Nb Implanted BaO as a Support for Gold Single Atoms
NNb Implanted BaO as a Support for Gold Single Atoms
Debolina Misra ∗ and Satyesh K. Yadav Department of Metallurgical and Materials Engineering,Indian Institute of Technology Madras, Chennai, 600036, India (Dated: October 1, 2020)Using first-principles modelling based on density functional theory we show that oxides implantedwith transition metal can act as support for Au single atoms, which are stable against agglomeration.In our previous work we have shown that implanted transition metal, doped in BaO is stable asinterstitial in various charge states by transferring the excess charge to an acceptor level close toVBM. Taking Nb as an example we show that single atom Au has its Fermi level close to the VBMof BaO and hence is able to accept charge from the dopant. This charge transfer process betweenNb and Au helps Au atoms bind strongly on the doped BaO support. We also show that thesecharged Au atoms repel each other and prefer to remain atomically dispersed preventing clusterformation. Substitutional doping of transition metals have earlier been reported to bind Au atoms.However, if doped at interstitial sites, they can bind more Au atoms; for example, 5 Au atoms canbe anchored per Nb dopant present in BaO interstitial, compared to 3 Au atoms when Nb is dopedat substitutional site. This work paves the way for an altogether new technique of stabilizing noblemetal single atoms on transition metal doped oxides.
I. INTRODUCTION
In spite of the tremendous growth in designing non-noble materials for catalysis, noble metals are still at theheart of catalysis research due to their highest catalyticactivity [1]. Noble metal nanoparticles (NPs) are oftenused due to the large surface-to-volume ratio and thehigher concentration of under coordinated-surface sitesthey expose, compared to larger particles with more ex-tended surfaces. Nevertheless, the high cost and lowabundance of noble metals still significantly hinder theuse of such NPs as catalysts [2, 3]. Single-atom cata-lysts (SACs), which consist of atomically dispersed metalatoms on a support, are promising solutions to reducethis cost. They maximize the surface-to-volume ratio,which can potentially lead to a way more efficient use ofthe noble metals and thus decrease the cost of catalystfabrication [4, 5]. Despite existing challenges in prepa-ration of stable and active SACs, the last decade hasseen several successful strategies for fabricating promis-ing SAC prototype [6–8].Nevertheless, atomically dispersed metals are often un-stable against agglomeration into larger metal nanoparti-cles [8–10]. Efficient SACs must therefore be catalyticallyactive and resistant to sintering, both of which stronglydepend on intricate metal-support interactions. Dopedoxides serve as very effective supports for catalysts. Thedopant first changes the electronic structure of the hostand can cause a charge transfer which is not only limitedto the direct interaction of the dopant and its nearby hostatoms, but can also be extended to the adsorbate, situ-ated on the surface of the host lattice [11]. Such processesof charge transfer between impurity, the host lattice andthe adsorbate are at the heart of heterogeneous catalysiswhere charge transfer to a proper adsorbate on the oxide ∗ [email protected] surface can reduce the barrier of a chemical reaction. Re-cently it has been shown that for a charge transfer to takeplace between a dopant and an adsorbate, a direct inter-action is not always necessary [12]. Some earlier studieshave revealed that adsorption properties of gold and itsgrowth pattern strongly depends on charge transfer togold atoms which in turn depends on the nature of thedopant and the host lattice [13, 14]. For example, Motransfers charge to Au in CaO and MgO and alters itsgrowth pattern on the oxide surface. However, presenceof Cr in both the oxides resulted in no charge transferbetween gold atom and Cr [15].To stabilize single atom Au, all the previous studieshave focused on substitutional doping of oxides with tran-sition metals (TMs). Here we conclusively show that in-terstitially doped oxides can act as better support as TMsat interstitial site can transfer more charge than at sub-stitutional site. We use density functional theory (DFT)to show: (1) Nb prefers to be in 5+ charge state and oc-cupy the interstitial site in BaO, (2) doped Nb transfersexcess electron to Au atoms adsorbed on the surface andbinds them strongly, and 3) Au atoms anchored to thesupport repel each other and resist cluster formation. II. METHODOLOGY
All the spin-polarized DFT calculations were per-formed employing projector-augmented wave (PAW)method [18] and a plane wave basis set with 500 eV en-ergy cut-off, as implemented in Vienna
Ab initio
Simula-tion Package (VASP) [16, 17]. Generalized gradient ap-proximation (GGA) was used to treat electronic exchangeand correlation, employing the Perdew, Burke, and Ernz-erhof (PBE) functional [19]. A k -point mesh of 4x4x4 wasused for achieving converged results within 10 − eV peratom. All the structures were fully relaxed using the con-jugate gradient scheme and relaxations were considered a r X i v : . [ c ond - m a t . m t r l - s c i ] S e p FIG. 1: Formation energy for the neutral and chargedNb interstitial in bulk BaO as a function of electronicchemical potential µ .converged when forces on each atom was smaller than0.02 eV/˚A. Calculation of density of states (DOS) wereperformed using linear tetrahedron method with Blochlcorrections [20] and a denser k-grid. For the bulk cal-culations, cubic super cell containing 32 formula units ofBaO was doped with Nb in all their possible charge stateswhich results in a dilute limit(3.1%) of doping. III. RESULTSA. Stability and preferred charge state of Nbdopant in BaO
First we attempt to understand charge state of Nbwhen doped in BaO and its electronic structure, to ascer-tain its potential to transfer charge to Au. Ionic radii ofNb in 6 coordinated environment are 0.70, 0.68 and 0.64 ,in +3, +4 and +5 charge states, respectively; while ionicradius of Ba is 1.35 . Our prediction model [21] suggeststhat Nb will be stable at interstitial in bulk BaO. Weindeed find that Nb is stable at interstitial site. We cal-culate defect formation energy E qf [22–25] of Nb in bulkBaO as a function of electronic chemical potential µ , us-ing the following equation. E qf = E qD − E B − η + q ( µ + E ref + ∆ V ) + E qcorr (1)Here E qD and E B are the total energies of the defect su-percell with charge q and the defect free host supercell,respectively. η is the chemical potential of the transitionmetal atom species. The ’-’ sign indicates the additionof defect in the host. E ref is a suitable reference energy,taken to be the valence band maximum (VBM), and ∆Vis the correction to realign the reference potential of thedefect supercell with that of the defect free supercell [26]. E qcorr is the correction to the electrostatic interaction andthe finite size of the supercell. In this work only the first-order monopole correction has been taken into account.Defect formation energy of Nb in bulk BaO as functionof electronic chemical potential is shown in Fig.1. Nb is - 2 0 2 4 6 801 02 03 0 Total DOS (States/eV)
E n e r g y ( e V )B a O
FIG. 2: Total density of states of Nb-doped BaO andpure BaO. The vertical dashed line indicates the Fermilevel.stable in a single charge state(5+) for the entire range ofelectronic chemical potential ( µ ) studied. Here µ variesfrom VBM up to the band-gap of the host oxide, obtainedfrom our DFT calculation. As we intend to use Nb dopedBaO as a support for single atom Au, we explored thefollowings : 1) change in Fermi-level position of dopedBaO with different charge states of Nb and 2) stabilityof Nb in BaO at various depth from the surface.Density of states for Nb-doped bulk BaO is shown inFig.2. For all the charge states of Nb except 5+, theFermi level lies in the antibonding state which indicatesthat these states are not stable. The Fermi level lies atthe VBM only for 5+ charge state of Nb in BaO whichsuggests that this state is the most stable one. Changein Fermi-level position for different charge states of Nb issimilar to substitutionally doped oxides, generally usedto stabilize single atoms [12, 15]. We can expect thatthe five valence electrons of Nb can be transferred to anacceptor when BaO is doped with Nb.Stability of Nb interstitials at various depth from thesurface of BaO has been investigated in order to traceits most preferred position. A 5 layer thick BaO slabFIG. 3: BaO { } surface with Nb in different layers;Nb at (a) subsurface (b) sub-subsurface and on surface(c) O-top and (d) hollow sites respectively.TABLE I: Relative energy of BaO { } surfacescontaining Nb at different layers Position Relative energy (eV)Nb Nb Subsurface (Fig.3a) 0.00 0.00Sub-subsurface (Fig.3b) 1.76 1.22On surface O-top (Fig.3c) 0.80 9.58On surface Hollow (Fig.3d) 1.50 6.09 containing { } surfaces (known to be one of the moststable surfaces of BaO) is used to model the surface. In-plane dimension of supercell is 2x2 units of BaO with a12 vacuum along [001] direction. Atoms in the bottomtwo layers of the slab is frozen to their bulk positions inall our calculations.Nb in different layers (subsurface, sub-subsurface andon surface O-top and hollow sites) of BaO { } surfaceis shown in Fig.3. The relative energies between all thesethree configurations for both neutral Nb and Nb in 5+charge state are listed in TableI.From TableI it can be seen that subsurface layer is en-ergetically the most preferred layer for Nb interstitial inboth neutral and 5+ charge states. However, the mostimportant observation one can make from this calcula-tion is, although Nb neutral is stable in the subsurfacelayer, the difference between configurations with Nb oc-cupying the subsurface interstitial and on surface O-topsite is very small (0.18 eV). Hence under high temper-ature, neutral Nb can easily escape from the subsurfacelayer and can occupy the surface O-top site. Nb occu-pying surface O-top site will further lead to the highlystable Nb O oxide formation as the formation enthalpyfor Nb O is -9.84 eV per Nb atom [27]. However, thescenario changes drastically once Nb reaches its most pre-ferred valence state 5+. The difference in energy between slabs containing Nb in subsurface and occupying O-topsite is nearly 10 eV. This huge difference in energy willprevent Nb from escaping the oxide subsurface layerand forming Nb O . Stability of Nb in subsurfaceBaO slab and the huge energy difference between Nb at subsurface and surface O-top sites provide an uniqueopportunity to use Nb doped BaO as a support for Ausingle atoms where transfer of excess charge from Nb toAu can cause a strong binding of Au on the support. B. Adsorption of Au on Nb-Doped BaO Surface
1. Preferred binding site of single Au adatom
The relaxed supercell that we used to calculate the en-ergy of Nb doped at subsurface site, has been taken tounderstand the adsorption of Au. While calculating Auadsorption on one side of the doped BaO surface, neces-sary dipole corrections were included. Keeping in mindHollow and O-top sites as the possible binding sites fornoble metal atoms on pure BaO slab, we explored var-ious possible binding sites for a Au atom on Nb-dopedBaO slab. The potential binding sites are named as Nb-top, near-Nb O-top (when the binding oxygen atom is apart of the oxygen tetrahedra formed around Nb), andfar O-top (when the binding oxygen atom is not a partof the tetrahedra formed around Nb) sites as shown inFig.4. Relative energies of Au single atom on these sitesare listed in TableIII. We also explored hollow, Ba-topand bridge sites, however, for hollow site, the Au atomends up at near-Nb O-top site, and in both the later casesthe Au atom ended up on the neighbouring O-top sitesindicating that these sites are not the preferred bindingsites for Au atoms on Nb-doped BaO support. Our cal-culations revealed that Au atom binds most strongly onnear-Nb O-top sites. For Au to get adsorbed on far O-topsites, it costs almost 0.29 eV more energy compared tonear-Nb O-top sites. Surface Nb site is also higher in en-ergy compared to the near Nb O-top sites. Hence oxygenatom near Nb, serves as the most preferred and strongestbinding site for Au adsorption on Nb-doped BaO.FIG. 4: Possible binding sites of a single Au atom onNb-doped BaO { } surface. Au occupies the (a) Nbtop (b) near-Nb O-top and (c) far O-top sitesrespectively.TABLE II: Relative energies of Au single atom adsorbedat different sites on Nb-doped BaO { } surface. Binding Site Relative energy (eV)Near-Nb O-top (Fig.4b) 0.00Nb top (Fig.4a) 0.09Far O-top (Fig.4c) 0.29
2. Stability of Au atoms
Stability of Au atom on pure and Nb-doped BaO sup-ports were examined in terms of Au adsorption energy(per atom) for 1, 2 and 5 Au atoms using the equation E ads = E sup + Au − E sup − nE Au (2)Here E sup + Au , E sup and E Au refer to the energies ofthe support (pure or Nb-doped BaO) with Au atom, onlysupport without Au and energy of Au single atom respec-tively and n is the number of Au atoms adsorbed on thesupport. For adsorption of five Au atoms on Nb-dopedBaO (001) slab, the adsorption sites are chosen in such away that two Au atoms occupy the near-Nb O-top sites(strongest binding site) and the rest three Au atoms oc-cupy the second nearest O-top sites. For pure BaO slab,adsorption energy of Au single atom is 1.88 eV and for 5Au atoms it is 1.80 eV per Au atom. However, from Ta-bleIII it is evident that, compared to bare BaO surface,Nb-doped surfaces bind the Au single atom substantiallywell. The adsorption energy is almost 2 eV stronger whenBaO is doped with Nb.This strong adsorption of Au atom could be due toa charge transfer to Au. We have calculated the excesscharge (difference in charge on the Au atom and numberof valence electrons) on Au atom using Bader decomposi-tion scheme,[28–30] and the results are listed in TableIII.Our calculations revealed that there is a transfer of al-most one electron from Nb to Au causing the Au atom tobind strongly on the support. Transfer of charge to Auis significantly low in case of pure BaO support. For Nb-doped BaO slab, Au atom acts as an acceptor of chargeaccepting excess electrons from Nb and the charge trans-fer process continues when number of adsorbed Au atomsis increased until a maximum of five Au atoms are ad-sorbed on the slab.
3. Cluster formation of Au atoms
In order to understand stability of adsorbed Au againstsegregation, we study interaction of two and five Auatoms adsorbed on pure BaO and Nb-doped BaO sur-face. We find that the charged Au atoms resist cluster-ing, one of the bottlenecks in fabricating single atom cat-alysts. Fig.5 shows the optimized structures for two Auatoms (a) on two different O-top sites and (b) together onsame O-top site forming Au cluster on pure BaO { } FIG. 5: Two Au atoms on pure BaO { } surface andtheir relative energy.FIG. 6: Various possible arrangements of two Au atomson Nb-doped BaO { } support with their relativeenergies.FIG. 7: Arrangement of five Au atoms on (a) pure and(b) Nb-doped BaO { } surfacessurface, along with their relative energies. Comparisonbetween these two arrangements revealed that formationof Au cluster is more energetically favourable on pureoxide slab with a d Au − Au separation of 2.54 ˚A, smallerthan Au-Au bond length of 3.15 ˚A, observed in Au clus-ters [32]. Similar trend has been observed for five Au SAson pure BaO slab; Au atoms show the tendency to formcluster with an average Au-Au bondlength of 2.74˚A, asshown in Fig.7(a). Hence, Au atoms bound on pure BaOslab are prone to form larger particles.On the other hand, for Nb-doped BaO slab, a com-pletely opposite trend has been observed. The Au atomsanchored on the doped oxide surface preferred to remainatomically dispersed and resist cluster formation. Figure6 depicts the possible configurations taken by two Auatoms on Nb-doped BaO slab: the optimized structuresfor two Au atoms (a) on two near-Nb O-top sites (b) oneAu on a near-Nb O-top sites and the other one on farO-top site. Figure6(c) is the initial structure of two Auatoms adsorbed on the same O-top sites. The relativeenergies between these configurations are also mentionedTABLE III: Adsorption energy ( | E ads | ), Bader charge q and average Au-Au bond length (d Au − Au ) for Au atom onNb-doped and pure BaO { } surfaces Nb-doped BaO { } pure BaO { } No. of Au atoms | E ads | (eV) q (e) d Au − Au (˚A) | E ads | (eV) q (e) d Au − Au (˚A)1 3.83 -0.84 1.88 -0.402 3.78 -0.83 4.64 2.47 -0.21 2.545 3.52 -0.73 4.49 1.80 -0.23 2.74 in Fig.6. We find that, Fig.6(a) is the most favoured con-figuration for two Au atoms on Nb-doped BaO supportwith a bond length of 4.64˚A, which is almost 1.5˚Alargerthan the maximum bond length reported (3.15˚A) [32].The third configuration (6c) is not stable and after op-timization it comes to configuration (6a). This clearlyshows that the Nb-doped BaO { } surface serves asan agglomeration-resistant support for Au SAs. For fiveAu SAs anchored on Nb-doped BaO (Fig.7(a)), the Auatoms seem to move away from each other indicating thatthey resist forming larger particle and prefer to stay dis-persed in the form of single atoms. Hence, it is evidentthat Nb-doping not only binds Au SAs more strongly butalso prevents cluster formation of the SAs.
4. Electronic structure
To explain why charge transfer takes place between Nband Au atoms, we calculate density of states (DOS) for(a) pure BaO (b) 1 Au atom adsorbed on pure BaO, (c)Nb-doped BaO, and (d) 1 Au, (e) 2 Au, and (f) 5 Auatoms adsorbed on Nb-doped BaO { } surfaces. Atomresolved DOS are shown in Fig.8; for Au atoms only 6 s states are shown. Being an insulator, pure BaO { } surface shows no defect states in the band gap region,and a single atom Au adsrobed on pure BaO has its 6 s orbital half-filled. Once Nb is doped in BaO, midgapstates, mostly carrying the d-orbital signature of Nb ap-pears in the bandgap region of BaO. These mid gap stateseases the charge transfer between Nb and Au. For BaOdoped with Nb, the highest occupied level of the systemis pushed to CBM of BaO. Once one or two single atom ofAu gets adsorbed on Nb doped BaO support, the highestoccupied level of the system remains in CBM with boththe 6 s states of Au now being occupied and degenerate,which is attributed to the charge transfer from Nb to Au.It is only after the adsorption of 5 Au atoms, the high-est occupied level comes down to VBM, indicating thatall the excess 5 electrons are transferred to Au from Nb.Unlike pure BaO support, the charge transfer betweenNb and Au is evident for the doped one. A single Au canaccommodate one extra electron from Nb in its s orbitaland this charge transfer process continues till all 5 Auatoms are adsorbed on the support accepting one elec-tron each from Nb. This in turn helps Nb to attain themost preferred charge state, 5+. A careful observation ofthe Fermi level position (blue dashed line) in DOS plots reveals that the Fermi level of the system which was ini-tially near CBM, comes down to VBM of BaO when allthe excess five electrons from Nb are donated to 5 Auatoms, indicating that 5 Au atoms bound on BaO { } slab with subsurface Nb interstitial is a well stable systemas a whole. IV. DISCUSSION
Charge transfer from substitutional dopant to adsor-bate for stabilizing single atom of Au has been reportedearlier [12]. However, two factors limit their effective-ness in charge transfer, as noted by Stavale et. al. [15]:1) change in the charge state of substitutionally dopedtransition metal in the host, 2) the electron that is tobe transferred to the adsorbate can also get captured bycationic vacancies or morphological defects in the dopedoxide. Our approach mitigates these limitations. Nbdoped at substitutional site in BaO would be able totransfer 3 electrons, while at interstitial site it will beable to transfer 5 electrons. As implantation can be doneon single crystalline BaO, creation of cationic vacanciesor morphological defects could be minimized.To our surprise, stability against segregation of Auwhen it accepts charge from the doped TM have notbeen shown in any of the previous studies. Our resultsexplain why formation of well dispersed two dimensionalAu layer on Mo doped CaO is favourable, as charge getstransferred from Mo to Au, while in case of Cr dopedMgO, a 3D island of Au forms, as Cr could not transferany charge to Au [15].Based on our calculations we propose following exper-imental steps to achieve well bound single atom Au onNb-doped BaO. First, single layer of Au can be dispersedon { } surface of single crystal BaO followed by im-plantation of Nb ions into the BaO surface. Once Nbgets doped into BaO at the subsurface layer, it transferscharge to Au and binds few of those gold SAs on nearNb surface oxygen sites of BaO. The loosely bound Aucould be washed away leaving only strongly bound singleAu atoms. V. CONCLUSIONS
Based on first principles calculations, we propose a newway to form single atom Au, supported on BaO. We showFIG. 8: Atom-resolved density of states for (a) pure BaO (b) 1 Au atom adsorbed on pure BaO and (c) Nb-dopedBaO { } surface; (d) 1 Au, (e) 2 Au and (f) 5 Au atoms adsorbed on Nb-doped BaO { } surface. For Au, only6 s states are shown. The blue dashed line indicates the Fermi level.that such charged Au atoms are also stable against ag-glomeration, which explains already reported formationof 2D gold layer on Mo doped CaO. To achieve such welldispersed Au, Nb needs to be implanted in BaO which isstable at interstitial site in BaO. This process of stabiliz-ing Au on an oxide supports provides a distinct advantageover the known methods involving substitutional dopingof oxide with transition metals using conventional chem-ical routes. Transition metal at interstitial site transfersmore charge to adsorbate on oxide surface. Also ion-implantation provides area- and concentration-controlleddoping as it is a well established doping route in elec-tronic industries to create n and p type semiconductors.Although our results deal with Nb doped BaO for bind-ing Au single atoms, it opens up the possibility to explore other combination of oxides, doped with different transi-tions metals to bind noble metal single atoms. ACKNOWLEDGMENTS
D.M. gratefully acknowledges the support from Indo-German Centre for Sustainability (IGCS) research forproviding the opportunity to visit Technical University ofMunich for three months and carry out most of this work,and also the institute post-doctoral fellowship providedby IIT Madras, India. D.M. and S.K.Y. acknowledgeDr. Somnath Bhattacharya’s help for providing accessto VASP source code. [1] J. Liu, “Catalysis by supported single metal atoms,”
ACSCatalysis , vol. 7, no. 1, pp. 34–59, 2017.[2] A. Bruix, Y. Lykhach, I. Matolnov, A. Neitzel, T. Skla,N. Tsud, M. Vorokhta, V. Stetsovych, K. evkov, J. Mys-liveek, R. Fiala, M. Vclav, K. C. Prince, S. Bruyre,V. Potin, F. Illas, V. Matoln, J. Libuda, and K. M. Ney-man, “Back cover: Maximum noble-metal efficiency incatalytic materials: Atomically dispersed surface plat-inum (angew. chem. int. ed. 39/2014),”
AngewandteChemie International Edition , vol. 53, no. 39, pp. 10546–10546, 2014. [3] J. Zhang, Y. Zhao, X. Zhao, Z. Liu, and W. Chen,“Porous perovskite lanio3 nanocubes as cathode catalystsfor li-o2 batteries with low charge potential,”
ScientificReports , vol. 4, 2014.[4] N. Cheng, S. Stambula, D. Wang, M. N. Banis, J. Liu,A. Riese, B. Xiao, R. Li, T.-K. Sham, G. A. Botton,L.-M. Liu, and X. Sun, “Platinum single-atom and clus-ter catalysis of the hydrogen evolution reaction,”
NatureCommunications , vol. 7, 2016.[5] X.-F. Yang, A. Wang, B. Qiao, J. Li, J. Liu, andT. Zhang, “Single-atom catalysts: A new frontier in het-erogeneous catalysis,”
Accounts of Chemical Research , vol. 46, no. 8, pp. 1740–1748, 2013. PMID: 23815772.[6] S. Sun, G. Zhang, N. Gauquelin, N. Chen, J. Zhou,S. Yang, W. Chen, X. Meng, D. Geng, M. N. Banis,R. Li, S. Ye, S. Knights, G. A. Botton, T.-K. Sham,and X. Sun, “Single-atom catalysis using pt/grapheneachieved through atomic layer deposition,” Scientific Re-ports , vol. 3, 2013.[7] B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang,Y. Cui, J. Liu, L. Jun, and T. Zhang, “Prediction of sitepreference of implanted transition metal dopants in rock-salt oxides,”
Nature Chemistry , vol. 3, no. 2, p. 12593,2011.[8] J. Jones, H. Xiong, A. T. DeLaRiva, E. J. Peterson,H. Pham, S. R. Challa, G. Qi, S. Oh, M. H. Wiebenga,X. I. Pereira Hern´andez, Y. Wang, and A. K. Datye,“Thermally stable single-atom platinum-on-ceria cata-lysts via atom trapping,”
Science , vol. 353, no. 6295,pp. 150–154, 2016.[9] J.-C. Liu, Y.-G. Wang, and J. Li, “Toward rational designof oxide-supported single-atom catalysts: Atomic disper-sion of gold on ceria,”
Journal of the American ChemicalSociety , vol. 139, no. 17, pp. 6190–6199, 2017. PMID:28406020.[10] X. F. Chen, J. M. Yan, and Q. Jiang, “Single layer ofpolymeric metalphthalocyanine: Promising substrate torealize single pt atom catalyst with uniform distribu-tion,”
The Journal of Physical Chemistry C , vol. 118,no. 4, pp. 2122–2128, 2014.[11] X. Shao, S. Prada, L. Giordano, G. Pacchioni, N. Nil-ius, and H.-J. Freund, “Tailoring the shape of metalad-particles by doping the oxide support,”
AngewandteChemie International Edition , vol. 50, no. 48, pp. 11525–11527.[12] S. Prada, L. Giordano, and G. Pacchioni, “Charging ofgold atoms on doped mgo and cao: Identifying the keyparameters by dft calculations,”
The Journal of PhysicalChemistry C , vol. 117, no. 19, pp. 9943–9951, 2013.[13] M. Sterrer, T. Risse, M. Heyde, H.-P. Rust, and H.-J. Freund, “Crossover from three-dimensional to two-dimensional geometries of au nanostructures on thinmgo(001) films: A confirmation of theoretical predic-tions,”
Phys. Rev. Lett. , vol. 98, p. 206103, May 2007.[14] D. Ricci, A. Bongiorno, G. Pacchioni, and U. Landman,“Bonding trends and dimensionality crossover of goldnanoclusters on metal-supported mgo thin films,”
Phys.Rev. Lett. , vol. 97, p. 036106, Jul 2006.[15] F. Stavale, X. Shao, N. Nilius, H.-J. Freund, S. Prada,L. Giordano, and G. Pacchioni, “Donor characteristics oftransition-metal-doped oxides: Cr-doped mgo versus mo-doped cao,”
Journal of the American Chemical Society ,vol. 134, no. 28, pp. 11380–11383, 2012. PMID: 22741775.[16] G. Kresse and J. Furthm¨uller, “Efficient iterative schemesfor ab initio total-energy calculations using a plane-wavebasis set,”
Physical Review B , vol. 54, no. 16, p. 11169,1996. [17] G. Kresse and J. Furthmller, “Efficiency of ab-initio totalenergy calculations for metals and semiconductors usinga plane-wave basis set,”
Computational Materials Sci-ence , vol. 6, no. 1, pp. 15 – 50, 1996.[18] P. E. Bl¨ochl, “Projector augmented-wave method,”
Phys-ical Review B , vol. 50, no. 24, p. 17953, 1994.[19] J. P. Perdew, K. Burke, and M. Ernzerhof, “Generalizedgradient approximation made simple,”
Physical reviewletters , vol. 77, no. 18, p. 3865, 1996.[20] P. E. Bl¨ochl, O. Jepsen, and O. K. Andersen, “Im-proved tetrahedron method for brillouin-zone integra-tions,”
Phys. Rev. B , vol. 49, pp. 16223–16233, Jun 1994.[21] D. Misra and S. K. Yadav, “Prediction of site prefer-ence of implanted transition metal dopants in rock-saltoxides,”
Scientific Reports , vol. 9, 2019.[22] C. Freysoldt, B. Lange, J. Neugebauer, Q. Yan, J. L.Lyons, A. Janotti, and C. G. Van de Walle, “Electron andchemical reservoir corrections for point-defect formationenergies,”
Phys. Rev. B , vol. 93, p. 165206, Apr 2016.[23] R. Ramprasad, H. Zhu, P. Rinke, and M. Scheffler, “Newperspective on formation energies and energy levels ofpoint defects in nonmetals,”
Phys. Rev. Lett. , vol. 108,p. 066404, Feb 2012.[24] C. Freysoldt, B. Grabowski, T. Hickel, J. Neugebauer,G. Kresse, A. Janotti, and C. G. Van de Walle, “First-principles calculations for point defects in solids,”
Rev.Mod. Phys. , vol. 86, pp. 253–305, Mar 2014.[25] S. Bajaj, G. S. Pomrehn, J. W. Doak, W. Gierlotka,H. jay Wu, S.-W. Chen, C. Wolverton, W. A. Goddard,and G. J. Snyder, “Ab initio study of intrinsic point de-fects in pbte: an insight into phase stability,”
Acta Ma-terialia , vol. 92, pp. 72 – 80, 2015.[26] C. G. Van de Walle and J. Neugebauer, “First-principlescalculations for defects and impurities: Applications toiii-nitrides,”
Journal of Applied Physics , vol. 95, no. 8,pp. 3851–3879, 2004.[27] M. A. R. and P. R. J., “Thermodynamic evaluation ofthe nb-o system,” 2006.[28] R. F. Bader,
Atoms in molecules . Wiley Online Library,1990.[29] S. Edward, K. S. D., S. Roger, and H. Graeme, “Im-proved grid-based algorithm for bader charge allocation,”
Journal of Computational Chemistry , vol. 28, no. 5,pp. 899–908, 2007.[30] W. Tang, E. Sanville, and G. Henkelman, “A grid-basedbader analysis algorithm without lattice bias,”
J. Phys.:Condens. Matter , vol. 21, p. 084204, 2009.[31] W.-L. Yim, T. Nowitzki, M. Necke, H. Schnars,P. Nickut, J. Biener, M. M. Biener, V. Zielasek, K. Al-Shamery, T. Klner, and M. Bumer, “Universal phe-nomena of co adsorption on gold surfaces with low-coordinated sites,”
The Journal of Physical ChemistryC , vol. 111, no. 1, pp. 445–451, 2007.[32] C. Zeng, Y. Chen, C. Liu, K. Nobusada, N. L. Rosi, andR. Jin, “Gold tetrahedra coil up: Kekul´e-like and doublehelical superstructures,”