Highly selective hydrogenation of CO2 into formic acid on a nano-Ni catalyst at ambient temperature: Process, mechanisms and catalyst stability

Highly selective hydrogenation of CO2 into formic acid on a nano-Ni catalyst at ambient temperature: Process, mechanisms and catalyst stability

Journal of CO2 Utilization 19 (2017) 157–164 Contents lists available at ScienceDirect Journal of CO2 Utilization journal homepage: www.elsevier.com...

2MB Sizes 0 Downloads 4 Views

Journal of CO2 Utilization 19 (2017) 157–164

Contents lists available at ScienceDirect

Journal of CO2 Utilization journal homepage: www.elsevier.com/locate/jcou

Highly selective hydrogenation of CO2 into formic acid on a nano-Ni catalyst at ambient temperature: Process, mechanisms and catalyst stability Chuan-Shu He, Li Gong, Jie Zhang, Pan-Pan He, Yang Mu* CAS Key Laboratory of Urban Pollutant Conversion, Collaborative Innovation Centre of Suzhou Nano Science and Technology, Department of Chemistry, University of Science and Technology of China, Hefei, China

A R T I C L E I N F O

Article history: Received 23 January 2017 Received in revised form 14 March 2017 Accepted 15 March 2017 Available online xxx Keywords: Carbon dioxide reduction Formic acid Hydrogenation Nano-Ni particle

A B S T R A C T

The hydrogenation of CO2 with gaseous hydrogen is currently believed to be the most commercially feasible synthetic method to resolve the serious worldwide greenhouse gas effects. However it is suffering from several disadvantages, such as expensive complex catalysts, and high energy consumption. In this study, a novel method for the hydrogenation of CO2 into formic acid with nanoNi catalyst is developed. The effects of various parameters on the capability of the catalyst for CO2 conversion were investigated. Furthermore, the HCO3 reduction process on the catalyst was rationalized theoretically with density functional theory simulations. The prepared nano-Ni was demonstrated to be an available catalyst for CO2 conversion into formic acid by employing H2 as a hydrogen source at ambient temperature and almost constant pressure. Moreover, the nano-Ni catalyst displayed good tolerance with pH variations in CO2 reduction. The formation of formic acid from CO2 reduction on nano-Ni particles was enhanced with an increase NaHCO3 concentration and catalyst dosage. Additionally, theoretical analysis elucidated that the hydrogenation of CO2 into formic acid on nano-Ni catalyst was favorable over attacking the C of HCO3 by the active H and hydroxyl group. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction While offering the majority of the current energy supply, fossil fuels also emit alarming levels of CO2, leading to serious worldwide greenhouse gas effects. CO2 is a thermodynamically and chemically stable compound [1], and thus, noticeable energy input and catalysis are necessary for its reduction [2,3]. To overcome this problem, many promising methods have already been previously proposed [4,5]. Catalytic conversion of CO2 into desired products combined with solar energy is regarded as the most promising approach. Although researchers have achieved obvious advancement in photocatalytic CO2 conversion, a low efficiency and expensive or complex catalyst make satisfying practical applications an elusive goal. The hydrogenation of CO2 with gaseous hydrogen is currently believed to be the most commercially feasible synthetic method [6,7]. Fischer–Tropsch (F–T) synthesis is an attractive alternative

* Corresponding author. E-mail address: [email protected] (Y. Mu). http://dx.doi.org/10.1016/j.jcou.2017.03.012 2212-9820/© 2017 Elsevier Ltd. All rights reserved.

approach for the conversion of CO2 and H2 into high quality hydrocarbons [8]. Fe- and Co-based catalysts are good choices for the F-T synthesis due to their good activities and selectivity [9]. In the Fe-based catalyst systems, Pt, an important promoter, could improve the CO2 uptake and selectivity towards olefins and longchain hydrocarbons [10]. When Al2O3 was added as a structural promoter in this system, enhancement of CO2 conversion as well as high selectivity to C2+ hydrocarbons were also observed [10]. It has been reported that Cu-based heterogeneous materials were able to successfully catalyze CO2 hydrogenation into CH3OH [1,2]. Moreover, many noble metals such as Pd, Ru and Rh could also be adopted as catalysts for formic acid production from CO2 conversion [11,12] as summarized in Table S1 in the supporting information (SI). Precious metals (Rh, Ru, Ir, Pd) are the most commonly used catalysts for CO2 conversion into formic acid as they contain homogeneous materials predominating with the water soluble ligands, such as phosphine ligands, pincer ligands, Nheterocyclic carbene ligands, proton-responsive ligands [13]. However, besides for the application of noble metals and noncheap organic ligands, the synthesis of these homogenous catalysts was complicated along with high energy consumption [14]. Special

158

C.-S. He et al. / Journal of CO2 Utilization 19 (2017) 157–164

solvents such as DMSO and MeOH were often used to stabilize the catalytic intermediate or exert and influence the entropy value [13]. Additionally, in order to achieve reasonable rates and conversions, many of these reactions should be conducted at elevated temperatures and/or pressures, leading to dramatic increase in energy consumption. In recent years, Ni-based catalysts have attracted intense attention because nickel is comparatively inexpensive and readily available, which is suitable for large scale industrial applications. It has been proven to efficiently break down the C–C bond and dissociate the hydrogen molecules [15,16]. It also has efficient activity close to the noble metals [4,17]. Recently, commercially available Ni powder catalyst was adopted for CO2 reduction with N2H4H2O as a hydrogen source, achieving a 50% yield of formic acid with 99% selectivity [18]. However, extra heat energy is required in order to maintain the reaction temperature at 300  C for the Ni powder catalyst. Compared to the powder catalyst, nanoparticles have higher specific surface area and better catalytic activity, which might be able to compensate for the extra heat required during the hydrogenation of CO2. However, to the best of our knowledge, there are no reported studies on CO2 hydrogenation into formic acid with nano-Ni particles as the catalyst. In this study, we have successfully demonstrated the prepared nano-Ni to be a highly efficient catalyst for the hydrogenation of CO2 into formic acid, a raw chemical that has the potential to compensate for the cost of the reaction, at ambient temperature and almost constant pressure. The effects of various parameters on the capability of the catalyst for CO2 conversion were investigated. Furthermore, the CO2 reduction process on Ni nanoclusters was rationalized theoretically by employing density functional theory (DFT) calculations. Additionally, the stability of nano-Ni catalyst as well as its recovery was also evaluated in this study.

overnight at 65  C under vacuum. The characterization of the synthesized nano-Ni particles is shown in the SI. 2.3. Batch tests for CO2 reduction

The chemical reagents (Sinopharm Chemical Reagent Co., Ltd, Shanghai, China) used in this study were of analytical grade, including formic acid, ethanol, N2H4H2O, NiCl26H2O, NaOH, NaHCO3, H2SO4 and HCl, except for 96% NaBH4. Nano-Ni and nanoNiO were purchased from Aldrich (Sigma–Aldrich Chemical Co., China). The reagents were used directly without any pretreatment. High purity compressed H2, N2, and CO2 gases were obtained from the Shangyuan Gas Products Co., Ltd (Nanjing, China). All chemical stock solutions were prepared using purified water and stored at 4  C.

Since the preliminary experiments showed that the major product was the same for hydrogenation of CO2 and NaHCO3with nano-Ni catalysts, NaHCO3 was adopted as the CO2 source for simplification in this study. After purging with N2 in advance for approximately 10 min to remove the dissolved oxygen, 35 mL of NaHCO3 solution was added into a 75 mL serum bottle first. Then, the reaction vial was flushed with high purity N2 for approximately 10 min to remove dissolved oxygen, followed by transferring it into the anaerobic chamber for the addition of Ni nanoparticles. When 5 min of aeration with high purity H2 was completed, the serum bottles were sealed with rubber stoppers and aluminum caps. For insurance, additional H2 was supplied into the bottles to confirm that enough H2 was added using an injector. Finally, the initial H2 pressure was 1.2–1.5 times the atmospheric pressure. In the typical reduction experiment of aqueous CO2, the reaction conditions were controlled at 10 mM NaHCO3, 0.5 M Ni and a pH of 6 with continuous H2 aeration, where the H2 was continuously added into the bottle until the initial pressure was achieved every 24 h. Then, the tests were conducted in the shaker at 35  C with a shaking speed of 220 rpm and shielded from light. To probe the differences of formic acid production caused by various parameters and attain the optimized parameters, pH values, catalysts and reactants were separately evaluated as variables, and the other conditions were controlled at the typical experiment conditions. The initial pH value in a range of 3.8–9.8 was adjusted by using HCl or NaOH solution after adding NaHCO3 into the solution in this study. To choose an optimum catalyst dosage, 0.05, 0.25, 0.5, 1.0, and 1.5 M Ni NPs were evaluated. In the tests of the catalyst stability, after each cycle, 0.5 M used Ni particles were collected through centrifugation and further used directly after 3-min of ultrasonic pretreatment. Concentrations of 1, 10, 25, 50, 100, 200, 300, 400, and 500 mM were separately set up as the initial NaHCO3 concentrations to explore the reduction capacity of Ni nanoparticles. The H2 aerated initial experiments were compared with continuous H2 aeration to estimate the effect of the reductant, where the H2 was aerated initially to 1.2–1.5 times the atmospheric pressure and no further H2 aeration was carried out. As for the test without H2, no H2 aeration was carried out after the addition of nano-Ni. In the H2 and CO2 adsorption tests, 1 g targeted nanoparticles were added into 35 mL deoxygenated water with keeping pH at around 8.5, following with initial H2 or CO2 aeration for about 5 min and then sealed operation. All tests were administered at least three times.

2.2. Catalyst synthesis and characterization

2.4. Analysis and calculations

Nano-Ni particles were freshly prepared by using the liquid phase reducing method, similar to the synthesis of nano-zero valent iron [19]. In brief, the reductant was prepared by dissolving 14.16 g of NaBH4 and 1 g of NaOH into 500 mL of purified water, which was sparged with N2 for 20 min in advance. Additionally, nickel chloride solution was prepared by adding 38.64 g of NiCl26H2O into a 400 mL mixture of ethanol and water (Vethanol/Vwater = 4:1), and then, it was transferred into a threenecked bottle for titration. A NaBH4 reducing agent solution was added drop-wise and slowly into a nickel chloride solution under the protection of highly purified N2 with magnetic stirring at 800 rpm at room temperature. The vigorous stirring was continued for an extra 30 min after the reductant titration was finished, and then, the products were washed 5 times with water and dried

The liquid products of CO2 reduction were determined by using high-performance liquid chromatography (HPLC) equipped with a UV detector and a 300  7.7 mm Hi-Plex H column at 55  C (Agilent Technologies, USA). The mobile phase was 5 mM H2SO4 at a flow rate of 0.6 mL/min, and the detection wavelength was 210 nm. H2 and CO2 were measured by a gas chromatograph (GC) (Model SP-6800A, Lunan Co., China) using a gas-tight syringe according to Sheng and Yu [20]. An Agilent 6890 N GC equipped with a flame ionization detector and Plot Q-19095P-Q04 column (Agilent Technologies, USA) was used to detect the generation of hydrocarbons in the headspace [21]. Both the total carbon (TC) and total organic carbon (TOC) were measured with a TOC analyzer (MultiN/C 2100, Analytik Jena, Germany). The simulation of NaHCO3 consumption and the determination of formic acid yield (FAY, %) are provided in the SI. In order to demonstrate the state of

2. Experimental 2.1. Chemicals

C.-S. He et al. / Journal of CO2 Utilization 19 (2017) 157–164

159

Fig. 1. Characterization of freshly prepared Ni nanoparticles: (a) SEM, (b) TEM, (c) HRTEM, and (d) XRD.

the art carefully, the turnover number (TON) of formic acid, which is the moles of formic acid produced per mole of nano-Ni used, was also applied to express the yield to explore the impacts of various parameters. 2.5. DFT calculations A Ni13 cluster was chosen for a Ni nanoparticle in DFT calculation since it exhibits a high binding energy as well as structural stability [22], and it has been successfully used in different theoretical studies [23–25]. A Ni13 nanocluster has one atom at the center and 12 other identical atoms on the spherical shell surface with a coordination number of 6, resulting in an Ih symmetric structure. The structure with Ih symmetry belong to a regular icosahedral structure, which has all the (1 1 1) facets on the surface. Therefore, the isolated Ni13 nanoparticle was optimized in a 15.0 Å cubic supercell in this study. DFT calculations were performed using plane-wave basis sets and ultrasoft pseudopotentials as implemented in the CASTEP module of Materials Studio [26]. The details for DFT calculations are provided in the SI. 3. Results and discussion 3.1. Properties of the nano-Ni catalyst As shown in Fig. 1a and b, the freshly synthesized nano-Ni was shown to be a nano-chain with an average size of approximately 15 nm, in the range of particle sizes (10–20 nm) for perfect activity according to a previous study [27]. The surface of the Ni nanoparticles was coated with a thick shallow layer from the HRTEM image (Fig. 1c). The original XRD peaks of the synthesized nanoparticles were clearly multi-fitted. According to the integrated XRD data (Fig. 1d), the diffraction peaks at 2u = 44.8 and 51.8 could be assigned to the characteristic (111) and (200) planes of face-centered cubic (fcc) Ni phase (JCPDS 04–0850) [28],

Fig. 2. (a) H2 adsorption on the freshly prepared and Aldrich nano-Ni particles, (b) CO2 adsorption on the freshly prepared and Aldrich nano-Ni as well as Aldrich nanoNiO (0.5 M Ni, 35  C).

160

C.-S. He et al. / Journal of CO2 Utilization 19 (2017) 157–164

coincident with the Ni supported by MgO for hydrogenation of CO2 into CH4 [29]. Two small diffraction patterns observed at 2u values of 34.0 and 60.8 were well indexed to the (100) and (111) planes of hexagonal Ni(OH)2 (JCPDS 14-0117) [30]. The reflection at 2u = 37.2 was attributed to the presence of a crystalline NiO phase (JCPDS 78-0643), while another prominent peak (43.3 ) of the NiO phase might be covered by the broad peak of Ni (111) [31]. Therefore, it is believed that the shell layer of nano-Ni particles was composed of a thin oxide phase. Additionally, the BET surface area and pore volume of the prepared nano-Ni catalyst were as high as 104.4 m2/g and 0.26 cm3/g, respectively, which can provide enough active sites for gas adsorption. CO2 conversion is determined by the amount of CO2 adsorbed and the selectivity is decided by the amount of H2 adsorbed, while Ni has been proved to be in favor of H2 adsorption [32]. As shown in Fig. 2, our freshly prepared Ni nanoparticles showed better performance for CO2 adsorption as compared to the commercial one, which might arise from the NiO layer formed on the surface of Ni nanoparticles for providing more active sites for CO2 adsorption [33]. As a consequent of high surface area and NiO layer, our prepared nano-Ni displayed superior ability for both H2 and CO2 adsorption than commercial nano-Ni, and thus may facilitate the hydrogenation of CO2 into useful hydrocarbon when using the prepared nano-Ni. Besides it is worth noting that no hydrogenation products were detected in the typical experiments with commercial nano-Ni as catalysts. 3.2. Typical reduction of CO2 on nano-Ni particles In a typical experiment (Fig. 3), the concentration of formic acid was rapidly increased from zero to approximately 5.12  0.11 mM

Fig. 3. Typical curve of formic acid formation from CO2 reduction catalyzed by Ni nanoparticles using H2 as a reductant (10 mM NaHCO3, 0.5 M Ni, initial pH 6, 35  C, continuous H2 aeration).

during the initial 75 h, indicating a high catalytic activity of Ni nanoparticles. However, the formation of formic acid improved slowly after 75 h. The highest concentration of formic acid reached 6.74  0.17 mM, resulting in a formic acid yield of 67.42 1.65%. This observation is in agreement with results for the CO2 reforming of CH4 via bare Ni catalyst, which may be due to a carbon layer that was formed on the Ni particles for deactivation during the reaction [34,35]. Fig. 3 presents the concentrations of NaHCO3 remaining in the aqueous reaction system, corresponding to the production of formic acid. It was found that the decay kinetics of NaHCO3 was

Fig. 4. Effects of various parameters on the yield of formic acid: (a) initial pH (10 mM NaHCO3, 0.5 M Ni, 35  C, continuous H2 aeration), (b) catalyst dosage (10 mM NaHCO3, initial pH 6, 35  C, continuous H2 aeration), (c) initial NaHCO3 concentration (0.5 M Ni, initial pH 6, 35  C, continuous H2 aeration), (d) H2 supply method (10 mM NaHCO3, 0.5 M Ni, initial pH 6, 35  C).

C.-S. He et al. / Journal of CO2 Utilization 19 (2017) 157–164

161

well described by the first-order reaction with a rate constant (k1) of 0.0468 h1 and regression coefficient (R2) of 0.8961, respectively. In this typical experiment, the amount of total carbon after the 48.5 h and 240 h reaction was 115.62  0.69 and 118.9  1.2 mg/L in the solution respectively, which was consistent with the added amount (10 mM NaHCO3, namely 120 mg/L TC). It's found that the measured TOC concentration after the reaction (48.5 h: 42.57  0.07 mg/L; 240 h: 82.0  2.0 mg/L) was comparable to that of formic acid (48.5 h: 43.75  0.41; 240 h: 78.4  1.0 mg/L). During the course of the experiment no hydrocarbon was detected in the gas phase. These results indicate an approximately 100% selectivity for the production of formic acid from CO2 using nano-Ni as the catalyst. 3.3. Impact of various parameters As shown in Fig. 4a, the formic acid yield was progressively but slowly decreasing with an increase in the initial pH in spite of the reaction time. Formic acid yields of 68.11  0.91% and 59.15  3.34% were achieved after 240 h at the lowest and highest pH values, respectively. As shown in Fig. 5, the first-order kinetic constant of NaHCO3 conversion and TON of formic acid were reduced from 0.00484 to 0.00388 h1 and 0.0136 to 0.0118 respectively when pH increased from 3.8 to 9.8, suggesting that solution pH had a slightly negative impact on the formation of formic acid. Different results were obtained by Jin et al. in zinc-catalyzed experiments, namely nearly the absence of formic acid without NaOH addition, that is in acidic pH values [36]. This is most likely due to the low dissolution of CO2 in water. As for the high reduction effect of CO2 in acidic condition in our studies, relatively higher BET surface areas of nano-Ni particles have resulted in good adsorption of CO2. In general, these results indicate that the activity of the nano-Ni catalyst was insensitive to pH values for CO2 conversion, and a faster or more effective rate could be achieved in acidic pH conditions. As shown in Fig. 4b, the formic acid yield increases linearly from 17.06  1.11% to 98.52  2.96% with an increase in catalyst dosage from 0.05 to 1 M and plateaued when the addition of Ni reached 1.5 M. Similarly, the kinetic constant of NaCO3 conversion remarkably increased from 0.00075 to 0.02364 h1 with the Ni dosage increasing from 0.05 M to 1 M but decrease slightly to 0.02110 h1 at 1.5 M. Conversely, the TON of formic acid (Fig. 5b) decreased gradually with the increase in the Ni dosage. Based on the reaction kinetics and TON of formic acid, the Ni dosage around the crossover point (0.6 M) in Fig. 5b would be the optimal dosage for CO2 hydrogenation in this case. The impact of the initial NaHCO3 concentration on the formic acid production is presented in Fig. 4c. The formic acid concentration increased from 0.998  0.045 to 51.77  2.46 mM when the initial NaHCO3 concentration was increased from 1 to 500 mM, resulting in a reduction of the formic acid yield from 99.84  4.02% to 10.35  0.49%. As illustrated in Fig. 5, the NaHCO3 consumption rate constant was typically reduced from 0.02110 to 0.00051 h1, as opposed to its initial concentration. Hydrogenation is thus slowing down, as indicated by the lower yield and smaller reaction rate constant at higher NaHCO3 concentrations. This finding may be observed because most of the adsorption sites are occupied by the HCO3 and few are available for H2 adsorption, which hinders the reaction [37,38]. Nevertheless, as can be seen from Fig. 5c, a higher NaHCO3 concentration would enhance the TON of formic acid. As illustrated in Fig. 4d, a relatively higher formic acid yield was observed when keeping an abundant H2 content compared to H2 only being supplied at the initial stage, where the H2 pressure decreased from 1.27 to 0.72 atm after 300 h (Figure S1). Moreover, the rate constant of NaHCO3 conversion and TON of formic acid

Fig. 5. Effect of initial pH value (a), Ni catalysts dosage (b) and NaHCO3 concentration (c) on the first-order reaction rate constant and the TON for the hydrogenation of NaHCO3 to formic acid.

with continuous aeration of H2 (0.0468 h1 and 0.0126) were slightly higher compared to the case with initial H2 supply (0.00381 h1and 0.0110). Interestingly, in the non-H2 supplied system, about 3% of NaHCO3 was reduced into formic acid (Fig. 4d), which might arise from the small amounts of H2 produced by Ni and H2O as preliminary experiments displayed (data not shown). As can be seen from Table S1, most of the maximum TON values reported are much higher than that in this study, but use of noble metal, dose of organic solvents or additives, high energy input due to achieve high temperature and pressure in those previous studies would not only greatly enhance the cost but also be environmentally unfriendly. In addition, only the amount of noble metals, 1.0– 6.0 wt% of the catalyst, was adopted to estimate the TON value for several solid catalysts such as Ru-TiO2, Ru(OH)[email protected] or activated carbon [39,40], resulting in a high TON value for CO2 reduction to formic acid. The maximum TON value for CO2 conversion using

162

C.-S. He et al. / Journal of CO2 Utilization 19 (2017) 157–164

Scheme 1. Pathways of CO2 reduction process on Ni13 nanocluster in aqueous solutions.

nano-Ni particles at 35  C in our study was comparable with that in two similar systems using micro-size Ni [18] or Zn [26] as the catalyst but with a much higher temperature (300  C) to initiate the hydrogenation process. 3.4. DFT calculations for CO2 reduction on Ni nanoparticles Aiming at understanding the elementary steps involved in formic acid formation from CO2 in the form of HCO3 and hydrogen on the nano-Ni catalyst, the adsorption and hydrogenation of CO2 on the Ni surface were studied using DFT calculations. The adsorption of CO2 and H2 as well as their co-adsorption on the Ni surface were initially explored, and the results are shown in the SI. Based on the reaction energetics and activation barriers of

individual elementary steps, possible reaction pathways for CO2 hydrogenation on the nano-Ni surface were mapped out and compared for the determination of more favorable pathways. For the different attack sites of HCO3, the reduction of HCO3 may proceed through two pathways with corresponding intermediates, as depicted in Scheme 1. The geometric structures of the optimized reaction intermediates of both pathways in the HCO3 reduction process are summarized in Fig. 6. In pathway 1, the C of HCO3 is attacked by the active H as illustrated in the State 3 of Fig. 6, and the hydroxyl group is leaving from the HCO3 molecule [41]. Finally, the other active sites of H2 combined with the hydroxyl group and formic acid are obtained together with a H2O, as shown in the State 4 of Fig. 6. Meanwhile, the active H attacks the O of HCO3 in pathway 2 (State 5 in Fig. 6.). Afterwards, a H2O molecule is formed

Fig. 6. Optimized structures of intermediates in the HCO3 hydrogenation process on Ni13 nanoclusters. HCO3 hydrogenation to formic acid: pathway 1, step 1 ! 2 ! 3 ! 4; pathway 2, step 1 ! 2 ! 5 ! 6 ! 4. From state 2 to state 7, oxidized nickel is formed on Ni13 by withdrawing an O atom from HCO3. Afterwards, oxidized nickel is reduced into nickel by active H atoms of H2, resulting in state 8. For convenience, we named the process of step 1 ! 2 ! 7 ! 8 as the NiO pathway.

C.-S. He et al. / Journal of CO2 Utilization 19 (2017) 157–164

163

Fig. 7. Calculated energy profile of the HCO3 reduction process on Ni13 nanoclusters. TS represents transition states.

with the H and hydroxyl group of the HCO3 molecule, with CO2 left on the Ni13 nanocluster as shown in the State 6 of Fig. 6. Then, the other active H of H2 combines with the leaving CO2 of HCO3, and formic acid is finally obtained (State 4 in Fig. 6). The free energies (G) of the intermediates in each pathway of the HCO3 hydrogenation process over Ni13 are given in Table S2 in detail, and the free energy changes of each step in the HCO3 reduction process are calculated in Table S7. To gain further insights into the HCO3 reduction process on Ni13 nanoclusters, the energy profiles for all steps through pathways 1 and 2 were plotted for comparison in Fig. 7. The process of HCO3 reduction to formic acid, namely from State 1 to State 4 in Fig. 7, was exothermic by 2.74 eV, which means that the process of the reduction of HCO3 to formic acid over Ni13 catalyst occurs spontaneously. In pathway 1, the rate-controlling step lays in the attack of C by the active H (i.e., step 2 ! 3) with an Ea of 3.24 eV and a DG of 0.55 eV. Meanwhile, in pathway 2, the attack of the O atom of HCO3 (i.e., step 2 ! 5) was the rate-controlling step, whereas in homogeneous catalyst cases the insertion of CO2 into the M–H bond (M = metal complex) was suggested to be the ratelimiting one [42]. The energy barrier and Gibb's free energy change of this step were 4.02 and 2.32 eV, respectively, which are both higher than these values of step 2 ! 3 in the pathway 1. In summary, the process of HCO3 reduction to formic acid over the Ni13 catalyst was likely to be more favorable to perform through pathway 1.

coexistence of Ni0 and Ni2+ on the surface of fresh Ni. However, the Ni0 disappeared after reacting with NaHCO3, proving the oxidation of Ni on the surface with H+ or an O atom from HCO3 (Fig. 6). In fact, there is no Ni3C peak displayed in Figure S2 and no extra C increase in the C1s spectrum (not shown here). Additionally, as shown in Figures S3 and S4, the freshly prepared Ni nanoparticles were coated by an oxide shell layer and a thin carbon phase, which was perhaps composed of the ethanol acting as a surfactant [37,45]. The oxygen elemental content was obviously enhanced after the reaction, which is caused by the oxidation of Ni. However, a relatively small increase in carbon was shown in Figures S3b and S4b, suggesting insufficient evidence for the formation of coke or other types of carbon on the face of catalyst after long exposure to CO2 hydrogenation conditions. In summary, the deactivation of Ni mainly resulted from the oxidation of Ni0 on the catalyst surface via reaction with H2O or H+, which was inferred by the detection of a small dosage of H2 in the coexistence of only Ni and water (data not shown here). Moreover, the oxide shell on the surface of metal particles has been found to make the catalyst more stable, offering the ability for CO2 reduction at a correspondingly slower rate after Ni oxidation [35]. Therefore, the oxidation of exposed Ni atoms might be responsible for lower but relatively steady CO2 conversion rate. H2 pre-treatment of reused Ni catalyst has been reported to lead to a partial reduction of the surface Ni2+ species, namely NiOxHy, forming Ni0 and making the catalyst reactivated [37]. In this study we calculated the thermodynamic and kinetic properties of the

3.5. Stability of the nano-Ni catalyst The cyclic stability of the catalyst is an important factor to be understood for practical applications [43]. As shown in Fig. 8, the catalyst showed a high catalytic activity for CO2 reduction with a formic acid yield of approximately 65% at 200 h in the first cycle. However, a reduction in the catalytic activity of Ni nanoparticles was found from the second cycle, and the formic acid yield was maintained in a range of 30% to 35% during the following three cycles. It is well-known that the hydrogenation process occurs only at metallic Ni0 sites rather than at other Ni-species, such as oxides, hydroxides or carbides [44]. The reduction of catalyst activity could be due to the deactivation of nickel caused by the oxide of Ni on the surface. In addition, the carbon layers formed on but not adhered strongly to the Ni particles during the reaction by the Boudouard Reaction (2CO ! C + CO2) are also suggested to be responsible for the deactivation [34,35]. As shown in Figure S2, the XPS peaks located at 852.2 and 869.4 eV, which correspond to the Ni0 state, and other peaks corresponding to the Ni2+ state indicate the

Fig. 8. Stability of the Ni catalyst with four cycles (10 mM NaHCO3, 0.5 M Ni, initial pH 6, 35  C, continuous H2 aeration).

164

C.-S. He et al. / Journal of CO2 Utilization 19 (2017) 157–164

process from the formation of oxidized nickel to the recovery of nickel (i.e., steps 1 ! 2 ! 7 ! 8 in Fig. 6) to demonstrate that the durability of the nano-Ni catalyst would be enhanced through H2 pretreatment. The possible recovery process is shown in Scheme S1 of SI and was named the NiO pathway for convenience. The thermodynamic properties of the NiO pathway are given in Table S2, with the reaction intermediates shown in Fig. 6. Consequently, the energy profile of the NiO pathway was plotted in Fig. 7 for ease of comparison with the HCO3 reduction process on the Ni13 nanocluster. It is shown that the entire process from oxidized nickel formation to the recovery of nickel with a DG of 1.07 eV could occur spontaneously on the Ni catalyst. The oxidized nickel formation (i.e., step 2 ! 7) was both thermodynamically and kinetically more favorable with an Ea of 2.86 eV and a DG of 0.27 eV, both of which were lower than the step 2 ! 3 (Ea: 3.26 eV, DG: 0.55 eV) in the reduction process of HCO3. After that, the O atom of oxidized nickel was combined with activated H2 (i.e., step 2 ! 7) forming H2O, and the Ni catalyst was recovered. This step also has a lower energy barrier of 3.10 eV than that of the ratecontrolling step of the HCO3 reduction process on the Ni13 nanocluster. Such relative low energy barriers indicate that the Ni catalyst tends to be oxidized during CO2 reduction on Ni nanoparticles in aqueous solutions, but Ni nanoparticles could also be easily recovered with H2-pretreatment. 4. Conclusions In this study, a novel method for the hydrogenation of CO2 at ambient temperature and almost constant pressure was successfully developed by employing nano-Ni as the catalyst. The prepared nano-Ni particles were demonstrated to be highly selective (>99%) for CO2 conversion into formic acid by employing H2 as a hydrogen source, resulting in about 99.84% yield using 1 mM NaHCO3 as reactant. Moreover, the maximum TON value for CO2 conversion using nano-Ni particles at 35  C in our study was comparable with that in those similar systems but with a much higher reaction temperature. In addition, it was found that nano-Ni particles as the catalyst displayed good stability to pH variation. DFT calculations elucidated that the hydrogenation of CO2 on nanoNi particles preferred to attack the C of HCO3 by the active H and hydroxyl group. The simple and energy-saving methodology developed in this study has the great potential for further practical applications. Acknowledgements The authors wish to thank the Natural Science Foundation of China (51538012 and 51478446), the Recruitment Program of Global Experts, and the Fundamental Research Funds for the Central Universities for financially supporting this study. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcou.2017.03.012.

References [1] J. Choi, M.J. Kim, S.H. Ahn, I. Choi, J.H. Jang, Y.S. Ham, J.J. Kim, S.K. Kim, Chem. Eng. J. 299 (2016) 37–44. [2] M.Y. Xing, F. Shen, B.C. Qiu, J.L. Zhang, Sci. Rep. 4 (2014). [3] S. Gao, Y. Lin, X.C. Jiao, Y.F. Sun, Q.Q. Luo, W.H. Zhang, D.Q. Li, J.L. Yang, Y. Xie, Nature 529 (2016) 68–71. [4] Y.J. Fan, S.F. Wu, J. CO2 Utilization 16 (2016) 150–156. [5] P. Gao, H.Y. Yang, L.N. Zhang, C. Zhang, L.S. Zhong, H. Wang, W. Wei, Y.H. Sun, J. CO2 Utilization 16 (2016) 32–41. [6] S. Saeidi, N.A.S. Amin, M.R. Rahimpour, J. CO2 Utilization 5 (2014) 66–81. [7] S. Kiatphuengporn, M. Chareonpanich, J. Limtrakul, Chem. Eng. J. 240 (2014) 527–533. [8] D. Schanke, A. Hilmen, E. Bergene, K. Kinnari, E. Rytter, E. Ådnanes, A. Holmen, Catal. Lett. 34 (1995) 269–284. [9] Y.L. Yao, D. Hildebrandt, D. Glasser, X.Y. Liu, Ind. Eng. Chem. Res. 49 (2010) 11061–11066. [10] P.S. Prasad, J.W. Bae, K.W. Jun, K.W. Lee, Catal. Sury. Asia 12 (2008) 170–183. [11] Q. Kang, T. Wang, P. Li, L.Q. Liu, K. Chang, M. Li, J.H. Ye, Angew. Chem. Int. Ed. 54 (2015) 841–845. [12] Q.G. Zhai, S.J. Xie, W.Q. Fan, Q.H. Zhang, Y. Wang, W.P. Deng, Y. Wang, Angew. Chem. 125 (2013) 5888–5891. [13] W.H. Wang, Y. Himeda, J.T. Muckerman, G.F. Manbeck, E. Fujita, Chem. Rev. 115 (2015) 12936–12973. [14] R. Tanaka, M. Yamashita, K. Nozaki, J. Am. Chem. Soc. 131 (2009) 14168–14169. [15] T. Fujimoto, S. Ogawa, T. Kanai, N. Uchiyama, T. Yoshida, S. Yagi, Int. J. Hydrogen Energ. 40 (2015) 11890–11894. [16] M. Aziz, A. Jalil, S. Triwahyono, M. Saad, Chem. Eng. J. 260 (2015) 757–764. [17] T. Jiang, Q. Huai, T. Geng, W.Y. Ying, T.C. Xiao, F.H. Cao, Biomass Bioenerg. 78 (2015) 71–79. [18] F.Y. Chen, G.D. Yao, Z.B. Huo, F.M. Jin, RSC Adv. 5 (2015) 11257–11260. [19] S.M. Hosseini, B. Ataie-Ashtiani, M. Kholghi, Desalination 276 (2011) 214–221. [20] G.P. Sheng, H.Q. Yu, Appl. Microbiol. Biotechnol. 74 (2007) 208–214. [21] J.F. Xie, Y.X. Huang, W.W. Li, X.N. Song, L. Xiong, H.Q. Yu, Electrochim. ACTA 139 (2014) 137–144. [22] E.K. Parks, B.H. Weiller, P.S. Bechthold, W.F. Hoffman, G.C. Nieman, L.G. Pobo, S. J. Riley, J. Chem. Phys. 88 (1988) 1622–1632. [23] D.H. Lim, J. Wilcox, J. Phys. Chem. C. 116 (2012) 3653–3660. [24] Z. Zhuang, S.A. Giles, J. Zheng, G.R. Jenness, S. Caratzoulas, D.G. Vlachos, Y. Yan, Nat. Commun. 7 (2016) 10141. [25] P. Cheng, S. Zhang, P. Wang, S. Huang, H. Tian, Comput. Theor. Chem. 1020 (2013) 136–142. [26] M. Segall, P.J. Lindan, M.a. Probert, C. Pickard, P. Hasnip, S. Clark, M. Payne, J. Phys.: Condens. Matter 14 (2002) 2717. [27] M. Ermakova, D.Y. Ermakov, G. Kuvshinov, L. Plyasova, J. Catal. 187 (1999) 77– 84. [28] L. Ren, K.S. Hui, K.N. Hui, J. Mater. Chem. A 1 (2013) 5689–5694. [29] T. Nakayama, N. Ichikuni, S. Sato, F. Nozaki, Appl. Catal. A 158 (1997) 185–199. [30] M.Z. Hu, J.H. He, M.Q. Yang, X.C. Hu, C.X. Yan, Z.X. Cheng, RSC Adv. 5 (2015) 26823–26831. [31] A. Larimi, S. Alavi, Fuel 102 (2012) 366–371. [32] Y. Liu, D.Z. Liu, Int. J. Hydrogen Energ. 24 (1999) 351–354. [33] T. Matsumoto, J. Kubota, J.N. Kondo, C. Hirose, K. Domen, Langmuir 15 (1999) 2158–2161. [34] Y. Kathiraser, U. Oemar, E.T. Saw, Z. Li, S. Kawi, Chem. Eng. J. 278 (2015) 62–78. [35] D.H. Kim, S.Y. Kim, S.W. Han, Y.K. Cho, M.-G. Jeong, E.J. Park, Y.D. Kim, Appl. Catal. A 495 (2015) 184–191. [36] F.M. Jin, X. Zeng, J.K. Liu, Y.J. Jin, L.Y. Wang, H. Zhong, G.D. Yao, Z.B. Huo, Sci. Rep. 4 (2014) 4503. [37] I. Czekaj, F. Loviat, F. Raimondi, J. Wambach, S. Biollaz, A. Wokaun, Appl. Catal. A 329 (2007) 68–78. [38] M. Seemann, T. Schildhauer, S. Biollaz, S. Stucki, A. Wokaun, Appl. Catal. A 313 (2006) 14–21. [39] P. Upadhyay, V. Srivastava, RSC Adv. 6 (2016) 42297–42306. [40] C. Hao, S. Wang, M.S. Li, L.Q. Kang, X.B. Ma, Catal. Today 160 (2011) 184–190. [41] J.Y. Ye, C.J. Liu, Q.F. Ge, J. Phys. Chem. C 116 (2012) 7817–7825. [42] P.G. Jessop, F. Joó, C.-C. Tai, Coord. Chem. Rev. 248 (2004) 2425–2442. [43] S. Vijayakumar, G. Muralidharan, J. Electroanal. Chem. 727 (2014) 53–58. [44] J. Liu, J.P. Lu, P.W. Chu, J.M. Blakely, J. Vac. Sci. Technol. A 10 (1992) 2355–2360. [45] Y.J. Jo, S.J. Oh, S.S. Lee, Y.H. Seo, B.H. Ryu, D.H. Yoon, Y.M. Choi, S.H. Jeong, J. Mater. Chem. C. 3 (2015) 4842–4847.