First-principles DFT insights into the mechanisms of CO2 reduction to CO on Fe (100)-Ni bimetals

Iron and nickel are known active sites in the enzyme carbon monoxide dehydrogenases which catalyzes CO2 to CO reversibly. The presence of nickel impurities in the earth abundant iron surface could provide a more efficient catalyst for CO2 degradation into CO, which is a feedstock for hydrocarbon fuel production. In the present study, we have employed spin-polarized dispersion-corrected density functional theory calculations within the generalized gradient approximation to elucidate the active sites on Fe (100)-Ni bimetals. We sort to ascertain the mechanism of CO2 dissociation to carbon monoxide on Ni deposited and alloyed surfaces at 0.25, 0.50 and 1 monolayer (ML) impurity concentrations. CO2 and (CO + O) bind exothermically i.e., − 0.87 eV and − 1.51 eV respectively to the bare Fe (100) surface with a decomposition barrier of 0.53 eV. The presence of nickel generally lowers the amount of charge transferred to CO2 moiety. Generally, the binding strengths of CO2 were reduced on the modified surfaces and the extent of its activation was lowered. The barriers for CO2 dissociation increased mainly upon introduction of Ni impurities which is undesired. However, the 0.5 ML deposited (FeNi0.5(A)) surface is promising for CO2 decomposition, providing a lower energy barrier (of 0.32 eV) than the pristine Fe (100) surface. This active 1-dimensional defective FeNi0.5(A) surface provides a stepped surface and Ni–Ni bridge binding site for CO2 on Fe (100). Ni–Ni bridge site on Fe (100) is more effective for both CO2 binding or sequestration and dissociation compared to the stepped surface providing the Fe–Ni bridge binding site.


Introduction
The levels of carbon dioxide in the atmosphere continues to increase as a result of anthropogenic activities like combustion of fossil fuels, leading to global warming and climate change [1]. CO 2 is an abundant and cheap carbon-one source, which could be a useful feedstock in the production of transportation fuels [2], industrial chemicals [3], and polymers [1]. However, due to the stability and inertness of the CO 2 molecule, catalysts are required for conversion [4,5]. Despite difficulties associated with CO 2 conversion industrially, anaerobic enzymes such as carbon monoxide dehydrogenases are known to reversibly catalyze the reduction of CO 2 to CO at ambient conditions of temperature and pressure [6]. CO 2 is said to anchor and receive electrons at the bridge site of iron and nickel in the Fe-Ni-S cluster in carbon monoxide dehydrogenases [7]. The catalytic CO 2 decomposition into CO has become an active field of research in catalytic chemistry as CO is the feedstock in the Fischer-Tropsch process for the production of long-chain hydrocarbon liquid transportation fuels [2,8].
Catalytic conversion of CO 2 to valuable industrial feedstock like CO is an attempt to ease the effects of CO 2 on our environment. Although experimental studies on CO 2 reduction on single crystal surfaces show activity for CO 2 chemisorption and reduction on bare Fe and Ni, including Ni (110) and Fe (111) [9], the energetics and mechanisms of CO 2 transformation to viable products like CO, methane, formic acid etc. on these bare metal surfaces were not well understood. The extent of CO 2 activation and dissociation on iron and nickel surfaces have been shown to be face specific experimentally [10][11][12][13][14].
This was later supported by other density functional theory (DFT) calculations whereby on iron the barrier to CO 2 dissociation on the low Miller index surfaces were of the trend Fe (100) ~ (111) < (110) [15]. Several computational studies have also been carried out to investigate the interactions of CO 2 with Ni, mostly employing the spin-polarized density functional theory-generalized gradient approximation (DFT-GGA) to understand the energetics on their various topologies [16][17][18][19][20][21][22][23][24]. CO 2 is reported to bind more strongly on iron than nickel while its decomposed species bind stronger to nickel than iron. Kinetically decomposition is observed to be favored on iron than nickel [21].
Iron and nickel are known active sites in the enzyme carbon monoxide dehydrogenases (CODH) and the presence of nickel impurity in earth-abundant iron could provide more active materials for CO 2 decomposition to CO. CO 2 interactions with bare and 1 ML deposited surfaces of the low Miller index surfaces of iron have been investigated previously [15], where nickel is seen to alter the ease of CO 2 dissociation. However, to the best of our knowledge, no theoretical studies have been carried out on alloys of iron and nickel and the concentration effect of nickel deposition on the activity of the iron surface has also not been explored. In this present study, we have employed dispersion-corrected spin-polarized-density functional theory calculations within the generalized gradient approximation (DFT-D2-GGA) to elucidate the mechanism of CO 2 reduction into carbon monoxide on the pristine Fe (100) facet, its nickel alloys and nickel deposited surfaces at varying concentrations of 0.25 ML, 0.5 ML and 1 ML.

Computational details
All calculations were carried out with the spin-polarized density functional theory method as implemented in the Quantum ESPRESSO package [25]. The generalized gradient approximation (GGA), with the Perdew, Burke, Ernzerhof (PBE) exchange-correlation functional [26] was used in all simulations. The surface was described by an asymmetric slab model, where periodic boundary conditions were applied to the central super-cell so that it is reproduced periodically throughout space. XcrysDen [27] software was employed for the visualization of structures and electron densities. The Fermi-surface effects were treated by the smearing technique of Fermi-Dirac, using a smearing parameter of 0.03 Ry. The energy threshold defining self-consistency of the electron density was set to 10 −6 eV. The Grimme's D3 correction was implemented for Vann der Waal's dispersion corrections.
Iron (100) surface was cleaved with the METADISE code [28] and a p(3 × 3) super-cell was employed for all calculations as the binding energy of CO 2 does not change significantly with increasing super-cell [15,29]. The slab was built to a thickness of three, made up of six atomic layers. A vacuum of 20 Å was introduced to the surface to prevent interactions between surfaces along the z-axis. The top three layers of the slab was relaxed in all calculations, which has been reported previously to be the converged structure of iron (100) [15,30,31]. All gaseous adsorbates were optimized in a cubic box of size 20 Å and allowed to relax in all calculations. Neighboring adsorbates in laterally repeating units of the slabs were more than 5 Å apart.
Using convergence tests, the kinetic energy cutoff of the plane wave basis set was set to 40 Ry and 320 Ry for the kinetic energy and charge density cut-off, respectively. The Monkhorst-pack K-points grid of (7 × 7 × 7), (5 × 5 × 1) and (1 × 1 × 1) were used for bulk, surface and adsorbate systems, respectively. The Climbing Image Nudged Elastic Band (CI-NEB) method was used to determine the energy barriers for dissociation. Vibrational modes were calculated whereby a single imaginary frequency was indicative of a transition state. Lowdin charge analysis was employed for charge density characterizations upon adsorption of CO 2 . Zero-point vibrational corrections were ignored in the binding energy estimations, as our previous studies shows it does not affect the qualitative view of the reaction on the surface.

CO 2 adsorption on pure and bimetallic surfaces
The computation parameters were first validated by calculating the bulk properties of iron. The unit cell of iron crystalizes in the body-centered cubic (BCC) form and our spin-polarized DFT-D3 calculations were able to reproduce the electronic properties of bulk iron [15,32].
where E Fe is energy of single iron atom, E Ni is energy of single nickel atom, n is the number of dopants and E slab is energy of the perfect Fe (100) surface.
As seen in Table 1, deposition is generally favored thermodynamically over alloying, nickel prefers to be segregated on iron than alloy at all concentrations from 0.25 to 1 ML. The high instability of the doped surfaces relative the deposited surfaces show that thermodynamically at 0.25 ML, 0.5 ML and 1 ML concentrations, nickel will be segregated on the surfaces than diffuse to form alloys with the Fe (100)  (2) Carbon dioxide adsorption was then studied on the pure and defective surfaces (see Fig. 2) and the binding energies of CO 2 on the various surfaces were calculated as follows; where E (slab+CO2) is the energy of the adsorbed system, E slab and E CO2 are the energies of the isolated surface and gaseous carbon dioxide, respectively.
The preferred CO 2 adsorption site on the clean Fe (100) surface has been reported to be the hollow site in the C 2V adsorption mode [15]. At the active site i.e., at the hollow site and in the C 2V preferred CO 2 adsorption state, we investigated the effect of doping on CO 2 binding and extent of activation. As shown in Fig. 2, CO 2 binding to bare Fe (100) is exothermic with adsorption energy of -0.87 eV. This is very consistent with earlier observations of − 0.9 eV [15], − 0.92 eV [33], − 0.7 eV [19]. Introduction of nickel into the bulk of iron (structure d) at point defect of 0.25 ML, increases the binding strength of CO 2 to − 0.92 eV. CO 2 coordinates to four iron atoms at the hollow site. Increasing the nickel concentration to 1D defect (structure e) increases the iron-CO 2 interactions at the hollow site to − 0.96 eV. Whiles at 2D defect (structure f), the iron-CO 2 interaction at the hollow site is decreased to − 0.84 eV, this is less than the binding on bare Fe (100). Comparing the electronegativity of nickel and iron, nickel is more electron withdrawing and increasing its concentration in the bulk of iron, lowers the electron density available at the surface for transfer into  the CO 2 moiety. The nickel electron withdrawing effect is felt at the surface with increasing concentration of nickel. Also increasing the concentration of nickel dopant at the sub-surface site introduces appreciable Ni properties at the surface, as bare Ni is known to bind CO 2 more weakly [34]. In Ni adsorption situations as shown in Fig. 2a, b, stepped surfaces are provided which facilitate lower CO 2 surface coordination (ɳ-CO 2 ) and lower binding energies due to the presence of nickel on surfaces compared to the alloyed surfaces in Fig. 2d-f. As seen for the nickel single ad-atom at 0.25 ML deposition, the binding takes place at the bridge of iron and nickel, and the strength of binding is weakened to − 0.33 eV compared to bare iron. Increasing nickel concentration at the surface decreases the binding strength of CO 2 further to − 0.19 eV for (FeNi 1 (B)) (c). Ni generally weakens CO 2 binding strength except in cases where the Ni effect is less felt on the surface. These results show that decreasing CO 2 surface coordination by the introduction of adatoms and higher electronegative atom effects like nickel on the surface weakens CO 2 binding.
From Table 1, the presence of nickel influences the Fermi levels and the work functions of the slabs. The work function which translates to their electrochemical potential or ability to transfer charges is reduced (as work function increase). The lower the electrochemical potential of the slab, the weaker the net charge it transfers to the CO 2 moiety and the lower the degree of activation of the molecule. It is seen that the lower the concentration of the Ni, the lower its impact on the work function. Hence to increase the charge mobility, less electronegative materials hold a better potential to reduce surface work functions, as Ni is more electronegative than Fe. Again, stepped surfaces with low CO 2 coordination number show stronger binding, for example comparing the monolayer deposited and single atom deposited surfaces i.e., FeNi 0.25 (A) and FeNi 1 (A). The net amount of charge gained by CO 2 molecule from the surface and the extent of CO 2 activation on the surfaces are seen to correlate, this is consistent with earlier studies [23].

CO 2 dissociation on pure and bimetallic surfaces
The reaction energies for CO 2 dissociation (E dis ) and the barriers for dissociation (E a ) were calculated with Eq. (5) and (6), respectively; where E products is the energy of the adsorbed dissociated system, E slab is the energy of isolated slab and E CO2 is the energy of isolated carbon dioxide molecule.
where E TS is energy of the transition state and E IS is the energy of the intermediate state i.e., adsorbed CO 2 .
To reduce surface interaction between adsorbed molecules, the decomposed species CO and O were optimized individually on the surfaces as well to determine the binding energies in Table 2. As reported in Table 2, the binding energy of decomposed CO 2 (E dis ), is generally favorable thermodynamically relative to the binding energy of CO 2 (E ads ). The thermodynamics of the dissociation steps were also calculated relative to the activated CO 2 moiety (Step dis ) and was found to be a thermodynamically favored step on all surfaces. The reaction barriers for the dissociation steps  were then computed for the reactions on the various surfaces. The energy profile diagram showing the energy transitions along the reaction coordinates are shown in Fig. 3. On bare Fe (100), a dissociation barrier of 0.53 eV was found. Earlier studies reported 0.22 eV [19] and 0.8 eV [35]. Comparing the energy barriers for the CO 2 dissociation step (E a in Table 2), generally nickel impedes CO 2 dissociation as higher barriers are encountered on the modified surfaces compared to the pure Fe (100) FeNi 1 (A). Monolayer deposited (1 ML) FeNi 1 (A) surface seems to be the most challenged surface for CO 2 dissociation kinetically, with a barrier of 4.16 eV. This is expected as it is the surface with most nickel atoms in the bulk has most pronounced nickel effects on the surface. Here the nickel behavior is predominating, as nickel surface provide higher decomposition barriers relative to iron [34]. The FeNi 0.5 (A) (1D defect) is promising for CO 2 dissociation kinetically where CO 2 is coordinated at a Ni-Ni bridge site. Although FeNi 0.5 (A) is the least stable of the deposited surfaces, its formation is thermodynamically favored and it is also the deposited surface that binds CO 2 the most and is most suitable for CO 2 sequestration.
Comparing the work function trends to the behavior of the modified surfaces (Fig. 4), as the work function increases, the fermi level, which shows the electrochemical potential of the surfaces is also seen to reduce. The fermi level shows a good correlation to the extent of CO 2 activation. The lower the work function, the higher the fermi level the more activated the CO 2 molecule on the surface as seen at work function of 3.80 eV for pure iron. This shows that to increase CO 2 activation, the fermi level and work function of the surface needs to be modified. The barriers of CO 2 dissociation did not correlate strongly with the work function or the degree of CO 2 activation. CO 2 binding and CO + O binding trends have a similar pattern. Surfaces that bind CO 2 strongly also bind CO + O relatively strongly when compared to other surfaces as seen on FeNi 0.25 (B) at 3.86 eV. Larger surface work functions are associated with weaker binding energies as seen around 5 eV.

Conclusion
The effect of Ni alloying and deposition on the ease of CO 2 direct dissociation has been studied using the DFT method. Nickel prefers to be segregated on iron than alloy at concentration of 0.25 ML up to 1 ML as alloying is seen to be unstable. The stabilities of the modified surfaces were of the order, FeN i 0 .25 (B) < FeNi 0.5 (B) < FeNi 1 (B) < F . CO 2 binds exothermically to bare Fe (100) surface (E ads = − 0.87 eV). Ni at the bulk site improves the binding of CO 2 and its applicability for CO 2 sequestration except at 1 ML doping, where the effect of bulk Ni is stronger on the surface. These results show that introduction of high amount of nickel in the bulk of iron weakens CO 2 binding. The work function and fermi level energy of the modified surfaces have a strong correlation to the CO 2 activation degree and binding trends. Thermodynamically, dissociation is favored on all surfaces probed. Kinetically, CO 2 dissociation is most favored on the FeNi 0.5 (A) surface, which is stepped and allows CO 2 to coordinate to two surface Ni atoms. Generally, the barriers for CO 2 dissociation are heightened compared to bare Fe (100), especially on the monolayer deposited (1 ML) FeNi 1 (A) surface. Ni deposition on Fe at 0.5 ML coverage could offer the most viable nickel-modified iron surface for CO 2 reduction and would provide a more reactive surface for CO 2 hydrogen unassisted splitting into CO (a feedstock essential for the Fischer-Tropsch process).