Selective Oxidation of Propylene on Cu2O(111) and Cu2O(110) Surfaces: A Systematically DFT Study.

Density functional theory calculations with a Hubbard U correction were used to investigate the selective oxidation of propylene on Cu2O(111) and Cu2O(110) surfaces, and the mechanism for the selective oxidation of propylene was discussed. On both surfaces, acrolein can be generated by two H-stripping reactions in the allylic hydrogen stripping path, while propylene oxide (PO), propanal, and acetone can be created through the propylene oxametallacycle intermediates in the epoxidation path. Our calculation results indicated that Cu2O has a high crystal plane-controlled phenomenon for the selective oxidation of propylene. It was found that the formations of propanal and acetone are unfavorable kinetically and acrolein is the main product on the (111) surface. On the (110) surface, the activation barrier of acrolein formation is too high to produce and PO becomes the favored product, which is different from the case of the (111) surface. Moreover, energetic span model analysis was carried out to discuss the selective oxidation of propylene on these two surfaces and confirm the above calculations. The present study can help people to design the proper crystal plane catalyst to get the target product of PO with high selectivity and activity in the selective oxidation of propylene.

Introduction

Propylene selective oxidation is a significant module in industrial catalysis[1] and the product propylene oxide (PO) is a crucial feedstock for the manufacture of various commodity chemicals.[2] Traditional chlorohydrin and organic hydroperoxide methods can either cause serious environmental pollution or produce large amounts of coproducers.[3] A profitable and environmentally friendly route, thus propylene selective oxidation catalyzed by metals or metal oxides, has received considerable attention. For example, Zheng and co-workers[4] discussed propylene epoxidation using molecular oxygen by the unsupported Cu and Ag catalysts with different oxidation states and discovering: (i) the metallic Cu catalyst exhibits much higher PO selectivity than others, (ii) the Cu0 species is the main active site for propylene epoxidation, and (iii) cuprous oxide has better performance than copper oxide on catalytic activity.

Experimentally, some catalysts (Au,[5,6] Ag,[7,8] Cu,[9−12] Cu2O,[13,14] and CuO[4,15]) have been widely used to study propylene oxidation. Cu-based catalysts are emerging with a range of functions as the oxidative alkane dehydrogenation,[16] catalyzing cyclization reactions[17−19] and so on.[20−25] Su et al.[11] investigated propylene epoxidation catalyzed by Cu/SiO2 with various promoters using molecular oxygen, and they announced that both Cu0 and Cu+ species in the Cu/SiO2 catalyst show epoxidation activity maybe caused by the strong bonding between propylene and Cu0 or Cu+ sites. For propylene oxidation, the metallic Cu can be easily oxidized to Cu2O and Cu2O is much stable at 900 K in vacuum with highly catalytic efficiency.[15] Therefore, Cu2O was exploited in the present study for the selective oxidation of propylene. Campbell discussed ethylene selective epoxidation catalyzed by Ag(111) and Ag(110) surfaces and proved that the activity of Ag(111) in the epoxidation route is about half that of Ag(110).[26] Schulz and Cox elucidated propylene oxidation influenced by different single-crystal facets of Cu2O (Cu+-terminated (100), oxygen-terminated (100), and (111)), which found that propylene selective oxidation favors acrolein and promoted with coordinately unsaturated surface lattice oxygen.[13] Hua et al. revealed the strong crystal plane-controlled selectivity of Cu2O catalysts in propylene oxidation in which the (110) facet has the highest selectivity of PO and the (111) facet favors acrolein formation.[14]

Theoretical calculations provide an indispensable and special view to study the catalytic activity. Theoretically, Torres et al. indicated that PO and propanal can be generated by propylene selective epoxidation through a metallacycle intermediate and the lower basicity of Cu favors epoxide formation than Ag.[27] Pulido et al. discussed the aerobic epoxidation mechanism of propylene over Ag(111) and (100) by density functional theory (DFT) calculations, which confirms that propylene epoxidation can also form acetone through the propylene oxametallacycle (OMP) intermediates.[28] Besides, the allylic hydrogen stripping (AHS) route of propylene has been illustrated by Lei et al. with a subnanometer size Ag nanoparticles on alumina supports.[29] Düzenli et al. studied propylene epoxidation within the mechanism over Cu2O(001) and CuO(001) surfaces, which revealed that acrolein is more favorable thermally on both facets and Cu+ species has higher activity.[30] For the copper-based catalysts, the DFT calculations confirmed that the CuO(110) facet is more active for PO formation than the CuO(111) facet[31] and the magnetic properties can be affected by the surface of Cu2O.[32] Considering the structure sensitivity and higher selectivity of PO formation over Cu2O as compared to that of CuO,[4] hence, propylene selective oxidation on Cu2O(111) and the results would be discussed in this study compared to the Cu2O(110). The related propylene oxidation mechanism is proposed in Scheme based on theoretical works.[29−31,33−39] The role of oxygen species on the Cu2O(111) facet is a key factor in the catalysis[40,41] and different kinds of oxygen in the selective oxidation of propylene on Cu2O(111) have been studied in our previous work.[42] The aim of this work is to investigate the roles and effects of different facets of Cu2O on the selective oxidation of propylene.

Scheme 1

Reaction Scheme for the Selective Oxidation of Propylene on Cu2O(111) and Cu2O(110) Surfaces

Calculation Methods and Models

Calculation Methods

The spin-polarized version of Vienna ab-initio simulation package (VASP)[43,44] coupled with the DFT + U method[45−47] was performed. The projector-augmented wave (PAW)[48,49] was used to describe the valence electron and inner core interactions, and the generalized gradient approximation (GGA) with a Perdew–Burke–Ernzerhof (PBE) exchange–correlation functional[50] was carried out. The electronic states were expanded in a plane-wave basis with a kinetic cutoff energy being 400 eV, and the value of (U-J) being 3.6 eV was selected to evaluate the on-site Coulomb interactions in the localized d orbital.[51−53] The Brillouin zone sampling was performed using a 2 × 2 × 1 Monkhorst-Pack k-point mesh.[54] The climbing image general nudged elastic band method developed by Jónsson et al. was approached to calculate the transition state (TS),[55,56] and the TS was confirmed by vibrational frequency analysis. The van der Waals correction was implemented for the weak interaction with the catalyst using the DFT-PBE-D3 method.[57,58] Moreover, we were dealing with the surface with a large dipole under periodic boundary conditions, although polarization corrections have been taken into account. The adsorption energy (Eads), activation energy (Ea), and reaction energy (ΔE) are calculated by the following formulas: Eads = Eadsorbate/substrate – Eadsorbate – Esubstrate, Ea = ETS – EIS, and ΔE = EFS – EIS, where Eadsorbate/substrate, Eadsorbate, Esubstrate, ETS, EIS, and EFS represent the energies of the adsorption system, adsorbate, substrate, TS, initial state (IS), and final state (FS), respectively.

Model Selection

Aiming at clarifying the crystal plane-controlled effect of cuprous oxide, a twelve-layer and a six-layer symmetrically periodic substrates by selecting a p (2 × 2) unit cell were used to model Cu2O(111) and Cu2O(110) surfaces, respectively (Figure ). An optimized lattice constant of 4.27 Å was applied in unit cell volume.[59] The types were obtained by cleaving the bulk in the 111 and 110 directions with a vacuum region of 18 Å. For Cu2O(111), the uppermost six layers were relaxed and four different chemical atoms existed on the surface, which were denoted as Cucus, Cucsa, Osuf, and Osub (Figure a). The Cucus is the single-coordinated unsaturated Cu atom, the Cucsa is the two-coordinated saturated Cu atom, the Osuf is the three-coordinated oxygen atom, and the Osub is the four-coordinated oxygen atom. For Cu2O(110) bulk, the uppermost three layers were relaxed and two different chemical atoms existed on the surface, which were denoted as Cu2f and O3f (Figure b). The Cu2f is a surface copper atom, which binds to two neighboring oxygen atoms, and the O3f is a surface oxygen atom, which binds to three neighboring Cu atoms.

Figure 1

Side view of the (a) Cu2O(111), (b) Cu2O(110), and (c) propylene molecule. Note: Cucus is the single-coordinated copper atom, Cucsa is the two-coordinated copper atom, Osuf is the three-coordinated oxygen atom, Osub is the four-coordinated oxygen atom, Cu2f is the two-coordinated copper atom, and O3f is the three-coordinated oxygen atom.

Results

Reaction Mechanism of Propylene Selective Oxidation on Cu2O(111) Surface

The adsorption properties of IS, intermediates, and FSs for propylene selective oxidation on Cu2O(111) were investigated first. The optimized adsorption configurations are exhibited in Figure S1a, and the adsorption energies are listed in Table S1 in the Supporting Information. The role of different oxygen species in the selective oxidation of propylene on Cu2O(111) was investigated in our previous study,[42] and the results would be discussed in this paper compared to the (110) surface. On the (111) surface, the structures of the TSs are summarized in Figure a, and the free-energy profiles are displayed in Figure a.

Figure 2

Calculated TS structures for propylene selective oxidation on (a) Cu2O(111) and (b) Cu2O(110) surfaces. Bond lengths are in pm.

Figure 3

Free-energy profiles of propylene selective oxidation on (a) Cu2O(111) and (b) Cu2O(110) at 433 K and 100 kPa. The black line shows the formation path of acrolein, the olive line and green line show the formation paths of PO and propanal via OMP1, and the red line and blue line show the formation paths of PO and acetone via OMP2, respectively. Note: The data in parentheses mean the energy barrier of related elemental step.

Reaction Mechanism of Propylene Selective Oxidation on Cu2O(110) Surface

The mechanism of propylene selective oxidation on Cu2O(110) was studied through two parallel routes: the AHS route and the epoxidation route. The first step between absorbed propylene and surface O3f can either create hydroxyl and allyl species or form OMP intermediates. The optimized adsorption configurations of reactant, intermediates, and products for propylene selective oxidation on (110) are summarized in Figure S1b, and the corresponding adsorption energies are listed in Table S1 in the Supporting Information. The structures of the TSs are displayed in Figure b, and the free-energy profiles are shown in Figure b, separately.

AHS Mechanism

The adsorption of propylene, allyl, C3H5O, and acrolein in the AHS route prefers the top site, 1,3-di-σ mode, Cu-three-fold type, and top site, respectively. The AHS mechanism included two H-stripping reactions and acrolein can be generated finally. The first step involves hydrogen abstraction of the methyl group by neighboring O3f forming allyl and hydroxyl groups. This hydrogen abstraction process is endothermic by 0.11 eV with an activation barrier of 0.65 eV, and at TS10, the distance of C3–H is 134 pm and the length of O–H is 128 pm, separately. The second step involves C3 of methylene addition to another neighboring O3f created C3H5O species through TS11, where the distance of C–O is 212 pm. Besides, this process releases 0.34 eV and has a barrier of 0.52 eV. DFT calculations indicate that the formation of acrolein prefers the second hydrogen abstraction from C3H5O species to surface O3f. The process of acrolein formation requires 0.43 eV and the calculated activation barrier at the PBE level is 1.89 eV via TS12, where the distance of breaking C–H is 141 pm and the length of forming O3f–H is 132 pm. Then, desorption of acrolein occurs with a barrier of 0.46 eV, and hydroxyl groups existing on the surface can form water via the diffusion of hydrogen, ending the AHS reaction. The rate-controlling step for AHS is the second hydrogen abstraction with a barrier of 1.89 eV.

Epoxidation Mechanism

Propylene, PO, propanal, and acetone are favored at the top adsorption, and OMP is absorbed in the 1,2-di-σ mode in the epoxidation path. The propylene epoxidation reaction can lead to the formation of PO, propanal, and acetone through two different intermediates. Adsorbed propylene acted on nearby O3f can create different OMP intermediates. First, the C1 atom binds to the O3f atom and the C2 atom connects to the copper atom forming OMP1 geometry, where the length of C1–O3f is 150 pm and the bond of C2–Cu is 203 pm. Second, the C1 atom binds to the copper atom and the C2 atom connects to the O3f atom, which can produce OMP2 with a C1–Cu distance of 199 pm and a C2–O3f length of 153 pm. The generation of OMP1 and OMP2 intermediates is endothermic by 0.22 and 0.18 eV with the corresponding barriers of 0.76 and 0.53 eV, respectively. At TS, the distance of C1–O is 193 pm in OMP1 and the length of C2–O is 191 pm in OMP2, separately.

The OMP1 intermediate can generate the final product PO in the epoxidation route, which involves the breaking of the C2–Cu bond and the formation of a (−C1OC2−) cycle. The reaction energy is 0.91 eV with an activation barrier of 1.57 eV, and the length of C2–O at TS15 is 191 pm. Meanwhile, OMP1 can also be oxidized to propanal by hydrogen shift from C1 to C2 via TS16, where the distance of C1–H is 133 pm and the length of C2–H is 158 pm. The shift process of hydrogen is endothermic by 0.11 eV and the calculated barrier is 1.39 eV. Clearly, the rate-controlling step for PO formation is the step from OMP1 to PO with a barrier of 1.57 eV and it is 1.39 eV for propanal formation.

The other intermediate OMP2 can form epoxide PO as well, where the bond of C1–Cu is broken and the O3f atom connected to the C2 atom is directly binded to the C1 atom. In this cyclizing process, the formation of PO requires 0.94 eV with an activation barrier of 1.41 eV and the length of the forming C1–O3f bond decreases to 175 pm at the TS17. In this process, the OMP2 → PO step with a barrier of 1.41 eV can be regarded as the rate-controlling step. Moreover, the OMP2 intermediate can also produce acetone by hydrogen shift from C2 to C1 and the shift process is exothermic by 0.45 eV. The calculated barrier for the formation of acetone is 1.62 eV via TS18, where the distance of C2–H is 127 pm and the length of C1–H is 152 pm. Obviously, the rate-controlling step is 1.62 eV for acetone formation.

From the above calculation results, we can find that the main product of propylene oxidation on the surface of Cu2O(110) might be propanal and PO, followed by acetone and acrolein if we considered the rate-controlling step only, and more discussion will be given in the following section.

Catalyst Recovery

Oxygen deficient can also be formed in propylene selective oxidation on these two surfaces, and molecular O2 is requisite at experimental conditions for the catalyst recovery. The lattice oxygen and the absorbed oxygen can be produced by a molecular O2 moiety bound in the hole in which the restoration process occurs easily. Then, the absorbed oxygen with higher activity becomes the oxidant participating in propylene selective oxidation, which has been discussed partially in our another paper.[42]

Discussion

Electronic Analysis for Adsorption

For deeper understanding the physical origin of why the adsorption of pertinent species (except C3H5O species) in propylene selective oxidation on the (111) surface is more favorable in comparison with the (110) surface, the electronic analysis of these pertinent species was performed based on the projected crystal orbital Hamilton population (COHP) method developed by Dronskowski et al.[60,61] The positive and negative values correspond to the bonding and antibonding states in the COHP diagram. As can be seen from Figure S3, the Fermi level lines appear between the bonding and the antibonding region. It was found that the population of the bonding region for these pertinent species (except C3H5O species) on the (111) surface is much larger than the (110) surface. The integrated COHP (ICOHP) relative to the Fermi level could reflect the orbital interaction between the adsorbate and substrate, and the more negative value corresponds to the stronger interaction. The ICOHP values for adsorbate–substrate interaction are exhibited in Table . For C3H5O species, however, the adsorption on the (110) surface is stronger than that on the (111) surface and can be also confirmed by COHP analysis. This may be caused by the reason that the oxygen atom in C3H5O species binds with the subsurface Cu atom on (110) and the interaction between them is much stronger (Figure S1b).

Table 1

ICOHP Value for the Interaction between Adsorbate and Substrate for Cu2O(111) and Cu2O(110) (eV)

adsorbate	ICOHP/Cu2O(111) substrate	ICOHP/Cu2O(110) substrate
C3H6	–0.65	–0.26
allyl	–0.88	–0.48
C3H5O	–0.59	–1.00
OMP1	–1.01	–0.66
OMP2	–1.03	–0.71
acrolein	–0.71	–0.24
PO	–0.86	–0.19
propanal	–0.94	–0.27
acetone	–0.93	–0.21

Free-Energy Diagrams

Free-energy diagrams for propylene selective oxidation on Cu2O(111) and Cu2O(110) surfaces are plotted in Figure . The free energies are determined by considering the entropy contributions of adsorption and desorption. We assume such contributions from the translational entropy, which is calculated as the formula:[62,63]S = 1.5R ln (2πMkBT) – 3R ln h + R ln (kBT/P) + 2.5R, where R, M, kB, T, h, and P are the ideal gas constant, molecular weight, Boltzmann constant, temperature, Planck constant, and pressure, respectively. The free-energy values in this method are approximate to the entropic energy calculated by Campbell and Sellers.[64] In the free-energy diagrams, the free energies are reported at 433 K and 100 kPa to keep in accordance with the experimental condition.[4]

For propylene selective oxidation on (111) depicted in Figure a, because of rather high barriers for propanal (2.21 eV) and acetone (2.02 eV) formation, free-energy diagrams of them are neglected. In the epoxidation path, the barriers of OMP1 and OMP2 formation from absorbed propylene are 1.22 and 0.90 eV and the barrier from OMP1 to PO being 1.59 eV is much high, which means that the formation route of PO from OMP1 is disadvantageous. The barrier of OMP2 formation is 0.30 eV lower than that of allyl (0.90 vs 1.20 eV), but the barrier of PO formation (1.92 eV) from OMP2 is so much high that the formation route of PO from OMP2 is hampered. All steps in the AHS path have relatively lower barrier than the epoxidation path, and the second H-stripping has a smaller barrier maybe caused by releasing much energy. Hence, propylene oxidation on the (111) surface prefers the AHS route-forming acrolein, which agrees well with the results discussed by Hua et al.[14]

For propylene selective oxidation on (110) depicted in Figure b, the generation of acrolein from C3H5O species in the AHS path has a barrier of 1.89 eV, which is unfavorable kinetically in the experiment. In the epoxidation path, the barriers of OMP1 and OMP2 formation from absorbed C3H6 on the (110) surface are 0.76 and 0.53 eV. To further understand the physical origin of why OMP1 is less favorable than OMP2 on (110), electronic analysis of TSs for C3H6 + O → OMP1 and C3H6 + O → OMP2 has been performed based on the projected COHP method developed by Dronskowski’s team.[60,61] In the COHP diagram (Figure ), the positive and negative values reflect the bonding and antibonding states. It was found that the population of the bonding region for the OMP2 system is much larger than the OMP1 system (the integrated COHP up to the Fermi level is −1.19 and −1.26 eV for OMP1 and OMP2, respectively, reading from the intersection of the Fermi level and the integral line), which means a strong interaction between the OMP2 and the (110) facet at the TS. Stronger interaction means that OMP2 is more stable on (110) at TS, which leads to a lower barrier. Thus, the generation of OMP2 is more preferred in comparison with OMP1 on (110) from geometrical and electronic structure points. The PO formation from OMP1 requires 0.91 eV with a barrier of 1.57 eV, and the generation of PO from OMP2 requires 0.94 eV with a barrier of 1.41 eV. The calculated results show that the process of PO formation can be influenced thermodynamically and the formation of PO from OMP2 is preferred. Moreover, the generation of acetone from OMP2 has a barrier of 1.62 eV, which means that acetone formation is relatively disadvantageous in comparison with PO formation from OMP2. The formation of propanal has a barrier of 1.39 eV, moderately, competing with PO formation from OMP2.

Figure 4

Projected crystal orbital Hamilton population (COHP) between C and surface Cu of the Cu2O(110) system at TSs: (a) C3H6 + O → OMP1 and (b) C3H6 + O → OMP2.

Energetic Span Model Analysis

The turnover frequency (TOF) was calculated with the energetic span model developed by the groups of Kozuch and Shaik.[65,66] This model was used to understand the catalytic performance of Cu2O for propylene selective oxidation at a temperature of 523 K14 and included three assumptions: (i) the TS theory is valid, (ii) a steady-state regime subsists, and (iii) the relaxation of the intermediates is fast. In the catalytic cycle of propylene oxidation, only one transition state and one intermediate determine the TOF, which called the TOF-determining transition state (TDTS) and the TOF-determining intermediate (TDI). For identifying the TDI and TDTS, the energetic span model offers a quantification of the influence of each intermediate and transition state on the TOF. Once the TDI and TDTS have been identified, the energetic span (δE), corresponding to the apparent activation energy of the full cycle, was defined approximately as:withThe apparent activation energies of acrolein, PO, propanal, and acetone formation are calculated to 2.21, 2.53, 3.17, and 2.63 eV on the (111) surface and the corresponding TOFs of them are 5.48 × 10–9, 4.52 × 10–12, 3.07 × 10–18, and 4.91 × 10–13 s–1, respectively. The lowest apparent activation energy and the highest TOF of acrolein mean that acrolein is the preferred product than others on the (111) surface. While on the (110) surface, 2.22, 1.59, 1.60, and 1.80 eV are contributed to the apparent activation energies for acrolein, PO, propanal, and acetone formation and the corresponding TOFs of acrolein, PO, propanal, and acetone are 4.39 × 10–9, 5.17 × 10–3, 4.14 × 10–3, and 4.90 × 10–5 s–1, respectively. The calculated results indicate that propylene oxidation on the (110) surface favors the formation of PO. In general, the energetic span model analysis demonstrated that acrolein is more advantageous on the (111) surface and PO is more favorable on the (110) surface qualitatively.

Analysis of the Difference between Cu2O(111) and Cu2O(110) Surfaces

Based on above analysis, we can see that a crystal plane-controlled phenomenon existed in propylene selective oxidation catalyzed by cuprous oxide and the main products in (111) and (110) surfaces have an essential distinction; thus, it is very meaningful to analyze the physical reasons causing such a difference.

In the AHS route, it is obvious that the H-stripping ability is related to the basicity of oxygen. In Figure , the basicity of the lattice oxygen on (111) and (110) was studied by projected density of states (PDOS) analysis.[67−69]Figure indicates that the states of the p orbitals of the lattice oxygen on (110) (−2.34 eV) are closer to the Fermi level than that of on the (111) surface (−3.76 eV), meaning that the basicity of the lattice oxygen on the (110) surface is higher, which confirms that the first H-stripping step on (110) (0.65 eV) has a lower barrier than that on the (111) surface (1.20 eV).

Figure 5

PDOS of the O on (a) Cu2O(111) and (b) Cu2O(110) surface.

In the epoxidation route, for further understanding the origin of the barrier variation for PO formation through OMP intermediates on (111) and (110) surfaces, the calculated geometries of the TSs were analyzed. The adsorption strength of OMP fragments on the (111) surface (−0.72 eV for OMP1 and – 0.98 eV for OMP2) is stronger than that on the (110) surface (−0.67 eV for OMP1 and −0.76 eV for OMP2). The geometry of TSs leading PO on (111) requires substantial elongation of Cu–C (122 or 77 pm) bonds in OMP1 or OMP2 (Figure a,b), and the bond lengths of Cu–C (59 or 43 pm) in the OMP1 or OMP2 (Figure c,d) are slight elongation for the TS configuration on (110). Substantial elongation of the configuration between OMPs and TSs requires larger barrier than slight elongation. As a result, the (111) surface demands more energy (1.59 and 1.92 eV) to overcome the barrier for these substantial elongation than the (110) surface (1.57 and 1.41 eV).

Figure 6

Schematic presentations of OMP1 and OMP2 intermediates with TSs leading to PO on (a, b) Cu2O(111) and (c, d) Cu2O(110) surfaces. Bond lengths are in pm.

Physical Origin of Energy Barrier

The PO-forming process can be divided into two key steps, which are the first and second C–O bond formation. The above results show a higher selectivity for the second C–O bond forming in the favored OMP1 path on (111). The formation of PO is largely determined by the first C–O bond formation, which is adversely competed against by the AHS reaction. To gain insight into the main factors affecting the barriers for PO formation on (111), the energy barrier decomposition scheme developed by Hu and Liu[70] was performed for the OMP1 path and the AHS path (see Figure a). The barrier can be decomposed into three terms, E = E + E + Eint, where E is the activation energy barrier, E(E) is the energy cost for the activation of reactant A(B) from the IS to the TS without reactant B(A), and E is the interaction energy between A and B at the TS. Here, A is the O and B is the C3H6. It can be found that the activation of the O is small and almost the same for both routes and the major part of the barriers was the C3H6 activation. The interactions between C3H6 and O were different in both paths. At the TSs, the interaction was attraction in the AHS path; inversely, the interaction was repulsion in the OMP1 epoxidation path. Attractive interaction may be caused by the reason that the H atom existed between C3H6 and O at the TS in the AHS path.

Figure 7

(a) Barrier decomposition analysis of the AHS path and the OMP1 epoxidation path on Cu2O(111), (b) the epoxidation path on Cu2O(111) and Cu2O(110). The black box represents E, the red box represents E, the blue box represents E, and the green box represents Eint. Negative (positive) values of Eint indicate attraction (repulsion) between A and B at the TS.

One key factor determining the epoxidation process was also analyzed by the above mentioned energy decomposition scheme. Here, the preferred second C–O bond formation step on (111) (i.e., OMP1 → PO) and (110) (i.e., OMP2 → PO) was chosen to analyze the reason why the epoxidation path is more favorable on (110) instead of the (111) surface. As seen in Figure b, we can know that the major portion of the activation energy barrier was the O activation on both surfaces in which the O moves beneath the C3H6 fragment. However, the activation of the C3H6 can be regarded as the main reason influenced the barriers because it can account for the large shift downward of C3H6 on the (111) surface.

Conclusions

The selective oxidation of propylene on Cu2O(111) and Cu2O(110) was explored in the present study by DFT calculations with a Hubbard U correction. The conclusions are summarized as follows:

(i) Propylene oxidation catalyzed by Cu2O had a strong crystal plane-controlled phenomenon and the main products were different for (111) and (110) surfaces. On the basis of our calculations, it was found that the adsorption strength of most reaction species on the (111) facet is stronger than that on the (110) facet. The projected crystal orbital Hamilton population (COHP) method analysis demonstrated that the orbital interaction between the adsorbate and substrate on the (111) surface has a more negative value (i.e., stronger interaction).

(ii) The energetic span model analysis results indicated that the apparent activation energy of the selective oxidation of propylene to acrolein is lower than that of PO, propanal, and acetone on the (111) surface. Thus, the selective oxidation of propylene on the (111) surface favors at the AHS route and acrolein is the main product.

(iii) On the (110) surface, the generation of acrolein was hampered by the high barrier of the second H-stripping step (1.89 eV). In the epoxidation route, crystal orbital Hamilton population method analysis indicated that the formation of OMP2 from propylene is more favorable than that of OMP1. Moreover, the activation energy barrier of OMP2 → PO was lower than OMP2 → acetone, which demonstrated that the OMP2 epoxidation route for the formation of PO was the most favorable route in propylene selective oxidation.