Engineering surface dipoles on mixed conducting oxides with ultra-thin oxide decoration layers

Improving materials for energy conversion and storage devices is deeply connected with an optimization of their surfaces and surface modification is a promising strategy on the way to enhance modern energy technologies. This study shows that surface modification with ultra-thin oxide layers allows for a systematic tailoring of the surface dipole and the work function of mixed ionic and electronic conducting oxides, and it introduces the ionic potential of surface cations as a readily accessible descriptor for these effects. The combination of X-ray photoelectron spectroscopy (XPS) and density functional theory (DFT) illustrates that basic oxides with a lower ionic potential than the host material induce a positive surface charge and reduce the work function of the host material and vice versa. As a proof of concept that this strategy is widely applicable to tailor surface properties, we examined the effect of ultra-thin decoration layers on the oxygen exchange kinetics of pristine mixed conducting oxide thin films in very clean conditions by means of in-situ impedance spectroscopy during pulsed laser deposition (i-PLD). The study shows that basic decorations with a reduced surface work function lead to a substantial acceleration of the oxygen exchange on the surfaces of diverse materials.


Supplementary Note 1: Correlations of different descriptors
To investigate the interchangeability of different descriptors and properties of oxidic decorations, we compared the ionic potential, the Smith acidity, the electronegativity and the bond ionicity of a variety of binary oxides (or their respective cations).In the following, the descriptors and their origins are briefly described: -The ionic potential is defined as the ratio of ionic charge and ionic radius and has the unit e/ Å1,2 .It is a measure of the charge density at the "surface" of an ion.Low ionic potential ions tend to be large ions with low charge, such as Li + or Sr 2+ , while high ionic potential ions tend to be small, high charge ions such as Cr 6+ or Si 4+ .The ionic potential is frequently used to describe the solubility of minerals and ionic interactions 3,4 .In this work, we used the formal charges of cations in the oxides and the crystal radii for the correct coordination proposed by Shannon 5 .
-The Smith acidity is a measure of an oxide's thermodynamic tendency to accept or release O 2− ions and is based on the formation enthalpy of an oxoacid salt from an acidic oxide and a basic oxide 6 , e.g.: CaO + CO 2 → CaCO 3 (1)   In this reaction, the CaO formally passes one O 2− ion to the CO 2 and leads to a compound with a Ca 2+ and a CO 2− 3 ion.Every oxide is then assigned a number a according to the empirical expression where A is the acidic oxide, B the basic oxide and h(A, B) the standard formation enthalpy of the oxoacid salt.Moreover, the stoichiometric equations are normalized such that exactly one O 2− ion is transferred in the reaction.To obtain the Smith acidity scale, a(H 2 O) is fixed to 0.
-The electronegativity of an oxide depends strongly on the electronegativity of the respective cation, which itself depends on the valence state of the ion in the compound.There have been several attempts to estimate electronegativities for cations in different valence states and to evaluate an electronegativity for a binary oxide [7][8][9] , for the comparison in this chapter, we use the electronegativity values proposed by Matar et al. 7 .
-The bond ionicity of an oxide describes the asymmetry of a metal-oxygen bond in an oxide compound.It is derived from the electronegativities of the oxide's constituents, so it is not surprising that it correlates well with the other metrics.In this study, bond ionicity values are taken from the work of Zhuravlev 10 .
In general, the comparison shows the expected result that the here explained metrics correlate very well with each other and are similarly well suitable to describe the properties of binary oxides.Differences emerge in their ease of use when considering more complicated oxides and in particular their surfaces.There, the Smith acidity, which is based on empirical data from binary oxides (and which also considers the structure of the products of reactions between binary oxides) is potentially not the best choice to describe the effects induced by modification processes.We suspect that real decorated surfaces deviate considerably from binary oxide structures and that it is the introduced cation that is decisive for the observed changes.Here we suggest that acidity and basicity (which are intuitive concepts and therefore desirable) are better correlated to other underlying ion-specific metrics such as the ionic potential or the electronegativity, which may be better suited to tackle more complicated problems.The ionic potential also has two further advantages: i) ionic radii are listed for different oxidation states, increasing the flexibility of the descriptor, ii) its constituents can be estimated by computational approaches, facilitating the synergy of experimental and theoretical studies.An overview of the quantitative correlations of the discussed metrics is given in the following figure (deviations for the ionic potential of SiO 2 might be due to the strong covalent character of the bond and our use of the formal 4+ charge): The work function is highest on SnO 2 decorated LSC and lowest for SrO decorated LSC.The main oxygen 1s species exhibits some asymmetry which is attributed to the metal-like electronic structure of LSC (it has previously been shown that this peak asymmetry correlates with the metallicity of the electronic structure of perovskite oxides 12 ).In addition, the main O 1s peak of SrO and SnO 2 decorated LSC is slightly broadened by the oxygen signature of the decoration, which is however not well distinguishable.The peak shape is further affected by the applied bias voltage (see below).Due to the strong overlap of the bulk and decorating oxide O 1s species, the peak asymmetry was optimized to match the envelope, rather than using two strongly covariant components.
When increasing the bias voltage, e.g. to 1000 mV (corresponding to ≈3.4 mbar), the peaks change slightly (see figure below for a measurement at 1.25 V).In particular, the work function of the SrO decorated LSC thin film increases more than for other surfaces and the previously SO 2− 4 related species appears to grow.However, the sulphur signal does not change accordingly during this process.The combination of these phenomena leads us to believe that the growing peak is related to peroxide species whose presence on the surface can be tuned by the application of sample bias.This is also in accordance with computational results which show that the work function tends to increase upon peroxide formation.This phenomenon has not yet been investigated in detail, but may be the first indication of the spectroscopic observation of peroxide species which take part in the oxygen exchange mechanism.Critical to this approach is the fact that measurement are performed in UHV and also at relatively low temperatures (450 • C).Thereby, the surface exchange is very slow and anodic polarization leads to a very high oxygen chemical potential at the working electrode surface, facilitating the formation of these peroxide adsorbates.In addition, the XPS signature is visible in UHV, because no SO 2− 4 adsorbates are present on the surface in these conditions.An in-depth exploration of this feature, however, goes beyond the scope of this study.

Supplementary Note 6: In-situ impedance spectroscopy during decoration
For a more in-depth discussion, we want to note here that detailed impedance spectroscopic investigations on thin films upon basic and acidic decoration have been presented by the authors in previous articles 11,17,18 .Generally, all i-PLD measurements follow a similar procedure, which has been outlined in previous studies 19 .The investigated samples consist of a YSZ single crystal electrolyte, Ti/Pt current collecting grids on both sides of the susbtrate and a 200 nm nanoporous LSC64 counterelectrode on one side.On the other side, the working electrode is grown during i-PLD.The temperature during deposition is controlled very precisely via the ohmic offset, which contains the electrolyte resistance (that is well known from literature 20 ), as well as resistive contributions from wiring and the current collecting grids (that are measured beforehand).Resulting impedance spectra usually consist of three major contributions.i) the above mentioned ohmic offset, ii) a mid-frequency semicircle, which corresponds to the surface exchange resistance coupled with the chemical capacitance of the working electrode, and iii) a low-frequency arc, which corresponds to the surface exchange resistance coupled with the chemical capacitance of the counter electrode (which is only visible for low working electrode resistances).The volume-related chemical capacitance of the counter electrode is much higher due to the higher thickness, leading to different characteristic frequencies of the two electrode contributions, and thus to well separable semicircles.It is worth mentioning, that for some samples, a small high-frequency shoulder appears, which is attributed to interfacial resistances between thin film and electrolyte.
As the working electrode feature is usually a nearly perfect semicircle, it is very unlikely that diffusion limitations affect the measurements (they would be visible as Warburg-type distortions on the high frequency side of the semicircle) and thus, the observed resistance is directly connected to the rate of the rate determining step of the oxygen exchange reaction.Our previous results have suggested that this rate determining step is related to charge transfer and dissociation 21 , but the details about this reaction mechanism are still not entirely clear.It is, however, very likely, that upstream reaction steps of the oxygen reduction reaction include adsorption of molecular oxygen and charge transfer onto the adsorbed oxygen.
During decoration, the only impedance contribution that is affected is the working electrode feature.Upon basic decoration, the surface exchange resistance decreases, suggesting faster oxygen exchange kinetics and vice versa for acidic decoration.In the figure below, exemplary impedance measurements of acidic and basic decoration for LSC64 and PCO are shown.
Supplementary Figure 8. a) Impedance spectra of La 0.6 Sr 0.4 CoO 3−δ in its pristine state as well as with 2 pls SrO and SnO 2 decoration.b) Impedance spectra of Pr 0.1 Ce 0.9 O 2−δ in its pristine state as well as with 2 pls SrO and SnO 2 decoration.

Supplementary Note 7: Deposition of thicker decoration layers
To evaluate the evolution of the oxygen exchange kinetics with growing thickness of a basic decoration layer, SrO was grown during i-PLD and the surface exchange resistance was tracked.Interestingly, the fastest kinetics were observed for decoration layers with a nominal thickness being slightly thinner than 1 monolayer (within experimental error, the optimal thickness is 1 monolayer).After that, the resistance starts to increase again.Mechanistically, we suggest that electronic interaction with the LSC bulk (which is essential for fast oxygen exchange) is still easy for one monolayer of SrO but gets increasingly difficult when depositing thicker layers.Preliminary results of a parallel study also suggest that the activation energy of the surface exchange resistance increases for thicker decoration layers, further supporting this hypothesis.
The same holds for PCO, where the fastest kinetics are reached at one monolayer.However, for PCO, the SrO layer can get relatively thick and still improve the kinetics of pristine PCO.This points towards the inherent differences between LSC and PCO, with LSC potentially being particularly active due to its electronic properties and easy electron transfer towards O 2 adsorbates.
More detailed studies of the activation energy of the oxygen exchange reaction with decoration thickness might be a viable opportunity to gain further insight into the underlying mechanism of the oxygen exchange reaction and are planned for future studies.Supplementary Figure 9. Evolution of the surface exchange resistance of LSC and PCO with the thickness of a growing SrO decoration layer.In both cases, the kinetics reach their fastest value at a decoration layer thickness of ≈1 monolayer.Afterwards, the kinetics continuously decrease and the resistance increases correspondingly.

LSC with SrO decoration
The SrO decoration was placed on the LSC slab, continuing the SrO termination.This means that the first two layers correspont to a rock-salt structure with the [100] direction rotated by 45 • compared to LSC.The Sr-Sr spacing in the decoration amounts along this direction amounts to 5.38 Å, compared to ≈ 5.16 Å in bulk SrO 22

LSC with SnO 2 decoration
The SnO 2 decoration was placed on the LSC slab as a BO 2 perovskite layer, yielding one nominal unit cell of SrSnO 3 on the surface.The diagonal Sn-Sn distance amounts to 5.39 Å, compared to ≈ 5.72 Å in bulk SrSnO 3 23 .

PCO pristine
The PCO structure is a 2x2x2 (111)-oriented and hexagonal cell with γ = 120 • and a cubic lattice parameter of 5.44 Å.One Ce atom in the surface and one Ce atom in the center were replaced with Pr to emulate a PCO10 stoichiometry.The SrO decoration was placed hexagonally on top of the PCO slab, with the O atoms placed above the Sr layer to emulate a (111) SrO layer.During relaxation, the oxygen atoms moved towards the PCO bulk, yielding a largely Sr terminated surface with a very low work function.However, this structure proved to be energetically unfavorable and a SrO 2 termination was identified as a more stable structure.The shortest Sr-Sr distance amounts to 3.86 Å vs. 3.65 Å in SrO bulk 22 .
24o SO 3 adsorbates were placed diagonally on the LSC slab surface in a tetrahedral configuration.S-O bond lengths are 1.46 Å in the SO 3 unit and 1.65 Å to the surface oxygen atom, compared to 1.49 Å in a SO 2− 4 anion24.
25r SnO 2 -decorated PCO, The fluorite structure was continued for one layer with Sn as the main cation.This leads to a shortest Sn-O distance of 2.32 Å, compared to 2.16 Å for a fluorite SnO 2 structure25.
24or the SrO 2 decoration, additional O atoms were placed on top of the relaxed O atoms in the SrO decoration.Up to a full SrO 2 layer, this proved to be energetically favorable.This also increased the work function considerably towards more resasonable values (with a better agreement with experimental values).The O-O bond length in the peroxide amounts to 1.50 Å, compared to 1.45-1.48ÅforbulkSrO226.PCO with SO 3 adsorbatesTwo SO 3 adsorbates were placed on two diagonally spaced top oxygen atoms of the PCO surface.S-O bond lengths are 1.47 Å in the SO 3 unit and 1.62 Å to the surface oxygen atom, compared to 1.49 Å in a SO 2− 4 anion24.