An interfactant for metal oxide heteroepitaxy: Growth of dispersed ZrO2(111) films on FeO(111) precovered Ru(0001)

We introduce an interfactant for zirconium oxide heteroepitaxy and propose that dispersed, epitaxial films of materials which are hard to crystallize are accessible by growing them on top of an ultrathin, polar FeO(111) film.

Metal-ceramic interfaces are of great interest for a large variety of technological applications.1

Oxide coatings serve as thermal barriers or corrosion protectors for high-quality alloy or metal materials, e.g., in jet engines.2

Fibre-matrix composites, sensor technology, light bulbs, and medical implants depend on the interaction at a metal-ceramic interface.

Heterogeneous catalysts consist often of metals in contact with oxides and the activity depends on the interaction between them.3

Ultrathin films and self-assembled nanostructures are currently attracting special interest for the development of electronic nanodevices and associated technologies4–7 since they often exhibit novel properties such as the giant magnetoresistance (GMR) effect.4,8

Zirconia is a material which is of interest for many applications in solid electrolyte fuel cells, in electrolytical cells for sensor and pump devices, in optical coatings, or in catalysis.9

The most stable modification of ZrO2 at room temperature is monoclinic.

However, for small particles or thin films or with dopants like Y2O3, the high-temperature tetragonal or cubic phases may be stabilized.10–13

So far, the surface chemistry of sulfated zirconia has not been investigated in detail, and suitable thin film systems accessible to electron-based surface science techniques have rarely been developed.14

For most applications, the growth of closed films is essential.

However, due to the energy balance Δγ = γf + γI – γSof the surface free energies of the substrate γS, the growing film γf and the interface energy γI,15 most metals form compact particles on oxides (Δγ > 0) whereas oxide films mostly are dispersed on metal substrates (Δγ < 0).

However, some oxides do not wet due to a high interface free energy caused by lattice mismatch.

Especially dispersed crystalline zirconia films are very difficult to obtain.

STM studies on Ag(100) showed poor dispersioned and no diffraction patterns were obtained.16

Also on Ru(0001), we were not able to obtain long-range ordered, epitaxial zirconia films by direct deposition of Zr on Ru(0001) in oxygen atmosphere or by subsequent oxidation.

The difficulties arise from the tremendous lattice mismatch with all transition metals due to the large lattice constant of ZrO2 which amounts to ∼3.6 Å for the cubic closed packed (111) and for the most stable tetragonal and monoclinic terminations.17

But epitaxial zirconia films have been grown on Pt(111)18 and Au(111)19 substrates despite a lattice mismatch of 25% and 30%, respectively.

Both studies showed broad (2 × 2) LEED reflexes of defective, cubic zirconia with limited long-range order.

An STM investigation of zirconia films on Pt(111) revealed a hillock morphology and film decomposition and dewetting beyond 1000 K.20

Deposition at 470 K yields significantly enhanced dispersion but beyond 1000 K the films decompose and become discontinuous.

Another problem is the high tendency for alloying of Zr with all metals which was also observed in the studies on Au(111) and Pt(111) after anneling to temperatures above 900 K.

Thus, the difficulty to obtain dispersed, long-range ordered zirconia films arises mainly from the generally large lattice mismatch and from the high tendency for alloying.

The first point leads to a high interface energy and a low dispersion in order to minimize the interface contact area.

In order to circumvent these problems, one may introduce a suitable third component like surfactants21 which may lower the surface or interface energy as recently demonstrated for the growth of Co films on a hydroxylated alumina interface.22

Alternatively, a suitable interface layer (or ‘interfactant’) between the substrate and the growing film may act as a diffusion barrier and may relieve misfit stress.23,24

Such interfactants have rarely been used so far.

A buffer layer may distribute the strain in the growing film.25

Or surface oxidation may generate a diffusion barrier between the substrate and a deposited metal film.26

A recent letter reports that Pd wets an ultrathin FeO(111) layer.

Even the structure of the Pd–FeO interface was determined.27

The growth and structure of FeO(111) films on Pt and Ru substrates has been studied in detail28–30 because such films are interesting for magnetic studies of ferrites31 or as model catalysts.32,33

On several substrates, thermally stable, ultrathin FeO(111) films can be grown which have a strong in-plane lattice expansion in the initial growth stages.29,34–37

On Ru(0001), FeO(111) films grow with a thickness up to 4 ML and a lattice constant expansion of up to 3.58 Å (bulk: 3.04 Å).37

This gave us the idea to use an FeO(111) precovered metal substrate as a substrate system for the growth of materials which are difficult to grow epitaxially in single-crystalline quality.

The growth of zirconium oxide on an oxide precovered metal involves two interfaces and requires additional consideration of the oxide–oxide interface.

For oxide on oxide growth, the epitaxial growth mode is mainly determined by the interfacial energies arising from the lattice mismatch of the oxygen sublattices.25,38,39

In general, the oxide stoichiometry is determined by the oxygen gas pressure.40

As epitaxial growth always takes place under nonequilibrium conditions, the oxidation kinetics is of high importance for the oxide phase and its crystallographic quality.25

For ultrathin oxide films, the growing phase may deviate from the expected bulk phases and interface-stabilized phases may form.30,41,42

From the energy balance, eqn. (1), it follows that dispersed multilayer systems are only accessible when for all layers the surface free energy of the deposited film is lower than that of the underlying material, otherwise clustering is anticipated.

In heteroepitaxy of mismatched materials, the situation is even worse due to significant contributions of the interface free energy.

Using a dense FeO(111) layer as an interfactant offers a way to circumvent these problems: The lattice constant of thin FeO(111) films on Ru(0001) expands considerably as it adjusts to the specific interface energetics,29,37 and fits significantly better with the lattice constant of zirconia films (3.59 Å).

Also on Pt(111), the FeO(111) lattice constant is expanded to values between 3.09 to ∼3.40 Å (depending on the coverage),35 and such a lattice constant expansion very likely applies to other metal substrates.

The stacking sequence of the FeO bilayers is not unambiguously clear on Ru(0001), but very likely, these films are oxygen-terminated as discussed in .ref. 29

This terminating oxygen layer provides a close packed oxygen matrix upon which the zirconium can be deposited and further closed packed oxygen layers with interstitial zirconium cations can grow in the cubic calcium fluoride type structure.

The chemical interaction of the oxygen termination may spread the zirconia film onto the substrate system so that a wetting ZrO2 film may be obtained.

As ZrO2 consists of alternating negatively and positively charged layers, dipole moments exist perpendicular to the surface and according to the classification introduced by Tasker43 such a type III polar surface can only be stabilized when charge compensated repeat units can be formulated so that the total dipole potential across the film vanishes, as shown schematically in Fig. 1 for the cubic phase along (111).

As emphasized by Noguera, the interface condition might be essential for the stabilization of polar structures,44,45 and specifically charge compensated repeat units of zirconia require the presence of an oxygen interface termination of the zirconia film to the substrate which is provided by the oxygen-terminated FeO(111) film.

At the same time, the polar FeO(111) film may be stabilized in this way as has been treated theoretically for metal deposits on the polar, isostructural MgO(111) where a strong interfacial electron redistribution leads to the stabilization of polar oxide surfaces by metal deposits.46

This mechanism has also been suggested for the wetting of Pd on FeO(111).27

The theoretical study by Goniakowski and Noguera also revealed that the adhesion energy for Zr on the isostructural MgO(111) is much higher on the close-packed O atoms and a much better wetting can be expected.46

Also for α-Al2O3(0001) a much stronger interaction with an oxygen termination is calculated.12

And for growth on Ni(111) a Ni–O surface termination with O–Zr–O repeat units on top is the energetically best configuration.13

Note that NiO(111) is isostructural with FeO(111).

Thus, the dense oxygen layer of the substrate system is beneficial for an electrostatic stabilization of the FeO(111) interface layer as well as the dispersed ZrO2(111) film.

The chemical interaction with this buffer layer spreads the zirconia film on the substrate while at the same time alloying with the substrate may be suppressed.

All our attempts to form a closed ZrO2 film with reasonable long-range order by direct deposition of Zr on clean Ru(0001) in oxygen atmosphere or by subsequent oxidation failed.

The LEED patterns showed broadened substrate spots, no adlayer spots and an increased diffuse background.

It was not tried to obtain STM images.

Therefore, we have grown zirconia films on FeO(111) precovered Ru(0001).

The experiments were performed in an ultra-high vacuum (UHV) chamber which is equipped with a scanning tunneling microscope (STM, Burleigh), a backview LEED optics with integrated Auger spectrometer (Omicron), an Ar+-sputter gun, two metal evaporators, ion gauges, and gas inlet systems.

Iron metal was evaporated from an iron wire wrapped around a resistively heated tungsten wire.

Zirconium metal was evaporated by electron bombardment from a zirconium wire.

All STM measurements were performed at room temperature in the constant current mode.

The Ru(0001) single crystal was cleaned by cycles of Ar+-sputtering and annealing to 1450 K. Onto the clean substrate (as monitored by Auger and LEED), about 1 ML of Fe was deposited and oxidized in 10–6 mbar O2 at 870 K for a few minutes with a final anneal to 1000 K. Zirconia films are prepared by deposition of Zr and subsequent oxidation in 10–6 mbar at 800 K for ∼15 min and finally heating to 1150 K with 10 K s–1 and immediate cooling in 10–6 mbar O2.

FeO(111) monolayer films on Ru(0001) produce a characteristic LEED pattern (Fig. 2a).

Details of FeO(111) films grown on Ru(0001) are presented elsewhere.29,37

This film with a lattice constant of 3.08 Å wets the Ru(0001) substrate (except for some cracks and vacancy islands, Fig. 2b).

It exhibits a hexagonal Moiré pattern with a 21.6 Å unit cell which results from coincidence of 7 FeO units with 8 Ru atoms.

In general, FeO(111) films on Ru(0001) are very well ordered with the atomic rows of FeO(111) aligned with those of the Ru(0001) substrate.

After deposition of Zr and subsequent oxidation on such an FeO(111) film and after heating to 1150 K with 10 K s–1 in 10–6 mbar O2, we obtained a LEED pattern as shown in Fig. 3a.

A (2 × 2) structure with respect to the Ru(0001) substrate from O/Ru(0001) is observed, and additionally a new hexagonal pattern with a lattice constant of about 6.9 Å occurs.

This pattern is similar to the (2 × 2) structures observed from cubic ZrO2(111) on Pt(111) and Au(111), but the lattice constant is reduced by ∼4% compared to cubic bulk ZrO2(111) (3.59 Å, i.e. 7.18 Å for a (2 × 2) structure).

LEED spots are much sharper than the broad spots reported in previous studies18–20 indicating much better long-range order.

The Auger spectrum shows reduced Ru signals, and signals of Zr and O (Fig. 4a).

Iron signals are almost absent.

Large scale STM images of such a film show characteristic round-edged holes with a depth of about 6–20 Å and an uniform diameter of ∼500–700 Å on otherwise very flat films (Fig. 3b).

The holes are multiples of 3 Å deep which fits well with the step heights of energetically equivalent surface terminations of cubic ZrO2(111) (d = 2.9 Å).

From the depth of the holes we can conclude that the averaged film thickness is about 6 Å.

Fig. 3e shows the distribution of the hole depth and gives an idea of the film thickness and homogeneity.

It was not possible to achieve atomic resolution on these films which may be due to their insulating character which requires high bias voltages.

We attribute one of the two superimposed LEED patterns to the surface region of the thick film and the other to the holes.

There, the well-known O/Ru(0001)-(2 × 2) structure may have formed,47 which we always observe on FeO-uncovered Ru terraces after growth of submonolayer FeO(111) films.29

The presence of this pattern shows that the underlying FeO(111) film underwent restructuring during annealing induced by the overlayer zirconia film leading to FeO-uncovered hole regions.

Thus, in the holes no FeO(111) buffer layer is present, and this indicates that the presence of the FeO(111) film is beneficial for the adhesion of the zirconia film while the O/Ru(0001)-(2 × 2) structure is not.

We can only speculate about the origin of the hole formation: It may be due to diffusion of FeO into the substrate leaving a submonolayer FeO(111) film with uncovered Ru regions.

Alternatively, a restructuring of the FeO(111) monolayer film may occur, since previous studies on Pt and Ru showed that the lattice constant of these films varies strongly as it adjusts to the electrostatic, thermodynamic and interfacial energy gain associated with the interaction at the specific interface.29

Reactive restructuring leading to hole formation has recently been observed during deposition of oxide films on other substrate systems, e.g., during the growth of MgO films on Ag(001)48 and NiO films on Cu(111).49

Next, we investigated whether the tendency for alloying is reduced in these films, i.e., whether still an FeO-associated “buffer layer” is present at the interface.

For this, the ZrO2(111) film was annealed to temperatures where previously alloying with the substrate occured.

The Auger intensities of Zr (Fig. 4b) and the ZrO2(111)-(2 × 2) and Ru(0001)-(2 × 2) LEED patterns get slightly weaker during prolonged annealing while the intensity of the Ru(0001)-(1 × 1) pattern gets more and more intense.

After ∼60 min annealing to 1200 K with or without oxygen we observed the LEED pattern shown in Fig. 3c.

It still shows the O/Ru(0001)-(2 × 2) pattern, very weak (10) spots from FeO(111), and some new rotated diffraction spots corresponding to a real space lattice constant of about 3.08 Å and thus a lattice constant as the initial ultrathin FeO(111) film, however, rotated by 15°.

Upon further annealing, this LEED pattern vanishes but still very weak unrotated main diffraction spots from FeO(111) and ZrO2(111)-(2 × 2) are seen.

Most studies on metal substrates showed that decomposition of ZrOx films starts beyond 900–1100 K, and we also observe an increase of the Ru Auger signal intensity.

However, even after 2 h annealing at 1200 K, STM shows that the morphology of these films has not changed at all (compare Fig. 3b with Fig. 3d) indicating a stable film structure and that decomposition mainly occurs via desorption of Zr and O. Although a direct chemical evidence for the presence of FeO at the interface is not available with our methods (Fe Auger signal intensities are too weak below the zirconia film), the observation of a lattice constant which matches a FeO(111) film even after annealing for long periods suggests that this layer separates the Ru substrate from the zirconia film.

The FeO(111) buffer layer seems to “float” between the zirconia and Ru.

However, we cannot exclude that decomposition and diffusion of the iron oxide may have occurred.

Such diffusion has been observed in an inverse Au(111)/ZrOx/FeO layer system.50

The new, ordered LEED structure may then be attributed to an ordered Ru/ZrOx interface structure as has been proposed to occur for zirconia films on Pt(111) after annealing beyond 1100 K.18

In this study, we have introduced a new interfactant for heteroepitaxial metal oxide growth by precovering a metal substrate with a monolayer FeO(111) film.

In this way, we have obtained dispersed, cubic ZrO2(111) films with improved long-range order compared to previous studies.

An effective interfactant requires (I) that the rigidity of the substrate is reduced, so that strain relieve can lead to a better epitaxy, (II) that it acts as a diffusion barrier, while on the other hand (III) the interaction with the substrate and also with the growing film has to be strong in order to achieve wetting of all layers.

In other words: An interfactant has to behave like a ‘glue’ sticking together the substrate and the film, a condition which is perfectly fulfilled for wetting, ‘floating’ FeO(111) films with their variable lattice constant adjustment and their strong adhesiveness which is due to a high tendency to reduce the polarity by an interfacial electron redistribution.

The development of such multi-layer oxide systems is highly relevant in all fields, where thin closed oxide layers are required.

We are aware that the presented study is preliminary and quantitative structural information by surface crystallography or electron microscopy are appreciated to verify the presented hypotheses, especially for the presence of the FeO(111) interface layer after deposition of zirconia and annealing.

However, even if Fe diffusion occurs in the zirconia film, this is no drawback as promotion of sulfated zirconia with Fe and Mn enhances the catalytic activity,51 and in this respect, the described substrate system may be an ideal candidate for model catalytic studies.

Such dissolution behaviour can be regarded as being due to a very low or even negative interface free energy,52 and thus from the point of view of the total energy balance, eqn. (1), would enhance the dispersion of the zirconia layer.

The novel concept of interfactant behavior is extended in this paper to oxide-on-oxide heteroepitaxy.

It presents an explanation for the unexpected wetting behavior on top of the polar FeO(111) interface layer.