Article pubs.acs.org/JPCC
Surface Morphology of Cu Adsorption on Different Terminations of the Hägg Iron Carbide (χ-Fe5C2) Phase Xinxin Tian,†,‡,§ Tao Wang,∥ Yong Yang,†,‡ Yong-Wang Li,†,‡ Jianguo Wang,† and Haijun Jiao*,†,∥ †
State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of Sciences, Taiyuan, 030001, China National Energy Center for Coal to Liquids, Synfuels CHINA Co., Ltd., Huairou District, Beijing, 101400, China § University of Chinese Academy of Sciences, Beijing, 100049, China ∥ Leibniz-Institut für Katalyse eV. an der Universität Rostock, Albert-Einstein-Strasse 29a, 18059 Rostock, Germany ‡
S Supporting Information *
ABSTRACT: Spin-polarized density functional theory computations have been carried out to investigate the surface morphology of Cun adsorption on the Fe5C2(100), Fe5C2(111), Fe5C2(510), Fe5C2(001), and Fe5C2(010) surface terminations in different surface Fe and C ratios. On the Fe5C2(100), and Fe5C2(510) surfaces, aggregation is thermodynamically more favored than dispersion, while dispersion is more favored than aggregation on the Fe5C2(111) surface for n = 2− 4, on the Fe5C2(010) surface for n = 2 and on the Fe5C2(001) surface for n = 2−4. The difference in structures and stability at low coverage depends on the stronger Cu−Fe interaction over the Cu−Cu interaction as well as the location of the adsorption sites. The adsorption energies do not correlate with the surface Fe and C ratios. Comparison among the most stable Fe(110), Fe3C(001), and Fe5C2(100) surfaces reveals that the Fe(110) surface has higher Cu affinity than the Fe3C(001) and Fe5C2(100) surfaces; and the carbide surfaces have close Cu affinities; in agreement with the experimental observations. On all these iron and carbide surfaces, two-dimensional monolayer surface adsorption configurations are energetically more favored than the adsorption of three-dimensional Cun clusters, and it can be expected that the adsorbed Cu atoms should grow epitaxially as a layer-by-layer mode at the initial stage. On the metallic Fe(110), Fe(100), Fe(111), and Fe3C(010) surfaces, the adsorbed Cu atoms are negatively charged; while on the Fe3C(100), Fe5C2(100), Fe5C2(111), Fe5C2(010), and Fe5C2(001) surfaces, the adsorbed Cu atoms are positively charged. On the Fe3C(001) and Fe5C2(510) surfaces, the adsorbed Cu atoms mainly interacting with surface Fe atoms are very slightly negatively charged. This trend is in line with their difference in electronegativity. Our results build the foundation for further study of the Cu-promotion effect in Fe-based FTS in particular and for metal-doped heterogeneous catalysis in general. version to metallic iron,23 lower reduction temperature24 (especially under H225), promote carburization rate,26−28 and increase the activity of FTS and water−gas shift (CO + H2O → CO2 + H2) reaction29−32 as well as suppress the formation of the χ-Fe5C2 phase under CO.33,34 In addition, synergistic effect was observed on copper and potassium copromotion.35 However, controversial results about copper promotion on hydrocarbon selectivity in FTS were reported.24,26,29,35 Recently, Blanchard et al.,36 prepared nanoiron carbides catalysts by using plasma spray technique and found that metallic copper promotion increases surface H2 concentration and inhibits the oxidation of carbides. Consequently, catalyst deactivation was almost completely suppressed. They also found that copper does not have significant effect on WGS
1. INTRODUCTION Fischer−Tropsch synthesis (FTS, CO + H2 → CxHy + H2O + CO2)1−3 is an essential technology in converting synthesis gas generated from coal, natural gas and biomass into clean fuels and value-added chemicals. The most attractive FTS catalysts for industrial application are iron-based,4−6 due to not only low price and high availability but also high resistance to contaminants (allowing the use of CO-rich synthesis gas) as well as low methanation activity at high temperature FTS (above 300 °C).7 In Fe-based FTS, iron carbides are considered as active catalysts and the most popular one is Hägg iron carbide (χFe5C2), which should be responsible for high activity8−10 and has received high theoretical attentions.11−21 Apart from iron carbides, promoters such as copper, potassium and silica or zinc oxide are usually added to improve the catalytic performance. The potential roles of copper promoter have been extensively studied experimentally.22 Copper can facilitate Fe3O4 con© 2015 American Chemical Society
Received: February 9, 2015 Revised: March 10, 2015 Published: March 11, 2015 7371
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Figure 1. Possible adsorption sites on the Fe5C2(100), Fe5C2(111), Fe5C2(510), Fe5C2(010) and Fe5C2(001) surfaces; one-fold (1F), 2-fold (2F), 3fold (3F), 4-fold (4F), 5-fold (5F), 6-fold (6F), and 7-fold (7F) sites (blue balls for Fe atoms and black balls for C atoms).
surface, van Steen et al.,43 found that atomic Cu on the ironrich χ-Fe5C2(100)0.25 surface enhances CO bonding to the surface at low CO coverage, and results in the red shift in the stretching frequency of adsorbed CO. Despite of the importance, to the best of our knowledge, no other mechanistic studies of Cu promotion were reported; and this might be partially due to the obscure of Cu existing way and the complex structures. Experimentally the structures of Cu species in promoted catalysts are difficult to image due to the rather low content and very high dispersion. A previous study in characterizing the surface of Fe and Fe−Cu FTS catalysts by Wachs et al.44 revealed that Cu agglomeration is stronger on the Fe5C2 surfaces than on pure Fe surface; and metallic iron surface should have higher Cu affinity than carbide surfaces. De Smit et al.,45 studied the surface of Cu-promoted Fe-based FTS catalyst by using synchrotron-based in situ X-ray photoelectron spectroscopy; and found that Cu is in metallic form after H2 reduction and spreads to the metallic iron surface when treated at 275 °C.
reaction. It was reported that less than 0.1 wt % copper content is sufficient to produce active FTS catalysts,37 the optimum copper content for the Fe−Cu−K/ZSM5 catalyst activity is 2 wt %, and higher copper content results in iron oxide segregation, which eventually lowers the catalytic performances.38 Theoretically, there are only few reports about Cu promotion on Fe-based FTS catalysts. The investigation of CO adsorption and dissociation on the basis Fe/Cu surface alloy model39,40 and Cu monolayer model on Fe surface41 showed that Cu can reduce CO dissociation activity. Under the consideration of different existing way of Cu species, Tian et al.,42 studied CO adsorption and dissociation on the clean as well as nCuadsorbed and nCu-substituted Fe(100) surfaces (n = 1−3) at different coverage. It is found that CO adsorption and dissociation are disadvantageous on Cu doped surfaces, and the restraining extent increases while Cu content increasing; and the nCu-substituted Fe(100) surface can suppress CO adsorption and dissociation more strongly than the nCuadsorbed Fe(100) surface. For Cu promotion on iron carbide 7372
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
under real FTS condition (under CO pretreatment, the relative chemical potential ΔμC = −7.46 to −7.18), the nonstoichiometric Fe5C2(100) is most stable with surface Fe/C ratio of 2.25 and 2.00, followed by Fe5C2(111) with surface Fe/ C ratio of 1.75 (nonstoichiometric), Fe5C2(510) with surface Fe/C ratio of 2.50 (stoichiometric) and 2.00 (nonstoichiometric); the nonstoichiometric Fe5C2(010)-1.25 termination is less stable, and the Fe 5 C 2 (001)-2.50 (stoichiometric), Fe5C2(001)-2.13 (nonstoichiometric) are least stable. Furthermore, under the consideration of only the stoichiometric terminations of the 11 facets observed from the XRD patterns, Pham et al.,61 found that the high Miller index Fe5C2(510) surface is the thermodynamically most stable stoichiometric facet and has the largest percentage among the exposed crystal facets. On the basis of these studies, we used Fe5C2(100)-2.25, Fe5C2(111)-1.75, Fe5C2(510)-2.50, Fe5C2(010)-1.25, and Fe5C2(001)-2.50 for this work (Figure 1). The total slab thickness and the number of relaxed atom layers were first tested based on Cu adsorption energy to choose a reasonable periodic slab model. These results are listed in the Supporting Information (Table S1). For modeling the Fe5C2(100) surface, the p(2 × 2) slab consisting of six Fe layers and two C layers (6Fe + 2C) with a slab thickness of 5.72 Å is used, where the top four Fe layers and two C layers (4Fe + 2C) are allowed to relax, and the other bottom layers are fixed in their bulk positions. To model the Fe5C2(111) surface, the p(2 × 2)slab consisting of nine Fe layers and six C layers (9Fe+6C) with a slab thickness of 4.49 Å is used, and the top seven Fe layers and five C layers (7Fe+5C) are allowed to relax, while the other bottom layers are fixed in their bulk positions. For the Fe5C2(510) surface, the p(2 × 1) slab consisting of three Fe layers and six C layers (3Fe + 6C) with a slab thickness of 5.63 Å is used, and the top two Fe layers and four C layers (2Fe + 4C) are allowed to relax, while the other bottom layers are fixed in their bulk positions. For the Fe5C2(010) surface, the p(2 × 1) slab consisting of five Fe layers and six C layers (5Fe + 6C) with a slab thickness of 5.59 Å is used, and the top three Fe layers and four C layers (3Fe + 4C) are allowed to relax, while the other bottom layers are fixed in their bulk positions. For the Fe5C2(001) surface, the p(2 × 2) slab consisting of 10 Fe layers and four C layers (10Fe + 4C) with a slab thickness of 4.15 Å is used, and the top eight Fe layers and three C layers (8Fe + 3C) are allowed to relax, while the other bottom layers are fixed in their bulk positions. According to the lattice sizes, 3 × 3 × 1 k-point grid sampling within the Brillouin zones was set. All slab models have a 15 Å vacuum region to exclude significant interaction between the slabs.
Theoretically, the morphology of Cu on the Fe and Fe3C surfaces has been studied.46,47 It is found that Cu favors aggregation on the Fe(110), Fe(100) and metallic Fe3C(010) surfaces; but dispersion on the Fe(111) surface. On the Fe3C(001) and Fe3C(100) surfaces with exposed Fe and C atoms, the adsorbed Cu atoms prefer to disperse at very low coverage, and aggregate along the iron regions at high coverage. van Steen et al.43 found that atomic Cu is more favorable on the iron-rich χ-Fe5C2(100)0.25 surface than in the surface; and the segregation of Cu from the surface yielding bulk Cu remains exergonic. Since the detailed structural information is essential for understanding Cu promotion and Fe5C2 is one of the active phases in FTS, we examined the Cu morphology on the Fe 5C 2(100), Fe5 C2 (111), Fe5C 2(510), Fe5C 2(001), and Fe5C2(010) surfaces. The choice of these surfaces are based on the previous study about the surface structure and stability of various facets of the χ-Fe5C2 phase.48,49 The aim of this work is to provide references for advanced experimental studies and more rational models for discussing Cu promotion, which is of pronounced importance for understanding the Fe-based FTS mechanisms.
2. COMPUTATIONAL DETAILS 2.1. Methods. All calculations were performed at the DFT level within the Vienna ab initio simulation package (VASP). 50,51 The projected augmented wave method (PAW)52 was used. The electron exchange and correlation energies were calculated using the generalized gradient approximation in the Perdew−Burke−Ernzerhof (GGA-PBE) functional.53,54 The plane wave basis was set up to 400 eV. Spin-polarization calculation was included for iron systems to correctly account for the magnetic properties. Geometry optimization was done when atomic force tolerance becomes smaller than 0.02 eV/Å and the energy difference was lower than 10−4 eV. Bader charge analysis55 was carried out for discussing charge transfer between the surface and the adsorbed Cu atoms. All transition states were estimated using the climbing-image nudged elastic band (CI-NEB) method,56,57 and their vibrational frequencies were also calculated. Adsorption energies and stepwise adsorption energies are used to describe the energetic properties of Cun on the surfaces. For Cun adsorption on the surface, the total adsorption energy [E(Cun/ads)] is defined according to eq 1 and the average adsorption energy [E(Cuads/av] is defined in eq 2, where E(Cu)n/slab is the energy of the Cun adsorbed system, E(slab) is the energy of the clean surface, n is the number of the adsorbed Cu atoms, and E(Cugas) is the total energy of an isolated Cu atom in gas phase. E(Cu n/ads ) = [E(Cu)n /slab − E(slab) − nE(Cu gas)]
(1)
E(Cuads/av) = E(Cu n /ads)/n
(2)
3. RESULTS AND DISCUSSION For Cun adsorption on these surfaces, aggregation (Cun-a) and dispersion (Cun-d) are examined. On the basis of our previous studies, there are two factors controlling the surface morphology, i.e.; the dominating Cu−Fe interaction and the coordination numbers of Cu to surface. Since aggregation of Cun on Fe and Fe3C surfaces is two-dimensional favored up to Cu13, and this thermodynamic preference can be extrapolated to monolayer (ML) adsorption, we consider mainly the twodimensional Cun (up to 1 ML) adsorption, only the 3D Cun structures at 1 ML coverage was included for comparison. It should be noted that the 1 ML coverage used in this work is not defined according to the ratio of the adsorbed Cu atoms to the surface Fe atoms; instead, it represents the largest number
The stepwise adsorption energy [ΔE(Cuads)] is defined according to eq 3. ΔE(Cuads) = E(Cu n + 1/ads) − E(Cu n /ads)
(3)
2.2. Models. The χ-Fe5C2 phase has a monoclinic structure with C2/c crystallographic symmetry.58 By using a 2 × 6 × 5 Monkhorst−Pack grid of k-points, the calculated lattice parameters (a = 11.545 Å, b = 4.496 Å, c = 4.982 Å, and β = 97.6°) and magnetic moment (1.73 μB) agree reasonably with the available experimental data.59,60 Zhao et al.49 found that 7373
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Table 1. Adsorption Energy [E(Cun/ads); eV], Average Adsorption Energy [E(Cuads/av); eV], Stepwise Adsorption Energy [ΔE(Cuads); eV], Coordination Number (CN) with Surface Fe and C Atoms, Number of Cu−Cu Bonds (NB-Cu), Range of Cu−Cu Distances (d, Å), and Average Bader Charge (q, e) for Cun (n = 1−13, 16) on the Fe5C2(100) Surface E(Cun/ads)
E(Cuads/av)
Cu1
−2.84
−2.84
Cu2-d Cu3-d Cu4-d Cu5-d Cu6-d Cu7-d
−5.69 −8.52 −11.35 −14.77 −18.17 −21.57
−2.84 −2.84 −2.84 −2.95 −3.03 −3.08
Cu2-a Cu3-a Cu4-a Cu5-a Cu6-a Cu7-a Cu8-a Cu9-a Cu10-a Cu11-a Cu12-a Cu13-a Cu16-a
−6.00 −9.15 −12.57 −15.52 −18.72 −21.88 −25.25 −28.75 −32.02 −35.42 −38.88 −42.55 −53.23
−3.00 −3.05 −3.14 −3.10 −3.12 −3.13 −3.16 −3.19 −3.20 −3.22 −3.24 −3.27 −3.33
Cu16-3D1 Cu16-3D2
−52.74 −51.71
−3.30 −3.23
ΔE(Cuads)
CN-Fe
4 Dispersed Mode −2.84 8 −2.83 12 −2.83 16 −3.42 20 −3.40 22 −3.40 26 Aggregated Mode −3.16 8 −3.15 12 −3.41 16 −2.96 18 −3.20 22 −3.16 26 −3.38 28 −3.50 30 −3.27 32 −3.40 36 −3.47 38 −3.67 42 48 3D Structures 46 36
d
q
CN-C
NB-Cu
2
0
0.22
4 6 8 9 9 11
0 0 0 2 5 7
0.23 0.23 0.24 0.20 0.15 0.13
4 6 8 7 9 10 9 8 5 7 7 6 0
1 2 4 6 7 8 11 15 18 20 24 27 40
2.53 2.56−2.58 2.60−2.63 2.43−2.66 2.51−2.67 2.50−2.64 2.49−2.65 2.50−2.68 2.47−2.64 2.47−2.65 2.46−2.66 2.48−2.62 2.50−2.58
0.17 0.16 0.15 0.10 0.12 0.13 0.10 0.08 0.06 0.07 0.06 0.06 0.04
3 5
40 45
2.39−2.78 2.48−2.63
0.05 0.05
energy is −9.15 and −12.57 eV, respectively. They are more stable than the dispersed Cu3-d and Cu4-d with all Cu atoms in remote 6F sites by 0.63 and 1.22 eV, respectively. For n = 5−7, the dispersed structures become partially aggregated. Although the fully aggregated structures are more stable than the partially aggregated ones; their difference decreases along with the cluster size from 0.75 to 0.31 eV for n = 5−7. It can be inferred that at higher coverage both modes will merge, and we therefore only considered the aggregation mode for n > 7. For n = 8−16, the newly added Cu atoms grow almost line by line along the valley direction. At n = 16, 1 ML (Cu16-a) structure has adsorption energy of −52.23 eV; and is more stable than the structures with one (Cu16-3D1) and four (Cu16-3D2) Cu atoms at the second Cu layer by 0.49 and 1.52 eV, respectively, indicating the dominant role of Cu−Fe interaction over Cu−Cu interaction. That Cu atoms prefer to aggregate on Fe5C2(100) surface is also further verified by the average adsorption energies and stepwise adsorption energies; and they are larger than that of single Cu atom adsorption, indicating the additional Cu−Cu interaction along with the dominant Cu−Fe interaction. 3.2. Cun on the Fe5C2(111) Surface. The Fe5C2(111) surface has three C atom layers (C1, C2 and C3) and four Fe atom layers (Fe1, Fe2, Fe3, and Fe4) exposed, and the layer distances are too small to distinguish (Figure 1b). There are 11 possible adsorption sites on Fe5C2(111) surface: 7-fold (7F; 2Fe1 + Fe3 + Fe4 + 2C1 + C2), 6-fold (6F; 2Fe1 + Fe2 + Fe4 + C1 + C3), 5-fold (5F; Fe1 + Fe2 + 2Fe3 + C3), 4-fold (4F1; Fe1 + 2Fe3 + C3 and 4F2; 2Fe1 + Fe2 + C1), 3-fold (3F1; 2Fe1 + C1, 3F2; Fe2 + Fe3 + C2, and 3F3; Fe1 + Fe3 + C1), 2-fold (2F; Fe1 + Fe2) and top (1F1 and 1F2) Fe sites. The results for Cun (n = 1−13, 16) adsorption on the surface are
of the adsorbed Cu atoms forming a monolayer structure on the Fe5C2 surface. 3.1. Cun on the Fe5C2(100) Surface. The Fe5C2(100) surface has an ordered wave-like structure with one first layer C and one first layer Fe (Fe1) on the peak of the wave as well as one second layer Fe (Fe2) in the valley of the wave (Figure 1a). Both first and second layer Fe atoms are in zigzag arrangement along the valley direction. The Fe5C2(100) surface has 6-fold (6F, 1Fe2 + 3Fe1 + 2C), 4-fold (4F, 2Fe2 + 1Fe + 1C), 3-fold (3F, 1C + 2Fe1), 2-fold (2F; 2Fe1) and one-fold (1F; 1Fe1) sites for adsorption. The results for Cun (n = 1−13, 16) adsorption on the surface are listed in Table 1, and the adsorption configurations are shown in Figure 2. The detailed adsorption structures and energies after optimization are given in the Supporting Information (Figure S1). The most stable adsorption site for one Cu atom is the 6F site with the adsorption energy of −2.84 eV; and the other adsorption sites are less stable; and this correlates with the coordination numbers; i.e., the 6F site has the highest coordination number. In addition, it is also found that the diffusion barrier of the adsorbed Cu atom along the valley from one 6F site to another 6F site via the 4F bridging site is only 0.28 eV, indicating the high mobility of the adsorbed Cu atom at the 6F site (Figure S2). For the more stable aggregated Cu2-a, both Cu atoms are located at the neighboring 6F sites with the Cu−Cu distance of 2.53 Å. The computed adsorption energy is −6.00 eV, which is larger than twice of single Cu atom adsorption (−5.68 eV); and the dispersed Cu2-d with two Cu atoms in remote 6F sites is less stable by 0.31 eV. Along with the valley, the most stable aggregated Cu3-a and Cu4-a have all Cu atoms located at the 6F sites with direct Cu−Cu interaction; and the adsorption 7374
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Figure 2. Structures for Cun (n = 1−13, 16) on the Fe5C2(100) surface (blue balls for Fe atoms, black balls for C atoms and red balls for Cu atoms).
aggregated one by 0.23 and 0.28 eV, respectively. For n = 13, the energy difference between two modes is only 0.03 eV, indicating that at higher coverage both modes will merge. Because of the cost of calculation, we used the smaller p(2 × 1) model for the 1 ML coverage adsorption. The 1 ML structure (Cu16-a) with the adsorption energy of −53.24 eV is more stable than the structures with one and four Cu atoms at the second Cu layer by 0.34 and 1.28 eV, respectively, again indicating the dominant role of Cu−Fe interaction over Cu− Cu interaction. 3.3. Cun on the Fe5C2(510) Surface. The Fe5C2(510) surface has one Fe atom and two C atom layers exposed (Figure 1c). There are 28 possible adsorption sites, the 4-fold (4F1−4F6) sites, 3-fold (3F1−3F11) sites, 2-fold (2F1−2F5) sites, and one-fold (1F1−1F6) sites. For the Cun growth (n = 1−17, 20), the adsorption energies and structure information are listed in Table 3, the adsorption configurations are shown in Figure 4. The detailed adsorption structures and energies are given in the Supporting Information (Figure S5). For single Cu atom, the most stable adsorption configuration on the surface is located at the 4F1 site with the adsorption energy of −3.27 eV. Because of the remote distance between two 4F1 sites, the migration of the Cu atom from one 4F1 to another 4F1 needs several steps and the most favorable route
listed in Table 2 and the adsorption configurations are shown in Figure 3. The detailed adsorption structures and energies after optimization are given in the Supporting Information (Figure S3). For single Cu atom, the most stable adsorption configuration on the surface is located at the 7F site with the adsorption energy of −3.23 eV. Because of the remote distance between two 7F sites, the migration of the Cu atom needs several steps and the most favorable route has energy barrier of 1.33 eV, revealing that the Cu atom at the 7F site is difficult to migrate (Figure S4). For Cu2 and Cu3, the dispersed modes with all Cu atoms remotely located at the 7F sites are more stable than the aggregated modes with direct Cu−Cu interaction by 0.77 and 0.41 eV, respectively. For Cu4, however, the dispersed mode with all Cu atoms remotely located at the 7F sites is only 0.02 eV more stable than the aggregated mode with four Cu atoms in a zigzag line. For n = 5−10, the aggregated mode becomes more stable than the dispersed ones, the corresponding energy difference is 0.09, 0.09, 0.16, 0.29, 0.67, and 0.43 eV, respectively. It is to note that the aggregated modes become partially dispersed since n = 7, and the dispersed modes become partially aggregated since n = 9. For n = 11−12, the more partially dispersed mode is more stable than the more partially 7375
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Table 2. Adsorption Energy [E(Cun/ads); eV], Average Adsorption Energy [E(Cuads/av); eV], Stepwise Adsorption Energy [ΔE(Cuads); eV], Coordination Number (CN) with Surface Fe and C Atoms, Number of Cu−Cu Bonds (NB-Cu), Range of Cu−Cu Distances (d, Å), and Average Bader Charge (q, e) for Cun (n = 1−13, 16) on the Fe5C2(111) Surface Cu1
E(Cun/ads)
E(Cuads/av)
−3.23
−3.23
Cu2-d Cu3-d Cu4-d Cu5-d Cu6-d Cu7-d Cu8-d Cu9-d Cu10-d Cu11-d Cu12-d Cu13-d
−6.78 −9.77 −12.91 −16.09 −19.26 −22.34 −25.40 −28.43 −31.77 −35.32 −38.50 −41.63
−3.39 −3.26 −3.23 −3.22 −3.21 −3.19 −3.18 −3.16 −3.18 −3.21 −3.21 −3.20
Cu2-a Cu3-a Cu4-a Cu5-a Cu6-a Cu7-a Cu8-a Cu9-a Cu10-a Cu11-a Cu12-a Cu13-a Cu16; p(2 × 1)
−6.01 −9.36 −12.89 −16.18 −19.35 −22.50 −25.69 −29.10 −32.20 −35.09 −38.19 −41.66 −53.24
−3.00 −3.12 −3.22 −3.24 −3.23 −3.21 −3.21 −3.23 −3.22 −3.19 −3.18 −3.20 −3.33
Cu16−3D1; p(2 × 1) Cu16−3D2; p(2 × 1)
−52.90 −51.96
−3.31 −3.25
ΔE(Cuads)
d
q
CN-Fe
CN-C
NB-Cu
4
3
0
0.38
8 12 16 21 26 31 36 40 42 44 46 47
6 9 12 13 14 16 18 18 18 19 21 21
0 0 0 0 0 0 0 2 5 9 11 13
0.35 0.40 0.40 0.36 0.34 0.33 0.33 0.30 0.27 0.24 0.22 0.21
7 12 13 19 20 24 26 25 30 38 31 40 39
3 3 5 6 6 9 10 11 14 15 14 17 12
1 2 3 4 7 7 10 11 12 13 15 17 44
2.58 2.55−2.64 2.45−2.67 2.45−2.69 2.43−2.67 2.43−2.66 2.44−2.68 2.37−2.55 2.35−2.62 2.40−2.77 2.40−2.56 2.38−2.76 2.38−2.70
0.27 0.22 0.23 0.22 0.19 0.22 0.19 0.17 0.18 0.18 0.15 0.17 0.07
39 37
13 16
44 38
2.40−2.69 2.34−2.69
0.08 0.10
Dispersed Mode −3.55 −2.99 −3.14 −3.18 −3.17 −3.09 −3.06 −3.02 −3.35 −3.54 −3.18 −3.13 Aggregated Mode −2.78 −3.36 −3.53 −3.29 −3.17 −3.15 −3.19 −3.41 −3.11 −2.88 −3.10 −3.47 3D Structures
atom layers exposed after relaxation (Figure 1d). There are 13 stable absorption sites, 6-fold (6F), 5-fold (5F1, 5F2, 5F3), 4fold (4F1, 4F2, 4F3, 4F4), 3-fold (3F1, 3F2, 3F3), 2-fold (2F), and 1-fold (1F) sites. The results for Cun (n = 1−7, 20) adsorption on the surface are listed in Table 4, the adsorption configurations are shown in Figure 5. The detailed adsorption structures and energies after optimization are given in the Supporting Information (Figure S7). On the Fe5C2(010) surface it is noted that the 6F sites are on the line bisecting two repeating units. The most stable adsorption site for one Cu atom is the 6F site with the adsorption energy of −3.53 eV. The migration of the Cu atom needs several steps and the most favorable route has energy barrier of 0.85 eV (Figure S8). For Cu2, the dispersion mode with two remotely located Cu atoms at the 6F sites (Cu2-d, −7.15 eV) is more stable than the aggregated one (Cu2-a, −6.88 eV) with one Cu atom at the 6F site and one at the 5F1 site. For Cu3 and Cu4, the fully aggregated modes along the bisecting line are much more stable than the fully dispersed and partially aggregated modes (Figure S7). On the basis of Cu4-a, we therefore considered only these aggregated adsorption configurations for Cu5, Cu6, and Cu7, with Cu atoms around the bisecting line, and they have very ordered arrangement of Cu atoms on the surface. On the basis of these ordered structures we explored the 1 ML structure (Cu20-a; −69.43 eV), which is more stable than the structure
has energy barrier of 0.51 eV, indicating its mobility (Figure S6). For Cu2, the dispersion mode (Cu2-d) and the aggregated mode (Cu2-a) have the same adsorption energy (−6.50 eV). In Cu2-d, both Cu atoms are remotely located at the 4F1 sites, while in Cu2-a, one Cu atom is located at the 4F1 site and one is located at the 3F2 site; and they form direct Cu−Cu bonding. For Cu3 and Cu4, the aggregated mode with Cu atoms in zigzag line is more stable than the fully dispersed mode by 0.63 and 1.37 eV, respectively. On the basis of the aggregated Cu3 and Cu4 modes, we computed the fully aggregated and partially dispersed Cu5, Cu6 and Cu7 modes; and the fully aggregated mode is more stable than the partially dispersed one by 1.23, 1.06, and 1.59 eV, respectively. On the basis of the aggregated Cu7 mode, we computed the aggregated modes for n = 8−17. The 1 ML structure (Cu20-a) has 20 Cu atoms on the surface and the adsorption energy is −68.94 eV. As expected, the Cu20-a structure is more stable than the structures with one and four Cu atoms at the second Cu layer by 0.66 and 1.99 eV, respectively, again indicating the dominant role of Cu−Fe interaction over Cu−Cu interaction. 3.4. Cun on the Fe5C2(010) Surface. On the basis of the above results, we only calculated the adsorption of Cun for n = 1−7 and the monolayer structure for n = 20 on the Fe5C2(010). The Fe5C2(010) surface has three C atom layers and two Fe 7376
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Figure 3. Structures for Cun (n = 1−13, 1 ML) on the Fe5C2(111) surface (blue balls for Fe atoms, black balls for C atoms and red balls for Cu atoms).
The most stable adsorption site for one Cu atom is also the 6F site with the adsorption energy of −3.53 eV. The more favorable migration path of the Cu atom from one 6F site to another 6F site needs two steps and the energy barrier is 0.74 eV (Figure S10). For Cu2, Cu3, and Cu4, there are one fully dispersed and two aggregated modes, respectively. The fully dispersed mode with all Cu atoms at the 6F sites is more stable than the aggregated ones by 0.30/0.30, 0.33/0.15, and 0.18/0.22 eV, respectively. For Cu5, Cu6, and Cu7, there are one partially dispersed and two aggregated modes, respectively. The partially dispersed mode with four Cu atoms at the 6F sites is more stable than the aggregated modes by 0.23/0.11, 0.26/0.22 and 0.17/0.36 eV, respectively. The small energy differences between the
with one and four Cu atoms in the second Cu layer by 0.70 and 2.82 eV, respectively. 3.5. Cun on the Fe5C2(001) Surface. As on the Fe5C2(010) surface we only calculated the adsorption of Cun for n = 1−7 and the monolayer structure for n = 20 on the Fe5C2(001) (Figure 1e). The Fe5C2(001) surface has two C atom layers and five Fe atom layers exposed on the surface after relaxation. There are nine possible sites for adsorption, 6-fold (6F), 4-fold (4F1, 4F2, 4F3), 3-fold (3F), 2-fold (2F1, 2F2), and 1-fold (1F1, 1F2) sites. The results for Cun (n = 1−7, 20) adsorption on the surface are listed in Table 5, the adsorption configurations are shown in Figure 6. The detailed adsorption structures and energies after optimization are given in the Supporting Information (Figure S9). 7377
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Table 3. Adsorption Energy [E(Cun/ads); eV], Average Adsorption Energy [E(Cuads/av); eV], Stepwise Adsorption Energy [ΔE(Cuads); eV], Coordination Number (CN) with Surface Fe and C Atoms, Number of Cu−Cu Bonds (NB-Cu), Range of Cu−Cu Distances (d, Å), and Average Bader Charge (q, e) for Cun (n = 1−17, 20) on the Fe5C2(510) Surface ΔE(Cuads)
E(Cun/ads)
E(Cuads/av)
−3.27
−3.27
Cu2-d Cu3-d Cu4-d Cu5-d Cu6-d Cu7-d
−6.50 −9.36 −12.17 −15.65 −19.19 −22.19
−3.25 −3.12 −3.04 −3.13 −3.20 −3.17
−3.23 −2.86 −2.82 −3.48 −3.53 −3.00
Cu2-a Cu3-a Cu4-a Cu5-a Cu6-a Cu7-a Cu8-a Cu9-a Cu10-a Cu11-a Cu12-a Cu13-a Cu14-a Cu15-a Cu16-a Cu17-a Cu20-a
−6.50 −9.99 −13.54 −16.88 −20.25 −23.78 −27.33 −30.73 −34.11 −37.35 −40.62 −44.04 −47.41 −50.76 −54.11 −57.76 −68.94
−3.25 −3.33 −3.38 −3.38 −3.37 −3.40 −3.42 −3.41 −3.41 −3.40 −3.38 −3.39 −3.39 −3.38 −3.38 −3.40 −3.45
−3.24 −3.49 −3.55 −3.35 −3.36 −3.53 −3.55 −3.40 −3.38 −3.25 −3.26 −3.42 −3.37 −3.35 −3.35 −3.65
Cu20-3D1 Cu20-3D2
−68.28 −66.95
−3.41 −3.35
Cu1
CN-Fe
4 Dispersed Mode 8 11 14 17 20 23 Aggregated Mode 7 11 14 18 22 25 28 31 34 37 40 43 46 49 52 53 58 3D Structures 58 47
d
q
CN-C
NB-Cu
0
0
−0.03
0 0 0 0 0 0
0 0 0 2 4 5
−0.04 −0.03 −0.04 −0.06 −0.07 −0.06
0 0 0 0 0 0 0 1 2 2 2 3 4 4 4 4 4
1 2 4 6 8 11 14 16 18 20 22 26 28 30 32 34 48
2.55 2.53−2.54 2.55 2.55−2.58 2.53−2.59 2.51−2.65 2.52−2.62 2.47−2.62 2.47−2.60 2.47−2.64 2.46−2.66 2.43−2.60 2.43−2.60 2.42−2.71 2.48−2.69 2.47−2.72 2.48−2.76
−0.05 −0.07 −0.09 −0.07 −0.05 −0.03 −0.03 −0.02 −0.02 −0.02 −0.02 −0.01 −0.01 −0.02 −0.03 −0.03 −0.02
3 2
45 51
2.42−2.72 2.31−2.78
−0.02 −0.02
more favored than aggregation for n = 2−4, while partial aggregation becomes more favored than dispersion at higher Cu coverage. On the Fe5C2(510)-2.50 surface, aggregation and dispersion have the same adsorption energy for n = 2, while aggregation is more favored than dispersion up to 1 ML Cu coverage. On the Fe5C2(010)-1.25 surface, dispersion is more favored than aggregation for n = 2, and aggregation becomes more favored than dispersion up to 1 ML Cu coverage. On the Fe5C2(001)-2.50 surface, dispersion is more favored than aggregation for n = 2−4, while partial aggregation becomes more favored for n = 5−7 and fully aggregated structures could be expected at higher coverage. The difference in structures and stability at low coverage depends on the stronger Cu−Fe interaction over the Cu−Cu interaction as well as the location of the most stable adsorption sites. For the most stable adsorption sites in close distance, it is possible to have Cu−Fe and Cu−Cu interaction and aggregation is more favored, while for the most stable adsorption sites in remote distances, it is only possible to have Cu−Fe interaction and dispersion is more favored. With the increase of Cu coverage, monolayer adsorption is the only possible form and is also more stable than the structures with double Cu layers, and this is because of the stronger Cu−Fe interaction. Despite the interaction of the adsorbed Cun clusters with the surface atoms and the Cu−Cu bonding, it is interesting to note that the Cu−Cu distances vary in a close range (Tables 1-5); and the average Cu−Cu distances are close to the value of fcc bulk Cu (2.56 Å). However, there is no direct correlation
dispersed and aggregated modes show that both modes can merge at high coverage. At 1 ML, the p(2 × 2) Fe5C2(001) can adsorb 20 Cu atoms. As expected, the 1 ML structure (Cu20-a; −71.21 eV) is more stable than the structure with one and four Cu atoms in the second Cu layer by 0.70 and 2.47 eV, respectively. 3.6. Discussion. In Fe-based FTS catalysts, the Hägg iron carbide (χ-Fe5C2) has been considered as one of the active phases. Under catalytic conditions, the χ-Fe5C2 phase has several surface terminations exposed. Since copper has been used as a component in promoting Fe-based FTS catalysis, we need to understand the surface morphology of Cu promoter as the first step for exploring the promotion mechanisms. Therefore, we computed the structures and stability of Cu adsorption on the nonstoichiometric Fe 5 C 2 (100)-2.25, Fe5C2(111)-1.75, and Fe5C2(010)-1.25 surfaces as well as the stoichiometric Fe5C2(510)-2.50 and Fe5C2(001)-2.50 surfaces in different surface Fe and C ratios. Along with the results on the Fe and Fe3C surfaces,46,47 we can show the surface differences among all these active phases in Fe-based FTS catalysts. For comparison, Table 6 list the surface energies,62,63 the average adsorption energies of Cun (n = 1, 13, 16 or 20) and Bader charge (q, e) for single Cu atom on Fe, Fe3C, and Fe5C2 surfaces. All these surfaces have common and different properties in Cu adsorption structures and stability. On the Fe5C2(100)-2.25 surface, aggregation is more favored than dispersion up to 1 ML Cu coverage. On the Fe5C2(111)-1.75 surface, dispersion is 7378
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Figure 4. Structures for Cun (n = 1−17, 20) on the Fe5C2(510) surface (blue balls for Fe atoms, black balls for C atoms and red balls for Cu atoms).
of the monolayer structures of Cu16 on the Fe5C2(111) and Cu20 on the Fe5C2(001) surfaces look like the Cu(111) surface. The Fe(110) and Fe(100) surfaces favor aggregation over dispersion up to 1 ML coverage, while the Fe(111) surface favors dispersion over aggregation up to 1 ML coverage. This is due to their differences in surface structures, for example, the Fe(110) and Fe(100) surfaces are more dense and have the most stable adsorption sites in close distances, and the Fe(111) surface is more open and has the most stable adsorption sites in remote places. Indeed, the layer-by-layer growth mode of Cu on the Fe(100) surface has been verified by experiment.64−66
between the Cu−Cu distances and the transferred charges. This is the same as found on the pure metallic Fe(110) and Fe(100) surfaces, where the adsorbed Cu atoms are negatively charged.46,47 Since the Fe5C2 surface terminations have different Fe/C ratios and arrangements, the surface structures determine the Cu−Cu distances of the adsorbed Cun clusters. On the fully Cu covered surfaces, the monolayer structures of the Cun clusters do not have the planar and regular structure of the body-centered cubic Cu(111) surface, since all these carbide surfaces are not planar and regular within the surface sizes. Instead, the monolayer structures of the Cun clusters are wave-shaped alternating (Figure S11), although the top views 7379
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Table 4. Adsorption Energy [E(Cun/ads); eV], Average Adsorption Energy [E(Cuads/av); eV], Stepwise Adsorption Energy [ΔE(Cuads); eV], Coordination Number (CN) with Surface Fe and C Atoms, Number of Cu−Cu Bonds (NB-Cu), Range of Cu−Cu Distances (d, Å), and Average Bader Charge (q, e) for Cun (n = 1−7, 20) on the Fe5C2(010) Surface E(Cun/ads)
E(Cuads/av)
−3.53
−3.53
Cu2-d Cu3-d Cu4-d
−7.15 −10.56 −14.09
−3.57 −3.52 −3.52
Cu2-a Cu3-a Cu4-a Cu5-a Cu6-a Cu7-a Cu20-a
−6.88 −10.70 −14.17 −17.63 −21.09 −24.42 −69.43
−3.44 −3.57 −3.54 −3.53 −3.52 −3.49 −3.47
Cu20-3D1 Cu20-3D2
−68.73 −66.61
−3.44 −3.33
Cu1
ΔE(Cuads)
CN-Fe
4 Dispersed Mode −3.62 8 −3.42 12 −3.53 14 Aggregated Mode −3.36 8 −3.82 10 −3.47 12 −3.46 16 −3.47 20 −3.33 24 56 3D Structures 54 52
d
q
CN-C
NB-Cu
2
0
4 5 7
0 1 3
2.49 2.51−2.59
0.36 0.31 0.28
3 6 8 9 10 11 24
0 2 4 5 6 7 52
2.47 2.55−2.56 2.58−2.59 2.45−2.67 2.44−2.67 2.43−2.64 2.45−2.68
0.3 0.31 0.29 0.27 0.26 0.25 0.14
23 20
51 46
2.38−2.76 2.25−2.77
0.14 0.13
0.37
Figure 5. Structures for Cun (n = 1−7, 20) on the Fe5C2(010) surface (blue balls for Fe atoms, black balls for C atoms and red balls for Cu atoms).
For the most stable surfaces, the adsorption energy of single Cu atom as well as the average adsorption energy at the highest coverage (Table 6) on Fe(110) surface (−3.30 vs −3.61 eV) is higher than that on the Fe3C(001) surface (−3.28 vs −3.35 eV) as well as on the Fe5C2(100) surface (−2.84 vs −3.33 eV). These results are in agreement with the experimental observations,44 where the metallic Fe surface has higher Cu affinity than the carbide surfaces with Fe and C atoms exposed. The lower Cu affinity of bulk Fe3C is also in line with the previous theoretical study based on the formation energies of (Fe0.917Cu0.083)3C67 and the partitioning enthalpy68 of Fen‑1Cu to Fe3q‑1CuCq. It is found that substituting Fe atom by Cu atom can destabilize the Fe3C phase, and during the competitive Cu dissolving in the body-centered cubic Fe and in Fe3C phases Cu prefers to partition to the Fe phase. Similarly, substituting Fe atom by Cu atom can also destabilize the Fe5C2 phase, and the partitioning enthalpies of Cu in Fe5C2 phase (0.18, 0.30, 0.66 eV/atom) is close to the Cu in Fe3C phase (0.24, 0.16 eV/
On the Fe3C(001) and Fe3C(100) surfaces with exposed iron and carbon atoms, the adsorbed Cu atoms prefer dispersion at low coverage, and aggregation within the iron regions at high coverage, while the metallic Fe3C(010) surface favors aggregation over dispersion up to 1 ML coverage. For single Cu atom adsorption of Fe, Fe3C and Fe5C2 phases (Table 6), the most stable Fe5C2(100)-2.25 surface has the lowest adsorption energy (−2.84 eV), while those (−3.23 and −3.27 eV, respectively) of the Fe 5 C 2 (111)-1.75 and Fe5C2(510)-2.50 surfaces are higher, and those (−3.53 and −3.53 eV, respectively) of the Fe 5 C 2 (010)-1.25 and Fe5C2(001)-2.50 surfaces are the highest. At 1 ML coverage, the Fe5C2(100)-2.25 and Fe5C2(111)-1.75 surfaces have the same average adsorption energy (−3.33 eV), and those of the Fe5C2(510)-2.50, Fe5C2(010)-1.25 and Fe5C2(001)-2.50 surfaces are higher (−3.45, −3.47, and −3.40 eV, respectively). It is noted that the adsorption energies do not correlate with the surface Fe and C ratios. 7380
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Table 5. Adsorption Energy [E(Cun/ads); eV], Average Adsorption Energy [E(Cuads/av); eV], Stepwise Adsorption Energy [ΔE(Cuads); eV], Coordination Number (CN) with Surface Fe and C Atoms, Number of Cu−Cu Bonds (NB-Cu), Range of Cu−Cu Distances (d, Å), and Average Bader Charge (q, e) for Cun (n = 1−7, 20) on the Fe5C2(001) Surface E(Cun/ads)
E(Cuads/av)
−3.53
−3.53
Cu2-d Cu3-d Cu4-d Cu5-d Cu6-d Cu7-d
−7.00 −10.44 −13.88 −17.20 −20.75 −24.13
−3.50 −3.48 −3.47 −3.44 −3.46 −3.45
Cu2-a1 Cu3-a1 Cu4-a1 Cu5-a1 Cu6-a1 Cu7-a1 Cu20-a
−6.70 −10.11 −13.70 −16.97 −20.49 −23.97 −71.21
−3.35 −3.37 −3.43 −3.39 −3.41 −3.42 −3.56
Cu2-a2 Cu3-a2 Cu4-a2 Cu5-a2 Cu6-a2 Cu7-a2
−6.70 −10.29 −13.66 −17.09 −20.53 −23.77
−3.35 −3.43 −3.41 −3.42 −3.42 −3.40
Cu20-3D1 Cu20-3D2
−70.51 −68.74
−3.53 −3.44
Cu1
ΔE(Cuads)
CN-Fe
5 Dispersed Mode −3.47 10 −3.44 15 −3.44 20 −3.32 23 −3.55 26 −3.38 30 Aggregated Mode 1 −3.17 8 −3.41 11 −3.59 16 −3.27 20 −3.52 24 −3.49 29 52 Aggregated Mode 2 −3.17 8 −3.59 13 −3.37 16 −3.43 18 −3.44 20 −3.25 22 3D Structures 50 46
d
q
CN-C
NB-Cu
1
0
0.21
2 3 4 5 5 5
0 0 0 2 4 5
2.51−2.66 2.56−2.67 2.54−2.73
0.19 0.17 0.15 0.12 0.08 0.06
1 2 3 3 4 5 12
1 2 3 5 5 5 60
2.56 2.58−2.61 2.50−2.62 2.51−2.68 2.54−2.67 2.56−2.74 2.38−2.77
0.02 0.05 0.05 0.07 0.04 0.05 0.02
2 3 4 4 4 5
1 2 4 7 10 12
2.46 2.53−2.64 2.49−2.60 2.47−2.67 2.51−2.67 2.46−2.77
0.08 0.10 0.07 0.05 0.04 0.03
11 10
57 55
2.39−2.75 2.39−2.76
0.02 0.02
qualitative correlation between the number of Cu−C interaction and the strength of the positive charge.
atom), indicating the comparative Cu affinity of these two phase.69 On the less stable surfaces, there are no such relationships and disorders have been found. Among all these surface, the Fe(111) surface has the highest adsorption energy of singe Cu atom (−3.77 eV). Despite the fact that all these adsorptions have negative adsorption energies, there are remarkable differences in charge transfer between surface and the adsorbed Cu atoms (Table 6). On the metallic Fe(110), Fe(100), and Fe(111) as well as Fe3C(010), the adsorbed Cu atoms are negatively charged; and this is associated with the difference in electronegativity between Cu and Fe atoms (1.80 for Fe; and 1.85 for Cu),70 where the stronger electronegative Cu atoms can attract electron from the less electronegative Fe atoms. On the Fe 3 C(100), Fe 5 C 2 (100), Fe 5 C 2 (111), Fe 5 C 2 (010), and Fe5C2(001) surfaces with Fe and C exposed, the adsorbed Cu atoms are positively charged; and this is because of the much higher electronegativity of C atom (2.54 for C) and the direct Cu−C interaction. However, it is also noted that the adsorbed Cu atoms are very slightly negatively charged on the Fe3C(100) and Fe5C2(510) surfaces, despite of the exposed C atom; and this is because that the adsorbed Cu atoms mainly interact with the exposed surface Fe atoms, and they have much less Cu−C interaction. For single Cu atom on the Fe5C2(100), Fe5C2(111), Fe5C2(010), and Fe5C2(001) surfaces, for example, the number of Cu−C bonding is 2, 3, 2 and 1, respectively, and the positive charge of the Cu atom is 0.22, 0.38, 0.37, and 0.21 e, respectively, while on the Fe5C2(510) surface, there is no direct Cu−C interaction and the Cu atom is slightly negatively charged (−0.03 e). This indicates a
4. CONCLUSIONS The configurations and stability of Cun adsorption up to one monolayer coverage on the Fe 5 C 2 (100), Fe 5 C 2 (111), Fe5C2(510), Fe5C2(010), and Fe5C2(001) surfaces with different surface Fe and C ratios have been systemically computed by using spin-polarized density functional theory method. Along with the previous studies on the Fe(110), Fe(100), and Fe(111) surfaces as well as on the Fe3C(001), Fe3C(100), and Fe3C(010) surfaces, it is able to outline some clues for advanced experimental studies in discussing Cu promotion, which is of pronounced importance for understanding Fe-based FTS mechanisms. This is because that iron and iron carbide phases have been considered as active phases in Fe-based FTS catalysts. On the Fe5C2(100)-2.25 and Fe5C2(510)-2.50 surfaces, aggregation is energetically more favored than dispersion up to 1 ML Cu coverage. On the Fe5C2(010)-1.25 surfaces, dispersion is more favored at very low coverage and aggregation becomes more favored at high coverage; on the Fe5C2(111)1.75 and Fe5C2(001)-2.50 surfaces, dispersion is more favored at very low coverage and partial aggregation becomes more favored at higher coverage and the monolayer structure can be expected. The difference in structures and stability at low coverage depends on the stronger Cu−Fe interaction over the Cu−Cu interaction as well as the location of the most stable adsorption sites. Different adsorption structures and stability have also been found on the Fe and Fe3C surfaces up to 1 ML Cu coverage, 7381
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
Figure 6. Structures for Cun (n = 1−7, 20) on the Fe5C2(001) surface (blue balls for Fe atoms, black balls for C atoms and red balls for Cu atoms).
Table 6. Surface Energy (γ, J/m2), Average Adsorption Energy of Cun (n = 1, 13, 16 or 20) and Bader Charge (q, e) for a Single Cu Atom on Iron and Iron Carbide Surfaces Fe γ E(Cu)d E(Cu13)d E(Cu16/20)d qd
Fe3C
Fe5C
(110)
(100)
(111)
(001)
(010)
(100)
(100)
(111)
(510)
(010)
(001)
2.37a −3.3 −3.61
2.44a −3.16 −3.64
2.60a −3.77 −3.77e
2.05b −3.28 −3.35
2.26b −3.39 −3.65
2.47b −3.58 −3.58
1.75c −2.84 −3.27 −3.33 0.22
1.91c −3.23 −3.20 −3.33 0.38
2.03c −3.27 −3.39 −3.45f −0.03
2.17c −3.53
2.55c −3.53
−3.47f 0.37
−3.56f 0.21
−0.11
−0.28
−0.27
−0.08
−0.13
0.16
a
Surface energies from ref 62. bSurface energies from ref 63. cSurface free energies of each Fe5C2 surface at 600 K and 30 atm in CO pretreatment from ref 48. dThe data of Fe and Fe3C phases are obtained from refs 46 and 47. eAverage adsorption energy of Cu9 in dispersion mode. fAverage adsorption energy of Cu20
the Fe(110) and Fe(100) surfaces favor aggregation, while the Fe(111) surface favors dispersion. On the Fe3C(001) and Fe3C(100) surfaces with exposed Fe and C atoms, the adsorbed Cu atoms prefer dispersion at low coverage, while aggregation within the iron regions at high coverage. The metallic Fe3C(010) surface favors aggregation over dispersion up to 1 ML coverage. On all these metal and carbide surfaces, Cu atoms prefer twodimensional surface adsorption configurations, while the adsorptions of three-dimensional Cu clusters are thermodynamically less favorable. It is therefore to expect that the
adsorbed Cu atoms almost epitaxially grow as a layer-by-layer mode at the initial stage. Among the most stable Fe(110), Fe 3 C(001), and Fe5C2(100) surfaces, which represent the most exposed surfaces in bulk catalysts, the Fe(110) surface has stronger Cu affinity than the Fe3C(001) and Fe5C2(100) surfaces; and this is in agreement with the experimental observations and the previous theoretical studies. Because of the different surface structures and compositions, the adsorbed Cu atoms are negatively charged on the Fe(110), Fe(100), Fe(111), and Fe3C(010) surfaces with only Fe atoms 7382
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
(5) Li, S.; Ding, W.; Meitzner, G. D.; Iglesia, E. Spectroscopic and Transient Kinetic Studies of Site Requirements in Iron-Catalyzed Fischer−Tropsch Synthesis. J. Phys. Chem. B 2002, 106, 85−91. (6) Davis, B. H. Fischer−Tropsch Synthesis: Reaction Mechanisms for Iron Catalysts. Catal. Today 2009, 141, 25−33. (7) Steynberg, A.; Dry, M. E. Fischer−Tropsch Technology; Studies in Surface Science and Catalysis 152; Elsevier: Amsterdam, 2004. (8) Herranz, T.; Rojas, S.; Pérez-Alonso, F. J.; Ojeda, M.; Terreros, P.; Fierro, J. L. G. Genesis of Iron Carbides and Their Role in the Synthesis of Hydrocarbons from Synthesis Gas. J. Catal. 2006, 243, 199−211. (9) de Smit, E.; Cinquini, F.; Beale, A. M.; Safonova, O. V.; van Beek, W.; Sautet, P.; Weckhuysen, B. M. Stability and Reactivity of ϵ-χ-θ Iron Carbide Catalyst Phases in Fischer−Tropsch Synthesis: Controlling μC. J. Am. Chem. Soc. 2010, 132, 14928−14941. (10) Yang, C.; Zhao, H.; Hou, Y.; Ma, D. Fe5C2 Nanoparticles: A Facile Bromide-Induced Synthesis and as an Active Phase for Fischer− Tropsch Synthesis. J. Am. Chem. Soc. 2012, 134, 15814−15821. (11) Cao, D. B.; Zhang, F. Q.; Li, Y.-W.; Jiao, H. Density Functional Theory Study of CO Adsorption on Fe5C2(001), -(100), and -(110) Surfaces. J. Phys. Chem. B 2004, 108, 9094−9104. (12) Cao, D. B.; Zhang, F. Q.; Li, Y.-W.; Jiao, H. Density Functional Theory Study of Hydrogen Adsorption on Fe5C2(001), Fe5C2(110), and Fe5C2(100). J. Phys. Chem. B 2005, 109, 833−844. (13) Cao, D. B.; Zhang, F. Q.; Li, Y.-W.; Jiao, H. Structures and Energies of Coadsorbed CO and H2 on Fe5C2(001), Fe5C2(110), and Fe5C2(100). J. Phys. Chem. B 2005, 109, 10922−935. (14) Cao, D. B.; Wang, S. G.; Li, Y.-W.; Jiao, H. What Is the Product of Ketene Hydrogenation on Fe5C2(001): Oxygenates or Hydrocarbons? J. Mol. Catal. A 2007, 272, 275−287. (15) Cao, D. B.; Li, Y.-W.; Wang, J.; Jiao, H. Chain Growth Mechanism of Fischer−Tropsch Synthesis on Fe5C2(001). J. Mol. Catal. A: Chem. 2011, 346, 55−69. (16) Steynberg, P. J.; van den Berg, J. A.; Janse van Rensburg, W. Bulk and Surface Analysis of Hägg Fe Carbide (Fe5C2): A Density Functional Theory Study. J. Phys.: Condens. Matter 2008, 20, 064238 (11pp). (17) Sorescu, D. C. Plane-Wave Density Functional Theory Investigations of the Adsorption and Activation of CO on Fe5C2 Surfaces. J. Phys. Chem. C 2009, 113, 9256−9274. (18) Petersen, M. A.; van den Berg, J. A.l; van Rensburg, W. J. Role of Step Sites and Surface Vacancies in the Adsorption and Activation of CO on χ-Fe5C2 Surfaces. J. Phys. Chem. C 2010, 114, 7863−7879. (19) Gao, R.; Cao, D. B.; Yang, Y.; Li, Y.-W.; Wang, J.; Jiao, H. Adsorption and Energetics of H2O Molecules and O atoms on the χFe5C2 (111), (−411) and (001) Surfaces from DFT. Appl. Catal., A 2014, 475, 186−194. (20) Gao, R.; Cao, D. B.; Liu, S. L.; Yang, Y.; Li, Y.-W.; Wang, J.; Jiao, H. Density Functional Theory Study into H 2O Dissociative Adsorption on the Fe5C2(010) Surface. Appl. Catal., A 2013, 468, 370−383. (21) Ozbek, M. O.; Niemantsverdriet, J. W. Elementary Reactions of CO and H2 on C-terminated χ-Fe5C2(001) Surfaces. J. Catal. 2014, 317, 158−166. (22) van Steen, E.; Claeys, M. Fischer−Tropsch Catalysts for the Biomass-to-Liquid Process. Chem. Eng. Technol. 2008, 31, 655−666. (23) Wielers, A. F. H.; Hop, C. E. C. A.; van Beijnum, J.; van der Kraan, A. M.; Geus, J. W. On the Properties of Silica-supported Bimetallic Fe/Cu Catalysts Part I. Preparation and Characterization. J. Catal. 1990, 121, 364−374. (24) O’Brien, R. J.; Xu, L.; Spicer, R. L.; Bao, S.; Milburn, D. R.; Davis, B. H. Activity and Selectivity of Precipitated Iron Fischer− Tropsch Catalysts. Catal. Today 1997, 36, 325−334. (25) de Smit, E.; Beale, A. M.; Nikitenko, S.; Weckhuysen, B. M. Local and Long Range Order in Promoted Iron-based FischerTropsch Catalysts: A Combined in Situ X-ray Absorption Spectroscopy/Wide Angle X-ray Scattering Study. J. Catal. 2009, 262, 244− 256.
exposed. On the Fe3C(001) and Fe5C2(510) surfaces with both Fe and C atoms exposed, the adsorbed Cu atoms mainly interacting with surface Fe atoms are very slightly negatively charged. On the Fe3 C(100), Fe 5C 2 (100), Fe 5 C 2 (111), Fe5C2(010), and Fe5C2(001) surfaces with both Fe and C atoms exposed, the adsorbed Cu atoms mainly interacting with both Fe and C atoms are positively charged. Such different behavior in charge transfer comes from the difference in electronegativity among Fe, C, and Cu atoms. On both iron and iron carbide surfaces, the strength of charge transfer decreases with the increase of the Cun size. It is therefore necessary to prepare all these less stable surfaces individually in exploring Cu adsorption and promotion in Fe-based FTS, which provide challenges to surface sciences and analytic techniques.
■
ASSOCIATED CONTENT
S Supporting Information *
Verification of the surface models (Table S1), various optimized Cun configurations on the Fe5C2(100) surface (Figure S1), diffusion pathways of single Cu atom on the Fe5C2(100) surface (Figure S2), various optimized Cu n configurations on the Fe5C2(111) surface (Figure S3), diffusion pathways of single Cu atom on the Fe5C2(111) surface (Figure S4), various optimized Cun configurations on the Fe5C2(510) surface (Figure S5), diffusion pathways of single Cu atom on the Fe5C2(510) surface (Figure S6), various optimized Cun configurations on the Fe5C2(010) surface (Figure S7), diffusion pathways of single Cu atom on the Fe5C2(010) surface (Figure S8), various optimized Cun configurations on the Fe5C2(001) surface (Figure S9), diffusion pathways of single Cu atom on the Fe5C2(001) surface (Figure S10), top and side views of the Cu(111) and Cun (1 ML) on the five Fe5C2 surfaces (Figure S11). This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the National Basic Research Program of China (no. 2011CB201406), the National Natural Science Foundation of China (no. 21273262& 21273266), and the Chinese Academy of Science and Synfuels CHINA Co., Ltd. We also acknowledge general financial support from the BMBF and the state of Mecklenburg-Vorpommern.
■
REFERENCES
(1) Fischer, F.; Tropsch, H. Ü ber die Herstellung Synthetischer Ö lgemische (Synthol) Durch Aufbau aus Kohlenoxid und Wasserstoff. Brennstoff Chem. 1923, 4, 276−285. (2) Fischer, F.; Tropsch, H. Die Erdölsynthese bei Gewöhnlichem Druck aus den Vergasungsprodukten der Kohlen. Brennstoff Chem. 1926, 7, 97−116. (3) Anderson, R. B. The Fischer−Tropsch Synthesis; Academic Press: Orlando, FL, 1984; p 3. (4) Govender, N. S.; de Croon, M. H. J. M.; Schouten, J. C. Reactivity of Surface Carbonaceous Intermediates on an Iron-based Fischer−Tropsch Catalyst. Appl. Catal., A 2010, 373, 81−89. 7383
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C
(47) Tian, X.-X.; Wang, T.; Yang, Y.; Li, Y.-W.; Wang, J.; Jiao, H. Structures and Energies of Cun Clusters on Fe and Fe3C Surfaces from Density Functional Theory Computation. Phys. Chem. Chem. Phys. 2014, 16, 26997−27011. (48) Zhao, S.; Liu, X. W.; Huo, C. F.; Li, Y. W.; Wang, J.; Jiao, H. Surface Morphology of Hägg Iron Carbide (χ-Fe5C2) from Ab Initio Atomistic Thermodynamics. J. Catal. 2012, 294, 47−53. (49) Zhao, S.; Liu, X.; Huo, C.; Li, Y.; Wang, J.; Jiao, H. Determining Surface Structure and Stability of ε-Fe2C, χ-Fe5C2, θ-Fe3C and Fe4C Phases under Carburization Environment from Combined DFT and Atomistic Thermodynamic Studies. Catal. Struct. React. 2015, 1, 44− 59. (50) Kresse, G.; Furthmüller, J. Efficiency of Ab-initio Total Energy Calculations for Metals and Semiconductors Using a Plane-wave Basis Set. Comput. Mater. Sci. 1996, 6, 15−50. (51) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab initio Total-energy Calculations Using a Plane-wave Basis Set. Phys. Rev. B 1996, 54, 11169−11186. (52) Blöchl, P. E. Projector Augmented-wave Method. Phys. Rev. B 1994, 50, 17953−17979. (53) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (54) Perdew, J. P.; Burke, K.; Ernzerhof, M. Errata: Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1997, 78, 1396. (55) Henkelman, G.; Arnaldsson, A.; Jónsson, H. A Fast and Robust Algorithm for Bader Decomposition of Charge Density. Comput. Mater. Sci. 2006, 36, 354−360. (56) Jónsson, H.; Mills, G.; Jacobsen, K. W. In Classical and Quantum Dynamics in Condensed Phase Simulations; Berne, B. J., Ciccotti, G., Coker, D. F., Eds.; World Scientific: Singapore, 1998, p385. (57) Henkelmann, G.; Uberuaga, B. P.; Jnsson, H. A Climbing Image Nudged Elastic Band Method for Finding Saddle Points and Minimum Energy Paths. J. Chem. Phys. 2000, 113, 9901−9904. (58) Steynberg, P. J.; van de Berg, J. A.; van Rensburg, W. J. Bulk and Surface Analysis of Hägg Fe Carbide (Fe5C2): a Density Functional Theory Study. J. Phys.: Condens. Matter 2008, 20, 064238(11pp). (59) Retief, J. J. Powder Diffraction Data and Rietveld Refinement of Hägg-Carbide, χ-Fe5C2. Powder Diffr. 1999, 14, 130−132. (60) Hofer, L. J. E.; Cohn, E. M. Saturation Magnetizations of Iron Carbides. J. Am. Chem. Soc. 1959, 81, 1576−1582. (61) Pham, T. H.; Duan, X.; Qian, G.; Zhou, X.; Chen, D. CO Activation Pathways of Fischer−Tropsch Synthesis on χ-Fe5C2(510): Direct versus Hydrogen-Assisted CO Dissociation. J. Phys. Chem. C 2014, 118, 10170−10176. (62) Huo, C. F.; Wu, B. S.; Gao, P.; Yang, Y.; Li, Y.-W.; Jiao, H. The Mechanism of Potassium Promoter: Enhancing the Stability of Active Surfaces. Angew. Chem., Int. Ed. 2011, 50, 7403−7406. (63) Chiou, W. C., Jr.; Carter, E. A. Structure and Stability of Fe3CCementite Surfaces from First Principles. Surf. Sci. 2003, 530, 87−100. (64) Wang, Z. Q.; Lu, S. H.; Li, Y. S.; Jona, F.; Marcus, P. M. Epitaxial Growth of a Metastable Modification of Copper with Bodycentered-cubic Structure. Phys. Rev. B 1987, 35, 9322−9325. (65) Heinrich, B.; Celinski, Z.; Cochran, J. F.; Muir, W. B.; Rudd, J.; Zhong, Q. M.; Arrott, A. S.; Myrtle, K.; Kirschner, J. Ferromagnetic and Antiferromagnetic Exchange Coupling in bcc Epitaxial Ultrathin Fe(001)/Cu(001)/Fe(001) Trilayers. Phys. Rev. Lett. 1990, 64, 673− 676. (66) Kamada, Y.; Matsui, M. Epitaxial Growth of Ni, Cu on bccFe(001) and Ni on fcc-Au(001). J. Phys. Soc. Jpn. 1997, 66, 658−663. (67) Shein, I. R.; Medvedeva, N. I.; Ivanovskii, A. L. Electronic Structure and Magnetic Properties of Fe3C with 3d and 4d Impurities. Phys. Status Solidi B 2007, 244, 1971−1981. (68) Ande, C. K.; Sluiter, M. H. F. First-principles Prediction of Partitioning of Alloying Elements between Cementite and Ferrite. Acta Mater. 2010, 58, 6276−6281. (69) Ande, C. K.; Sluiter, M. H. F. First-Principles Calculations on Stabilization of Iron Carbides (Fe3C, Fe5C2, and η-Fe2C) in Steels by Common Alloying Elements. Metall. Mater. Trans. A 2012, 43A, 4436−4444.
(26) Li, S.; Li, A.; Krishnamoorthy, S.; Iglesia, E. Effects of Zn, Cu, and K Promoters on the Structure and on the Reduction, Carburization, and Catalytic Behavior of Iron-based Fischer−Tropsch Synthesis Catalysts. Catal. Lett. 2001, 77, 197−205. (27) Rao, K. R. P. M.; Huggins, F. E.; Huffman, G. P.; Gormley, R. J.; O’Brien, R. J.; Davis, B. H. Mossbauer Study of Iron Fischer- Tropsch Catalysts During Activation and Synthesis. Energy Fuels 1996, 10, 546−551. (28) Zhang, C. H.; Yang, Y.; Teng, B. T.; Li, T. Z.; Zheng, H. Y.; Xiang, H. W.; Li, Y.-W. Study of an Iron-Manganese Fischer−Tropsch Synthesis Catalyst Promoted with Copper. J. Catal. 2006, 237, 405− 415. (29) O’Brien, R. J.; Davis, B. H. Impact of Copper on an Alkali Promoted Iron Fischer−Tropsch Catalyst. Catal. Lett. 2004, 94, 1−6. (30) Boellaard, E.; van der Kraan, A. M.; Sommen, A. B. P.; Hoebink, J. H. B. J.; Marin, G. B.; Geus, J. W. Behaviors of Cyanide-Derived CuxFe/Al2O3 Catalysts during Fischer−Tropsch Synthesis. Appl. Catal., A 1999, 179, 175−187. (31) Hayakawa, H.; Tanaka, H.; Fujimoto, K. Studies on Precipitated Iron Catalysts for Fischer−Tropsch Synthesis. Appl. Catal., A 2006, 310, 24−30. (32) Bukur, D. B.; Mukesh, D.; Patel, S. A. Promoter Effects on Precipitated Iron Catalysts for Fischer−Tropsch Synthesis. Ind. Eng. Chem. Res. 1990, 29, 194−204. (33) Chonco, Z. H.; Ferreira, A.; Lodya, L.; Claeys, M.; van Steen, E. Comparing Silver and Copper as Promoters in Fe-based Fischer− Tropsch Catalysts Using Delafossite as a Model Compound. J. Catal. 2013, 307, 283−294. (34) Chonco, Z. H.; Lodya, L.; Claeys, M.; van Steen, E. Copper Ferrites: A Model for Investigating the Role of Copper in the Dynamic Iron-Based Fischer−Tropsch Catalyst. J. Catal. 2013, 308, 363−373. (35) Soled, S. L.; Iglesia, E.; Miseo, S.; Derites, B. A.; Fiato, R. A. Selective Synthesis of α-Olefins on Fe-Zn Fischer-Trpsch Catalysts. Top. Catal. 1995, 2, 193−205. (36) Blanchard, J.; Abatzoglou, N. Nano-iron Carbide Synthesized by Plasma as Catalyst for Fischer−Tropsch Synthesis in Slurry Reactors: The Role of Iron Loading and K, Cu Promoters. Catal. Today 2014, 237, 150−156. (37) Kölber, H.; Ralek, M. The Fischer−Tropsch Synthesis in the Liquid Phase. Catal. Rev. 1980, 21, 225−274. (38) Bae, J. W.; Park, S. J.; Kang, S. H.; Lee, Y. J.; Jun, K. W.; Rhee, Y. W. Effect of Cu Content on the Bifunctional Fishcher-Tropsch Fe-CuK/ZSM Catalyst. J. Ind. Eng. Chem. 2009, 15, 798−802. (39) Elahifard, M.; Fazeli, E.; Joshani, A.; Gholami, M. Ab-Initio Calculations of the CO Adsorption and Dissociation on Substitutional Fe−Cu Surface Alloys Relevant to Fischer−Tropsch Synthesis: bcc(Cu)Fe(100) and fcc-(Fe)Cu(100). Surf. Interface Anal. 2013, 45, 1081−1087. (40) Zhao, Y. H.; Li, S. G.; Sun, Y. H. CO Dissociation Mechanism on Cu-Doped Fe(100) Surfaces. J. Phys. Chem. C 2013, 117, 24920− 24931. (41) Zhao, X. H.; Li, Y.-W.; Wang, J.; Huo, C. F. CO Adsorption, CO Dissociation, and C-C Coupling on Cu Monolayer-Covered Fe(100). J. Fuel Chem. Technol. 2011, 39, 956−960. (42) Tian, X.-X.; Wang, T.; Yang, Y.; Li, Y.-W.; Wang, J.; Jiao, H. Copper Promotion in CO Adsorption and Dissociation on the Fe(100) Surface. J. Phys. Chem. C 2014, 118, 20472−20480. (43) van Steen, E.; Claeys, M. Promoting χ-Fe5C2(100)0·25 with Copper − a DFT study. Catal. Struct. React. 2015, 1, 11−18. (44) Wachs, I. E.; Dwyer, D. J.; Iglesia, E. Characterization of Fe, FeCu, and Fe-Ag Fischer−Tropsch Catalysts. Appl. Catal. 1984, 12, 201−217. (45) de Smit, E.; de Groot, F. M. F.; Blume, R.; Hävecker, M.; KnopGericke, A.; Weckhuysen, B. M. The Role of Cu on the Reduction Behavior and Surface Properties of Fe-based Fischer−Tropsch Catalysts. Phys. Chem. Chem. Phys. 2010, 12, 667−680. (46) Tian, X.-X.; Wang, T.; Yang, Y.; Li, Y.-W.; Wang, J.; Jiao, H. Adsorption Structures and Energies of Cun Clusters on the Fe(110) and Fe3C(001) Surfaces. J. Phys. Chem. C 2014, 118, 21963−21974. 7384
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385
Article
The Journal of Physical Chemistry C (70) Allen, L. C. Electronegativity is the Average One-Electron Energy of the Valence-Shell Electrons in Ground-State Free Atoms. J. Am. Chem. Soc. 1989, 111, 9003−9014.
7385
DOI: 10.1021/acs.jpcc.5b01324 J. Phys. Chem. C 2015, 119, 7371−7385