Space charge governs the kinetics of metal exsolution

Nanostructured composite electrode materials play a major role in the fields of catalysis and electrochemistry. The self-assembly of metallic nanoparticles on oxide supports via metal exsolution relies on the transport of reducible dopants towards the perovskite surface to provide accessible catalytic centres at the solid–gas interface. At surfaces and interfaces, however, strong electrostatic gradients and space charges typically control the properties of oxides. Here we reveal that the nature of the surface–dopant interaction is the main determining factor for the exsolution kinetics of nickel in SrTi0.9Nb0.05Ni0.05O3–δ. The electrostatic interaction of dopants with surface space charge regions forming upon thermal oxidation results in strong surface passivation, which manifests in a retarded exsolution response. We furthermore demonstrate the controllability of the exsolution response via engineering of the perovskite surface chemistry. Our findings indicate that tailoring the electrostatic gradients at the perovskite surface is an essential step to improve exsolution-type materials in catalytic converters.

To determine the position of the thin film-to-substrate interface however, one relies on the Ni-and Nbsignals (intensity drop with tailing feature).The interface positions are determined to be located at the point where the intensity decreased to about half of the initial signal (on the log scale, typically corresponding to a drop of relative intensity by a factor of 10-100 with respect to the maximum).The sputter times may slightly vary due to small variations in the sputter area and small differences in the sputter behavior as a result of the different sample processing (e.g.oxide vs. metallic state of Ni).Please note that the accumulation width is estimated based on the sputter rate obtained from the oxide thin film and hence potential deviations in the sputter rate of the metallic nanoparticles were neglected.

Supplementary Note 3: Oxidizing pre-annealing
If the samples are pre-annealed in oxidizing conditions, considerable differences in the exsolved Ni volume are detected (after reducing thermal treatment) by investigations of the surface morphology (Fig. S3a), which reflects the retarding influence of a negative (blocking) surface potential established under oxidizing conditions.X-ray diffraction analysis reveals no changes in the thin film diffraction pattern after the oxidizing annealing, which is mainly determined by the doping level of the host lattice and the defect structure (Fig. S3b). 1,2Typically, a reducing thermal treatment at high temperatures (> 600°C) results in a relaxation of the host lattice due to exsolution of Ni-dopants to the surface and the formation of metallic nanoparticles in the oxide bulk.Throughout our in-situ spectroscopy and ex-situ topological investigations addressing the blocking space charge effect, the thin film samples were processed at low temperatures of T = 400°C and oxygen partial pressures of p(O 2 ) = 0.1 mbar.In order to rule out the significant segregation of host cations under oxidizing conditions that may influence the exsolution response by the formation of a secondary phase at the perovskite surface, we show the ratio of the integrated peak areas of the Ti 2p and Sr 3d core-level spectra recorded at different points of time during the oxidizing annealing step of protocol 2 of our NAP-XPS investigations (cf.Fig. 2a,b) in Fig. S4a.As can be seen, no considerable changes of the total peak area ratio are detected over the entire annealing time indicating that no dense cover layer was formed i.e. that no segregation of either cation is induced by the oxidizing treatment.For comparison, we calculate the theoretical attenuation of the Ti 2p XPS core-level intensity assuming the presence of a dense SrO surface layer of different thickness.For this purpose the inelastic mean free path was calculated by the QUASES-IMFP software using the TPP2M formula. 3On the basis of the IMFP, the Ti 2p core-level intensity can be estimated to be attenuated to 63% of the intensity acquired from a pristine STNNi surface if an SrO blanketing layer of only 1 nm thickness was present.If an SrO surface layer of 2 nm thickness was present at the STNNi surface an attenuation of the Ti 2p intensity to 39% of the initial intensity would be expected.Furthermore, Fig. S4 shows a representative AFM scan (Fig. S4b) and the according line profile across the terrace step structure of an STNNi thin film after oxidizing annealing (Fig. S4c).As can be seen, the terrace step structure is preserved after the oxidation and the step height is ~0.4 nm equal to the height of the asprepared perovskite unit cell.Therefore, no indications for extended cation segregation under oxidizing conditions can be detected.As we show in the main manuscript the time span of 60 min oxidizing treatment already results in a significant suppression of the exsolution response (Fig. 2d), indicating that not a stoichiometric effect but a space charge effect is present.
In Fig. S5 we show the synthesis of an STNNi thin film sample with an engineered SrO-termination layer (one atomic layer in thickness) and the morphology evolution during metal exsolution.After deposition of a 20 nm thick STNNi thin film with monolayer precision via RHEED-PLD (Fig. S5a), an SrO-termination layer is fabricated by in-situ ablation of a ceramic SrO2 target (Fig. S5b), i.e. without exposure to air when changing between the target materials.In this way, the surface termination can be controlled, synthesizing a SrO-terminated sample to investigate its' influence on the exsolution response.Notably, the native surface of as-deposited STNNi is likely to exhibit a mixed SrO/TiO2 termination.5][6][7] Here, the intensity increase of the diffraction spots relative to the intensity of the central specular spot indicates the formation of an SrO surface termination layer (Fig. S5c, d).Therefore, RHEED-PLD enables the precise deposition of one monolayer of SrO by in-situ monitoring of the relative intensity change, which is shown in Fig. 5d.As can be seen in Fig. S5e, the growth of STNNi with an engineered SrO termination layer can be achieved with smooth surface morphology of the thin films, yielding ideal model samples for studying the impact of SrO surface termination on the exsolution behavior.Importantly, the SrO-terminated STNNi thin film fabricated by sequential deposition of STNNi and SrO represents a surface with altered termination layer, while the formation of negatively charged strontium vacancies during the fabrication is negligible compared to samples that have experienced thermal oxidation.In contrast, oxidizing pre-annealing results in formation of a strong electrostatic field at the surface as well as space charge driven enrichment of trace amounts of Sr species, as discussed in our main manuscript.After reducing annealing of the sample at equal conditions to the oxygen pre-annealing study presented in our manuscript (5 h, 400°C, 4% H2/Ar), the surface morphology is investigated by atomic force microscopy (Fig. S2f).Our findings show that a SrOtermination layer, fabricated by PLD, cannot suppress the exsolution response at the STNNi surface.While a high density of exsolved nanoparticles is visible at the SrO-terminated STNNi surface, the introduction of a negative surface potential results in a strong suppression of nanoparticle exsolution considering the same time-temperature window (cf.Fig. 2d and Figure S3a).This indicates that not a change in the termination chemistry results in a suppression of the exsolution process, but the space charge region that is introduced upon oxidizing annealing.Yet, a clear influence of SrO-termination on the exsolution and nucleation dynamics on the surface is apparent (as expected), where a larger nanoparticle density of smaller average size, but with comparable sum of the nanoparticle volume around 1E-22 m3ꞏµm-2 is detected as for sample that have not experienced oxidizing pre-annealing, which resulted in a drop of exsolution volume by one order of magnitude.

Supplementary Note 5: Redox response of STNNi
In Fig. S6a an extended measurement protocol applied during NAP-XPS investigations is displayed including multiple consecutive reducing and oxidizing steps (the first part is equal to protocol 1 as discussed in the main manuscript, Fig. 2a).After introduction of hydrogen to the NAP-XPS chamber, formation of metallic Ni species is apparent based on the emergence of a low-binding energy signal.By switching of the ambient gas from hydrogen to oxygen, the nanoparticles are oxidized and the metallic signal is absent.After a second reduction step, again a significant metal signal is detected.2][13][14] Notably, oxidation typically results in an increase of the nanoparticle volume, which may result in an overall increase of the Ni signal intensity in comparison to the initial formation of the nanoparticles.Depending on the ambient atmosphere, significant shifts in the binding energy relative to UHV conditions (position denoted by dashed lines) can be observed based on the Ti 2p, Sr 3d, O 1s and Nb 3d core-level spectra (Fig. S6b).Please note that the redox equilibria are not activated at room temperature i.e., UHV conditions and hence the binding energy position at UHV conditions is used only as a relative reference value.Since the spectra are obtained from the perovskite support, the evident shift in binding energy is not related to the nanoparticle size as observed for the metallic Ni 2p signal.The apparent peak shifts also cannot be understood in terms of sample charging due to insufficient supply of electrons from the thin film bulk.Such limited charge compensation would result in a continuous shift toward larger binding energies and for n-STO would be expected to be most pronounced in oxidizing conditions, whereas the observed shift upon oxidation is in the opposite direction.The shift in binding energy hence is directly related to the formation of space charge regions at the perovskite surface.The active surface Schottky-equilibrium results in the formation of surface strontium vacancies in oxidizing conditions which yield an electron depletion layer close the surface and a repelling electric field for acceptor-type dopants, such as Ni.For the oxygen exchange equilibrium at the surface, we moreover use ∆ surface ∆ bulk 1.4 eV.This condition reflects an energetically favoured formation of oxygen vacancies in the surface-near region, well discussed in the literature and essentially allows the inversion of the space charge potential at low oxygen activity, where oxygen vacancies become the dominant defect species in highly donor doped SrTiO 3 .
The dedicated formulation of surface equilibria results in differing defect concentrations in bulk and surface at a given temperature and oxygen activity.As a result of this, a redistribution of defects via diffusion and drift is triggered which essentially leads to a balanced space charge equilibrium, in which diffusive currents driven via concentration gradient and drift currents driven via local electric fields vanish.This equilibrium moreover corresponds to the minimum in Gibbs free energy of the system, defining the thermodynamic equilibrium state of the space charge layer.In order to solve the space charge potential ϕ(x) we selfconsistently solve eqs.( 2), ( 4) and ( 7) together the boundary conditions of 0 2 Here, equations ( 9), (10) reflect Gauss law evaluated at the surface (x = 0) and far from the surface, where the electric field vanishes (global charge neutrality).At the surface, the electric field is given by the surface charge Here, c denotes the lattice constant of SrTiO3 and the surface concentrations  •• 0 ,  0 obey the surface equilibria defined in equation ( 7) and ( 4), considering a reduced reduction enthalpy.Equation (11)  reflects the Poisson equation, whereas the local charge density is given by the sum of local defect concentrations within the space charge layer.
All equation can be solved numerically using a finite-element approach, revealing the established surface space charge potential ϕ(x) as shown in the main paper, as well as all defect concentration profiles within the surface layer.Table S1 summarizes all numerical values used for space charge calculations as available in the listed literature.

Fig. S2|
Fig. S2| Depth-profiling of the cation distribution over the thin film thickness by secondary ion mass spectrometry comparing STNNi samples in the as-prepared and reduced state (cut from the same thin film) after reducing thermal treatment at different temperatures (4% H2/Ar, t = 30h).(a) Schematic sketch of the sputter process.An early increase of the Ni signal and respectively a delayed increase of the Sr, Ti and Nb signals indicates accumulation of Ni at the thin film surface after the reducing thermal treatment (left).Sputter profile obtained from STNNi after reducing treatment at T = 800°C (shown in Fig. 1c in the main manuscript), including the sputter profile of Pt (right panel) (b) A nickel accumulation zone of similar thickness is visible after reduction at different temperatures of T = 600°C, T = 700°C and T = 800°C.The samples are equal to Fig. 1 of the main manuscript.

Fig. S3|
Fig.S3|(a) Surface morphology of STNNi samples which were reduced (reducing conditions: 4% H2/Ar, T = 400°C, t = 5 h) after oxidizing pre-annealing (oxidizing conditions: p(O2)= 0.108 mbar, T = 400°C).The AFM scan size is 2x2 µm 2 and scale bars 1 µm respectively.After oxidizing treatment, the volume of nickel particles exsolved to the surface upon reducing treatment is significantly decreased.(b) X-ray diffractograms obtained from the oxidized thin film samples.No changes in the position of the thin film reflection are visible.Thus, only the surface region is affected by the thermal treatment while the crystallographic properties remain unchanged.

Fig. S4|
Fig. S4| (a) Ratio of the integrated Ti 2p and Sr 3d peak areas recorded at different points of time during oxidizing treatment of STNNi by near ambient pressure X-ray photoelectron spectroscopy.A Shirley-type background was subtracted.(b) Atomic force microscopy of the STNNi surface after oxidizing annealing at T = 400°C and p(O2)= 0.108 mbar for 60 min.After oxidizing annealing at T = 400°C and p(O2)= 0.108 mbar for 60 min the STNNi surface exhibits a smooth surface morphology with a distinct step terrace structure.The scale bar denotes 2 µm.(c) A line profile extracted from (b) reveals a terrace step height of ~0.4 nm equal to the perovskite unit cell and well comparable to the as-prepared state of the sample.

Fig. S5|
Fig. S5| Synthesis of epitaxial STNNi thin films with an engineered SrO surface termination layer.(a) In-situ RHEED monitoring of the PLD of 52 monolayers (20 nm) of STNNi in a layer-by-layer deposition mode.(b) In-situ RHEED monitoring of the PLD of one monolayer of SrO.(c) RHEED surface electron diffraction pattern obtained from the TiO2-terminated STO substrate (left), the as-deposited STNNi surface (center) and after deposition of one monolayer of SrO (right).(d) The average intensity ratio between the first order diffraction spots and the specular spot during SrO deposition revealing a change in their relative intensity.(e) Atomic force microscopy imaging of the as-deposited SrO-terminated STNNi surface and (f) of the same surface after reducing annealing.The reducing annealing was performed in a continuous flow of a 4% H2/Ar gas mixture at T = 400°C for t = 5 h.

Fig. S6|
Fig. S6| Near ambient pressure x-ray photoelectron spectroscopy.(a) Extended NAP-XPS measurement protocol showing the Ni 2p3/2 core-level spectrum under different redox conditions.The binding energy was corrected to the Ti 2p position.(b) Representative XPS core-level spectra of the Ti 2p, Sr 3d, O 1s, and Nb 3d regions recorded under UHV conditions and under ambient gas atmosphere.No correction of the binding energy position is applied for (b).The dashed line denotes UHV position.Sample conditions are denoted by the color code given below.(c) Based on shifts in the binding energy (determined from (b)) relative to ultra-high-vacuum conditions, the formation of a surface space charge region can be observed.The space charge potential is different for oxidizing and reducing conditions as it depends on the respective electric and ionic reconstruction in the surface region and reproducibly forms under repeated redox cycling.(d) Detailed comparison of representative Sr 3d core-level spectra recorded under repeated redox cycling, where a reversible change in the width of the photoemission signals is visible for different annealing environments that accompanies the binding energy shift depicted in (b,c).Red arrows highlight changes in the intensity valley between the Sr 3d3/2 and Sr 3d5/2 doublet.

Fig. S7|
Fig. S7| Representative AFM scans showing the surface morphology of stack samples, combining an STNNi bottom layer (20nm) and a top layer of four monolayers.Nb-doped SrTiO3 (donor-type) and undoped SrTiO3 (acceptor-type) is deposited as top layer material.Surface morphology is shown for samples in the as-prepared state and after reducing thermal treatment (reducing conditions: 4% H2/Ar, T = 400°C, t = 5 h), comparing samples without oxidizing pre-annealing and after pre-annealing (oxidizing conditions: p(O2)= 0.108 mbar, T = 400°C).The exsolution response is determined by the redox chemistry of the top layer material, i.e. a passivation effect is detected for the donor-doped (pre-oxidized) surface.The AFM scan size is 2x2 µm 2 and scale bars 1 µm respectively.