1
Low-Dimensionality and Epitaxial Stabilization in Metal Supported Oxide Nanostructures: MnO on Pd(100) MnO
Cesare Franchini† and Francesco Allegretti‡
†Faculty of Physics, Computational Materials Physics, Universität Wien and Center for Computational Materials Science A-1090, Wien, Austria
‡Institute of Physics, Surface and Interface Physics, Karl-Franzens University Graz, A-8010 Graz, Austria
We present a survey of the growth and structure of manganese oxide nanolayers on a Pd(100) substrate, investigated in two different thickness regimes through a plethora of surface science techniques (STM, AFM, LEED, SPA-LEED, XPS, XAS, and HREELS) and state-of-the-art theoretical tools (DFT and hybrid DFT approaches). In the high thickness regime, above 10-15 monolayers, depending on the preparation conditions different films with specific growth direction and stoichiometry are formed. At low and intermediate pressure ( 110 mbar and 2-510 mbar, respectively) the oxide structures are already bulk-like in terms of their in-plane lattice constant and can be assigned to MnO(111) and MnO(100), respectively. At high pressures ( 510 mbar), MnO layers (001)-oriented are obtained by oxidation of MnO(100). To explore the epitaxial (geometric) relationships that favor the growth of the different oxide phases, we have investigated the atomistic details of different oxide/oxide and oxide/metal interfaces. In particular, we have addressed the issue of the stability of the MnO(001)/MnO(001) interface and determined the phase stability diagram of MnO/Pd(100) phases at a Mn coverage of about 1 ML. In the latter low thickness regime we have identified nine different two-dimensional (2D) phases, which are novel in terms of their structural and electronic properties. These nanophases can be classified according to similar building block units and described either as O-Mn-O MnO(111)-like trilayers or in terms of metal-deficient MnO(100)-like monolayers, and therefore we argue that they mediate the epitaxial growth of MnO thicker films on Pd(100) by providing structurally graded interfaces. Moreover, the formation of O or Mn vacancies drives the transition between 2D phases with similar structural units but different lattice periodicity, indicating that ion vacancies, mixed valence states and substoichiometry lie at the basis of the architectural flexibility in the monolayer regime. Interestingly, the latter concepts play a major role in the more complex class of functional oxides such as the manganites, of which binary manganese oxides are the simplest parent compounds.
10.1 Introduction
Since several decades, the study of transition metal oxides represents a field of active and intense research. Thanks to the uniquely rich spectrum of structural, electronic and magnetic properties, which cover the range from metallic behavior to magnetic insulators and encompass spectacular phenomena as superconductivity, charge ordering and colossal magnetoresistance, these compounds have attracted the interest of physicists, solid state chemists and materials scientists. Renewed interest has been fuelled by the advances in the synthesis processes of oxide materials, which now allow control over the structure and stoichiometry at the level of a unit cell. This has opened up new far-reaching perspectives for both fundamental studies and technological applications, in particular in the field of functional and multi-functional oxides, many of which incorporate in their structure one or more transition metal atom species. Functional oxide materials have unique physical and chemical properties, which can be suitably controlled and modified by means of external stimuli, such as changes of temperature and pressure, electric or magnetic fields and the adsorption of foreign atoms or molecules, thereby enabling the development of new functionalities. As a result, a wide range of intriguing phenomena can be observed, ranging from superconductivity to ferroelectricity, piezoelectricity and exotic magnetic behaviors. The ability to tailor and tune the properties of these functional materials thus guarantees a high potential for applications in micro- and nano-electronics, spintronics, heterogeneous catalysis, gas sensing, energy harvesting, etc.
In addition to these developments, novel properties can also arise by the scaling down of the dimensions of oxide-based devices. At the nanoscale, in fact, effects arising from the reduced dimensionality, the geometric confinement, and the proximity of surfaces and interfaces as well as effects arising from the coupling of the oxide to the supporting material (so-called substrate-induced effects, such as electronic hybridization, charge transfer and elastic strain) may convey new physical and chemical properties and therefore new functionalities to oxide systems. This potential is indeed reflected in the enormous impulse given to fundamental and applied research in the fields of nanoscience and nanotechnology. Among low-dimensional nanostructured oxide systems, a prominent role is played by transition metal oxide ultrathin films supported on metal surfaces. Just to name a few technological applications, they are typically employed as gate dielectrics and tunnelling barrier layers in conventional and novel electronic devices [9, 63], as protective layers in corrosion prevention and inhibition [42], as gas sensor materials [34, 71], as support surfaces in the field of heterogeneous catalysis [22]. In fundamental scientific studies, on the other hand, thin films of transition metal oxides grown on metal single crystal surfaces constitute preferred model systems for the elucidation of emerging phenomena at the atomistic scale. Not only the presence of the metal substrate allows circumventing charging problems arising from the insulating character of many oxides, but also it may lead - through the active participation in the elastic and electronic coupling - to the stabilization of novel hybrid systems, whose structural, electronic and magnetic properties bear no correspondence to those of bulk oxides [60].
With this general frame in mind, in this chapter we focus on the study of ultrathin manganese oxide films epitaxially grown on Pd(100), providing a brief review of a combined theoretical and experimental investigation performed by our research groups at the University of Vienna and at the Karl-Franzens University of Graz. All calculations presented in the present chapter have been performed using the Vienna ab initio Simulation Package (VASP) [46, 47] in the framework of density functional theory (DFT) [45] and hybrid DFT [40].
Due to the ability of manganese to assume different oxidation states, ranging from +2 to +7 [39], Mn oxides in bulk compounds exhibit a number of stoichiometries with a complex phase diagram [28]. Their architecture is dictated by the ability of the Mn/O complexes to assemble by corner-sharing, edge-sharing or double corner-sharing, such that more than thirty Mn oxide mineral phases occur in nature [64]. As a consequence, Mn oxides display a richness of behaviors which render them attractive systems in applications of heterogeneous catalysis [10] (for example as electro-catalysts in fuel cells [72, 77]) as well as in applications as electrode materials in solid state batteries [8] and in the environmental waste treatment [21, 75]. Moreover, Mn oxides are parent compounds of a particular class of functional oxides, the manganites, the properties of which are determined by the complex interplay among spin, orbital and lattice degrees of freedom, which results in outstanding phenomena [66] such as giant magnetoresistance [76], metal-insulator transitions and orbital orderings[41]. These spectacular properties promise to find application in the fields of magnetic recording and novel spintronic devices. As to the study of Mn oxides in reduced dimensions, an increasing effort has been devoted in the last decade, which has revealed the richness of unusual behaviors and an even higher degree of complexity relative to bulk Mn oxides. For example, the ferromagnetic behavior of small MnO nanoparticles, anomalous with respect to the observed antiferromagnetic ordering in the bulk phase, has been first predicted by DFT [59] and then confirmed by subsequent experiments [50]. Recently, ferromagnetism of MnO and MnO nanowires has also been observed [56], whereas unconventional exchange bias interaction has been reported for oxide coated Mn nanoparticles with a MnO shell [73], the latter effect being attributed to an unusual spin alignment sequence at the interface.
To understand the unconventional properties of low-dimensional Mn oxides, fundamental studies on model systems are desirable, which require a detailed knowledge of the oxide structures at the atomistic level in the effort to elucidate the structure-properties relationship. So far, these studies on model systems have mainly focused on the growth of MnO thin films on the Ag(001) surface. Due to the relatively large in-plane lattice constant of MnO(001) (=3.14 Å), the template has been chosen to ensure a reasonable overlayer-substrate matching (=2.89 Å), which is expected to favor better epitaxial growth. Indeed, despite the still large mismatch (about 9%) films of good quality have been obtained [55, 74]. Further investigations cast light on the evolution of the structural and electronic properties with the oxide thickness due to the partial release of the epitaxial strain [57, 58, 19]. However, no detailed studies of the oxide-Ag(001) interface in terms of the Mn oxide phases formed at coverage below or about 1 monolayer (ML) are available in the literature. Interface-stabilized MnO monolayers have instead been reported on two different substrates, Rh(100) [61] and Pt(111) [38], but the literature is still sparse and the unambiguous structural assignment of the oxide phases has not been yet accomplished. In this context, we have recently performed a joint experimental-theoretical investigation aimed to extensively characterize the growth and structure of MnO ultrathin films on a Pd(100) substrate. The film thickness range extended from a few Å, corresponding to the interface-stabilized phases of the monolayer regime, up to 30-50 Å, where bulk-like behavior sets in. The detailed theory-experiment comparison for the Mn oxide systems, which will be presented in the next section, proved a decisive factor for the unambiguous assignment of the oxide phases, and it enabled to identify the interface-stabilized oxide structures that mediate the epitaxial growth. Moreover, it proved highly beneficial, in that it allowed assessing the reliability of density functional theory (DFT) and post-DFT approaches applied to the Mn oxides. This is a very important issue in the light of the well-known difficulties encountered by standard DFT in dealing with strongly correlated materials [1].
10.2 Growth of MnOLayers on Pd(100)
The growth of Mn oxides on Pd(100) has been the subject of a series of recent papers [3, 16, 4, 51, 26, 27], which explore in a rather systematic way the phase stability diagram at different oxide thicknesses. In particular, it has been shown that below 1 ML a complex surface phase diagram with a multitude of novel Mn oxide structures develops, the oxide structures being characterized by specific structural building blocks and vibrational properties [51, 26, 27]. In the high thickness regime, upon deposition of 20-30 ML, MnO(100) films exhibiting good long range order can be grown epitaxially [3], despite the considerable lattice mismatch with the substrate (14%). The MnO(100) structure can be preferentially converted either into MnO(111) [3, 4] or MnO(001) [16] depending on the combination of temperature and applied oxygen pressure. The evolution of the physical properties of the manganese oxide layers on Pd(100) with respect to the changes in film thickness provides a unique method to separate surface (2D) and bulk (3D) effects. In fact, to a decrease of film thicknesses from the high coverage (multilayer) to the low coverage ((sub)monolayer) regime corresponds an increase of the surface to bulk ratio, and low dimensional effects become more distinguishable. The fundamental characteristics of the MnO/Pd(100) system are discussed in details in Sec. 2.1 and 2.2 and can be summarized as follows:
MnO/Pd(100)
Low coverage regime, 0.75 monolayer (see Sec. 2.1). Ultrathin layers of variable MnO stoichiometry, only 1-2 ML thick, can be formed at the metal-MnO interface. At least nine different 2D MnO phases on Pd(100) are found, which are novel in terms of the known Mn oxide bulk crystal structures and which are stabilized by the metal-oxide interface and by the confinement in the direction perpendicular to the surface. These low-dimensional interface-stabilized phases may mediate the epitaxial growth of thicker layers by providing structurally graded interfaces.
High coverage regime, 20-30 layers (see Sec. 2.2). MnO(100) with bulk-like in-plane lattice constant is stable in a wide range of pressure and temperature. Through the appropriate tuning of temperature and oxygen pressure the MnO(100) films can be transformed into MnO(111) (annealing at elevated temperatures or reactive evaporation at lower oxygen pressures [ mbar]) or transformed into MnO(001) surfaces (high temperature oxidation at relatively high oxygen pressure [ mbar]).
10.2.1 Low Coverage Regime
The experimental phase stability diagram of Mn oxides on Pd(100) below 1ML is depicted schematically in Fig. 1, where the various Mn oxide nanolayer phases are ordered as a function of the oxygen pressure and of the oxygen chemical potential during the preparation procedure, and they are represented by their corresponding scanning tunneling microscopy (STM) profile.
Fig. 1 Experimental schematic phase stability diagram of the interfacial Mn oxides, presented as a function of the oxygen pressure and of the oxygen chemical potential . The nominal coverage of Mn on Pd(100) is 0.75 ML. From Ref. [51] with permission
The manifold Mn oxide phase diagram comprises nine different nanophases which are characterized by specific windows in the parameter space of the thermodynamic variables temperature (in the total range: 600-800 K) and oxygen pressure (in the total range: 510-510 mbar). We distinguish:
1. The oxygen-rich regime (510 mbar > < 510 mbar)
(a) Two hexagonal phases (HEX-I and HEX-II), which are both obtained at high pressures
2. The oxygen-intermediate regime (510 mbar > < 110 mbar)
(a) A c(42) structure and a stripe phase described as a uni-axially compressed c(42), which are both stabilized at intermediate pressures
(b) Two structures which were called chevrons (CHEV-I and CHEV-II), because of their STM appearance
3. The oxygen-poor regime: (110 mbar > < 510 mbar)
(a) Two reduced phases with complex structures, named waves and labyrinth
(b) At the most reducing conditions, a third hexagonal phase (HEX-III), commensurate with the Pd(100) substrate (2) along one of the two directions.
The extraordinary architectural flexibility of the interfacial Mn oxides on Pd(100) and the electronic properties of the novel 2D phases can be rationalized and understood through a synergic combination of experimental [STM, low energy electron diffraction (LEED), high-resolution electron energy loss spectroscopy (HREELS) and X-ray photoemission spectroscopy (XPS)] and computational (DFT and hybrid functionals) techniques. Indeed, in modern surface science theoretical calculations have become an efficient complementing tool to the experimental observations. Theoretical models based on educated guesses of possible structures can be tested and directly compared with the experiments in order to clarify the structural aspects and provide an atomistic interpretation of the measured properties. In the present case, the joint experimental-theoretical analysis reveals that the two oxygen-rich phases (HEX-I and HEX-II) can be described in terms of O-Mn-O trilayers with MnO(111)-like structures (see Sec.2.1.1), whereas the intermediate oxygen regime (c(42) and chevrons) is based on a compressed MnO(100)-like monolayer model (see Sec.2.1.2). At low oxygen chemical potentials the waves and labyrinth structures show in STM very complex unit cells whose link to the other phases is less clear but seems still to be related to a MnO(100)-like wetting layer, while the additional hexagonal phase HEX-III is of uncertain attribution (see Sec.2.1.3).
Fig. 2 HREELS phonon spectra of the four Mn oxide submonolayer phases at high and intermediate oxygen partial pressures. From the top to the bottom: HEX-I, HEX-II, c(42) and CHEV-I. The statistical uncertainty in the peak position is 0.5 meV. For every Mn oxide phase, measurements performed on samples freshly prepared in different days agree within 1 meV. Adopted from Ref. [51]
The recurrent MnO(100)- and MnO(111)-like structural features are reflected in the phonon-loss spectrum displayed in Fig. 3(b)@. As expected, the common building blocks shared between the different oxide phases result in similar phonon losses. All spectra exhibit a clear single peak structure. The phonon loss is centered at around 70 meV for the two HEX-I and HEX-II phases and shifts down to 44-45 meV for the c(42) and CHEV-I phases. Interestingly, the phonon spectra (not shown) of the CHEV-II, waves and HEX-III phases, which are obtained by further lowering the O chemical potential, are also characterized by a single peak at 43-45 meV. These findings provide clear indication of two distinct regimes, a MnO(111)-like regime comprising phases with a higher energy phonon loss (70 meV) and a MnO(100)-like regime with a single phonon loss around 44 meV.
10.2.1.1 MnO(111)-like Phases (Oxygen-Rich Regime)
In this pressure regime two Mn oxide phases have been detected, HEX-I and HEX-II, which are characterized by hexagonal or quasi-hexagonal (i.e. distorted) symmetry linked to that of MnO(111), and which display a phonon spectrum with a single loss peak centered at around 70 meV.
HEX-I
For the sake of clarity, it is instructive to recall the structure of MnO(111), which is made out of alternating O and Mn hexagonal layers with ABC stacking sequence, each layer having in-plane lattice constant =3.14 Å [see Fig.(a)].
Fig. 3 (a) Real and (b) reciprocal lattice for a (11) MnO(111)-like hexagonal structure on a Pd(100) surface. The lattice parameter of the overlayer is assumed to be the measured bulk value = 3.14 Å, whereas =2.75 Å. In (a) only one of the two symmetry domains is reported for clarity. (c) SPA-LEED two-dimensional pattern measured at E=90 eV, for the undistorted HEX-I phase. (d) LEED pattern of the distorted HEX-I phase, recorded at E=96 eV. (e) Real and (f) reciprocal lattice associated with the distorted hexagonal MnO(111)-like phase. Only one symmetry domain is shown in (e) for clarity, for which the lattice parameter in the direction of the distortion ([011]) is =2.94 Å. In total, four symmetry domains contribute to the reciprocal lattice in (f): two are obtained from panel (e) with at either +60 or -60 from , and the other two are obtained by rotating by 90 relative to the substrate mesh. (g) Hard sphere model simulating in the real space the moiré pattern that originates from the interference of the quasi-hexagonal oxide lattice of panel (e) with the square mesh of the Pd(100) substrate. Figure adapted from Ref. [26] and Ref. [51] with permission