Effect of Single-Layer MoS2 on the Geometry, Electronic Structure

Subscriber access provided by Iowa State University | Library

Article 2

Effect of Single-Layer MoS on the Geometry, Electronic Structure, and Reactivity of Transition Metal Nanoparticles Takat B Rawal, Duy Le, and Talat S Rahman J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.7b00036 • Publication Date (Web): 20 Feb 2017 Downloaded from http://pubs.acs.org on February 22, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Effect of Single-Layer MoS2 on the Geometry, Electronic Structure, and Reactivity of Transition Metal Nanoparticles Takat B. Rawal1, Duy Le1, and Talat S. Rahman1,2,* 1

Department of Physics, University of Central Florida, Orlando, FL 32816, USA

2

Max-Planck Institut für Festkörperforschung, Heisenbergstr. 1, D-70569 Stuttgart, Germany

ABSTRACT We present results of ab initio density functional theory (DFT) based calculations of the geometry, electronic structure, and reactivity of sub-nanometer-sized (29-atom) transition metal nanoparticles (NPs) (Cu29, Ag29, and Au29) supported on single-layer MoS2. As compared to its pristine form, defect-laden MoS2 (with a S vacancy row) has relatively larger effect on the above properties of the NPs. The NPs bind more strongly on defect-laden than on pristine MoS2 (in the order Cu29>Ag29>Au29), confirming the important role of vacancies in stabilizing the NPs on the support. The presence of vacancies also leads to an increase in charge transfer from the NPs to MoS2 (with the same elemental trend as for their binding energy), and to a shift of the d-band center of the NPs further towards the Fermi level, in turn influencing their propensity towards chemical activity. We examine the adsorption and dissociation of O2 as the prototype reactions, and find that there is no barrier for O2 to adsorb on top of an atom at the NP apex, where the frontier orbitals are localized, and that the dissociation channel proceeds through a chemisorbed state. The presence of the support leads to increase in the number of sites at which O2 can adsorb with similar binding energy (Ag29>Au29) as for pristine MoS2. Since in the case of defect-ladensupported NPs, 16 atoms, as opposed to 10 atoms for the pristine-supported NPs, participate in forming the interface, there are more atoms in direct contact with the support. In addition, the electron affinity of the defect-laden is higher than that of pristine because of the presence of unoccupied Mo d states, near the Fermi level, inside the band gap. Therefore, the larger charge transfer from the defect-laden support can be traced to the presence of mid-gap states (Figure S6b-c), which are not present in the band structure of pristine MoS2. The mid-gap states, dominantly exposed Mo d orbitals, are introduced by S vacancies on MoS2 (Figure S6b-c), and participate in strong electronic interaction between the s-d states of NPs (e.g. see Figure 4). Such electronic interaction is a critical factor that affects the catalytic activity of the NPs.28, 61 The Bader charge distributions for both unsupported and supported NPs, summarized in Figure S7, also show interesting variation in the charge on atoms at different parts of the NPs. The two types of support considered here bring these changes. Upon the formation of the interface, the ionization energy of the NPs and the electron affinity of the MoS2 supports align such that the charge transfer takes place from the NPs to the supports, as described earlier. Since defect-laden steals more charge from the NPs than the pristine MoS2, the former has larger effect the latter on

11

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 35

the charge redistribution in the NPs that can be understood by comparing the charges redistributed on atoms at the interface. More details can be found from Figure S8, and description in SI Sec. 6. While the presence of the support has only a minor effect on the charge distribution of atoms at the apex of the Cu29 and Ag29, the effect of the support is more pronounced for Au29 (for the charge gain by atoms at the apex; see Figure S7). This is because the Au atoms at the interface lose relatively smaller charge than Cu or Ag atoms at the interface. From Figure 3b it is apparent that the charge transfer from NPs to the support follows the elemental trend, Cu29>Ag29>Au29, which correlates with that of the BE of NPs to the support, when defect-laden, but not when pristine. This inconsistency may be understood as follows: first, the charge transfer between NPs and the support is mainly governed by the difference between the ionization energy of the NP and the electron affinity of the support. Replacing the pristine support by a defect-laden one leads to increase in the electron affinity of the support, as defect states appear below the conduction band of MoS2 (Figure S6b), resulting in the larger charge transfer but no significant effect on the trend. Second, for pristine MoS2 the vdW contribution to the total energy is larger for Au29 than for Ag29 (Figure S1b), resulting in the smaller BE of Ag29 that of Au29. The characteristics of the electron withdrawal by the MoS2 substrates are expected to affect the electronic structure of the NPs, and hence their reactivity. For demonstrating the effect of the support-induced electronic structural changes on the reactivity of the NPs we present in section 3.2 results of our examination of adsorption and dissociation of O2 on these model systems.

12


Page 13 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3.2. Adsorption and dissociation of O2 on MoS2-supported NPs 3.2.1. O2 adsorption energetics

Figure 5. Schematic representation of O2 adsorption at apex of Cu29 supported on: (a) pristine MoS2 and (b) defect-laden MoS2. The upper and lower panels indicate the top and side views, respectively.

We find that on unsupported Cu29, O2 prefers to bind at the NP apex with BE of 2.4 eV, which is about 0.4 eV more than that for the next preferred site. On supported Cu29, the apex atoms are still the preferred adsorption site with BE of 2.63 eV for pristine and of 2.34 eV for defect-laden MoS2. However, several other sites become favorable with BE within a few meV of that on the preferred site, particularly for defect-laden MoS2. Thus one effect of the support is to increase the number of sites at which O2 prefers to adsorb on the NP, thereby enhancing its potential to dissociate and react further. Details of O2 adsorption geometry, preferred site hierarchy, and the bond lengths of interest when O2 adsorbs on the NPs, unsupported and supported, are summarized in SI (Figures S9-S11 and Table S2). Since we find the adsorption sites hierarchy to be more or less similar on Au29 and Ag29, to keep matters simple, we present here detailed analysis of adsorption of O2 at only the apex atoms, as shown in Figure 5. 13



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 35

A quick glance at the BE of O2 on unsupported and supported Cu29, Au29 and Ag29, summarized in Figure 6a, shows Cu29 to be the least affected by the support. The BE of O2 on Cu29, Ag29 and Au29, at the NP apex are 2.40, 1.13, and 0.64 eV, respectively. These values change a bit when these NPs are supported on MoS2 because of the modified electronic structure of the NPs. On NPs supported on pristine MoS2, they increase to 2.63, 1.38, and 0.79 eV, respectively (i.e. by 9, 22, and 23%, respectively). On NPs supported on defect-laden MoS2, they slightly decrease to 2.34, 0.96, and 0.56 eV, respectively (i.e. by 2, 15, and 12%, respectively). This analysis indicates that effect of the MoS2 supports on the BE of O2 is noticeable for Ag29 and Au29. In addition, we find the following trend for the BE of O2 for all NPs: Cu29>Ag29>Au29, regardless the type of the MoS2 support, indicating that among the three considered NPs, O2 binds most strongly on Cu29.

14


Page 15 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. Comparison of (a) Binding energy (BE), (b) O-O bond-length, (c) charge transfer (Δ Q), and (d) the dissociation barrier of O2 adsorbed on M29.

It is interesting that when these NPs are supported on MoS2, O2 does not adsorb at the interfacial sites, but at the NP apex – a characteristic different from that on metal-oxides supports,29-35 which provide active interfacial sites for molecular adsorptions and reactions. These results point to the striking difference between MoS2 and metal-oxide as support for the nanoparticles considered here. We had started this work with the hypothesis that the 15



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 35

nanoparticles will enhance the reactivity of MoS2, particularly when defect-laden. However, after the detailed analysis presented here, we are led to the conclusion that the reactivity has to be associated to the composite system, i.e. the NP-MoS2 hybrid and not the individual constituents. It is should also be noted that compared to the low-index surface counterparts, NPs offer enhanced O2 binding energy. On Cu(111), Cu(110), and Cu(100) surfaces, for example, the BEs of chemisorbed O2 were found to be 0.50 eV,64 1.64 eV (for O2[001]) and 1.52 eV (for O2[11ത0]),65 and 0.83 eV,66 respectively. They are much smaller than our calculated values and also another reported value (1.85 eV)67 for O2 adsorption on Cu38. On Ag(111), Ag(110), and Ag(100) surfaces, the BEs of chemisorbed O2 were reported to be 0.17 eV,68 0.41 eV,69 and 0.4 eV,70 respectively, while that on Ag38 was also found to be larger (0.89 eV).67 Previous studies7172

have shown that O2 does not bind on the Au low index surfaces due to the fact that the HOMO

of Au does not overlap with the π* orbital of O2.72 In contrast, O2 chemisorbs on Au NPs,73-74 thereby promoting oxidation reactions.73, 75 Note that our BE of O2 on Au29 (0.64 eV) is larger than that previously reported (0.48 eV)76. This discrepancy can be traced to the usage of different DFT functionals. It is worth mentioning a well-known problem of DFT in predicting the BE of an isolated O2 molecule,77-79 and subsequently that of O2 on a solid surface.80 This is because common functionals of DFT fail to describe the ground state of an isolated O2 which is a triplet (most stable), with its 2π∗ molecular orbital half filled with two electrons.79 Although the BE of O2 is overestimated by DFT, our interest here is in the relative trend of the BE of O2 on the NPs at different sites which is not affected by this issue. Similarly, the activation energy barriers are not affected because of the error cancelation. Thus, major conclusions drawn from this work are not impacted by the failure of DFT in reproducing the experimental value of the BE of O2. 16


Page 17 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3.2.2. Analysis of electronic density of states Figure 7 compares the projected density of states (PDOS) onto the d orbitals of the four TM atoms at the NP apex of unsupported and MoS2-supported Cu29, Ag29, and Au29. In all three cases, interaction of the NPs with the MoS2 support and the charge transfer between them lead to shift of the d-states of one of the active sites towards the Fermi level. Note that the shift is more noticeable for defect-laden than pristine MoS2 and correlates with the charge transfer from the NP to MoS2, as discussed earlier. Also, the shift of the d-band center is largest, i.e. 0.62 eV, for Au29 when supported on defect-laden MoS2, while it is 0.30 eV and 0.37 eV, respectively, for supported Cu29 and Ag29. Such energy shift towards the Fermi-level has been predicted to lead to increase in the reactivity of metal-oxide-supported NPs.81-83 As we shall see, the above changes in the electronic structure do tune the activity of NPs, particularly for Au29, for O2 dissociation.

Figure 7. Electronic density of states projected onto d orbitals of the four TM atoms at the NP apex of (a) Cu29 (b) Ag29 and (c) Au29: unsupported NP (upper panel); supported by pristine MoS2 (middle panel) and by defect-laden MoS2 (lower panel). Calculated d-band center of the occupied d-states is depicted in each case.

17



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 35

More pronounced effect of the support can be gleaned from the hybridization of O p and NP d orbitals plotted in Figure 8(a) for the case of Cu29 (we find similar trend for Ag29 and Au29). Note the shift of the hybridized orbitals towards the Fermi level when Cu29 is supported by pristine MoS2 (middle panel). These states shift further when Cu29 is supported on defect-laden MoS2 (bottom panel). In each case, these Cu states hybridize with the O p states, as depicted in Figure 8(a). The atom-resolved density of states presented in Figure 8(b) further attest to the higher concentration of charge at the apex atoms, particularly when the NP is supported by MoS2 (compare Figure 8b i, ii, and iii). Here, the density of states has been integrated over the energy range mentioned in the figure caption.

Figure 8. (a) Density of states projected on the d orbitals of the four Cu atoms at the apex (as shown, in inset) of Cu29: unsupported (upper panel), pristine-MoS2-supported (middle panel), and defect-laden-MoS2-supported (bottom panel), together with the p-orbitals of O2 adsorbed on the respective NPs. Zero is set at the Fermi energy (EF). The p-d hybridized states formed by overlapping O p and Cu d orbitals are indicated by arrow. (b) Atom- resolved density of states (fraction of the total density of states): (i) unsupported Cu29 (integrated from -1.40 eV to -0.85 eV with reference to EF), (ii) Cu29/pristine-MoS2 (-0.55 eV to EF); (iii) Cu29/defect-laden-MoS2 (18


Page 19 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


0.22 eV to EF); (iv) O2 on Cu29 (-1.20 eV to -0.85 eV); (v) O2 on Cu29/pristine-MoS2 (-0.43 eV to EF); and (vi) O2 on Cu29/defect-laden-MoS2 (EF to 0.70 eV).

The support-induced changes in the electronic structure of the adsorbed O2 are accompanied by changes in their O-O bond. From the summary in Figure 6b of the O-O bond length of adsorbed O2 on unsupported and MoS2 supported NPs, a common trend emerges: it is significantly elongated from that (1.232 Å) in the gas-phase. The O-O bond length on unsupported Cu29, Ag29, and Au29 are 1.601, 1.516, and 1.528 Å, respectively. When these NPs are supported on pristine and defect-laden MoS2, the bond length decreases slightly (1.577, 1.498, and 1.500 Å, respectively, compared to be 1.569, 1.464, and 1.480 Å). In addition, our results indicate that the O-O bond is more elongated for O2 on Cu29 as compared to that on Ag29 or Au29, regardless whether these NPs are supported on MoS2 or are unsupported. The bond lengths (>1.464 Å), summarized above resemble those of oxygen with peroxo-like character,69,

84

which are very

active intermediates in several oxidation reactions.85-86

3.2.3. Charge transfer to adsorbed O2 The charge transfer from Cu29, Ag29, and Au29, both unsupported and supported, to the adsorbed O2, summarized in Figure 6c, shows that O2 gains relatively more charge from Cu29 than from Ag29 or Au29, holding the trend Cu29>Ag29>Au29. It decreases somewhat when the NPs are supported on MoS2: about 2% reduction on pristine MoS2 and between 2.5 and 8.6% on defectladen MoS2. Such reduction of the charge transfer to O2 is related to the fact that the NPs also donate charge to MoS2, as already discussed in Section 3.1.2. Note that our calculated charge transfer (1.19e) for O2/Cu29 is larger than that (0.67e) reported earlier for O2/Cu20.87 This

19



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 35

discrepancy arises not from the computational approach (Bader analysis), rather from the bonding characteristic (overlap of the orbitals) of adsorbed O2 in two cases: Cu29 and Cu20. On Cu29, O2 is bonded to four Cu atoms, whereas on Cu20 it is bonded to two Cu atoms. On the other hand, the differences in our calculated charge transfer (1.05e) for O2/Ag29 obtained here using Bader analysis with that obtained for O2/Ag13 (0.47e)88 and O2/Au29 (0.3e)76 from Mulliken Population analysis is related to the usage of two different approaches. It has been already reported that the Mulliken population analysis yields less partial charge on each atom than Bader analysis does.89

3.2.4. Dissociation of O2 The minimum energy paths (MEPs) for O2 dissociation on the NPs supported on defect-laden MoS2, presented in Figure 9, shows some interesting features. The activation energy barrier for O2 dissociation on Cu29, Ag29, and Au29 are, respectively, 0.11, 1.16, and 0.50 eV. Interestingly, the barrier turns out to be higher for Ag29 which is puzzling. On the other hand, the enigmatic nature of the interaction of oxygen with Ag surfaces continues to be the subject of investigation.69 Note that for O2 dissociation on unsupported Cu29, Ag29, and Au29, the energy barriers are 0.02, 0.58, and 0.25 eV, respectively, as compared to 0.09, 0.97, and 0.46 eV on pristine-MoS2-supported ones. The MEPs for O2 dissociation on these systems can be found in Figure S12. The results presented here indicate that MoS2 supports may have a negative impact on the dissociation of O2, which may be counter-intuitive if our intuition is based on experiments on nanoparticles supported on well-known metal-oxide substrates,91-93 in which the interface is expected to play a major role. As discussed in Section 3.2.1 and elaborated in SI (Figures S9-

20


Page 21 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


S11 and Table S2), the positive effect of the MoS2 support lies in increasing the number of adsorption sites for O2 on the NPS. Since molecules can adsorb with similar binding energies at a number of sites, reactions can occur at multiples sites and blocking of some would still leave others available. The dramatic increase in activation energy barriers for O2 dissociation on the MoS2-supported NPs can be traced to the shrinking of O-O bond and to the electronic effects. The charge transfer from the NPs to the support makes the particles positively charged, thereby reducing the charge donation to the adsorbed O2, consequently making O-O bond contracted as compared to that on unsupported NPs. Therefore, the activation barriers for O2 dissociation on the NPs, summarized in Figure 6d, inversely correlates with the O-O bond lengths, Figure 6b, such that the longer the O-O bond the smaller the activation barrier. As summarized in Figure 6d, the elemental trend for the barriers for O2 dissociation turn out to be Cu29Au29 for defect-laden and of Cu29>Au29>Ag29 for pristine. Strong electronic interaction between NPs and defect-laden MoS2 is facilitated by mid-gap states introduced by S vacancies, which also transform the planar MoS2 structure into the corrugated one with v-shaped structure that nicely couples with the boat-shaped NP via two triangular facets. These results suggest that S vacancies not only serve as the anchoring sites for NPs to cluster but also play a role in stabilizing the NPs on MoS2. Defect-laden MoS2 is, therefore, a good support material for dispersing the NPs. (ii) Both pristine and defect-laden MoS2 help shift the d-band center of the NPs further towards the Fermi level. This effect is most noticeable for defect-laden-MoS2-supported Au29, followed by Ag29. In addition, more charge transfer takes place from NPs to the defect-laden than to pristine MoS2 with the trend as Cu29>Ag29>Au29. (iii) Defect-laden MoS2 is responsible for increasing the number of active sites at which O2 adsorb with similar binding energy (

Effect of Single-Layer MoS2 on the Geometry, Electronic Structure

Recommend Documents