Few-monolayer yttria-doped zirconia films: Segregation and phase stabilization
Peter Lackner, Amy J. Brandt, Ulrike Diebold, Michael Schmid
FFew-monolayer yttria-doped zirconia films: Segregation and phasestabilization
Peter Lackner, Amy J. Brandt,
1, 2
Ulrike Diebold, and Michael Schmid a) Institute of Applied Physics, TU Wien, 1040 Vienna, Austria Department of Chemistry and Biochemistry, University of South Carolina, Columbia, SC 29208,USA (Dated: 22 January 2020)
For most applications, zirconia (ZrO ) is doped with yttria. Doping leads to the stabilization of the tetragonal or cubicphase, and increased oxygen ion conductivity. Most previous surface studies of yttria-doped zirconia were plaguedby impurities, however. We have studied doping of pure, 5-monolayer ZrO films on Rh(111) by x-ray photoelectronspectroscopy (XPS), scanning tunneling microscopy (STM), and low-energy electron diffraction (LEED). STM andLEED show that the tetragonal phase is stabilized by unexpectedly low dopant concentrations, 0.5 mol% Y O , evenwhen the films are essentially fully oxidized (as evidenced by XPS core level shifts). XPS also shows Y segregation tothe surface with an estimated segregation enthalpy of − ± I. INTRODUCTION
Zirconia (ZrO ) is an oxide with high thermal stability andfavorable properties for many applications. ZrO is usuallydoped with yttria, which enhances the mechanical strength(fracture toughness). This is related to the control of thecrystallographic modification by yttria doping and forms thebasis for applications as an engineering material. By intro-ducing oxygen vacancies into the lattice, yttria doping alsoincreases the oxygen ion conductivity, which makes yttria-doped ZrO a solid-state electrolyte with high ionic (thoughvery low electronic) conductivity at high temperatures, pro-viding the basis for its use in gas sensors and solid oxide fuelcells (SOFCs). Due to the small size of Zr cations, oxygen-oxygen repulsion governs the stability of zirconia structures;with decreasing temperature, pure zirconia undergoes trans-formation from a cubic structure via tetragonal ( T < ◦ C)to monoclinic ( T < ◦ C), in order to maximize O–Odistances while maintaining short Zr–O bonds. Already inthe beginning of the 20 th century, it was found that non-monoclinic phases can be stabilized by doping, e.g., withmagnesia, thoria, or – most commonly – yttria (YSZ, yttria-stabilized zirconia). Added trivalent Y + replaces tetravalentZr + ; for charge compensation, one oxygen vacancy is cre-ated per two Y atoms. These vacancies form the basis forthe high oxygen ion conductivity at high temperatures. Dis-tortions around the vacancy and an overall lattice expansiondecrease the average O–O repulsion, stabilizing the phasesotherwise present at high temperatures only. Below 1.5 mol%Y O , zirconia remains monoclinic. Concentrations above7.5 mol% Y O stabilize the cubic phase. Between 1.5 mol%and 7.5 mol%, a more complicated behavior is reported. De-pending on the preparation parameters, either a mixture of thecubic and monoclinic phases, tetragonal and cubic phases, oronly the tetragonal phase is formed.
Tetragonal ZrO canaccommodate strain by changing the crystal phase and/or ori-entation, which leads to the above mentioned high mechanical a) Electronic mail: [email protected] strength (tetragonal zirconia polycrystals TZP, zirconia tough-ened ceramics ZTC).
In nanoparticles, the tetragonal andthe cubic phase can form at lower dopant concentrations oreven without doping. This stabilization is mainly due to theintroduction of oxygen vacancies.
Upon annealing YSZ at high temperatures in ultrahigh vac-uum (UHV), or O , yttrium can diffuse and segregateto the surface and grain boundaries. This can lead to lo-cal phase transformations, as the yttrium content in some re-gions increases while it decreases in others.
Segregationis a well-studied topic for YSZ, as both, impurity (mainly Si,but also Ca and Na) and Y dopant segregation can influencematerial properties such as the oxygen exhange or the selec-tivity in chemical reactions, as shown for the example of for-mate oxidation. Many studies agree that surface segregationin YSZ is dominated by impurities.
The surface regionthen usually consists of silicates, with the subsurface regionenriched in yttria; it seems that the interface stabilizes the yt-tria below.
Only few experimental results are available foryttria segregation with little influence from impurities, yetyttria surface segregation is also found in these. A study onYSZ single crystals with ALD-deposited surface layers of in-creased Y content showed that oxygen incorporation, a typ-ical rate-limiting step for SOFCs, is increased with increas-ing yttrium concentration near the surface, while a silicon-containing surface layer leads to a decrease. As studies ofY segregation on clean surfaces are rare, neither the concen-tration profile nor the impact on applications are completelyunderstood. Density functional theory (DFT) studies do notagree whether Y enrichment should occur in the uppermostlayer, or in the layer immediately below ; a recent surfacex-ray diffraction study suggests surface enrichment.For a controlled atomistic study of the surface region ofYSZ, single crystals can be used. These are typically cu-bic with a doping level of 8–10 mol% Y O . As YSZ is anelectronic insulator with a wide band gap, surface sciencestudies at room temperature (RT) are difficult. Morrow et al.used high-temperature scanning tunneling microscopy (STM)to study the surface at 300 ◦ C; the existence of a tunnelingcurrent at this temperature was attributed to both, electronic a r X i v : . [ c ond - m a t . m t r l - s c i ] J a n and ionic currents. At RT, the insulating nature of the ma-terial can be circumvented by measuring on thin films. In thepresent work, we build on our previous work on pure zirconiathin films grown on Rh(111) single crystals; these films havebeen thoroughly characterized.
We showed that filmswith a thickness of five monolayers (ML) or more form bulk-like structures – either the tetragonal or the monoclinic phase– depending on the annealing temperature. ZrO films an-nealed at temperatures below 730 ◦ C are tetragonal. Whenannealing the films at 850 ◦ C, holes down to the Rh sub-strate appear in the films, i.e., the zirconia starts dewetting thesubstrate. Simultaneously, the films transform to the mono-clinic phase. Low energy electron diffraction (LEED) andSTM can be used to quickly characterize the film structure,as tetragonal films exhibit a ( × ) periodicity with respectto cubic ZrO (111), while monoclinic films show a distorted ( × ) structure (angles differ from 60 and 120 ◦ ). In STMimages, the tetragonal structure exhibits rows with a distanceof 0.63 nm; the Zr–Zr distance of 0.36 nm within the rows ismore difficult to resolve. X-ray photoelectron spectroscopy (XPS) showed thattetragonal films are slightly reduced (off-stoichiometry < Oxygen vacancies stabilize the tetragonal phase andare positively charged w.r.t. the unperturbed lattice (V •• O inKröger-Vink notation). This charge shifts the electrostaticpotential, resulting in substantial binding energy ( E B ) differ-ences ( ∆ E B ≥ . / levels of reduced,tetragonal zirconia were found between 183.4 and 182.6 eV,and for the monoclinic film between 181.8 and 181.6 eV. Itshould be noted that this shift, induced by different align-ment of the oxide bands with respect to the Fermi level, isopposite to the usual chemical shift caused by different ox-idation states; it is of purely electrostatic origin. The Zr inthe films remains in the 4+ charge state. When tetragonalfilms start dewetting the Rh substrate, the Rh substrate be-comes exposed and can act as a catalyst for O dissociation; the activated oxygen oxidizes the tetragonal zirconia film, andit transforms to the thermodynamically stable phase of stoi-chiometric ZrO , i.e., monoclinic zirconia.In section III A of this work we show how to create yttria-doped 5 ML-thick zirconia films by Y deposition on ZrO and annealing in oxygen. The stability of tetragonal zirco-nia thin films is substantially increased by Y incorporation;already 0.5 mol% Y O is sufficient to prevent formation ofthe monoclinic phase. Our approach provides a well-definedmodel system for yttria-doped ZrO , without impurities andwith a well-defined Y content. In section III B, the electronicstructure of the films is studied with XPS and compared to un-doped zirconia films. In section III C, XPS results are used toextract the Y segregation behavior. II. EXPERIMENTAL METHODS
We used a two-chamber UHV system with STM, LEED,and XPS capabilities in the analysis chamber ( p base < × − mbar). Sample preparation was conducted in the prepa- ration chamber ( p base < − mbar), which is connected tothe analysis chamber via a gate valve. The preparation cham-ber contains an electron-beam evaporator for Y (OmicronEFM 3), a UHV-compatible Zr sputter source, an Ar + sput-ter gun, and an e-beam heating stage for the sample. The setupis described in more detail in Ref. 28; details about the XPSmeasurement setup in this UHV system as well as the data fit-ting procedure using the program CasaXPS are found in Ref.12. LEED images were processed with dark frame and flatfield correction to suppress artifacts such as the grid struc-ture of the LEED optics.5 ML-thick ZrO thin films were sputter-deposited ( p Ar ≈ . × − mbar, p O2 = . × − mbar) at room tempera-ture on a Rh(111) single crystal substrate (diameter 9 mm,thickness 2 mm; from MaTecK, Germany). The films werethen post-annealed at 720 ◦ C in O to form an ordered, tetrag-onal structure. Unless noted otherwise, annealing was al-ways done in O at p O2 = × − mbar for 10 min. Todope these films with Y, either 0.05 ML or 0.2 ML (corre-sponding to 0.5 and 2 mol% Y O with respect to films with5 ML thickness) were deposited from an Y wire (1 mm di-ameter, 99.9% purity, handled in protective Ar atmosphereto avoid excessive oxidation or ignition). We define a de-posited monolayer (ML) as one atom per Zr atom in the sur-face layer (8 . × cm − ). Yttrium was deposited in an O background ( p O2 = × − mbar) at various temperatures,see below. As the sample could not be heated at the posi-tion in front of the evaporator, the samples were preheated;then, Y was deposited while cooling. The deposition rate ofY was measured by a water-cooled, retractable quartz crystalmicrobalance moved to the sample position before the actualdeposition. Additionally, we also verified the Y depositionrate in an experiment where we evaporated a defined amountof Y on Rh(111) and measured the coverage with STM. Sam-ple temperatures were measured with a thermocouple attachedto the sample holder and calibrated at T > 800 ◦ C using adisappearing-filament pyrometer. At lower T , the tempera-ture differences between the sample and sample holder areextrapolated. We estimate the temperatures obtained by thisprocedure to be accurate within ± ◦ C. III. RESULTSA. Structure: Stabilization of the Tetragonal Phase
The standard preparation of tetragonal yttria-doped filmsstarted with an ordered, closed, 5 ML-thick zirconia film inthe tetragonal phase (annealed at T = ◦ C). After confirm-ing the successful preparation of the structure with LEED andSTM, sub-monolayers of yttrium were deposited on top of thefilm at T = ◦ C in O . STM measurements revealed clusterformation at this temperature; Figure 1a shows the surface di-rectly after deposition of 0.05 ML Y. The inset of Figure 1a aswell as the LEED pattern in Fig. 1b confirmed the unchangedtetragonal structure of the film after deposition (cf. Ref. 28).The height of the clusters ( ≈ . (111) or Y O (111). We could R h111_2019_180 U bias = +3.2 V, I t = 1.24 nA (b) nm (a) 400 °C nm U bias = +3.6 V, I t = 0.17 nA nm R h111_2019_235 (c) 750 °C U bias = +3.0 V, I t = 0.16 nA R h111_2019_165 nm U bias = +3.2 V, I t = 0.27 nA R h111_2019_249 eV 70 eV R h111_2019 / R h111_2019 / (b) 400 °C (d) 850 °C FIG. 1. Preparation of 5 ML-thick yttria-doped ZrO films. (a) STM image of clusters after the deposition of 0.05 ML of Y on tetragonalzirconia. The small-area image (inset) shows that the typical row structure of tetragonal ZrO is unaffected by the deposition. (b) The LEEDimage of this surface confirms that the tetragonal phase remains stable after deposition. (c) STM image after annealing the film in (a,b) at750 ◦ C. The film is still tetragonal and not yet broken, and larger islands appear at the surface. The row structure is still present, and a largenumber of point defects appear (inset). (d) After annealing at 850 ◦ C, Rh spots (orange arrow) indicate that the film dewets the substrate, whilethe tetragonal structure is unchanged. (Blue arrows mark weak spots originating from a different Rh crystallite near the edge, and the red arrowpoints at a spot of the ( × ) O/Rh superstructure. ) not resolve the atomic structure of the clusters, probably be-cause of their small size (flat areas < ◦ C. However, after an-nealing 0.05 ML of Y on tetragonal zirconia at T = ◦ C(Fig. 1c), larger islands formed at the surface. In STM, theirsurface appeared like that of the ZrO film, with rows char-acteristic for tetragonal ZrO . The triangular and hexago-nal shapes of the islands are also found for undoped zirconiafilms. We therefore conclude that the material of the clus-ters had diffused into the film, resulting in zirconia (proba-bly doped) being expelled to the surface, forming a partial6 th layer. In principle, part of the island material might alsohave spilled out from holes in the film (which do not reachthe substrate at this temperature). Since the area fraction ofthe islands (5%) is compatible with the amount of depositedyttria, it is unlikely that the material in the islands originatesfrom the holes. Also, the islands are not preferentially locatedclose to these holes, as typically observed for material flowing out from holes. Diffusion of Y into the film was confirmedby XPS, see below.When annealing at even higher temperatures ( T = ◦ C;Figure 1d), holes in the film revealed the Rh(111) substratebelow. This can be inferred from the appearance of brightRh spots in LEED (marked with orange arrows in Figure 1d)as well as STM images (not shown). For a pure zirconiafilm, access to Rh would lead to full oxidation of the film andthus to a transformation to the monoclinic phase. However,with added Y, dewetting and exposing the Rh substrate wasnot accompanied by such a phase transformation; LEED stillshowed the ( × ) pattern of tetragonal ZrO , not the morecomplex pattern of the monoclinic phase. Considering thelow amount of Y deposited, this result came somewhat un-expected. Assuming a homogeneous distribution of Y acrossall five layers of the film, the material is doped with only 1at% of Y or 0.5 mol% Y O . This is quite low compared with > . ◦ C in O ) remained tetragonal upto the highest annealing temperature we tested (950 ◦ C; notshown).Although the film remained tetragonal, the number of pointdefects at the surface increased after Y deposition and anneal-ing, as becomes apparent by comparing the STM images inthe insets of Figs. 1a and b. Two types of point defects ap-peared: Bright species and a few dark holes. It is tempting tointerpret the bright species as Y ions; nevertheless it is alsopossible that the bright features are due to an electronic effectcaused by oxygen vacancies.In a different experiment, a monoclinic ZrO film was pre-pared by annealing at 850 ◦ C. The nominal film thickness was5 ML; due to the holes where the substrate was uncovered(dewetting), the local film thickness was 6–9 ML. On thismonoclinic film, 0.2 ML of Y were deposited at 550 ◦ C inO . The deposition did not change the structure of the film.However, post-annealing at 850 ◦ C in 5 × − mbar O ledto a complete transformation back to the tetragonal phase, seeFigure 2. Without Y, such an additional annealing step wouldnot change the structure of the film. A monoclinic → tetrag-onal transformation is also possible without Y-doping by an-nealing under highly reducing conditions (950 ◦ C in UHV ),which introduce oxygen vacancies. With added yttrium, thephase transition also occurs under oxidizing conditions and atlower temperatures. B. Photoelectron Spectroscopy: Core level shifts
X-ray photoelectron spectroscopy (XPS) was measured forfreshly prepared, tetragonal zirconia films, after deposition ofY, and after each of several annealing steps. Results from afilm with 0.2 ML Y are shown in Figure 3.For the pure, tetragonal zirconia films used in this study,we found the Zr 3d / and O 1s levels at binding energies of E B = 183.3 eV and 530.9 eV, respectively – in the range ob-served previously for 5 ML-thick tetragonal zirconia films. The binding energies are at the high side of this range, how-ever, pointing towards a rather high oxygen vacancy concen-tration (in the order of ≈ These binding energies wereunaffected by Y deposition at T = 400 ◦ C. Similar to purezirconia films, annealing at higher temperatures led to holesin the films reaching down to the substrate (local dewetting).This resulted in an oxidation of the film by oxygen spilloverfrom the metal, and thus the binding energy shifted to lowervalues. The onset temperature of dewetting and oxidation ofthe film depends on the details of the preparation; for Y de-position at T ≈ ◦ C, oxidation was encountered only afterpost-annealing at 750 ◦ C. However, when Y was deposited atsufficiently high temperatures (T = 550 ◦ C), as in the experi-ment shown in Figure 3, we found a few holes reaching downto the substrate (local dewetting) already after deposition. Forthis film, all XPS core levels (Zr, Y, and O) can be fitted bytwo contributions, one fixed to the binding energies of the re-duced film (same Zr and O binding energies as before Y depo- eV70 eV R h111_2019 / R h111_2019 / (a)(b) FIG. 2. LEED patterns of the transformation of a monoclinic filmto the tetragonal phase. (a) Original monoclinic zirconia film, withtypical elongated and multiple spots due to the unit cell angles de-viating from 60 ◦ . The orange arrow marks a Rh spot, and the redarrows points at a spot of the ( × ) O/Rh superstructure. (b) Afterdeposition of 0.2 ML Y and annealing at 850 ◦ C, the film transformsto the tetragonal phase. Some LEED spots are slightly smeared out,possibly due to remaining small, monoclinic areas. sition), and one at a lower E B . This indicates partial oxidationof the film. Additional annealing at 750 ◦ C in 5 × − mbarO converted the whole film into an essentially fully oxidizedstate (Zr 3d / at 181.9 eV).For Y 3d / , a typical E B value of the main component (cor-responding to the reduced film) of 158.5 eV was found afterdeposition. As expected for deposition in 5 × − mbar O ,no metallic Y component was present. The core level shiftsbetween the initially reduced films and the more oxidizedfilms are similar for the Y 3d / and Zr 3d / levels: Fromthe reduced to the oxidized film after annealing at 750 ◦ C, thebinding energies decreased by 1.2 and 1.3 eV, respectively. At880 ◦ C, Y 3d / shifted by an additional − . ◦ C led to another E B shift of − . − . (a) Zr 3d (b) Y 3d (c) O 1s I n t en s i t y ( kc p s ) Raw data Reduced Interface Oxidized Background Envelope
Binding energy (eV) C l ean F il m + Y a t ° C A nnea l ed ° C (d) STM
50 nm255075 50751001251508910
RhRhRh4 th ML C l ean F il m + Y a t ° C A nnea l ed ° C no m e t a lli c Y FIG. 3. XPS measurements and fits for 0.2 ML Y/ZrO , (a) Zr 3d, (b) Y 3d, and (c) O 1s region. (d) STM images for these preparations. Thetop panels are before Y deposition, the middle panels are after Y deposition at 550 ◦ C, and the lower panels are after annealing at 950 ◦ C inO . noisy due to the low coverage, so the peak fitting is less ac-curate. O 1s shifted by 0.2 eV less than Zr 3d / and Y 3d / in the 0.2 ML preparation, and 0.3 eV less in the 0.05 MLpreparation, when comparing the preparations before Y de-position and after annealing at 750 ◦ C (and above). None ofthese shifts were accompanied with changes of the crystallo-graphic structure, which was always tetragonal, as confirmedby STM and LEED.
C. Y Segregation
The intensity ratio of Y 3d and Zr 3d peaks can be used totrack the diffusion of Y in the film. Using this ratio avoidsstrong effects from dewetting, which leads to increased filmthickness for the remaining film and therefore lower intensi-ties (due to the very similar kinetic energy of the photoelec-trons, both, the Y and the Zr signals are equally affected atleast in case of a homogeneous Y distribution). Since the Y/Zrratio depends on the exact amount of Y deposited, in this sec-tion we only discuss changes of the Y/Zr intensity ratio beforeand after annealing; in contrast to the Y/Zr ratios themselvesthese changes are insensitive to inaccurate initial Y coverages.After annealing a tetragonal zirconia film with 0.2 ML of Yat 750 ◦ C, the Y 3d/Zr 3d intensity ratio dropped to 93% of theoriginal ratio, yet then remained nearly constant with 90% af-ter annealing at 880 ◦ C and to 91% after annealing at 950 ◦ C.Due to the low intensity of the Y 3d doublet, the changes ob-served above 750 ◦ C lie within the error bars. A film with 0.05 ML Y deposition showed a similar intensity ratio drop to 91%after annealing at 880 ◦ C. The drop in intensity could eitherindicate that yttria diffused from the clusters, which formed atthe surface during Y deposition (see above), into the film, orit diffused over the surface to form larger, 3D yttria clusters,thick enough to attenuate some of the Y 3d signal. STM con-firmed the first interpretation, as no 3D clusters were found atthe surface after annealing at elevated temperatures. Thus, itcan be assumed that Y is incorporated into the film.To estimate the distribution of Y in the film, we compare themeasured intensity ratios Y 3d/Zr 3d to ratios simulated withthe program SESSA. For these simulations, we assumed thatthe overall Y content of the film does not change upon anneal-ing. For the oxygen concentrations in the layers, full oxida-tion was assumed, i.e. x ZrO + y YO . , and the bandgap ofthe oxide layers was set to 5 eV. We did not simulate the par-tially filled first layer as such but rather took a single layer(0.3 nm thick) with appropriate average concentrations as thefirst layer (see bottom right of Fig. 4 for 0.2 ML deposited on5 ML ZrO ), since, in our experience, this leads to better ac-curacy.When distributing 0.2 ML YO . homogeneously in all lay-ers (resulting in a composition of Zr . Y . O . , leftmostpoint in Fig. 4), compared with the same amount of YO . in the first layer only, the Y/Zr intensity ratio drops to 73%.Comparison with the experimental drop to ≈
91% shows that,on average, Y does not diffuse deep into the film, but mostY stays near the surface of the film. Assuming Y enrichmentin the first layer only and a constant concentration in the lay-ers below, the experimental change of the Y/Zr intensity ratiocan be reproduced with Zr . Y . O . in the first layer andZr . Y . O . below (see Fig. 4). The experimental re-sult is not compatible with Y enrichment in the second layeronly, and the surface being pure ZrO , as suggested in a DFTstudy : Moving all Y to the second layer would cause a de-crease of the Y intensity to 85%, a stronger decrease than ob-served experimentally, and the decrease would be even morewhen assuming some Y in the layers below.To examine whether Y stays near the surface due to kineticlimitations, we have performed an experiment where the ini-tial position of Y was below the surface: 2.4 ML of zirco-nia were deposited with the usual parameters, followed by 0.2ML Y deposition at room temperature in O , and an additional2.4 ML of zirconia. When annealing this film at 750 ◦ C, theY 3d/Zr 3d ratio increased by 24%, and stayed constant af-ter annealing at 880 ◦ C and 950 ◦ C. This again demonstratesY segregation to the surface region, and the simulation canreproduce this intensity ratio for essentially the same surfaceconcentration as above, Zr . Y . O . in the first layer andZr . Y . O . below. The good agreement with the con-centration profile obtained after yttria deposition at the top andannealing indicates that Y diffusion is fast enough for equili-bration. The high level of agreement also means that Y doesnot float up during room-temperature deposition of the upper2.4 ML zirconia, but it gets buried by the zirconia (otherwisethe increase of the Y/Zr intensity ratio would be less). st layer0.0350.0400.045 Y / Z r i n t en s i t y r a t i o
20% Y4.2 MLpure ZrO ≈
4% Yin all 5.2layers Rh(111) experimentalY/Zr XPS ratioafter annealing(91% of initial)initial stateafter yttriadeposition
FIG. 4. Simulated Y 3d/Zr 3d intensity ratios (blue) after yttria depo-sition (rightmost point), and with different degrees of Y enrichmentin the surface layer. The total Y content is the same in all cases.The leftmost data point assumes homogeneous distribution of Y inthe 5.2 ML thick film. The decrease of the Y/Zr intensitiy ratio uponannealing as found experimentally is given by a horizontal, red line,with the orange area symbolizing the uncertainty. The correspondingconcentration profiles (assuming that only the first layer can have adifferent Y concentration than the others) are shown at the bottom.
IV. DISCUSSION
After deposition of yttrium on zirconia films at 5 × − mbar O , the small size of the clusters found by STMsuggests the formation of small Y O aggregates; we have noevidence for immediate diffusion of Y into the film at 400 ◦ C.The enthalpy of formation of Y O is ≈ − Thiscorresponds to − . − bar at 400 ◦ C. Therefore, metallic Ycan be safely excluded (in agreement with XPS showing nometallic Y). After deposition or annealing at higher tempera-tures, larger, triangular or hexagonal islands appear, with thesurface structure typical for tetragonal zirconia. This is ac-companied by a lowering of the XPS peak ratios of Y 3d/Zr3d, which shows that Y is incorporated into the film. How-ever, Y stays preferably in the surface region of the film, i.e.,the topmost layer, even after annealing at temperatures of upto 950 ◦ C. We cannot exclude some Y enrichment in the layerbelow, but the results do not support Y enrichment in the sec-ond layer only, as suggested in a DFT study. When Y isplaced in the middle layer of the film already during deposi-tion, it also segregates to the surface region after annealing.In contrast to previous experimental studies, our films containno traces of impurities (down to the detection limit of XPS). Since it has been suggested that impurities like Si, Na or Caenhance yttrium segregation, we consider our study the firstconfirmation of Y surface segregation where the influence ofimpurities can be definitely ruled out.We can use the Langmuir-McLean equation to estimatethe segregation enthalpy of the YO . /ZrO system. Assumingthat only the surface layer is enriched in Y, from the concen-trations mentioned above we obtain ∆ H segr = − ± O to 5 ML-thick zirconia films is suf-ficient to stabilize them in the tetragonal phase. It should benoted that this is an average concentration; due to Y segre-gation the surface concentration is substantially higher, whilethe subsurface layers contain less than 0.5 mol% Y O . Forcomparison, in bulk zirconia, more than 1.5 mol% is neededfor the stabilization of non-monoclinic phases. In Ref. 12, theamount of oxygen vacancies stabilizing the tetragonal phasein 5 ML-thick zirconia films was estimated to be (cid:46) O would berequired. There can be different reasons why the tetragonalstructure is more stable in thin films than in the bulk: (i) Itis possible that the surface energy of the tetragonal phase islower, as suggested previously. Calculated values of the sur-face energies for the relevant surface orientations are verysimilar for the two phases, however. (ii) The interface to thesubstrate below may stabilize the tetragonal phase. (iii) Eventhe oxidized tetragonal yttria-doped films may still contain ad-ditional oxygen vacancies (beyond those introduced by Y dop-ing), stabilizing the tetragonal phase. Currently, we cannotdecide whether (ii) and/or (iii) are the main factors extendingthe stability of the tetragonal phase to lower Y concentrations.As for pure ZrO films, all XPS core level shifts dependon the oxygen vacancies present in the film. In the current caseof yttria-doped films, we have to distinguish between oxygenvacancies caused by doping (keeping the Zr x Y y O x + . y stoi-chiometry) and additional oxygen vacancies, making the filmoxygen-deficient. Our experiments show that all levels shift tohigher binding energies in the presence of additional oxygenvacancies; the E B values of the fully oxidized films are notinfluenced by the existence of Y-induced O vacancies in thefilm. At first glance, this may seem unexpected as both typesof oxygen vacancies are positively charged V •• O . The differ-ence lies in the fact that the crystal remains electrically neutralwhen there are only doping-induced O vacancies, while anyadditional vacancies lead to an excess of positive charge, and,hence, a positive electrostatic potential of the film. As dis-cussed previously, this is the reason for the increased bind-ing energy of all species in oxygen-deficient films. We haveobserved that the shift of the O 1s lines upon oxidation of thefilm is slightly less (by 0.2–0.3 eV) than the shift of the Zr andY 3d lines. A similar behavior occurs when oxidation goeshand in hand with the tetragonal → monoclinic phase trans-formation and was explained with a changing band gap or thechange of the local structure. Compared to the electrostaticcore level shifts of more than 1 eV, surface core level shifts ofZr in ZrO are predicted to be much smaller ( ≈ .
15 eV, Ref.38), and we consider it likely that the same is true for Y inyttria-doped ZrO . V. CONCLUSIONS
We have studied yttria doping by Y deposition on high-purity, 5 ML-thick zirconia films. Annealing at 750 ◦ C in5 × − mbar O leads to an equilibrium distribution of theY. In this state, the surface is strongly Y-enriched; the XPSresults are not compatible with pure ZrO in the surface layerand Y in the second layer only as suggested previously. Dop-ing levels of 0.5 mol% Y O (averaged over the film thick-ness, higher Y content at the surface and less in the layersbelow) are sufficient to stabilize the tetragonal phase even foran oxidized film. We could also transform monoclinic ZrO films to the tetragonal phase via Y-doping (2 mol% Y O ).The films showed similar XPS core level shifts as pure ZrO films, with high binding energies caused by the positive elec-trostatic potential when the films carry net positive charge dueto a concentration of positive oxygen vacancies (V •• O ) exceed-ing the one introduced by the doping. Thin yttria-doped zirco-nia films provide a good basis for further studies of importantproperties of this material, such as phase stability under moreextreme conditions, reactivity, or diffusion of dopants as wellas impurities, which can be added intentionally in a controlledfashion. VI. ACKNOWLEDGEMENTS
This work was supported by the Austrian Science Fund(FWF) under project number F4505 (Functional Oxide Sur-faces and Interfaces – FOXSI). R. C. Garvie, R. H. Hannink, and R. T. Pascoe, “Ceramic steel?” Nature , 703 (1975). R. H. J. Hannink, P. M. Kelly, and B. C. Muddle, “Transformation tough-ening in zirconia-containing ceramics,” J. Am. Ceram. Soc. , 461–487(2000). N. Q. Minh, “Ceramic Fuel Cells,” J. Am. Ceram. Soc. , 563–588 (1993). J. B. Goodenough, “Oxide-ion electrolytes,” Annu. Rev. Mater. Res. ,91–128 (2003). E. H. Kisi and C. J. Howard, “Crystal structures of zirconia phases and theirinter-relation,” Key Eng. Mater. , 1–36 (1998). O. Ruff and F. Ebert, “Beiträge zur Keramik hochfeuerfester Stoffe. i. DieFormen des Zirkondioxyds,” Z. Anorg. Allg. Chem. , 19–41 (1929). T. Götsch, W. Wallisch, M. Stöger-Pollach, B. Klötzer, and S. Penner,“From zirconia to yttria: Sampling the YSZ phase diagram using sputter-deposited thin films,” AIP Adv. , 025119 (2016). For the relation between mol% Y O and Y concentration or Y/Zr ratio inthe film, see Ref. 7. H. G. Scott, “Phase relationships in the zirconia-yttria system,” J Mater Sci , 1527–1535 (1975). M. Chen, B. Hallstedt, and L. J. Gauckler, “Thermodynamic modeling ofthe ZrO –YO . system,” Solid State Ionics , 255–274 (2004). S. Shukla and S. Seal, “Mechanisms of room temperature metastable tetrag-onal phase stabilisation in zirconia,” Int. Mater. Rev. , 45–64 (2005). P. Lackner, Z. Zou, S. Mayr, U. Diebold, and M. Schmid, “Using photo-electron spectroscopy to observe oxygen spillover to zirconia,” Phys. Chem.Chem. Phys. , 17613–17620 (2019). V. Vonk, N. Khorshidi, A. Stierle, and H. Dosch, “Atomic structure andcomposition of the yttria-stabilized zirconia (111) surface,” Surf. Sci. ,69–76 (2013). G. S. a. M. Theunissen, A. J. A. Winnubst, and A. J. Burggraaf, “Segre-gation aspects in the ZrO –Y O ceramic system,” J. Mater. Sci. Lett. ,55–57 (1989). M. de Ridder, R. G. van Welzenis, A. W. D. van der Gon, H. H. Brongersma,S. Wulff, W.-F. Chu, and W. Weppner, “Subsurface segregation of yttria inyttria stabilized zirconia,” J. Appl. Phys. , 3056–3064 (2002). J. Nowotny, T. Bak, M. K. Nowotny, and C. C. Sorrell, “Charge transferat oxygen/zirconia interface at elevated temperatures: Part 3: Segregationinduced interface properties,” Adv. Appl. Ceram. , 165–173 (2005). G. Witz, V. Shklover, W. Steurer, S. Bachegowda, and H.-P. Bossmann,“Phase evolution in yttria-stabilized zirconia thermal barrier coatings stud-ied by Rietveld refinement of X-ray powder diffraction patterns,” Journal ofthe American Ceramic Society , 2935–2940 (2007). C.-C. Chao, J. S. Park, X. Tian, J. H. Shim, T. M. Gür, and F. B. Prinz, “En-hanced oxygen exchange on surface-engineered yttria-stabilized zirconia,”ACS Nano , 2186–2191 (2013). M. de Ridder, R. G. van Welzenis, H. H. Brongersma, and U. Kreissig,“Oxygen exchange and diffusion in the near surface of pure and modifiedyttria-stabilised zirconia,” Solid State Ionics , 67–77 (2003). J. Lahiri, A. Mayernick, S. L. Morrow, B. E. Koel, A. C. T. van Duin, M. J.Janik, and M. Batzill, “Modification of active sites on YSZ(111) by yttriasegregation,” J. Phys. Chem. C , 5990–5996 (2010). A. E. Hughes, “Segregation in single-crystal fully stabilized yttria-zirconia,” J. Am. Ceram. Soc. , 369–378 (1995). R. Chaim, D. G. Brandon, and A. H. Heuer, “A diffusional phase transfor-mation in ZrO2–4 wt% Y O induced by surface segregation,” Act. Metall. , 1933–1939 (1986). A. Bernasik, K. Kowalski, and A. Sadowski, “Surface segregation in yttria-stabilized zirconia by means of angle resolved X-ray photoelectron spec-troscopy,” J. Phys. Chem. Sol. , 233–239 (2002). A. D. Mayernick, M. Batzill, A. C. T. van Duin, and M. J. Janik, “A reac-tive force-field (ReaxFF) Monte Carlo study of surface enrichment and stepstructure on yttria-stabilized zirconia,” Surf. Sci. , 1438–1444 (2010). X. Wang, “Yttrium segregation and surface phases of yttria-stabilized zir-conia (111) surface,” Surf. Sci. , L5–L9 (2008). T. Götsch, E. Bertel, A. Menzel, M. Stöger-Pollach, and S. Penner, “Spec-troscopic investigation of the electronic structure of yttria-stabilized zirco-nia,” Phys. Rev. Materials , 035801 (2018). S. L. Morrow, T. Luttrell, A. Carter, and M. Batzill, “High temperaturescanning tunneling microscopy of purely ion conducting yttria stabilizedzirconia (YSZ),” Surface Science , L78–L81 (2009). P. Lackner, Z. Zou, S. Mayr, J.-I. J. Choi, U. Diebold, and M. Schmid,“Surface structures of zro films on Rh(111): From two layers to bulk ter-mination,” Surf. Sci. , 180–187 (2019). P. Lackner, J. I. J. Choi, U. Diebold, and M. Schmid, “Substoichiometricultrathin zirconia films cause strong metal–support interaction,” J. Mater.Chem. A , 24837–24846 (2019). P. Lackner, J. I. J. Choi, U. Diebold, and M. Schmid, “Construction andevaluation of an ultrahigh-vacuum-compatible sputter deposition source,”Rev. Sci. Instrum. , 103904 (2017). R. Koller, W. Bergermayer, G. Kresse, C. Konvicka, M. Schmid,J. Redinger, R. Podloucky, and P. Varga, “The structure of the oxygen-induced c ( × ) reconstruction of V(110),” Surf. Sci. , 16–28 (2002). L. Köhler, G. Kresse, M. Schmid, E. Lundgren, J. Gustafson, A. Mikkelsen,M. Borg, J. Yuhara, J. N. Andersen, M. Marsman, and P. Varga, “High-coverage oxygen structures on Rh(111): Adsorbate repulsion and site pref-erence is not enough,” Phys. Rev. Lett. , 266103 (2004). W. Smekal, W. S. M. Werner, and C. J. Powell, “Simulation of electronspectra for surface analysis (SESSA): a novel software tool for quantitativeAuger-electron spectroscopy and X-ray photoelectron spectroscopy,” Surf. Interface Anal. , 1059–1067 (2005). V. Swamy, H. J. Seifert, and F. Aldinger, “Thermodynamic properties ofY O phases and the yttrium–oxygen phase diagram,” J. Alloys Comp. ,201–207 (1998). M. Schmid and P. Varga, “Segregation and surface chemical ordering-anexperimental view on the atomic scale,” in
Surface Alloys and Alloy Sur-faces , The Chemical Physics of Solid Surfaces, Vol. Volume 10, edited byD. Woodruff (Elsevier, 2002) pp. 118–151. R. C. Garvie, “Stabilization of the tetragonal structure in zirconia micro-crystals,” J. Phys. Chem. , 218–224 (1978). A. Christensen and E. A. Carter, “First-principles study of the surfaces ofzirconia,” Phys. Rev. B , 8050–8064 (1998). H. Li, J.-I. J. Choi, W. Mayr-Schmölzer, C. Weilach, C. Rameshan, F. Mit-tendorfer, J. Redinger, M. Schmid, and G. Rupprechter, “Growth of anultrathin zirconia film on Pt Zr examined by high-resolution X-ray photo-electron spectroscopy, temperature-programmed desorption, scanning tun-neling microscopy, and density functional theory,” J. Phys. Chem. C119