Hui-Xin Zhang, Zheng-Qing Huang, Tao Ban, Xue Su, Bolun Yang, Chun-Ran Chang. DFT Studies of CO Reaction Behavior on α-Fe2O3(001) Oxygen-Vacancy Surface in Chemical Looping Reforming[J]. Chinese Journal of Chemical Physics , 2024, 37(1): 116-124. DOI: 10.1063/1674-0068/cjcp2304028
Citation:
Hui-Xin Zhang, Zheng-Qing Huang, Tao Ban, Xue Su, Bolun Yang, Chun-Ran Chang. DFT Studies of CO Reaction Behavior on α-Fe2O3(001) Oxygen-Vacancy Surface in Chemical Looping Reforming[J]. Chinese Journal of Chemical Physics , 2024, 37(1): 116-124. DOI: 10.1063/1674-0068/cjcp2304028
Hui-Xin Zhang, Zheng-Qing Huang, Tao Ban, Xue Su, Bolun Yang, Chun-Ran Chang. DFT Studies of CO Reaction Behavior on α-Fe2O3(001) Oxygen-Vacancy Surface in Chemical Looping Reforming[J]. Chinese Journal of Chemical Physics , 2024, 37(1): 116-124. DOI: 10.1063/1674-0068/cjcp2304028
Citation:
Hui-Xin Zhang, Zheng-Qing Huang, Tao Ban, Xue Su, Bolun Yang, Chun-Ran Chang. DFT Studies of CO Reaction Behavior on α-Fe2O3(001) Oxygen-Vacancy Surface in Chemical Looping Reforming[J]. Chinese Journal of Chemical Physics , 2024, 37(1): 116-124. DOI: 10.1063/1674-0068/cjcp2304028
Shaanxi Key Laboratory of Energy Chemical Process Intensification, School of Chemical Engineering and Technology, Xi’an Jiaotong University, Xi’an 710049, China
2.
Shaanxi Key Laboratory of Low Metamorphic Coal Clean Utilization, School of Chemistry and Chemical Engineering, Yulin University, Yulin 719000, China
Chemical looping reforming of methane to syngas (CO and H2) is one of the most promising routes for methane utilization, where the further reaction of CO on oxygen carrier surfaces is a primary determinant of CO selectivity. In this work, the effects of oxygen vacancy (VO) on CO desorption, CO oxidation, and CO dissociation are systematically studied by using density functional theory calculations. Our calculated results reveal that increasing VO concentration can weaken CO desorption at Fe sites due to the enhanced localization of electrons in the Fe atoms. Also, the increase in VO concentration from 1/12 ML to 1/6 ML leads to a dramatic increase of activation energy in the CO oxidation from 0.64 eV to 1.10 eV. Moreover, the increase in VO concentration is conducive to CO dissociation, but the dissociation is still almost impossible due to the high reaction energies (large than 3.00 eV). Considering these three reaction paths, CO desorption can proceed spontaneously at reaction temperatures above 900 K. Increasing VO concentration can improve the selectivity of syngas production due to the less favorable CO oxidation compared with CO desorption at high VO concentrations (1/6 ML). This work reveals the microscopic mechanism that CO selectivity rises in the CLRM as the degree of Fe2O3 reduction increases.
Conversion of methane (CH4) to syngas (CO + H2) through reforming reactions is considered as an energy/carbon-intensive process [1-3]. Among various routes, the chemical looping reforming of methane (CLRM) is one of the most promising techniques for using carbon and hydrogen elements in CH4 to produce high-quality syngas due to its environmental friendliness compared with the conventional reforming process [4, 5]. The CLRM reaction consists of redox reactions taking place in two interconnected reactors: a fuel reactor and an air reactor, where the oxygen carriers (OCs) in the fuel reactor transfer lattice oxygen to methane to generate syngas at first, and then the reduced oxygen carriers in the air reactor are regenerated by re-oxidization in air, completing one closed-cycle. In contrast to conventional reforming processes, this technology process is not only a mildly exothermic reaction but also avoids pure O2 as the feedstock to cause the risk of gas mixing explosion [6-9].
For the CLRM, the selection of oxygen carriers and the control of the reaction selectivity are two critical issues receiving extensive attention. Among various oxygen carriers, Fe-based oxygen carriers are the most promising for practical use because of their high oxygen-carrying capacity, good recyclability, long-term stability, and low production cost [10-12]. The CO selectivity of CLRM is related to the sufficient oxygen supply capacity of the oxygen carrier, favorable thermodynamics of fuel conversion to syngas, high reactivity of reduction and oxidation reactions, and negligible coke deposition [13, 14]. Especially, the reaction behavior of CO on the surface of α-Fe2O3 oxygen carrier is the most critical for CLRM to syngas. Generally, CO may undergo three reaction pathways on the surface of α-Fe2O3 oxygen carrier: CO desorption into the gas phase, CO oxidation to CO2, and CO dissociation to form carbon deposition [15]. Therefore, the reaction behaviors of CO on Fe-based oxygen carriers are extremely significant for improving the selectivity of producing syngas in CLRM.
According to previous studies, oxygen vacancy (VO) can greatly affect the reaction behavior of CO. Lin et al. studied the behaviors of different α-Fe2O3 facets for CO oxidation and dissociation at different oxidation states and proposed that rationally controlling VO concentration and crystal surface of iron oxide can optimize chemical-looping combustion performance [15]. According to literature reports, the (001) plane with Fe–O–Fe termination of Fe2O3 examined theoretically and experimentally is the most stable facet under high temperature and oxygen-free environment, similar to the reaction conditions in a chemical looping fuel reactor [16, 17]. Therefore, we choose the Fe2O3(001) as the objective plane. Our previous studies also found that increasing the surface oxygen vacancy concentration of Fe2O3(001) can promote surface oxygen migration and CH4 dissociation, and then matching the rate of these two reactions can further improve the syngas selectivity in CLRM [18]. Kang et al. found that high CO selectivity catalyzed by Fe2O3/Al2O3 species in garnet was originated from enhanced VO formation ability [10]. However, the studies at present still lack systematic investigations on the reaction behaviors of CO on α-Fe2O3 surface, the understanding of the VO effects at molecular-level on the reaction behaviors of CO remains unclear, which is important for the optimization of CLRM process and the further enhancement of CLRM efficiency. Therefore, it is desirable to study in detail the microscopic reaction mechanism of CO on oxygen-vacancy α-Fe2O3 surface, especially the evaluation of the difficulties of CO desorption, oxidation, and dissociation.
Herein, we carried out the density functional theory (DFT) calculations to study the effect of VO on the CO desorption, oxidation, and dissociation process on the α-Fe2O3(001). Our DFT calculations demonstrated that surface oxygen vacancies influence the three pathways of CO on Fe2O3 by affecting the electron distribution on the surface Fe sites. The CO desorption has an absolute advantage among the three pathways and high oxygen vacancy concentrations can improve the selectivity of CLRM to syngas. This work attains a fundamental understanding of the effect of oxygen vacancies on CO selectivity in CLRM.
II.
COMPUTATIONAL DETAILS
All of the spin-polarized DFT calculations were performed by using the density functional theory [19] with the Vienna ab initio simulation package (VASP) [20]. To describe electron exchange-correlation potential, the spin-polarized Perdew-Burke-Ernzerhof (PBE) version of the generalized gradient approximation (GGA) was applied [21, 22]. The core electrons were described by the projector-augmented wave (PAW) [23] pseudopotentials proposed by Kresse and Joubert [24] with a plane-wave energy cutoff of 400 eV. Unless otherwise stated, Fe2O3 mentioned in this work refers to α-Fe2O3. The calculations in this work adopted the hexagonal Fe2O3 bulk unite antiferromagnetic configuration (↑↓↓↑) with the lowest energy structure in previous theoretical work (FIG. S1 in Supplementary materials, SM) [25]. The DFT+U approach [26] was used to correct the on-site Coulomb and exchange interaction of the strongly localized Fe 3d electrons with an effective U = 4 eV, in consistence with the literatures [27, 28]. The weak interactions between the adsorbates and the surfaces were considered by calculating the van der Waals dispersion forces using the DFT-D3 method of Grimme [29]. The first Brillouin zone k-point sampling was generated by the Monkhorst-Pack scheme with a 2 × 2 × 2 mesh for the Fe2O3 bulk and a 2 × 2 × 1 mesh for the Fe2O3 surfaces with a vacuum space of 15 Å [30].
The calculated Fe2O3 bulk lattice parameters are a = b = 5.02 Å and c =13.71 Å, which are very close to the experimental values (a = b = 5.04 Å and c = 13.77 Å) [31]. The p(2 × 2) Fe2O3(001) surface with Fe–O–Fe termination was built and used for the calculations of surface reactions due to its lowest surface energy of 0.18 eV/Å2 (FIG. S2 in SM). The surface supercell contains 72 oxygen atoms and 48 iron atoms. Moreover, to study the effect of oxygen vacancy (VO) concentration on the selectivity of carbon monoxide production, we constructed Fe2O3(001) models of stoichiometric, one VO and two oxygen vacancies surfaces, respectively. The atoms of the bottom three layers were fixed, while those of the remaining layers with adsorbates were allowed to relax. The antiferromagnetic Fe2O3 slabs were modeled by assigning the same spin direction to bulk along the c-axis from bottom to top [32]. A dipole correction was introduced to correct the dipole moment in the c-axis direction [33].
For CO oxidation barrier calculations, the climbing-image nudged elastic band (CI-NEB) combined with the minimum-mode following dimer method was used to find the minimum energy path and the transition states [34, 35]. The vibration analyses of all transition states (TS), initial states (IS), and final states (FS) were performed: the transition state structures were determined by a single imaginary frequency, and the structures of the initial state and the final state were determined by all positive frequencies. The desorption energy (ΔEdesorption), activation energy (Ea), and dissociation energy (ΔEdissociation) were calculated using the following equations:
ΔEdesorption=E(CO)+Esurface–ECO-surface
(1)
Ea=ETS1–ECO-surface
(2)
ΔEdissociation=EC-surface+EO-surface–ECO-surface
(3)
where ECO, Esurface, ECO-surface, ETS1, EC-surface, and EO-surface are the energies of CO in the gas phase, α-Fe2O3(001) surface, the surface with adsorbed CO, transition state of TS1, the surface with adsorbed C atom, and the surface with adsorbed O atom on the surface, respectively.
The Gibbs free energies are calculated as follows:
H=EDFT+EZPE+EH
(4)
G=H−TS=EDFT+EZPE+EH−TS
(5)
where EDFT is the electronic energy of the system optimized by DFT calculations, EZPE is zero-point vibrational energy, and H and S are the enthalpy and entropy, respectively. EH is the contribution of enthalpy to energy from 0 K to a certain temperature. The pressure adopted for calculating the Gibbs free energy was 1.0 atm. The VASP post-processing program, VASPKIT, was used to calculate Gibbs free energy based on DFT calculations [36].
The TS1 was identified as the transition state of the rate-limiting step in most cases as described later. The projected crystal orbital Hamilton population (pCOHP) curves were analyzed by using the LOBSTER program. The pCOHP was calculated by dividing the energy of the band structure into orbital-pair interactions, which can be used to reflect the bonding strength [37, 38]. Atomic charges were evaluated by Bader charge analysis using the atom-in-molecule scheme [39].
III.
RESULTS AND DISCUSSION
A
Surface properties of Fe2O3(001)
The catalytic performance of the oxygen carrier can be determined by the electronic structure of the active sites, which is related to the surface oxygen vacancies. During the chemical looping combustion process, the VO concentration of Fe2O3(001) surface continuously increases over time. To study the effects of VO concentration on surface reaction, three Fe2O3(001) surfaces with different oxygen vacancies were constructed as shown in FIG. S3 in SM, i.e. stoichiometric Fe2O3(001) surface, reduced surface with one oxygen vacancy (Fe2O3(001)_1VO, VO concentration: 1/12 ML), and reduced surface with two oxygen vacancies (Fe2O3(001)_2VO, VO concentration: 1/6 ML), respectively. The introduction of oxygen vacancies can alter the valence state of metal elements on oxide surfaces [40], and then the change valence states of surface active sites, further affect the catalytic reaction process. Hence, the Bader charges of Fe atoms on the top atomic layer of Fe2O3(001) surfaces with different VO concentrations were calculated and analyzed (FIG. S3 in SM and Table I). The calculated results indicate that with the increase of VO concentration, the average atomic charge of top-layer Fe atoms on the Fe2O3(001) surface decreases from 1.569 e to 1.397 e owing to the enhanced electron localization at the Fe site caused by O deficiency, similar to the experimental results of Rioult et al. that increased VO concentration leads to parts of Fe3+ ions changing to Fe2+ ions [41].
Table
I.
Calculated atomic charges (Q) and their mean values of the four Fe atoms on three Fe2O3(001) surfaces.
The Fe valence state of Fe2O3(001) will inevitably affect its surface electronic structure, which can alter the activities of these surfaces for CO desorption. To study the desorption properties of CO, the adsorption of CO on the surface was first investigated as the inverse process of desorption. The possible adsorption sites for CO are the Fe sites, the O sites, and the bridge sites between Fe and O atoms on the Fe2O3(001) surfaces. Our calculated results found that the most stable sites for CO adsorption are the Fe sites on the surface (FIG. S4 in SM), consistent with the previous studies [42, 43]. The optimized configurations and corresponding desorption energies are shown in FIG. 1(a). The calculated results show that the surface Fe atom is actively bonded with the C atom of CO. Additionally, the d-transition metal Fe mainly has the valence electrons at d-orbitals, which mainly participate in surface reactions. The d-band center of Fe2O3(001)_2VO was closer to the Fermi level than that of Fe2O3(001) and Fe2O3(001)_1VO. The closer the d-band center is to the Fermi level (FIG. 2(b)), the easier the electron donation ability of the catalyst to the adsorbate. Meanwhile, the CO desorption energy is 0.64 eV on Fe2O3(001), 1.05 eV on Fe2O3(001)_1VO, and 1.17 eV on Fe2O3(001)_2VO, respectively. The calculated results indicate the desorption of CO generated in CLRM becomes more difficult with the increase of the surface VO concentration from 0 to 1/6 ML. However, after considering the effects of temperature on the desorption, the CO desorption becomes spontaneous at the experimental conditions (above 900 K) [44] due to the negative free energy of CO desorption (Table S1 in SM).
Figure
1.
(a) CO desorption energy with the corresponding optimized structure and (b) electron density difference (Δρ = ρ(CO-surface)−ρ(surface)−ρ(CO)) maps of CO on three Fe2O3(001) surfaces. The atomic positions of surface and CO are identical with those in the CO-surface. The electron-density isosurfaces are plotted at 0.02 e/Bohr3.
Figure
2.
(a) Projected density of states of gas-phase and absorbed CO. (b) DOS of Fe 3d for CO adsorption on three Fe2O3(001) surfaces. The values correspond to the d-band center of the Fe on different surfaces. The partial charge density isosurfaces are plotted at 0.008 e/Bohr3.
To further explore the electron transfer between CO and Fe2O3(001) surface, the electronic structure analyses, including the electron density difference and projected density of states, are also conducted. As shown in FIG. 1(b), the electron density difference maps show that the electron accumulation between the Fe atom and CO increases obviously along with the increase of surface VO, which can be ascribed to the fact that there are more electrons on the Fe atom of the surface containing oxygen vacancies (Table I), resulting in the formation of a strong covalent bond with the C atom of CO. This can be used to explain the enhancement of CO absorption ability with the increase of VO concentration. Then, the PDOS in FIG. 2(a) shows that the energy of the CO molecular orbital ranges from −12.0 eV to 5.0 eV before CO adsorption, and the bonding form of the CO molecular orbital can be divided into 4σ, 1π, and 5σ. The occupied 5σ orbitals of CO interact with the d orbitals of the Fe atom on three Fe2O3(001) surfaces to generate new orbitals at about −7.0 eV. Interestingly, on the two reduced surfaces, the unoccupied 2π* orbitals of CO interact with the occupied d orbitals of Fe atoms to form new orbitals at −0.9 eV, which enhances the CO adsorption on reduced Fe2O3(001) surfaces. Moreover, the Fe atoms binding with CO on reduced Fe2O3(001) surfaces are in lower oxidation states, which is indicated by the occupied d orbitals (at −0.9 eV) close to the Fermi level in FIG. 2(b) and smaller Bader atomic charges in Table I. Overall, the above calculation results show that with the increase of surface VO, the change of Fe electronic structure leads to enhanced CO adsorption and thus makes CO desorption more difficult.
C
CO oxidation on Fe2O3(001)
Apart from desorption, the adsorbed CO may be oxidized to CO2 via surface oxygen. To understand the difficulty of CO oxidation to CO2, we calculated the CO oxidation reaction under different concentrations of surface oxygen vacancies. Taking the CO oxidation on the Fe2O3(001) surface as an example (FIG. 3), a CO molecule is first adsorbed on the Fe site of the surface. The adsorbed CO interacts with the surface O atom, forming a stable adsorbed species *COO after passing through TS1. Then, *CO2 is further formed through TS2, and finally CO2 desorbs to form a gaseous molecule. The calculated CO oxidation path is consistent with the reaction path calculated by Huang et al. [16]. Similarly, the CO oxidation processes on Fe2O3(001)_1VO and Fe2O3(001) _2VO surfaces is also calculated (FIG. S5 and S6 in SM) and show that the reaction paths do not significantly change with the increase of vacancy concentration.
Figure
3.
Optimized geometric structures of the initial state, TS, and intermediates during the CO oxidation process on the Fe2O3(001) surface.
On this basis, we summarize the energy changes of CO oxidation on three surfaces in the potential energy diagram (FIG. 4). From the thermodynamic data, CO is more preferentially adsorbed with increasing the concentration of oxygen vacancies (−0.64, −1.10, and −1.17 eV). From the kinetic data, the rate-limiting step for CO oxidation on three surfaces is the process from *CO to *COO through TS1. Moreover, the reaction barrier changes from 0.39 eV to 0.64 eV and then to 1.10 eV with the increase of VO concentration, indicating that CO oxidation becomes more difficult with the increase of VO concentration. The increased oxygen vacancies will block CO to capture surface oxygen, which is responsible for unfavorable kinetics. Here, we carry out an analysis of the transition state of the rate-limiting step to further study the reason for unfavorable kinetics caused by increasing VO concentration (FIG. S7 in SM). The bond distance between the carbon of CO and the surface oxygen becomes larger (from 1.77 Å to 2.66 Å) with the increase of VO concentration, indicative of weak oxygen-grabbing power. Alternatively, the ICOHP (integration of pCOHP to Fermi level) for C–O bond in transition states on three surfaces are −2.68 eV (Fe2O3(001)), −2.04 eV (Fe2O3(001)_1VO), and −0.10 eV (Fe2O3(001)_2VO), respectively, suggesting the increase of VO concentration makes it difficult to capture oxygen dynamically.
Figure
4.
Potential energy diagram of the CO oxidation reaction taking place on the stoichiometric and oxygen vacancy Fe2O3(001) surface at 0 K. The asterisk (*) represents the adsorbed state.
To unravel the degree of CO dissociation with various concentrations of oxygen vacancies, the dissociation energies of CO on three kinds of surfaces were calculated. First, CO dissociates into C and O atoms at two Fe sites on stoichiometric Fe2O3(001) in FIG. 5(a). Whereas, the O atom will preferentially fill the VO to form a relatively stable structure on reduced Fe2O3(001) displayed in FIG. 5 (b) and (c). The dissociation energies of CO on stoichiometric Fe2O3(001), Fe2O3(001)_1VO and Fe2O3(001)_2VO are 9.76, 4.20 and 3.14 eV, respectively. The calculated results show that CO dissociation is relatively more easily in thermodynamics enabled by the increased oxygen vacancies in Fe2O3(001). However, compared with the CO desorption energy and CO oxidation activation barrier, the dissociation energy is much higher. It can be expected that the calculated CO dissociation energy barrier is bound to be higher, which indicates that CO dissociation is very difficult to occur compared with desorption and oxidation. Furthermore, this also suggests that the carbon deposition in CLRM to synthesis gas may be more likely caused by the direct dissociation of CH4 rather than the dissociation of CO.
Figure
5.
Initial state and final state of CO dissociation on the stoichiometric and oxygen vacancy Fe2O3(001) surface. The values in electronvolts correspond to the dissociation energies of each adsorbate. CO before and after dissociation is labeled by black circles. The oxygen vacancies are labeled by blue circles. The balls in larger and smaller diameters are in the first and other atomic layers, respectively.
E
Comparison of CO desorption, oxidation, and dissociation
Based on the above studies, the Gibbs free energy of CO desorption, and the Gibbs free energy of activation in CO oxidation and CO dissociation on the stoichiometric Fe2O3(001), Fe2O3(001)_1VO and Fe2O3(001)_2VO are plotted as shown in FIG. 6. CO desorption is more likely to happen in these three processes, which also indicates that Fe2O3 is indeed an ideal oxygen carrier for the CLRM of CH4 to syngas. The increased oxygen vacancies render it difficult for CO to oxidize and desorb, which would be difficult to judge its selectivity by a simple comparison of energy values. To qualitatively describe the effect of increased oxygen vacancies on CO selectivity in the future, a method can be used by calculating the difference between the Gibbs free energy of activation in CO oxidation and the Gibbs free energy of CO desorption on three surfaces (Table S1 in SM). The calculated results show that the difference value on Fe2O3(001)_2VO surface is much larger than that on stoichiometric Fe2O3(001) and Fe2O3(001)_1VO surface, which indicates that high VO concentration (1/6 ML) can improve CO selectivity to some extent verified by experiment [45].
Figure
6.
Gibbs free energy change of CO desorption, oxidation, and dissociation on the stoichiometric and oxygen-vacancy Fe2O3(001) surfaces at 600, 900, and 1200 K.
In summary, we have investigated the influence of VO concentration on CO desorption, oxidation, and dissociation on the Fe2O3(001) surface, and present an in-depth analysis using DFT. The main conclusions are drawn as follows: (i) the increase of oxygen vacancy concentration weakens CO desorption due to the enhanced electron localization at the Fe site; (ii) the increase in oxygen vacancy concentration also leads to a sharp increase in the CO oxidation energy barrier; (iii) the increase in oxygen vacancy concentration is conducive to CO dissociation, but CO dissociation is difficult due to the high dissociation energy; (iv) at the actual reaction temperatures (above 900 K), CO desorption has an absolute advantage, and high oxygen vacancy concentration (1/6 ML) can improve the selectivity of CLRM to syngas. Overall, this work implies that the CO selectivity in the CLRM can be improved by adjusting the VO concentration of the oxygen carrier.
ACKNOWLEDGMENTS
This work is supported by the National Natural Science Foundation of China (No.22038011, No.22078257, No.22108213, and No.52176142), the China Postdoctoral Science Foundation (2021M692548), and the Joint Fund of the Yulin University and the Dalian National Laboratory for Clean Energy (Grant YLU-DNL Fund 2022001). Chun-Ran Chang also acknowledges the Young Talent Support Plan of Shaanxi Province. The calculations were performed by using the HPC Platform at Xi’an Jiaotong University.
Supplementary materials: The initial spins for the four Fe atoms in Fe2O3 primitive cell; the structures of α-Fe2O3 bulk, Fe2O3(001), Fe2O3(001)_1VO and Fe2O3(001)_2VO surfaces; the optimized structures of CO adsorbed at Fe, Fe–O bridge of the Fe2O3(001) surface; the optimized structures of initial states, transition states and intermediates during the CO oxidation process; the calculated data of desorption free energy, oxidation free energy barrier, and dissociation free energy are shown.
K. J. Warren and J. R. Scheffe, J. Phys. Chem. C 123, 13208 (2019). doi: 10.1021/acs.jpcc.9b01352
[8]
J. J. Huang, W. Liu, W. T. Hu, I. Metcalfe, Y. H. Yang, and B. Liu, Appl. Energ. 236, 635 (2019). doi: 10.1016/j.apenergy.2018.12.029
[9]
A. Shafiefarhood, N. Galinsky, Y. Huang, Y. G. Chen, and F. X. Li, ChemCatChem 6, 790 (2014). doi: 10.1002/cctc.201301104
[10]
Y. Kang, M. Tian, C. Huang, J. Lin, B. Hou, X. Pan, L. Li, A. I. Rykov, J. Wang, and X. J. Wang, ACS Catal. 9, 8373 (2019). doi: 10.1021/acscatal.9b02730
[11]
J. Adanez, A. Abad, F. Garcia-Labiano, P. Gayan, and L. F. de Diego, Prog. Energy Combust. Sci. 38, 215 (2012). doi: 10.1016/j.pecs.2011.09.001
[12]
Y. Kang, M. Tian, Y. H. Wang, Y. T. Wang, C. D. Huang, Y. Y. Zhu, L. Li, G. J. Wang, and X. D. Wang, ACS Sustainable Chem. Eng. 6, 12884 (2018). doi: 10.1021/acssuschemeng.8b02262
[13]
L. B. Yan, C. J. Lim, G. X. Yue, B. Z. Fang, B. S. He, and J. R. Grace, Thermochim. Acta 648, 52 (2017). doi: 10.1016/j.tca.2016.12.003
[14]
Q. Zafar, T. Mattisson, and B. Gevert, Ind. Eng. Chem. Res. 44, 3485 (2005). doi: 10.1021/ie048978i
J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson, D. J. Singh, and C. Fiolhais, Phys. Rev. B 46, 6671 (1992). doi: 10.1103/PhysRevB.46.6671
G. Henkelman and H. Jónsson, J. Chem. Phys. 111, 7010 (1999). doi: 10.1063/1.480097
[35]
B. J. Berne, G. Ciccotti, and D. F. Coker, Classical and Quantum Dynamics in Condensed Phase Simulations: Proceedings of the International School of Physics, Singapore: World Scientific, (1998).
[36]
V. Wang, N. Xu, J. C. Liu, G. Tang, and W. T. Geng, Comput. Phys. Commun. 267, 108033 (2021). doi: 10.1016/j.cpc.2021.108033
[37]
V. L. Deringer, A. L. Tchougréeff, and R. Dronskowski, J. Phys. Chem. A 115, 5461 (2011). doi: 10.1021/jp202489s
[38]
R. Nelson, C. Ertural, J. George, V. L. Deringer, G. Hautier, and R. Dronskowski, J. Comput. Chem. 41, 1931 (2020). doi: 10.1002/jcc.26353
DownLoad:
