0.6Al0.3(OH)2(CO3)0.15·nH2O, were prepared via coprecipitation. An aqueous solution (4.5 mL) of 2 M Mg(NO3)2·6H2O and 1 M Al(NO3)3·9H2O was added dropwise to an aqueous solution (6.7 mL) of 1.2 M Na2CO3 and 4.7 M NaOH under vigorous stirring. The inorganic salts used in the synthesis were purchased from Sigma-Aldrich. The measured pH of the solution was 10. The resulting white suspension was aged at 6782

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research premixed gas (BOC) for 1 h. The system was then purged in flowing Ar for 2 h to remove physisorbed CO2, and the temperature was then increased to 1023 at 50 K min−1. The argon used in the experiments was purified using moisture and oxygen traps. A mass spectrometer (ESS GeneSys II with heated capillary sample line) was used to monitor the CO2 concentration in the quartz tube exit gas stream. Prior to each TPD experiment, the response was calibrated by injecting 2 mL of the premixed CO2 gas. 2.5. CO2 Adsorption Measurements. A thermogravimetric analyzer (PerkinElmer, TGA4000) was used to determine the adsorption capacity of the samples. An amount (∼15 mg) of adsorbent was calcined in situ as described earlier. After activation, the temperature was decreased from 673 to 573 K at 10 K min−1 and held for 10 min. The feed was then switched to a 20% CO2/Ar premixed gas (BOC), and held usually for 1 h. The adsorption capacity of the materials was determined from the change in mass during the adsorption step. Effects due to changes in gas viscosity and gas density were accounted by measuring the response of silicon carbide and then subtracting it from the adsorbent response. Under the operating conditions used this blank response was very small. In multicycle tests, the adsorption step was carried out at 573 K for 1 h flowing the premixed CO2 gas and the desorption step was performed at 673 K during 30 min flowing nitrogen. The flow rate was kept constant at 20 mL min−1 during the experiment. The adsorption capacities reported in the equilibrium isotherms were taken as the average of at least three measurements under the same operating conditions. The data used in the multicycle isotherms correspond to the capacities measured from adsorbents exposed to cyclic operation with a mixture 20% CO2/Ar at 573 K followed by a desorption/purge step at 673 K in flowing nitrogen (60 mL min−1) for 30 min. After a near-constant capacity was obtained (typically 20 cycles), the adsorption isotherm was measured continuing the adsorption−desorption cycles but increasing successively the concentration of CO2 in the adsorption step using premixed gases. For the multicycle adsorption isotherms, the standard deviation in the mean was calculated from the adsorption capacity after 20 cycles obtained in at least three different experiments. This error was assumed to be the same for the adsorption capacities obtained at the different CO2 partial pressures. The experimental adsorption isotherms were fitted to Freundlich (1), Toth (2), and Langmuir (3) models, applying a nonlinear least-square method using OriginPro8.6.

Table 1. Composition of the As-Synthesized Adsorbents sample namea

wt % GO (nom.)

wt % GO (actual)

Mg/Al (mol/mol)

wt % Na

pure LDH LDH99 LDH20 LDH7 LDH3 LDH2 LDHMW LDH20‑MW

0 1 5 13 25 33 0 5

0 2 4 9 17 20 0 6

2.2 2.3 2.2 2.2 2.2 2.1 2.2 2.2

0.02 0.02 0.07 0.05 0.40 0.54 2.80 1.90

a

LDHx, where x is the nominal LDH-to-GO weight ratio of the sample.

(2 L of water per 150 mg of GO). This shows that it is not feasible to reduce the residual sodium in the samples below the trace levels given in Table 1. These samples can be regarded as being essentially sodium-free. The TGA patterns of the Na-free adsorbents are shown in Figure 1. The unsupported LDH exhibited three stages of

Figure 1. Thermograms of unsupported LDH, GO, and LDH/GO hybrid samples.

weight loss corresponding to its dehydration (below ∼483 K), partial dehydroxylation and decarbonation (between ∼483 K and ∼733 K), and further decarbonation (above ∼733 K). The TGA of the pure GO presents a weight loss at around 500 K due to the decomposition of functional groups and the oxidation debris and a weight loss at around 750 K that has been attributed to the combustion of the (damaged) graphitic regions.31 In the hybrids, the weight loss at high temperature correlates with the GO content. Therefore, the actual GO weight percentage in the LDH/GO materials can be estimated from the residue of each sample. The TGA residue of pure LDH is composed of ∼60 wt % mixed solid oxides, whereas the residue of GO can be considered to be negligible. The nominal and actual weight loadings are very close for the lower GO content samples (LDH99 and LDH20). However, the higher GO loading adsorbents have considerably less GO than expected (Table 1). This is in agreement with previous findings showing that the loss of oxidative debris during filtering becomes more significant as the GO content in the sample increases.27 The minimum washed as-synthesized materials had actual sodium contents of 1.9 wt % for LDH20‑MW and 2.8 wt % for LDHMW, according to ICP measurements, Table 1. In situ STEM-EDS analysis confirmed that magnesium, aluminum, and sodium are distributed evenly in the samples (Supporting

3. RESULTS AND DISCUSSION 3.1. Characterization of the Adsorbents. 3.1.1. Chemical Composition. The compositions of the as-synthesized LDHs and hybrids are given in Table 1. The average Mg/Al ratio of the adsorbents, determined by ICP, was found to be 2.2 ± 0.06, which is close to the intended ratio of 2. In the thoroughly washed materials, sodium was detected only at trace levels, indicating that the NaOH and Na2CO3 used during the synthesis were successfully removed. However, it can be observed that the sodium content, although very small, increases with the GO loading. This is caused by the attraction between the sodium ions and the highly negatively charged GO sheets during the LDH coprecipitation. In fact, it was found that a hydrothermally treated GO (i.e., stirring the GO in water for 12 h at 333 K and pH 10) contained a small amount of sodium after it was thoroughly washed during the filtration step 6783

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research Table 2. Physicochemical Properties of the As-Synthesized and Activated Adsorbents as-synthesized adsorbents

a

activated adsorbents

sample namea

SBET (m2/g)

Vpore (cm3/g)

crystallite size (003) (nm)

sample namea

SBET (m2/g)

Vpore (cm3/g)

crystallite size (200) (nm)

pure LDH LDH99 LDH20 LDH7 LDH3 LDH2 LDHMW LDH20‑MW GO

112 116 106 113 143 114 53 70 55

0.46 0.62 0.54 0.54 0.47 0.55 0.24 0.38 0.06

21 18 19 16 11 16 29 24

pure LDO LDO99 LDO20 LDO7 LDO3 LDO2 LDOMW LDO20‑MW GO

223 205 199 219 170 157 73 99 81

0.51 0.83 0.62 0.64 0.54 0.66 0.31 0.40 0.15

3.2 3.2 3.1 3.1 2.9 2.8 3.9 3.3

LDHx and LDOx, where x is the nominal LDH- or LDO-to-GO weight ratio of the sample.

Information, Figure S1). The content of graphene oxide of LDH20‑MW is close to the intended nominal value and is slightly higher than that of the corresponding thoroughly washed hybrid. This coincides with the limited loss of carbon debris due to the minimum washing. The TGA patterns of the LDH and LDH20 hybrid with deliberately remaining sodium are very close to those of the corresponding pure materials (Supporting Information, Figure S2). 3.1.2. Morphological Characterization. N2 physisorption was used to investigate the textural properties of the adsorbents. All as-synthesized and activated materials exhibited type IV isotherms with H1 hysteresis loops.32 A narrow hysteresis that extends over the whole range of adsorption is observed, caused by void slits originated within agglomerates of LDH platelets.32 Representative BET isotherms are shown in the Supporting Information (Figure S3). The surface areas of the sodium-free LDHs and hybrids are high, Table 2, compared to other reported values for similar synthesis methods.25,27 The surface area of the materials was observed to increase with the extent of washing during the filtration step. Upon calcination, the surface area of the adsorbents increases, but the increment is lower for the hybrids with high content of GO (LDO3 and LDO2). As suggested in our previous work, GO restacking becomes more significant as the amount of intercalating hydroxide layers decreases.27 For the as-synthesized and calcined samples that deliberately contain sodium residues, the surface areas and pore volumes were found to be significantly lower than those of the corresponding alkali free materials, Table 2. This decrease is possibly related to blockage/filling of the pores with sodium species. 3.1.3. Crystallographic Characterization. The crystallinity of the thoroughly and minimally washed LDHs was investigated by XRD (Figures 2 and 3). All of the LDH containing samples display the characteristic reflections corresponding to 2D hydrotalcite-like materials with carbonates in the interlayer (JCPDS no. 14-191) and can be indexed accordingly.33 The pure GO shows reflection lines at 11.3° and 42.3°, corresponding to the (002) and (100) planes. In our previous work based on LDH/GO hybrids, we observed that the sharp GO peak at 11.3° completely disappears after hydrothermal treatment and washing. Instead, a broad peak appears at 27.1°, which is associated with disordered restacking of partially oxidized GO sheets.27 However, this reflection is observed only at very high loadings of GO and therefore is absent from the patterns of the GO-supported LDHs presented here. For the thoroughly washed hybrids, the intensity of the LDH reflections and crystallite size decreased slightly with the amount of GO (Figure 2 and Table 2). In general, the

Figure 2. Representative XRD diffraction patterns of as-synthesized unpromoted adsorbents. Characteristic reflections of LDH (⧫) and GO (□).

Figure 3. Representative XRD diffraction patterns of as-synthesized sodium-promoted adsorbents. Characteristic reflections of LDH (⧫) and NaNO3 (○).

dispersion (surface area) was found to correlate with the crystallite size. The crystallite sizes of the as-synthesized adsorbents prepared deliberately to contain residual sodium (LDHMW and LDH20‑MW) are larger than those of the thoroughly washed samples, Figure 3 and Table 2. This suggests that extensive washing during the filtration step decreases the final crystallite size of the material, probably due to erosion. After calcination, all sodium-free samples (LDH and hybrids) show broad reflection peaks ascribed to the diffraction planes of periclase (MgO, JCPDS no. 45-946); the aluminum compounds are thought to be well-dispersed or in an amorphous phase, Figure 4.18 The calcination of the pure GO (as-synthesized and after 6784

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research

NaNO3 into NaNO2 and O2.34 In fact, during CO2-TPD of unsupported LDOMW, masses corresponding to oxygen desorption are detected with a maximum at 950 K. No oxygen evolution is observed in the TPD profile of the hybrid (LDO20‑MW) that contained sodium residues deliberately (see below). This is likely to be related to a lower concentration of NaNO3 crystallites in the sample and to the fast consumption of any released oxygen in the combustion of GO. TEM images of the unsupported and supported LDHs show small, crystalline, hexagonal platelets of LDH. No GO nanosheets were found in the TEM images of the hybrids (see Supporting Information, Figure S4). It is assumed that for these low−medium loadings the GO nanosheets are wellintegrated with the LDH particles. In agreement with the XRD results, Table 2, larger crystallites were observed for the minimum washed samples, especially in the case of unsupported LDH (Supporting Information, Figure S4). 3.1.4. Temperature-Programmed Desorption of CO2. Temperature-programmed desorption of CO2 was used to investigate the influence of graphene oxide on the density and strength of the basic sites of the LDOs. The desorption profile of the unsupported LDO shows three overlapping peaks with maxima at 423, 543, and 813 K, Figure 6. The high-temperature

Figure 4. Representative XRD diffraction patterns of activated unpromoted adsorbents. Characteristic reflections of LDO (⧫) and calcined GO (□).

hydrothermal treatment) led to diffraction lines corresponding to graphitic (002) and (100) lattice planes.27 In the hybrids, these reflections were observed only at relatively high loadings of support (i.e., LDO3 and LDO2). For the minimum-washed as-synthesized adsorbents, an additional diffraction line representative of NaNO3 (104) is observed at 29.4°, the intensity of which is higher for the unsupported LDH.18 The intensity of this peak is related to the amount of crystalline sodium nitrate remaining in the sample. However, not all of the residual sodium may be in the form of crystalline agglomerates, and some may be dispersed over the LDH. In addition, as discussed above, some sodium ions are retained by the support during the synthesis due to interaction with GO and cannot be removed. The diffraction patterns after calcination of the adsorbents that intentionally contain sodium are presented in Figure 5. The XRD pattern of hybrid

Figure 6. CO2-TPD of pure LDO, LDO20, and LDO2 following pretreatment at 673 K and adsorption at 313 K (dark lines). The blanks were subjected to pretreatment only (light lines).

state can be deconvoluted into two states centered at 813 and 963 K (Supporting Information, Figure S5). These basic sites of increasing strength have been assigned previously to the formation of bicarbonates, bidentate, and monodentate carbonates, respectively,35,36 and correspond to desorption energies in the range 100−200 kJ/mol (assuming a preexponential factor of 1013 and negligible readsorption).37,38 Weaker adsorption states and physisorbed CO2 are not observed in the TPD due to the initial purging. After the pretreatment at 673 K, the low and intermediate energy states are the main contributors to the CO2 chemisorption on the LDO at the pressure and temperature used. The CO2-TPDs of pure LDO and the LDO/GO hybrids containing 4 wt % of GO (LDO20) and 20 wt % of GO (LDO2) are presented in Figure 6. The distribution of basic sites in the hybrid containing a relatively low amount of GO, LDO20, is very close to that of pure LDO. The presence of 20 wt % of GO in the LDO did not modify the distribution of low and intermediate energy states, but the desorption peak at high temperature is more dominant. This peak corresponds to the evolution of remaining carbonates from the LDH structure and

Figure 5. Representative XRD diffraction patterns of activated sodiumpromoted adsorbents. Characteristic reflections of LDO (⧫) NaNO3 (○).

LDO20‑MW shows the diffraction peaks of LDO exclusively, whereas unsupported LDOMW additionally exhibits the characteristic peaks of NaNO3. The calcination temperature is above the melting point of sodium nitrate, 581 K, and therefore significant mobility is expected, facilitating agglomeration. It is likely, therefore, that the GO present in LDO20‑MW suppresses the agglomeration of sodium nitrate to form crystallites. It is also important to note that the NaNO3 reflections observed in the XRD pattern of LDOMW are slightly shifted to lower 2θ angles, which may be caused by the partial decomposition of 6785

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research Table 3. Density of Basic Sites: CO2-TPD sample namea

wt % GO actual

total CO2 evolved (μmol/g)

CO2 evolvedb (minus blank) (μmol/g)

CO2 evolvedb (minus blank) (μmol/g LDO)

CO2 evolvedb (minus blank) (μmol/m2)

pure LDO LDO20 LDO2 LDO20‑MW

0 4 20 6

600 690 710 920

440 410 360 450

440 430 450 480

1.96 2.07 2.29 4.57

a

LDOx, where x is the nominal LDO-to-GO weight ratio of the sample; -Na refers to residual sodium. bValues obtained by subtracting the corresponding blank from the total CO2 evolved

sodium in the minimum washed samples (∼2 wt %) approximately doubles the surface density of basic sites. In general, use of intermediate adsorption temperatures and low regeneration gradients are more favorable for temperature swing applications. Consequently, in this study, adsorption and desorption temperatures of 573 and 673 K were used to assess the multicycle stability of the adsorbents and is reported below. Under CO2 adsorption at low pressure, mainly the desorption state centered at 543 K and part of the state at 813 K are cyclically populated and depopulated, whereas the lowtemperature sites are nearly empty. However, as the CO2 pressure during adsorption at 573 K is increased, weaker states become populated and ultimately physisorption contributes to the adsorption capacity.16,30 3.2. CO2 Adsorption Measurements. The CO2 adsorption capacity of the materials at a fixed partial pressure was determined from the weight gain of the samples exposed to the adsorptive gas in a thermogravimetric analyzer. This provided a convenient measure of capacity for the purposes of comparison. All of the adsorbents were activated in situ and showed fast adsorption kinetics at 573 K, requiring 30 min to achieve ca. 90% of their equilibrium capacity. For practical purposes, the values of CO2 capacity here reported correspond to those obtained after 60 min of CO2 exposure at 200 mbar (Supporting Information, Figure S8). The adsorption capacities in the first exposure to the adsorptive gas (first-contact capacity) are given in Table 4. Under the operating conditions

to the thermal decomposition of GO (Supporting Information, Figure S6). Values for the CO2 chemisorption capacity determined by TPD are given in Table 3. These values were obtained by subtracting the corresponding blank, i.e., the CO2 evolved during the temperature program from the activated samples after pretreatment at 673 K that were not exposed to CO2 adsorption (see Figure 6 and Table 3). The similar density and distribution of basic sites of the unsupported and supported LDOs revealed by CO2-TPD suggests that the presence of GO, in the range of loadings studied, does not modify significantly the nature of the adsorption sites available after calcination at 673 K. This does not preclude the possibility that higher loadings of GO than studied here and/or alternative preparation methods may result in significantly modified sites. The desorption profile from LDO20‑MW is shown in Figure 7, and the corresponding CO2 evolution is given in Table 3. The

Table 4. First-Contact Adsorption Capacities of the Adsorbents at 573 K and PCO2 = 200 mbar Figure 7. CO2-TPDs of LDO20 and LDO20‑MW following pretreatment at 673 K and adsorption at 313 K (dark lines). The blanks were subjected to pretreatment only (light lines).

number of basic sites measured by TPD that are available after pretreatment at 673 K is ∼10% higher for the minimumwashed sample with sodium residues left. A similar TPD profile was obtained for the minimum-washed unsupported LDO sample (Supporting Information, Figure S7). Contrary to GO, which acts mainly as an efficient support for LDOs, sodium alters considerably the distribution of the basic sites. The release of bicarbonates formed on Brønsted OH− groups (low temperature) is maintained, whereas the adsorption in the intermediate temperature state corresponding to bidentate carbonates bonded to metal−oxygen pairs decreases. In addition, the number of monodentate carbonates adsorbed on low-coordination oxygen anions (high temperature) is significantly higher for the sodium-promoted samples. The very strong basic sites that desorb CO2 at 1115 K are possibly related to the decomposition of bulk sodium carbonates at the surface of the mixed oxides.39 Overall, the presence of residual

sample namea

wt % GO actual

pure LDO LDO99 LDO20 LDO7 LDO3 LDO2 LDOMW LDO20‑MW

0 2 4 9 17 20 0 6

mol CO2/kg LDO 0.29 0.28 0.29 0.32 0.30 0.30 0.54 0.58

± ± ± ± ± ± ± ±

0.01 0.03 0.03 0.03 0.02 0.03 0.02 0.02

mol CO2/kg ads 0.29 0.28 0.28 0.29 0.25 0.24 0.54 0.54

± ± ± ± ± ± ± ±

0.01 0.03 0.03 0.03 0.02 0.03 0.02 0.02

a

LDOx, where x is the nominal LDO-to-GO weight ratio of the sample.

used, the contribution of GO to the CO2 uptake of the hybrids was negligible. The fact that the intrinsic adsorption capacities (i.e., per mass of LDO) of the sodium-free LDO and LDO/GO hybrids are the same within the experimental error is consistent with the TPD results, which indicate that the density of available basic sites after activation is essentially the same for the unsupported and supported LDOs. Moreover, the results indicate that the traces of sodium retained by the samples after extensive washing have no noticeable influence on the 6786

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research capacities measured. It should be noted that the CO2 capacity determined by TGA at 200 mbar is less than the total uptake in TPD since only a fraction of the CO2 is in equilibrium, as shown by isotopic exchange measurements.36 As expected, the adsorption capacity normalized with respect to the total mass of adsorbent decreases slightly with the small amount of GO in the hybrid. In contrast to the above thoroughly washed (i.e., sodiumfree) samples, it is observed that the presence of sodium residues in the minimum-washed samples (LDOMW and LDO20‑MW) promotes the first-contact adsorption capacities of the corresponding unpromoted materials, Table 4. Since the surface area of the promoted adsorbents is lower than that of the sodium-free samples, it can be concluded that the enhancement is attributable to the presence of additional interactions generated by the sodium cations. As discussed above, the TPD profiles in Figure 7 show that the presence of intentionally left sodium increases the number of basic sites corresponding to the high-temperature state. After activation, part of these stronger sites would be available for adsorption at 573 K and therefore it is likely that they are responsible for the increase in the observed adsorption capacity under the conditions used. These findings are consistent with other studies that have shown that the presence of alkali ions on LDHs generates stronger basic sites with increased activity for CO2 adsorption and base-catalyzed reactions.40,41,20,39 Meis et al. suggested that the stronger basicity is related to defects introduced in the LDO surface by the presence of alkali ions.20 Additionally, Walspurger et al. proposed that the interaction between the aluminum centers in the mixed oxides and the alkali ions plays a crucial role in the formation of strongly basic sites that reversibly adsorb CO2 at 573−773 K.39 Other studies have found relatively low heats of adsorption for potassiumpromoted LDOs under dry conditions, indicating that physisorption or very weak chemisorption is important in the CO2 adsorption properties of LDOs between 573 and 753 K as well as at higher pressures, as noted above.12,16 The first-contact adsorption isotherms at 573 K were determined for the sodium-free samples at CO2 partial pressures between 50 and 990 mbar. The experimental data can be fitted adequately to the Freundlich model (1), as shown in Supporting Information Figure S9.

q = k[pCO ]1/ n 2

Table 5. Toth Parameters for First-Contact Isotherms at 573 K from TGA under Dry Conditions Toth isotherma m (mol kg )

log b (bar−1)

t (−)

pure LDO LDO20 LDO3 LDO20‑MW

1.14 0.88 0.63 4.10

4.01 3.53 3.10 9

0.17 0.21 0.25 0.09

LDOx, where x is the nominal LDO-to-GO weight ratio of the sample.

(1)

Figure 8. Representative first-contact adsorption isotherms at 573 K (shown with standard errors).

mbPco2 (1 + (bPco2)t )1/ t

adsorbent

a

Freundlich isotherms are associated with a heterogeneous distribution of adsorption sites, which is consistent with the variety of surface energies revealed by CO2-TPD for the adsorbents after activation. Upon consecutive adsorption− desorption cycles, it was observed that gradually the materials become more uniform (see the following section). Consequently, it is desirable to capture this transition in the isotherm description, and to this end, the Toth model (2) appears to be a suitable choice. q=

−1

Toth parameters fitted for the thoroughly washed LDO and LDO20 are similar, indicating that the presence of GO does not modify significantly the nature of the adsorption sites. The first-contact adsorption isotherm of the minimumwashed hybrid (i.e., containing residual sodium) can also be described by the Freundlich and Toth equations. The corresponding t parameter is lower than the one fitted for the thoroughly washed samples, Table 5, and indicates the

(2)

The Toth isotherm includes an additional parameter, t, which is associated with the surface heterogeneity of the adsorbents. The fitting of the experimental isotherms to eq 2 results in t values well below 1, Table 5 and Figure 8. Such deviation from unity reflects the initial diversity of adsorption sites available in the first exposure to the adsorptive gas. It is observed that the 6787

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research influence of the deliberately left sodium ions on the basic strength distribution. Additionally, the presence of sodium causes an increase in the capacity parameter, m, and in the gas− solid interaction parameter, b. 3.3. Adsorption−Desorption Thermal Cycling. Continuous adsorption−desorption cycles were carried out to assess the regeneration and stability of the adsorbents under dry conditions. The CO2 capacity of the pure LDO and LDO/GO hybrid drops markedly during the first cycles and becomes stable gradually, Figure 9. This indicates that during the first-

Figure 9. Normalized CO2 adsorption capacity for LDO/GO samples over 20 adsorption−desorption cycles (adsorption 573 K; desorption 673 K).

Figure 10. CO2-TPDs of (a) pure LDO (first-contact and multicycles) and (b) LDO3 (first-contact and multicycles).

contact between the adsorbent and the adsorptive gas some CO2 is irreversibly chemisorbed on the fresh material and then reversible adsorption dominates with parallel thermal sintering. Similar trends have been reported for studies dealing with adsorption of CO2 on LDO containing adsorbents at relatively high temperatures (573−753 K) and under dry conditions.12,16,26 The stability of the pure LDO is markedly enhanced by the addition of graphene oxide and increases with the GO content in the hybrid, Figure 9. For example, the adsorption capacity of the unsupported LDO after 20 cycles is 60% of its initial value, whereas the hybrid with the highest content of GO studied (LDO2) reduced its initial capacity by only 10%. All of the hybrids exhibited a minor loss of weight during the first cycles, and the weight remained practically constant thereafter. The loss of mass is related to the partial thermal decomposition of the GO and therefore it becomes more significant for high GO loadings. Remarkably, this does not affect the multicycle stability of the adsorption capacity, as observed in Figure 9. The CO2-TPD of the pure LDO after cycling shows a significant decrease in the number of low and intermediate energy states, whereas the strongest sites are affected only slightly. This leads to a decrease in the heterogeneity of the adsorption sites on unsupported LDO, Figure 10a. On the contrary, in the presence of graphene oxide, there is a small loss of intermediate and high energy states in similar proportions and therefore the initial distribution of adsorption sites is better preserved, Figure 10b. The minimum washed adsorbents containing residual sodium show multicycle profiles very close to those of the corresponding sodium-free samples, Figure 11. This indicates that relatively low contents of sodium have no effect on the stability of the unsupported and supported LDOs yet greatly enhance their adsorption capacity. The present results confirm that the enhancement in the intrinsic capacity of the

Figure 11. Normalized CO2 capacity for sodium-promoted samples over 20 adsorption−desorption cycles (adsorption 573 K; desorption 673 K).

unpromoted LDO/GO hybrids previously reported27 was probably associated with the presence of some sodium remaining in the samples from the synthesis. [In ref 27, it was reported that sodium was absent based on single-point EDS measurements. It is possible that the distribution of sodium was inhomogeneous. ICP, as used here, is intrinsically more sensitive and gives an average concentration.] The adsorption isotherms at 573 K of the thermally cycled materials were determined after near-constant capacities were obtained in continuous adsorption−desorption cycles. Then, the CO2 partial pressure of the gas in the adsorption step was successively increased as described in the Experimental Section. The adsorption isotherms can be described by the Langmuir (3) model, as shown in Supporting Information Figure S10. The Toth eq 2 also provides a description of the adsorption 6788

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research

and therefore the experimental data can also be fitted to the Langmuir equation. In general, either description is independent of the presence of sodium in the minimum-washed materials and is in agreement with the changes in the chemisorption sites observed by CO2-TPD, which were found to evolve toward a homogeneous distribution after multiple thermal cycles, Figure 10. The parameters of the Toth equation fitted to the adsorption data of the different materials, Table 6, reveal that while for all of the adsorbents the value of t tends to unity after multicycle experiments, lower values are preserved for the LDO/GO hybrids. The minimum-washed adsorbents containing residual sodium exhibit higher monolayer capacities, m, and a larger interaction parameter, b. A direct comparison between the adsorption capacities obtained in this study for unsupported and GO supported LDOs and capacities from the literature for LDO-based adsorbents is difficult due to the wide range of preparation methods, chemical compositions, pretreatment procedures, and adsorption conditions used in previous studies. However, the CO2 capacity for the unsupported LDO here reported is within the range of values typically found in the literature for Mg−Al− CO3 LDOs (0.1−0.6 mol CO2/kg).1,42 In addition, first-contact adsorption capacities of the LDO/GO hybrids are comparable to results obtained from LDO coated zeolites when differences in adsorption temperature and loading are taken into account.23 A comparison of the first-contact and multicycle CO 2 adsorption capacities of the LDO/GO hybrids and LDO supported on Al2O3 and MWCNTs is given in Table 7. These materials were tested under the same experimental conditions of the present study, i.e., dry gas at 200 mbar of CO2 and 573 K.26 Also included in Table 7 is first-contact adsorption data for LDO supported on CNF, as it is one of the few references to the use of other nanostructured carbon supports.25 The adsorption capacity of LDO/CNF was determined at a lower temperature, 523 K, and in the presence of steam, so higher values of capacity per mass of LDO were obtained. As observed in Table 7, a significant increase in the intrinsic adsorption capacity of pure LDO can be achieved by using relatively high contents of high surface area supports (see values for CNF, alumina, and MWCNTs). However, high dispersion leads to low adsorption capacities per total volume of adsorbent, which may limit the commercial use of these materials. For LDO/MWCNTs hybrids, an optimum LDO loading (about 40 wt % MWCNTs) in terms of adsorption capacity per total mass of adsorbent was found after adsorption−desorption cycling.26 Graphene oxide appears to be especially effective at supporting LDO compared to MWCNTs, since a better compromise between adsorption capacity per total volume of adsorbent and amount of support used can be achieved after the same number of cycles. This high mass efficiency can be attributed to the obvious compatibility in terms of charge and geometry between GO sheets and the LDO platelets. The increase in the CO2 adsorption capacity of the pure LDO and the LDO/GO hybrids arising from the presence of residual sodium coincides with previously reported enhancements in the presence of sodium and other alkali ions.20,43,18,1 For instance, Meis et al.20 found that the adsorption capacity of a LDO/CNF hybrid was nearly doubled by leaving residual sodium ions in the adsorbent, Table 7. The absolute value of intrinsic capacity of the Na-LDO/CNF is higher than that of the minimum-washed LDO/GO hybrid (LDO20‑MW) here reported. This can be attributed to the higher loading of alkali,

isotherm, Figure 12, and is able to capture changes in behavior upon thermal cycling in the parameters, as noted above. q=

mbPCO2 1 + bPCO2

(3)

Figure 12. Representative multicycle adsorption isotherms at 573 K (shown with standard errors).

The thermally cycled LDO samples exhibit t values that tend toward 1, Table 6. As the t value approaches unity, the Toth isotherm reduces to the description of an ideal uniform surface Table 6. Toth Parameters for Multicycle Isotherms at 573 K from TGA under Dry Conditions Toth isotherma adsorbent

m (mol kg−1)

log b (bar−1)

t (−)

pure LDO LDO20 LDO3 LDO20‑MW

0.19 0.27 0.47 0.42

1.46 1.39 2.27 1.70

0.96 0.92 0.33 0.86

a

LDOx, where x is the nominal LDO-to-GO weight ratio of the sample. 6789

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research Table 7. Comparison with Other LDO Adsorbents authors ref 20 ref 20 ref 20 ref 26 ref 26 ref 26 present work present work present work

adsorbents

support (wt %)

mol CO2/kg LDO firstcontact

mol CO2/Lads firstcontacta

relative capacity at 20 cycles (%)

mol CO2/Lads at 20 cycles

LDO LDO/CNF LDO/CNF [12 wt % Na, residual] LDO LDO/Al2O3 LDO/MWCNTs LDO

0 90 90

0.10 1.30 2.20

0.10 0.13 0.22

unknown unknown unknown

unknown unknown unknown

0 80 38 0

0.28 0.43 0.42 0.29

0.28 0.19 0.19 0.29

45 52 65 60

0.13 0.10 0.12 0.18

LDO/GO [LDO2]

20

0.30

0.26

91

0.24

LDO/GO [LDO20‑MW]

6

0.58

0.59

73

0.41

a

The particle density of the adsorbents was determined experimentally except for the LDO/CNF hybrid, which was estimated by considering the density of pure CNF. For the alkali-promoted samples, the density was assumed to be the same as that of the corresponding unpromoted adsorbent. Experimental conditions: ref 20: PCO2 = 50 mbar, 523 K, wet conditions, TSA regeneration at 773 K under dry conditions; ref 26: PCO2 = 200 mbar, 573 K, dry conditions, TSA regeneration at 673 K under dry conditions.

The first-contact CO2 adsorption data for the sodium-free and -promoted LDO and LDO/GO samples is better described by the Freundlich isotherm, but after multiple thermal cycling the isotherms, they tend to fit the Langmuir model as the surface becomes more uniform. These trends in the gradual homogenization of the surface of the adsorbents can be correctly predicted by the Toth model in which the t parameter is found to depend on the number of cycles and the GO content.

support (CNF), and the operating conditions used (523 K, wet adsorption). However, Na-LDO/CNF exhibits a much lower adsorption capacity per mass of adsorbent than the LDO20‑MW hybrid due to dilution with high loadings of sodium and CNF. The enhanced CO2 adsorption capacities per mass/volume of sodium-promoted LDO/GO hybrid and the superior stability demonstrated in the present work are important characteristics for application of this hybrid, assuming that graphene oxide becomes readily available as is expected. It is notable that only low loadings of GO are needed to obtain reasonable thermal stability, since it is likely that the adsorbents in their commercial form will require the addition of binders to improve their mechanical properties. As a consequence, alkalipromoted LDO/GO hybrids appear to be promising candidates for CO2 adsorption applications such as sorption-enhanced hydrogen production, particularly considering that other types of adsorbents (e.g., zeolites, carbons, supported ammines, and CaO) exhibit shortfalls in the temperature range required regarding their various tolerance to water and other competing species, CO2 adsorption capacity, and stability and regenerability.1



ASSOCIATED CONTENT

S Supporting Information *

Figure S1: STEM-EDS image of LDHMW. Figure S2: Thermograms of sodium-promoted samples. Figure S3: Representative nitrogen adsorption−desorption isotherms of the as-synthesized and activated adsorbents. Figure S4: Representative TEM images of the as-synthesized adsorbents. Figure S5: Deconvolution of the CO2-TPD of the pure LDO. Figure S6: CO2-TPD of hydrothermally treated GO. Figure S7: CO2-TPDs of pure LDO and LDOMW. Figure S8: Representative CO2 adsorption profiles of the activated adsorbents at 573 K and CO2 partial pressure of 200 mbar. Figures S9 and S10: Representative first-contact and multicycle isotherms at 573 K (Langmuir and Freundlich). Tables S1−S4: Parameters for first-contact and multicycle isotherms at 573 K (Langmuir, Freundlich, and Toth). The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.iecr.5b01215.

4. CONCLUSIONS The thermal stability of LDO is markedly enhanced by the addition of modest amounts of graphene oxide, causing the CO2 adsorption capacity of the LDO/GO hybrids to be better preserved over thermal cycling. The CO2 adsorption capacity per mass of LDO does not change significantly in the presence of these low contents of GO, and TPD suggests that GO does not modify the nature and density of the CO2 adsorption sites. Compared to other supports for LDO such as MWCNTs or alumina, GO appears to be especially effective as a support since a better compromise between adsorption capacity per total volume of adsorbent and the amount of support used is achieved. The presence of small amounts of residual sodium in minimum-washed materials (∼2 wt %) was found to almost double the CO2 adsorption capacity without compromising the stability imparted by the GO. TPD shows that incorporation of sodium ions on the adsorbents generates a greater density of stronger basic sites, which seem to be responsible for the increase in the CO2 capacity observed at 573 K and 200 mbar.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

Concepts for the research program were conceived by all authors. Experiments were devised by D.C. and M.S. together with D. I., who carried them out and performed data analysis. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest. 6790

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research



(11) Radha, S.; Navrotsky, A. Energetics of CO2 adsorption on Mg− Al layered double hydroxides and related mixed metal oxides. J. Phys. Chem. A 2014, 118, 29836−29844. (12) Ding, Y.; Alpay, E. Equilibria and kinetics of CO2 adsorption on hydrotalcite adsorbent. Chem. Eng. Sci. 2000, 55, 3461−3474. (13) van Dijk, H. A. J.; Walspurger, S.; Cobden, P. D.; van den Brink, R. W.; de Vos, F. G. Testing of hydrotalcite-based sorbents for CO2 and H2S capture for use in sorption enhanced water gas shift. Int. J. Greenhouse Gas Control 2011, 5, 505−511. (14) Boon, J.; Cobden, P. D.; van Dijk, H. A. J.; Hoogland, C.; van Selow, E. R.; van Sint Annaland, M. Isotherm model for hightemperature, high-pressure adsorption of and on K-promoted hydrotalcite. Chem. Eng. J. 2014, 248, 406−414. (15) Tsuji, M.; Mao, G.; Yoshida, T.; Tamaura, Y. Hydrotalcites with an extended Al3+-substitution: synthesis, simultaneous TG-DTA-MS study, and their CO2 adsorption behaviors. J. Mater. Res. 1993, 8, 1137−1142. (16) Hutson, N. D.; Attwood, B. C. High temperature adsorption of CO2 on various hydrotalcite-like compounds. Adsorption 2008, 14, 781−789. (17) Wang, Q.; Tay, H. H.; Ng, D. J. W.; Chen, L.; Liu, Y.; Chang, J.; Zhong, Z.; Luo, J.; Borgna, A. The effect of trivalent cations on the performance of Mg-M-CO3 layered double hydroxides for hightemperature CO2 capture. ChemSusChem 2010, 3, 965−973. (18) León, M.; Díaz, E.; Bennici, S.; Vega, A.; Ordóñez, S.; Auroux, A. Adsorption of CO2 on hydrotalcite-derived mixed oxides: sorption mechanisms and consequences for adsorption irreversibility. Ind. Eng. Chem. Res. 2010, 49, 3663−3671. (19) Oliveira, E. L. G.; Grande, C. A.; Rodrigues, A. E. CO2 sorption on hydrotalcite and alkali-modified (K and Cs) hydrotalcites at high temperatures. Sep. Purif. Technol. 2008, 62, 137−147. (20) Meis, N. N. A. H.; Bitter, J. H.; de Jong, K. P. On the influence and role of alkali metals on supported and unsupported activated hydrotalcites for CO2 sorption. Ind. Eng. Chem. Res. 2010, 49, 8086− 8093. (21) Ding, Y.; Alpay, E. High temperature recovery of CO2 from flue gases using hydrotalcite adsorbent. Process Saf. Environ. Prot. 2001, 79, 45−51. (22) Ram Reddy, M. K.; Xu, Z. P.; Diniz da Costa, J. C. Influence of water on high-temperature CO2 capture using layered double hydroxide derivatives. Ind. Eng. Chem. Res. 2008, 47, 2630−2635. (23) Othman, M. R.; Rasid, N. M.; Fernando, W. J. N. Mg−Al hydrotalcite coating on zeolites for improved carbon dioxide adsorption. Chem. Eng. Sci. 2006, 61, 1555−1560. (24) Aschenbrenner, O.; McGuire, P.; Alsamaq, S.; Wang, J.; Supasitmongkol, S.; Al-Duri, B.; Styring, P.; Wood, J. Adsorption of carbon dioxide on hydrotalcite-like compounds of different compositions. Chem. Eng. Res. Des. 2011, 89, 1711−1721. (25) Meis, N. N. A. H.; Bitter, J. H.; de Jong, K. P. Support and size effects of activated hydrotalcites for precombustion CO2 capture. Ind. Eng. Chem. Res. 2009, 49, 1229−1235. (26) Garcia-Gallastegui, A.; Iruretagoyena, D.; Mokhtar, M.; Asiri, A. M.; Basahel, S. N.; Al-Thabaiti, S. A.; Alyoubi, A. O.; Chadwick, D.; Shaffer, M. S. P. Layered double hydroxides supported on multi-walled carbon nanotubes: preparation and CO2 adsorption characteristics. J. Mater. Chem. 2012, 22, 13932−13940. (27) Garcia-Gallastegui, A.; Iruretagoyena, D.; Gouvea, V.; Mokhtar, M.; Asiri, A. M.; Basahel, S. N.; Al-Thabaiti, S. A.; Alyoubi, A. O.; Chadwick, D.; Shaffer, M. S. P. Graphene oxide as support for layered double hydroxides: enhancing the CO2 adsorption capacity. Chem. Mater. 2012, 24, 4531−4539. (28) Iruretagoyena, D.; Shaffer, M. P.; Chadwick, D. Adsorption of carbon dioxide on graphene oxide supported layered double oxides. Adsorption 2014, 20, 321−330. (29) Do, D. D. Adsorption Analysis: Equilibria and Kinetics; Imperial College Press: London, UK, 1998. (30) Hutson, N. D.; Speakman, S. A.; Payzant, E. A. Structural effects on the high temperature adsorption of CO2 on a synthetic hydrotalcite. Chem. Mater. 2004, 16, 4135−4143.

ACKNOWLEDGMENTS The authors are grateful to Xiaowen Huang, Raul Montesano, and Ainara Garcia-Gallastegui for discussions and assistance. D.I. thanks CONACyT and SEP for the scholarships awarded and EPSRC for additional funding.



ABBREVIATIONS LDH = layered double hydroxide LDO = layered double oxide GO = graphene oxide CNF = carbon nanofiber MWCNTs = multiwalled carbon nanotubes subscript MW = minimum washing XRD = X-ray diffraction TGA = thermogravimetric analysis TEM = transmission electron microscopy SEM = scanning electron microscopy ICP-OES = inductively coupled plasma optical emission BET = Brunauer, Emmett, and Teller CO2-TPD = temperature-programmed desorption of CO2 m = monolayer capacity b = gas−solid interaction parameter in Toth and Langmuir isotherms k = prefactor in Freundlich isotherm n = fitting parameter in Freundlich isotherm PCO2 = partial pressure of CO2 in the feed q = adsorption capacity q0 = adsorption capacity in the first cycle t = exponent constant in Toth equation



REFERENCES

(1) Choi, S.; Drese, J. H.; Jones, C. W. Adsorbent materials for carbon dioxide capture from large anthropogenic point sources. ChemSusChem 2009, 2, 796−854. (2) Wang, Q.; Luo, J.; Zhong, Z.; Borgna, A. CO2 capture by solid adsorbents and their applications: current status and new trends. Energy Environ. Sci. 2011, 4, 42−55. (3) Bhatta, L. K. G.; Subramanyam, S.; Chengala, M. D.; Olivera, S.; Venkatesh, K. Progress in hydrotalcite like compounds and metalbased oxides for CO2 capture: a review. J. Cleaner Prod. 2015, 103, 171−196. (4) Harrison, D. P. Sorption-enhanced hydrogen production: a review. Ind. Eng. Chem. Res. 2008, 47, 6486−6501. (5) van Selow, E. R.; Cobden, P. D.; Verbraeken, P. A.; Hufton, J. R.; van den Brink, R. W. Carbon capture by sorption-enhanced water−gas shift reaction process using hydrotalcite-based material. Ind. Eng. Chem. Res. 2009, 48, 4184−4193. (6) Halabi, M. H.; de Croon, M. H. J. M.; van der Schaaf, J.; Cobden, P. D.; Schouten, J. C. Kinetic and structural requirements for a CO2 adsorbent in sorption enhanced catalytic reforming of methanePart I: Reaction kinetics and sorbent capacity. Fuel 2012, 99, 154−164. (7) Ding, Y.; Alpay, E. Adsorption-enhanced steam-methane reforming. Chem. Eng. Sci. 2000, 55, 3929−3940. (8) Reijers, H. T. J.; Boon, J.; Elzinga, G. D.; Cobden, P. D.; Haije, W. G.; Brink, R. W. v. d. Modeling study of the sorption-enhanced reaction process for CO2 capture. II. Application to steam-methane reforming. Ind. Eng. Chem. Res. 2009, 48, 6975−6982. (9) Jang, H. M.; Lee, K. B.; Caram, H. S.; Sircar, S. High-purity hydrogen production through sorption enhanced water gas shift reaction using K2CO3-promoted hydrotalcite. Chem. Eng. Sci. 2012, 73, 431−438. (10) Debecker, D. P.; Gaigneaux, E. M.; Busca, G. Exploring, tuning, and exploiting the basicity of hydrotalcites for applications in heterogeneous catalysis. Chem.Eur. J. 2009, 15, 3920−3935. 6791

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792

Article

Industrial & Engineering Chemistry Research (31) Rourke, J. P.; Pandey, P. A.; Moore, J. J.; Bates, M.; Kinloch, I. A.; Young, R. J.; Wilson, N. R. The real graphene oxide revealed: stripping the oxidative debris from the graphene-like sheets. Angew. Chem. 2011, 50, 3173−3177. (32) Rouquerol, F.; Rouquerol, J.; Sing, K. Adsorption by Powders and Porous Solids: Principles, Methodology, and Applications; Academic Press: London, UK, 1999. (33) Millange, F.; Walton, R. I.; O’Hare, D. Time-resolved in situ Xray diffraction study of the liquid-phase reconstruction of Mg−Al− carbonate hydrotalcite-like compounds. J. Mater. Chem. 2000, 10, 1713−1720. (34) Freeman, E. S. The kinetics of the thermal decomposition of sodium nitrate and of the reaction between sodium nitrite and oxygen. J. Phys. Chem. 1956, 60, 1487−1493. (35) Debecker, D. P.; Gaigneaux, E. M.; Busca, G. Exploring, tuning, and exploiting the basicity of hydrotalcites for applications in heterogeneous catalysis. Chemistry 2009, 15, 3920−3935. (36) Di Cosimo, J. I.; Apesteguıa, C. R.; Ginés, M. J. L.; Iglesia, E. Structural requirements and reaction pathways in condensation reactions of alcohols on MgyAlOx Catalysts. J. Catal. 2000, 190, 261−275. (37) Falconer, J. L.; Schwarz, J. A. Temperature-programmed desorption and reaction: applications to supported catalysts. Catal. Rev.: Sci. Eng. 1983, 25, 141−227. (38) Redhead, P. A. Thermal desorption of gases. Vacuum 1962, 12, 203−211. (39) Walspurger, S.; Boels, L.; Cobden, P. D.; Elzinga, G. D.; Haije, W. G.; van den Brink, R. W. The crucial role of the K+-aluminium oxide interaction in K+-promoted alumina- and hydrotalcite-based materials for CO2 sorption at high temperatures. ChemSusChem 2008, 1, 643−650. (40) Di Cosimo, J. I.; Díez, V. K.; Apesteguía, C. R. Base catalysis for the synthesis of α,β-unsaturated ketones from the vapor-phase aldol condensation of acetone. Appl. Catal., A 1996, 137, 149−166. (41) Abelló, S.; Medina, F.; Tichit, D.; Pérez-Ramírez, J.; Rodríguez, X.; Sueiras, J. E.; Salagre, P.; Cesteros, Y. Study of alkaline-doping agents on the performance of reconstructed Mg−Al hydrotalcites in aldol condensations. Appl. Catal., A 2005, 281, 191−198. (42) Gao, Y.; Zhang, Z.; Wu, J.; Yi, X.; Zheng, A.; Umar, A.; O’Hare, D.; Wang, Q. Comprehensive investigation of CO2 adsorption on MgAl-CO3 LDH-derived mixed metal oxides. J. Mater. Chem. A 2013, 1, 12782−12790. (43) Yang, J.-I.; Kim, J.-N. Hydrotalcites for adsorption of CO2 at high temperature. Korean J. Chem. Eng. 2006, 23, 77−80.

6792

DOI: 10.1021/acs.iecr.5b01215 Ind. Eng. Chem. Res. 2015, 54, 6781−6792