Creating single-atom Pt-ceria catalysts by surface step decoration

Abstract

Single-atom catalysts maximize the utilization of supported precious metals by exposing every single metal atom to reactants. To avoid sintering and deactivation at realistic reaction conditions, single metal atoms are stabilized by specific adsorption sites on catalyst substrates. Here we show by combining photoelectron spectroscopy, scanning tunnelling microscopy and density functional theory calculations that Pt single atoms on ceria are stabilized by the most ubiquitous defects on solid surfaces--monoatomic step edges. Pt segregation at steps leads to stable dispersions of single Pt²⁺ ions in planar PtO₄ moieties incorporating excess O atoms and contributing to oxygen storage capacity of ceria. We experimentally control the step density on our samples, to maximize the coverage of monodispersed Pt²⁺ and demonstrate that step engineering and step decoration represent effective strategies for understanding and design of new single-atom catalysts.

ARTICLE

Received 4 Sep 2015 | Accepted 20 Jan 2016 | Published 24 Feb 2016

Filip Dvok1,*, Matteo Farnesi Camellone2,*, Andrii Tovt1, Nguyen-Dung Tran2,3, Fabio R. Negreiros2,w, Mykhailo Vorokhta1, Tom Skla1, Iva Matolnov1, Josef Mysliveek1, Vladimr Matoln1 & Stefano Fabris2,3

Single-atom catalysts maximize the utilization of supported precious metals by exposing every single metal atom to reactants. To avoid sintering and deactivation at realistic reaction conditions, single metal atoms are stabilized by specic adsorption sites on catalyst substrates. Here we show by combining photoelectron spectroscopy, scanning tunnelling microscopy and density functional theory calculations that Pt single atoms on ceria are stabilized by the most ubiquitous defects on solid surfacesmonoatomic step edges. Pt segregation at steps leads to stable dispersions of single Pt2 ions in planar PtO4 moieties incorporating excess O atoms and contributing to oxygen storage capacity of ceria. We experimentally control the step density on our samples, to maximize the coverage of monodispersed Pt2 and demonstrate that step engineering and step decoration represent effective strategies for understanding and design of new single-atom catalysts.

DOI: 10.1038/ncomms10801 OPEN

Creating single-atom Pt-ceria catalysts by surface step decoration

1 Charles University in Prague, Faculty of Mathematics and Physics, V Holeovikch 2, Prague 18000, Czech Republic. 2 CNR-IOM DEMOCRITOS, Istituto Ofcina dei Materiali, Consiglio Nazionale delle Ricerche, Via Bonomea 265, Trieste 34136, Italy. 3 SISSA, Scuola Internazionale Superiore di Studi Avanzati, Via Bonomea 265, Trieste 34136, Italy. * These authors contributed equally to this work. w Present address: Universidade Federal do ABC. Av. dos

Estados, 5001 Bairro Bangu, Santo Andr SP CEP 09210-580, Brasil. Correspondence and requests for materials should be addressed to J.M. (email: mailto:[email protected]

Web End [email protected] ) or to S.F. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10801

Single-atom catalysts represent the limiting realization of supported metal catalysts with metal load ultimately dispersed as single atoms1,2. This maximizes the utilization

of supported metals and helps development of sustainable catalytic technologies for renewable energies and environmental applications with reduced precious metal contents3,4. A central prerequisite for understanding and knowledge-based design of single-atom catalysts is the identication of specic adsorption sites on catalyst supports that provide the stabilization of single metal atoms under reaction conditions at elevated temperatures and pressures. For oxide supports, understanding specic adsorption sites presently concentrates on low-index oxide facets59. Single-atom catalysts are, however, nanostructured large-area materials; thus, a question arises whether single supported atoms can be stabilized at defect sites of nanostructured oxide supports.

Highly dispersed platinum (Pt) ions on ceria qualify as single-atom catalysts2 and hold a promise of radical reduction of Pt load in critical large-scale catalytic applicationshydrogen production3, three-way catalytic converters10 and fuel cells11. Ceria surfaces provide a limited amount of low coordinated surface sites where Pt2 ions can adsorb and remain stable in real applications2,3,10,11. Recent studies on large-area ceria samples identify the necessity of nanostructuring the ceria substrates for obtaining supported Pt2 ions2,4 and propose a square-planar PtO4 unit as a Pt2-containing surface moiety4. In the present model study on the single crystalline CeO2(111) surface, we demonstrate

that single-ion dispersions of Pt2 are stabilized at monolayer (ML)-high ceria step edges. Pt2 ions at step edges are located in

PtO4 units that can be considered the elementary building blocks of Pt2 /ceria single-atom catalysts. The PtO4 units incorporate excess

O and can act as oxygen source for redox reactions. Besides clarifying the nature of Pt2 stabilization on ceria, our study demonstrates the importance of step edgesthe most common surface defects on oxide supports12for single-atom catalyst stabilization. We experimentally adjust the step density on the ceria supports for maximizing the load of monodispersed Pt2 ions.

This identies step engineering13 and step decoration14,15 as advanced techniques for designing new single-atom catalysts.

ResultsPt deposits on highly dened CeO2(111) surfaces. The experiments were performed on model CeO2(111) surfaces prepared as 20 to 40 thick ceria lms on Cu(111) using procedures that allow adjusting the density of ML-high steps16 and the density of surface oxygen vacancies on ceria surface17. On these highly dened surfaces, we deposit 0.06 ML of Pt, anneal at 700 K in ultra-high vacuum (UHV) and observe stabilization of Pt2 species and/or nucleation of Pt clusters with scanning tunnelling microscopy (STM) and with photoelectron spectroscopy (PES). Deposition and annealing of Pt on CeO2(111) surfaces containing low concentrations of defects

ML-high steps and surface oxygen vacancies (Fig. 1a)yield

4

2

0

4

2

0

4

2

0

a b

c

Pt 4f, hv =180 eV

Pt2+

Pt0

d e

f

Intensity (103 counts)

g

h

i

78

76

74

72

70

E B (eV)

Figure 1 | Nucleation of Pt and stabilization of Pt2 on ceria surfaces containing controlled amount of surface defects. (ac) CeO2(111) surface with low density of surface oxygen vacancies and ML-high steps. (df) CeO1.7 surface with increased density of surface oxygen vacancies. (gi) CeO2(111) surface with increased density of ML-high steps. (a,d,g) STM images of clean surfaces before deposition of Pt. (b,e,h) STM images after deposition of0.06 ML Pt and annealing at 700 K in UHV. All STM images 45 45 nm2, tunnelling current 2575 pA, sample bias voltage 2.53.5 V. Scale bar, 20 nm (a).

(c,f,i) PES spectra of the Pt deposit after annealing. All PES spectra were acquired with photon energy hn 180 eV (black points). Fits indicate metallic

(Pt0, blue line) and ionic (Pt2 , red line) contributions to Pt 4f signal. EB is the photoelectron binding energy.

2 NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10801 ARTICLE

a

metallic Pt0 clusters (Fig. 1b,c) coexisting with ionic Pt2 species (Fig. 1c). To determine whether the charge of the supported

Pt species is selectively induced by a specic defect type, we repeat the experiment varying independently the amount of surface O vacanciesup to 0.16 ML, creating CeO1.7 surface (Fig. 1df)and the amount of ML-high steps on the CeO2(111) surfaceup to 0.15 ML (Fig. 1gi). We observe that surface oxygen vacancies do not promote the dispersion of Pt2 species but lead to small metallic Pt0 clusters (Fig. 1e,f)18. On the other hand, the increased step density leads to almost complete oxidation of the Pt deposit to Pt2 (Fig. 1h,i) proving that step edges selectively promote the stabilization of Pt2 species. Detailed STM images allow to exclude formation of three-dimensional and two-dimensional PtOx clusters (Supplementary Fig. 1), and allow to conclude that Pt2 species are incorporated in the ceria step edges.

Nucleation of Pt0 clusters and stabilization of Pt2 species represent concurrent processes. Differently to Pt0 clusters (Fig. 1b,e), Pt2 species at the step edges are not discernible in empty states STM imaging (cf. Fig. 1g,h without and with Pt deposit) because of their electronic structure. STM imaging in occupied states on metal-supported ceria is unavailable19.

The possibility to adjust the density of ML-high steps on the model CeO2(111) surfaces16 allows us to obtain a quantitative correlation between the step density and the amount of Pt2 species. We prepare CeO2(111) samples with step density between 0.06 and 0.20 ML16 and deposit 0.06 or 0.18 ML Pt at 300 K. Parameters of the prepared samples are summarized in Supplementary Table 1. After annealing at 700 K the amount of Pt stabilized in the form of Pt2 is determined by PES. For quantication, all relevant parametersthe density of ceria steps, deposited amount of Pt and amount of stabilized Pt2 are expressed in ML where 1 ML corresponds to the density of Ce atoms on the CeO2(111) surface, that is, 7.9 1014 cm 2. The

density of steps is dened as the density of Ce atoms located at the ceria step edges20.

The amount of Pt2 ions as a function of the step density is plotted in Fig. 2a. For the higher Pt coverage 0.18 ML, the analysis reveals a linear dependence between the amount of stabilized Pt2 ions and the step density (Fig. 2a, blue symbols), conrming the activation of Pt oxidation to Pt2 and the localization of

Pt2 at the surface steps. The highest step density 0.20 ML allows converting up to 80% of the Pt deposit to Pt2 . The degree of oxidation of the Pt deposit increases with decreasing the amount of deposited Pt to 0.06 ML (Fig. 2a, black symbols). In this case, up to 90% of Pt converts to Pt2 . The concentrations of ceria step edges and the amount of Pt2 stabilized on the surface obey a classical supply-and-demand scenario characteristic for single-atom catalysts1,4,10: when sufcient step edges, the amount of oxidized Pt2 is limited by the amount of deposited

Pt. Otherwise, the amount of oxidized Pt2 is limited by the step density regardless of the amount of deposited Pt. The Pt deposit exceeding the available step sites cannot be oxidized and nucleates as metallic Pt0 clusters on the surface. Besides the high oxidative power of the step edges towards Pt, our quantitative analysis reveals also the capacity of ceria step edges to accommodate a high density of Pt2 ions. Up to 0.16 ML of Pt2 ions can be stabilized by the sample containing 0.18 ML of Pt deposit and0.20 ML of steps (Fig. 2a, blue symbols). This corresponds to 80% of the step-edge sites being occupied by Pt2 . In the whole range of the step densities between 0.06 and 0.20 ML, the occupation of the step-edge sites by Pt2 varies between 50 and 80%.

Stability and charge state of Pt2 /CeO2(111) samples. The necessity of annealing the Pt deposit on ceria in UHV at 700 K for obtaining Pt2 stabilization in our experiment indicates the

activated nature of Pt segregation at the ceria steps and oxidation, and implies considerable thermal stability of Pt2 ions on ceria.

High-temperature annealing represents a prerequisite for obtaining Pt2 ions also in the experiments on large-area nanostructured ceria samples3,4. Once created, Pt2 ions remain stable on repeated annealing at 700 K in UHV. The Pt2 ions in our experiment also remain stable on adsorption and thermal desorption of CO in UHV (Supplementary Fig. 2), or on exposure to air at ambient conditions (Supplementary Fig. 3).

Parallel to the charge state of the Pt deposit we determine the charge state of the CeO2 support, in particular the concentration of surface Ce3 ions that is indicative of reduction of the ceria surface. Contrary to the case of stabilizing Ni2 ions on ceria21, we observe that Pt oxidation during annealing is not accompanied by a corresponding reduction of CeO2(111)

surface (Fig. 2b). This rules out the direct participation of ceria into the observed Pt oxidation at steps and indicates the involvement of other oxidizing agents in the Pt2 stabilization, such as excess oxygen atoms. In the UHV environment of our experiments, the eligible source of excess oxygen can be water adsorbing in sub-ML amounts from background atmosphere (Supplementary Fig. 4) and undergoing dissociation on reduced ceria and Pt/ceria substrates22,23. In the large-area Pt2 /CeO2 catalysts displaying high concentration of Pt2 ions and exceptional redox reactivity, excess O atoms may be incorporated during the synthesis that proceeds in air3,10.

0.2

0.1

0

Pt/CeO2

0.18 ML Pt0.06 ML Pt

Pt2+ coverage (ML)

Ce3+ coverage (ML)

b

0.1

0 0 0.1 0.2Step density (ML)

Figure 2 | Capacity of stepped CeO2(111) surface to accommodate Pt2 . (a) Amount of Pt2 stabilized on CeO2(111) substrates with different density of steps for 0.18 ML (blue symbols) and 0.06 ML (black symbols) of deposited platinum. Pt not stabilized in the form of Pt2 remains metallic.

Lines represent guides to the eyes. Blue line is a linear t of 0.18 ML Pt data. Black line represents the maximum achievable amount of Pt2 in the case of 100% oxidation of Pt for 0.06 ML of deposited Pt. (b) Reduction of the ceria surface accompanying the stabilization of Pt2 ions determined by resonant

PES expressed as a coverage of the surface by Ce3 ions. Lines represent guides to the eyes from a and indicate the Pt2 concentration. The Ce3 concentration is lower or equal to the Pt2 concentration on all samples.

NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10801

Segregation of Pt at CeO2(111) step edges. Ab initio density functional theory (DFT) calculations allow to interpret the above experimental results. We calculate the segregation thermodynamics and the atomic and electronic structures of Pt atoms in representative adsorption sites on CeO2(111) surfaces. The results

for the lowest-energy congurations are summarized in Fig. 3 and Table 1. The model adsorption sites include oxygen vacancies (Fig. 3a), regular sites (Fig. 3b) and Pt clusters24 (Fig. 3c) on the CeO2(111) terrace, as well as two low-energy ML-high steps, which we label following ref. 25 as step I (Fig. 3d) and step II (Fig. 3e). A detailed list of the systems considered in the DFT analysis is reported in the Supplementary Note 1. The steps I and II represent the preferred types of steps at the CeO2(111) surfaces at temperatures o1,000 K (ref. 25). On our experimental samples, the steps I and II appear in equal proportion as evidenced from the absence of triangularly shaped islands in Fig. 1a,g25,26.

In agreement with the experiment, our calculations predict the preferential segregation of Pt adatoms at the steps I and II, independently on the local step geometry and stoichiometry. The binding energies of Pt at the steps are 1.63.4 eV higher than at stoichiometric or defective (111) terraces (Table 1). This driving force for Pt segregation at the steps is in qualitative agreement with recent calorimetric studies of other metal clusters on CeO2(111)27,28, see the Supplementary Discussion. The particular binding energy and the charge state of Pt atoms at the steps depend on both the local step geometry and stoichiometry (Table 1). In the following, we show that a good agreement with all the experimental observations can only be achieved when considering segregation of Pt at steps in the presence of excess O (calculations denoted as O), while Pt segregation at stoichiometric steps (denoted S) exhibits signicant discrepancies.

Pt at stoichiometric CeO2(111) step edges. Pt segregation on step I-S yields Pt2 species that are coordinated by four lattice O atoms in a characteristic PtO4 planar unit (Fig. 3d). The PtO4 unit is remarkably similar to that one proposed for Pt-doped ceria nanoparticles4 and for surface reconstructions of Pdceria systems29. Instead, the different atomic structure of the step IIS edge prevents the formation of PtO4 units, hinders the full Pt oxidation to Pt2 and yields weakly oxidized PtD species (Fig. 3e). Calculation results presented in Fig. 3dg correspond to the Pt coverage at the steps 1/3 (1 Pt atom per 3 Ce step-edge atoms). For interpreting the capacity of the ceria step edges to accommodate a high density of Pt2 ions, we calculate the adsorption of Pt at the ceria steps with increasing Pt coverage at the steps (Fig. 4), ranging from 1/3 to 1 (1 Pt atom per 1 Ce step-edge atom). On the step I-S, the maximum coverage of Pt2 species is 2/3 (Fig. 4a). Higher Pt2 coverages are unattainable and lead to nucleation of metallic Pt clusters, due to the large strain buildup resulting from long sequences of interconnected PtO4 step units (Fig. 4b). On the step II-S, metallic Pt0 species appear already for a coverage higher than 1/3 (Fig. 4c). Thus, on

d

c

b

(111)

Pt O

Ce4+

Ce3+

a

(111)

CeO2(111)

a b c

Pt @ step I-S

Pt @ step I-O

Pt @ step II-S

d

e

*

*

Pt2+

Pt+

Pt2+

Pt @ step II-O

f

g

*

*

Pt2+

Figure 3 | Pt adsorption sites on the CeO2(111) surface obtained from DFT calculations. (a) Pt adatom in a surface O vacancy, (b) on the stoichiometric CeO2(111) terrace and (c) supported Pt6 cluster. (d) Pt adatom at the stoichiometric step I (step I-S) and (e) at the stoichiometric step II (step II-S). (f) Pt adatom at the step I with excess O (step I-O) and (g) at the step II with excess O (step II-O). Binding energies and Bader charges are summarized in Table 1. (dg) The * symbol denotes the O atom removed to calculate the O vacancy formation energy reported in Table 2.

Table 1 | Properties of Pt on CeO2(111) obtained from DFT calculations.

Pt adsorption site Binding energy (eV) Formal charge Bader charge (e) Pt @ O vacancy 2.8 PtD 10.9

Pt @ CeO2(111) 3.3 PtD 9.7 Pt6 @ CeO2(111) 4.4 Pt0 9.9 Pt @ step I-S 5.0 Pt2 9.2

Pt @ step II-S 5.1 PtD 9.7 Pt @ step I-O 6.6 Pt2 8.6

Pt @ step II-O 6.7 Pt2 9.0 Pt bulk 5.5* Pt0 10.0

DFT, density functional theory; Pt, platinum; I-S and II-S, stoichiometric step types I and II; I-O and II-O, step types I and II with excess O atoms.

Results for the binding energies per atom, formal charges and Bader charges of Pt adatoms at the adsorption sites displayed in Fig. 3ag. The binding energy for the Pt cluster is the total binding energy divided by the number of Pt atoms. For reference, Pt bulk values are given in the last line.*Bulk cohesive energy.

4 NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10801 ARTICLE

Pt step coverage

2/3

1

a

b

Pt @ step

*

I-S

Pt2+Pt2+

Pt2+Pt2+Pt0

c

II-S

Pt0Pt0

d

e

I-O

*

Pt2+Pt2+

Pt2+Pt2+Pt2+

f

g

II-O

*

Pt2+Pt2+

Pt2+Pt2+Pt2+

Figure 4 | Capacity of the CeO2(111) step edges to accommodate Pt2 ions obtained from DFT calculations. Calculated top views of the Pt binding to the steps I-S (a,b), step II-S (c), step I-O (d,e) and the step II-O (f,g) for Pt step coverage 2/3 (a,c,d,f) and 1 (b,e,g). At the step I-S, the limiting coverage of Pt2 is 2/3 (a), additional Pt attaches to step edge as

Pt0 (b). At the step II-S, the Pt2 coverage is 0. Pt atoms attach as weakly ionized PtD and readily form metallic dimers (c) and clusters. On both steps I-O and II-O, excess oxygen can stabilize ionic Pt2 at step edges as single ions appearing isolated or in groups up to 100% step coverage (dg).

The * symbol denotes the O atom removed to calculate the O vacancy formation energy reported in Table 2.

Table 2 | Minimum energy to remove an O atom from the CeO2(111) surface obtained from DFT calculations.

O vacancy site Pt step coverage O vacancy formation energy (eV)

Pt @ step II-S 1/3 3.8 Pt @ step I-S 1/3 3.3 CeO2(111) 2.5 step II-S 0 2.0

Pt @ step I-O 1/3 1.9 Pt @ step II-O 1/3 1.9 step I-S 0 1.8 Pt @ step I-O 1 1.8 Pt @ step II-O 1 1.7

DFT, density functional theory.

Results for steps I and II, and for Pt at steps I and II in the presence and absence of excess O. The energies for Pt @ steps were calculated by removing the O atom marked by the * symbol in Figs 3 and 4.

samples with equal proportion of the stoichiometric steps I and II, ab initio calculations predict maximum Pt2 coverage at the steps (r33% of the step-edge sites) and maximum conversion of the Pt deposit to Pt2 (r33% of deposited Pt) that are well below the experimental values (5080% of step-edge sites, up to 90% of deposited Pt, cf. Fig. 2a).

Most importantly, the calculations on the stoichiometric steps predict that Pt segregation, oxidation and the formation of the Pt2 species are always accompanied by the reduction of surface

Ce atoms from Ce4 to Ce3 (denoted in gray in Figs 3 and 4). The resulting concentration of the Ce3 ions exceeds that of the

Pt2 ions by a factor of 2. This is in stark contrast with the resonant PES measurements on our samples showing that the concentration of Ce3 is considerably lower than the concentration of Pt2 after annealing the samples (Fig. 2b). This indicates that Pt is preferentially oxidized by other mechanisms than the

Pt0/Ce4 redox couple.

Pt2 at CeO2(111) step edges with excess O. Agreement between the theory and the experiment can be achieved when taking into account the step edges in the presence of an excess of O atoms. Irrespective of the local step geometry and Pt coverage at the steps, we nd that excess O atoms readily bind to Pt at the ceria steps and drive a rearrangement of the step morphology forming ionized Pt2 species incorporated in the planar PtO4 moieties on both steps I and II (Fig. 3f,g). In the presence of excess of oxygen, Pt atoms bind stronger to the ceria step edges, with calculated binding energies up to 6.7 eV, which are higher than at the stoichiometric steps edges by B1.6 eV, and which are also higher than the cohesive energy of bulk metallic Pt (Table 1). This condition, which determines the stability of the Pt2 species at steps with respect to metallic Pt clusters, is fullled only in the presence of excess oxygen at the steps. The computed electronic structure and density of states of the PtO4 moieties at the steps

I-O and II-O (Supplementary Figs 5 and 6) conrm that the Pt0-Pt2 oxidation results from the ionic PtO bond in the

PtO4 planar units, and that Ce3 ions do not form in agreement with the experimental evidence (Fig. 2b). The calculated maximum coverage of Pt2 at the steps I-O and II-O is 100%

(Fig. 4e,g and Supplementary Table 2), as interconnected assemblies of the PtO4 units can optimally t the periodicities of both steps I and II at calculated Pt coverages at the step edges 1/3, 2/3 and 1 (Figs 3f,g and 4dg). The presence of excess oxygen at steps therefore explains also the maximal Pt2 ionization experimentally measured on the ceria-supported catalysts.

The stabilization of excess oxygen in the PtO4 moieties by the Pt2 ions suggests an oxygen source for redox reactions and hence provides a link between the presence of highly dispersed ionic Pt species on ceria and the increased redox reactivity of Pt2/CeO2 single-atom catalysts3. The oxygen buffering capacity of ceria-based catalysts is associated with easy oxygen vacancy formation. We calculate the vacancy formation energy on the clean CeO2(111) terrace, on the stoichiometric steps and on the

Pt-decorated steps (Table 2). Compared with the energy of 2.5 eV calculated on the CeO2(111) terrace, the energies required to remove an oxygen atom bound to Pt at the stoichiometric steps are 3.3 eV (from the PtO4 unit at the step I-S) and 3.8 eV (from the PtO4 unit at the step II-S; see Table 2). Pt segregation at the stoichiometric step edges yields the formation of strong PtO bonds and therefore hinders ceria O-buffering. Much lower energies are instead needed to remove the excess O incorporated in the PtO4 units at steps I-O and II-O, where the O vacancy formation energy can be as low as 1.7 eV, lower or comparable to the values for the stoichiometric steps without Pt (2.01.8 eV;

NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10801

Table 2)30. This indicates that the dispersed Pt2 ions can enhance the oxygen storage capacity of ceria-based catalysts by assisting the reversible storage of excess O atoms.

The single-ion nature of Pt2 in the PtO4 units is preserved at all coverages, even when densely packed at the ceria step edges as interconnected PtO4 units. Indeed, the Pt charge state, its local electronic structure and the O vacancy formation energy of the densely packed PtO4 units are comparable to that of the isolated

PtO4 units at the steps (Supplementary Fig. 7 and Table 2). The PtO4 units exhibit a large adaptability in stabilizing at different types of surface step edges, resulting in high effectiveness and capacity of the ceria surface to accommodate the Pt2 ions.

Regardless of the particular organization on the surface, the PtO4 units are also always accessible to the reactants. Thus, the square-planar PtO4 units carrying monodispersed Pt2 ions can be considered elementary building blocks of single-atom Pt-ceria catalysts.

DiscussionOn large-area samples, Pt2 ions on ceria show exceptional reactivity with minimum Pt load in important applications: watergas shift reaction3, hydrogen oxidation on the anode of proton-exchange membrane fuel cell11 and in the three-way catalyst converter10. In these applications, Pt2 ions on ceria exhibit long-term stability under realistic reaction conditions of elevated temperatures and ambient pressure of reactant gases3,31,32. Pt2 in large-area samples is routinely identied with PES. Complementary measurements with extended X-ray absorption ne structure (EXAFS)33 and high-resolution transmission electron microscopy3,4,31,32 conrm the absence of three-dimensional Pt or PtOx clusters and, in agreement with the advanced PES measurements34, identify Pt2 as highly dispersed surface species on ceria32.

Our present study identies the stabilization of monodispersed Pt2 ions with one particular defect site on the ceria surfacethe monoatomic step edgeand excludes the stabilization of Pt2 on the oxygen vacancies. Monodispersed Pt on ceria is observed to be effective in incorporating excess oxygen even in the unfavourable conditions of UHV experiment. Excess oxygen and Pt2 arrange in the square-planar PtO4 moieties decorating different types of the surface steps at coverages up to one PtO4 per one step-edge Ce atom. The excess oxygen can be easily detached, indicating enhancement of the redox properties of ceria loaded with the Pt2 ions. Adjusting the step density and the

Pt load on the model CeO2(111) surface allows maximizing the coverage of Pt2 , while suppressing the nucleation of metallic Pt0 clusters. In the present experiment, we achieve surface coverage of Pt2 0.05 ML. A further increase of the completely monodispersed ionized Pt2 coverage to 0.1 ML can be expected.

Step edges on ceria have been previously identied as preferred nucleation sites for supported metal clusters20,21,27,28,3537. Our present study highlights the property of the step edges on ceria to provide specic structural and electronic environments for selective formation of monodispersed, thermally and chemically stable Pt2 ions. The step edges represent intrinsic defects ubiquitously present on nanostructured ceria surfaces4,38; our results are thus applicable for the interpretation of the properties and the optimization of the Pt2 load on large-area ceria supports3,10,11. More generally, the step edges may represent a common type of adsorption sites providing stabilization for monodispersed metal atoms and ions in any oxide-supported single-atom catalysts15,39. Our results therefore introduce important concepts of step reactivity40 and step engineering13,14 in understanding the stability, the activity and in designing new single-atom catalysts.

Methods

Experiment. The experiments were performed on surface science apparatuses in Surface Science Laboratory in Prague (STM, laboratory X-ray PES (XPS) withhn 1,487 eV (Al Ka), low-energy electron diffraction) and at the Materials Sci

ence Beamline in Trieste (PES with hn 221,000 eV (synchrotron), laboratory

XPS with hv 1,487 eV (Al Ka) and low-energy electron diffraction).

Preparation of the ceria substrates. The ceria layers and their Pt loading were prepared using the same procedures and parameters in both laboratories, and investigated by surface science methods in situ without exposing to air. The procedures and parameters of all samples are summarized in Supplementary Table 1. The ceria layers were prepared by deposition of Ce metal (Ce wire 99.9%, Goodfellow Cambridge Ltd) from Ta or Mo crucible heated by electron bombardment on clean Cu(111) substrate (MaTecK GmbH) in a background pressure of 5 10 5 Pa of O2 (5.0, Linde AG). The growth rate of CeO2 was 6 ML

per hour. Varying densities of 1 ML high steps on the prepared CeO2 layers were obtained by growth of CeO2 at constant substrate temperature 423 or 523 K (Method I in Supplementary Table 1) or linearly increasing substrate temperature from room temperature to 723 K (Method II in Supplementary Table 1)16. For experiments in Fig. 1, the ordered, fully oxidized layer of CeO2 (Fig. 1a,b) and the ordered reduced layer of CeO1.7 (Fig. 1d,e) were obtained by approach published in ref. 17 that yields the lowest step density. In this approach, rst, fully reduced Ce2O3 layer is prepared by depositing metallic Ce on a CeO2 layer and annealing in vacuum. Subsequently, the Ce2O3 layer is exposed to a controlled dose of O2 at5 10 5 Pa and annealed to obtain desired stoichiometry CeO1.7 or CeO2

(Method III in Supplementary Table 1). The CeO2 surface imaged in Fig. 1g,h was obtained by depositing 0.3 ML CeO2 on the CeO2(111) substrate as in Fig. 1a, forming small ML-high islands. This homoepitaxy of CeO2 on CeO2 yields clearly arranged samples with high step density (Method IV in Supplementary Table 1).

Characterization of the ceria substrates. The thickness of the ceria layers was determined from the attenuation of the substrate Cu 2p3/2 XPS signal measured at hn 1,487 eV. For calculations, we used inelastic mean free path of electrons in

CeO2 11.2 . The thickness of the ceria layers was set between 20 and 40 or 7 and12 ML with 1 ML corresponding to 3.1 , the distance between Ce(111) atomic planes of CeO2. In this range of thickness, the coverage of the Cu substrate by ceria ranges between 97 and 100 % (ref. 17). For determining the density of 1 ML high steps, we use a semi-automated procedure when the rst step outlines are marked in STM images manually. Step outlines are then mapped onto a properly scaled and rotated hexagonal mesh of surface Ce atoms. The atoms that are closest to the outlines are automatically identied as step-edge atoms and their density evaluated in ML. The error in determining the density of 1 ML high steps is estimated to be 10 % and is marked in Fig. 2a.

Preparation of the Pt deposit. Pt was deposited on the ceria layers from a Pt wire(99.99%, MaTecK GmbH) heated by electron bombardment. Pt was deposited on the sample surface at 300 K and subsequently stabilized by increasing the sample temperature to 700 K at the rate 2 K s 1. Both Pt deposition and annealing proceeded in the UHV background pressure 5 10 8 Pa or below. The thermal

treatment supports the ionization of Pt to Pt2 .

Characterization of the Pt deposit. The amount of Pt was calculated from the deposition time after calibrating the constant evaporation rate of the Pt evaporator. The evaporation rate was determined by a Quartz Crystal Microbalance and/or in a dedicated experiment from the thickness of 4-ML-thick Pt layers on CeO2(111)/

Cu(111) determined by attenuation of the substrate Cu 2p3/2 XPS signal measured at hn 1,487 eV. This dedicated experiment was used to correlate Pt evaporation rates

between the two experimental apparatuses. For calculations, we used inelastic mean free path of electrons in Pt 8 . The fraction of Pt2 after thermal treatment was determined by tting the ionic Pt2 and neutral Pt0 component in the PES Pt 4f spectrum measured at hn 180 eV (cf. Fig. 1c,f,i). The error in determining the Pt

and Pt2 amounts on the studied samples is 20 % and is marked in Fig. 2a. This error represents the calibration error of the Pt evaporation rate.

Resonant PES. Reduction of the ceria surface after deposition of Pt and thermal treatment was determined with resonant PES of Ce 4f state. We determine the so-called resonant enhancement ratio (RER) as dened in refs 41,42 from measurements of intensities of Ce3 and Ce4 components of valence-band resonant PES Ce 4f spectra of CeO2 measured off-resonance (hn 115 eV) and on-

resonance (hn 121.4 eV for the Ce3 component and hn 124.8 eV for the Ce4

component). The value of resonant enhancement ratio represents an upper estimate of the concentration of Ce3 ions on the ceria surface and is plotted in Fig. 2b17,42.

STM imaging. STM measurements were performed with commercial PtIr tips (Unisoku). STM imaging of CeO2(111) and Pt/CeO2(111) lms was available only

6 NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10801 ARTICLE

via unoccupied states. We used sample voltages 2.53.5 V and tunnelling currents 2575 pA.

Theory. All calculations were based on the DFT and were performed using the spin-polarized GGA U approach43, employing the PerdewBurkeErnzerhof

exchange-correlation functional44 and ultrasoft pseudopotentials45. The spin-polarized KohnSham equations were solved with a plane-wave basis set and the Fourier representation of the charge density was limited by kinetic cutoffs of 40 and 320 Ry, respectively. The Quantum-ESPRESSO computer package was used in all the calculations46. In the Hubbard U term, the occupations of the f-orbitals were dened in terms of atomic wave function projectors and the value of the parameter U was set to 4.5 eV, following our previous studies47,48.

Slab models. The ceria (111) surfaces were modelled with periodic (3 3) slabs

being three CeO2 ML thick and separated by more than 10 of vacuum in the direction perpendicular to the surface. The Brillouin zone was sampled at Gamma point. In the present work, we considered two low-energy ML-high steps that we label following ref. 25 as step I and step II. The edge of both step I and step II steps are oriented along the 1 10 direction. These surface steps were modelled with

vicinal surfaces described with monoclinic periodic slabs separated by 410 of vacuum in the direction perpendicular to the (111) terrace. The dimensions of the cells were 17.97 11.67 2 along the 1 12 and 1 10 directions (step I) and

15.72 11.67 2 along the 1

12 and 1

10 (step II). All the vicinal surfaces slabs

included three CeO2 ML. This thickness was shown to be sufcient to calculate the structural and thermodynamic properties of these steps25. The complete set of surface structures and systems considered in this work is listed in the Supplementary Note 1. All these systems were structurally optimized according to the HellmannFeynman forces. During the geometry optimization, the atomic positions of the lowermost CeO2 ML were constrained, as well as those of the Ce atoms in the central ML, except for the Ce atoms below the step edge.

Energetics. Binding energies were computed as 1/NPt (Eslab NPt EPt E

slab/Pt),

where Eslab/Pt is the total energy of the ceria slab containing NPt atoms of Pt, Eslab is the total energy of the corresponding relevant (stoichiometric, reduced or oxidized) Pt-free ceria slab and EPt is the total energy of a Pt atom in vacuum. The energies required to form an oxygen vacancy Ov were calculated as (Eslab/Ov EO2 Eslab), where E

slab/

Ov and Eslab are the total energies of the ceria supercell with and without the O vacancy, respectively, whereas EO2 is the total energy of a gas-phase O2 molecule compensated for the known overbinding predicted by (semi)local functionals for O2.

References

1. Yang, X. et al. Single-atom catalysts: a new frontier in heterogeneous catalysis. Acc. Chem. Res. 46, 17401748 (2013).

2. Flytzani-Stephanopoulos, M. & Gates, B. C. Atomically dispersed supported metal catalysts. Annu. Rev. Chem. Biomol. Eng. 3, 545574 (2012).

3. Fu, Q., Saltsburg, H. & Flytzani-Stephanopoulos, M. Active nonmetallic Au and Pt species on ceria-based water-gas shift catalysts. Science 301, 935938 (2003).

4. Bruix, A. et al. Maximum noble-metal efciency in catalytic materials: atomically dispersed surface platinum. Angew. Chem. Int. Ed. 53, 1052510530 (2014).

5. Qiao, B. et al. Single-atom catalysis of CO oxidation using Pt1/FeOx. Nat. Chem. 3, 634641 (2011).

6. Novotn, Z. et al. Ordered array of single adatoms with remarkable thermal stability: Au/Fe3O4(001). Phys. Rev. Lett. 108, 216103 (2012).

7. Parkinson, G. S. et al. Carbon monoxide-induced adatom sintering in a Pd-Fe3O4 model catalyst. Nat. Mater. 12, 724728 (2013).

8. Bliem, R. et al. Subsurface cation vacancy stabilization of the magnetite (001) surface. Science 346, 12151218 (2014).

9. Li, F., Li, Y., Zeng, X. C. & Chen, Z. Exploration of high-performance single-atom catalysts on support M1/FeOx for CO oxidation via computational study. ACS Catal. 5, 544552 (2015).

10. Hatanaka, M. et al. Ideal Pt loading for a Pt/CeO2-based catalyst stabilized by a PtOCe bond. Appl. Catal. B Environ. 99, 336342 (2010).

11. Fiala, R. et al. Proton exchange membrane fuel cell made of magnetron sputtered PtCeOx and PtCo thin lm catalysts. J. Power Sources 273, 105109 (2015).

12. Gong, X.-Q., Selloni, A., Batzill, M. & Diebold, U. Steps on anatase TiO2(101). Nat. Mater. 5, 665670 (2006).

13. Barth, J. V, Costantini, G. & Kern, K. Engineering atomic and molecular nanostructures at surfaces. Nature 437, 671679 (2005).

14. Vang, R. T. et al. Controlling the catalytic bond-breaking selectivity of Ni surfaces by step blocking. Nat. Mater. 4, 160162 (2005).

15. Gong, X., Selloni, A., Dulub, O., Jacobson, P. & Diebold, U. Small Au and Pt clusters at the anatase TiO2(101) surface: behavior at terraces, steps, and surface oxygen vacancies. J. Am. Chem. Soc. 130, 370381 (2008).

16. Dvok, F. et al. Adjusting morphology and surface reduction of CeO2(111) thin lms on Cu(111). J. Phys. Chem. C 115, 74967503 (2011).

17. Ducho, T. et al. Ordered phases of reduced ceria as epitaxial lms on Cu(111).J. Phys. Chem. C 118, 357365 (2014).18. Zhou, Y., Perket, J. M. & Zhou, J. Growth of Pt nanoparticles on reducible CeO2 (111) thin lms: effect of nanostructures and redox properties of ceria. J. Phys. Chem. C 114, 1185311860 (2010).

19. Shao, X., Jerratsch, J.-F., Nilius, N. & Freund, H.-J. Probing the 4f states of ceria by tunneling spectroscopy. Phys. Chem. Chem. Phys. 13, 1264612651 (2011).

20. Lu, J.-L., Gao, H.-J., Shaikhutdinov, S. & Freund, H.-J. Morphology and defect structure of the CeO2(111) lms grown on Ru(0001) as studied by scanning tunneling microscopy. Surf. Sci. 600, 50045010 (2006).

21. Zhou, Y. & Zhou, J. Interactions of Ni nanoparticles with reducible CeO2(111) thin lms. J. Phys. Chem. C 116, 95449549 (2012).

22. Mullins, D. R. et al. Water dissociation on CeO2(100) and CeO2(111) thin lms.J. Phys. Chem. C 116, 1941919428 (2012).23. Bruix, A. et al. A new type of strong metal-support interaction and the production of H2 through the transformation of water on

Pt/CeO2(111) and Pt/CeOx/TiO2(110) catalysts. J. Am. Chem. Soc. 134, 89688974 (2012).24. Negreiros, F. R. & Fabris, S. Role of cluster morphology in the dynamics and reactivity of subnanometer Pt clusters supported on ceria surfaces. J. Phys. Chem. C 118, 2101421020 (2014).

25. Kozlov, S. M., Vines, F., Nilius, N., Shaikhutdinov, S. & Neyman, K. M. Absolute surface step energies: accurate theoretical methods applied to ceria nanoislands. J. Phys. Chem. Lett. 3, 19561961 (2012).

26. Torbrgge, S., Cranney, M. & Reichling, M. Morphology of step structures on CeO2(111). Appl. Phys. Lett. 93, 073112 (2008).

27. James, T. E., Hemmingson, S. L. & Campbell, C. T. Energy of supported metal catalysts: from single atoms to large metal nanoparticles. ACS Catal. 5, 56735678 (2015).

28. James, T. E., Hemmingson, S. L., Ito, T. & Campbell, C. T. Energetics ofCu adsorption and adhesion onto reduced CeO2(111) surfaces by calorimetry.

J. Phys. Chem. C 119, 1720917217 (2015).29. Colussi, S. et al. Nanofaceted Pd-O sites in Pd-Ce surface superstructures: Enhanced activity in catalytic combustion of methane. Angew. Chem. Int. Ed. 48, 84818484 (2009).

30. Kozlov, S. M. & Neyman, K. M. O vacancies on steps on the CeO2(111) surface. Phys. Chem. Chem. Phys. 16, 78237829 (2014).

31. Fiala, R. et al. PtCeOx thin lm catalysts for PEMFC. Catal. Today 240, 236241 (2015).

32. Hatanaka, M. et al. Reversible changes in the Pt oxidation state and nanostructure on a ceria-based supported Pt. J. Catal. 266, 182190 (2009).

33. Nagai, Y. et al. Sintering inhibition mechanism of platinum supportedon ceria-based oxide and Pt-oxide-support interaction. J. Catal. 242, 103109 (2006).

34. Matoln, V. et al. Platinum-doped CeO2 thin lm catalysts prepared by magnetron sputtering. Langmuir 26, 1282412831 (2010).

35. Zhou, J., Baddorf, A. P., Mullins, D. R. & Overbury, S. H. Growth and characterization of Rh and Pd nanoparticles on oxidized and reduced CeOx(111) thin lms by scanning tunneling microscopy. J. Phys. Chem. C 112, 93369345 (2008).

36. Zhou, Y., Perket, J. M. & Zhou, J. Growth of Pt nanoparticles on reducible CeO2(111) thin lms: effect of nanostructures and redox properties of ceria.

J. Phys. Chem. C 114, 1185311860 (2010).37. Zhou, Y. & Zhou, J. Growth and sintering of Au Pt nanoparticles on oxidized

and reduced CeOx(111) thin lms by scanning tunneling microscopy. J. Phys. Chem. Lett. 1, 609615 (2010).38. Sayle, T. X. T., Parker, S. C. & Sayle, D. C. Oxidising CO to CO2 using ceria nanoparticles. Phys. Chem. Chem. Phys. 7, 29362941 (2005).

39. Castellani, N. J., Branda, M. M., Neyman, K. M. & Illas, F. Density functional theory study of the adsorption of Au atom on cerium oxide: effect of low-coordinated surface sites. J. Phys. Chem. C 113, 49484954 (2009).

40. Zambelli, T., Wintterlin, J., Trost, J. & Ertl, G. Identication of the active sites of a surface-catalyzed reaction. Science 273, 16881690 (1996).

41. Matoln, V. et al. Water interaction with CeO2(111)/Cu(111) model catalyst surface. Catal. Today 181, 124132 (2012).

42. Mullins, D. R. The surface chemistry of cerium oxide. Surf. Sci. Rep. 70, 4285 (2015).

43. Cococcioni, M. & de Gironcoli, S. Linear response approach to the calculation of the effective interaction parameters in the LDA U method. Phys. Rev. B 71,

035105 (2005).44. Perdew, J. P. J., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 38653868 (1996).

45. Vanderbilt, D. Soft self-consistent pseudopotentials in a generalized eigenvalue formalism. Phys. Rev. B 41, 78927895 (1990).

46. Giannozzi, P. et al. QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials. J. Phys. Condens. Matter 21, 395502 (2009).

NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10801

47. Fabris, S., de Gironcoli, S., Baroni, S., Vicario, G. & Balducci, G. Taming multiple valency with density functionals: a case study of defective ceria. Phys. Rev. B 71, 041102 (2005).

48. Fabris, S., Vicario, G., Balducci, G., De Gironcoli, S. & Baroni, S. Electronic and atomistic structures of clean and reduced ceria surfaces. J. Phys. Chem. B 109, 2286022867 (2005).

Acknowledgements

This work was supported by Czech Science Foundation (contract numbers 15-06759S and 13-10396S), and by the European Union via the FP7-NMP-2012 project chipCAT under contract number 310191 and the EU FP7 COST action CM1104. A.T. acknowledges the support of the Grant Agency of the Charles University, contract number 2048514. S.F. acknowledges the support provided by the Humboldt Foundation through a Friedrich Wilhelm Bessel Research Award. The high-performance computing resources were gratefully provided by ISCRA initiative of CINECA. CERIC-ERIC consortium is acknowledged for nancial support.

Author contributions

F.D., A.T., M.V., T.S., I.M., J.M. and V.M. designed and performed the experiments. M.F.C., N.-D.T., F.R.N. and S.F. performed the DFT calculations. All authors interpreted

the experimental and computational results. J.M., S.F., T.S., I.M. and V.M. wrote the manuscript. F.D., A.T., J.M., I.M., V.M. and S.F. provided funding.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Dvok, F. et al. Creating single-atom Pt-ceria catalysts by surface step decoration. Nat. Commun. 7:10801 doi: 10.1038/ncomms10801 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

8 NATURE COMMUNICATIONS | 7:10801 | DOI: 10.1038/ncomms10801 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications