Emergent Metal-Insulator Transitions Associated with Electronic Inhomogeneities In

Emergent Metal-Insulator Transitions Associated with Electronic Inhomogeneities In

1

Emergent Metal-Insulator Transitions Associated with Electronic Inhomogeneities in Low-Dimensional Complex Oxides

An-Ping Li† and Thomas Z. Ward‡

†Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

‡Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

A prominent feature of complex oxides is the coexistence of competing electronic phases. The separation of metallic and insulating phases are believed to be responsible for a variety of emergent transport phenomena, including quantum criticality in ruthenates and colossal magnetoresistance in manganites. Interestingly, the phase boundaries between neighboring phases can often be displaced by small perturbations such as chemical doping, heating, stress, and electric or magnetic field, leading to intriguing metal-insulator transitions (MIT). The association of the emergent MIT with electronic inhomogeneities is particularly pronounced in low-dimensional materials which are uniquely suited to studying the MIT and phase evolutions in response to modification of the order parameters. Here, we present a few examples to illustrate the intimate interplay between emergent MIT and the competing electronic phases in functional metal oxide materials, including a percolative MIT near the critical temperature of the Mott transition in a Mn-doped bilayer ruthenate Sr3Ru2O7 crystal surface, and the abrupt conductance changes and reemergent MIT in manganite nanowires of La5/8-xPrxCa3/8MnO3. This experimental research has benefited from new developments in the fabrication and characterization of low-dimensional oxide materials and nanostructures. A rare glimpse of the microscopic phase separation, the dynamic phase percolation, and the strain-tuned MIT has been provided. The results indicate the critical role of electron-lattice interactions in phase separation and suggest that the origin of phase coexistence is much more strongly influenced by strain than local chemical inhomogeneity, both for ruthenates and manganites.

4.1. Introduction

It is becoming increasingly clear that the exotic properties displayed by correlated electronic materials such as colossal magnetoresistance (CMR) in manganites, high-Tc superconductivity in cuprates, and non Fermi liquid behavior in heavy-fermion compounds are intimately related to the coexistence of competing nearly degenerate states which couple with active degrees of freedom—charge, lattice, orbital, and spin states [1, 2]. The striking phenomena associated with these materials are due in large part to spatial electronic inhomogeneities or nanoscale phase separation [3]. In many of these hard materials, the functionality is a result of the soft electronic components that lead to self-organization [4]. These materials require a new approach to their study. With so many layers of electronic correlation driving the observable self-organized properties, it is vital to find methods which will allow for single ingredients to be tuned independently while at the same time offering the ability to recognize small local shifts that drive the larger macroscopic properties. By using low-dimensional materials and multi-length scale transport probing techniques, it is possible to isolate individual contributions from specific interactions through their inherent length scales. In this chapter, we will discuss several novel approaches on complex oxides that are helping to untangle these interactions.

While much work has been done on colossal electroresistance, colossal magnetoresistance, high Tc superconductivity and the metal-insulator transition in complex materials; there is yet no known theoretical model that is capable of fully explaining any one of these behaviors let alone a unifying understanding capable of explaining the effects of complexity on emergent behavior as a whole. The one thing that many of the materials exhibiting these properties share is electronic phase separation (PS) [5]. For this reason, a better understanding of electronic phase separation will have far reaching implications across a wide range of materials. Many systems with correlated electrons, such as manganites, ruthenates, cobaltates and cuprates, have been shown to contain inherent electronic inhomogeneity near a phase transition. The fact that electronic phase separation is present in so many materials of such varying character raises many questions as to its origin and its role in correlated electronic systems.

A particularly interesting approach is to use spatial confinement techniques on complex oxides. In transport measurements on unconfined systems where device size is larger than the inherent electronic phase domains, current bypasses regions of high resistance in favor of regions with lower resistance, because the probing electrons will follow the path of least resistance. By confining complex materials exhibiting emergent phenomena to length scales smaller than the electronic phase domains that reside within them, it is possible to simultaneously probe multiple resistive regions [Figure 1a]. This method allows for a much more complete view of the phases residing in a material and gives vital information on phase formation, movement and fluctuation. Since these phase separated regions also posses varied properties, this technique promises to lead to unexpected functionalities for future device applications.

Fig.1 (a) Diagram of effects of confinement on transport in a phase separated material. (b) Electronic correlation lengths offer new length scales beyond statistical and quantum limits to exploit

Similarly, spatial confinement and local probes should make it possible to probe the delicate interplay between spin, charge, orbital and lattice interactions that drive the mechanisms governing emergent physical phenomena in correlated systems. Each layer of interaction that contributes to the observable self organized behaviors is thought to carry with it a unique correlation length scale [6-8]. By confining these materials and locating characterization probes across a spectrum of lengths, it should be possible to create systems in which individual contributions from specific interactions can be manipulated through their correlation lengths [Figure 1b]. These studies will be invaluable in unraveling these systems since varied correlation groupings can be experimentally produced through doping, strain and degree of confinement.

4.2. Experimental Approach

4.2.1. Fabrication of Spatially Confined Oxide Nanostructures:

While there are several methods to directly write confined structure using scanning tunneling microscopes [9], atomic force microscopes [10] and self-assembly during growth [11, 12], it is often desirable to start with a macroscopic sample and etch to a desired geometry. This allows for measurements to be taken at macro-scales and repeated through successive confinements. There are several methods at our disposal to spatially confine thin films. These methods include: focused ion beam milling (FIB), atomic force microscopy (AFM) lithography, electron beam lithography (e-beam), direct laser writing, and optical lithography.

The FIB and AFM methods mechanically remove atoms and do not require a mask or photoresist layer to be put down[10, 13]. AFM lithography uses direct probe tip contact to mechanically remove material. This technique can shape down to 10 nm; however it is very slow, and there is a danger of disrupting the crystalline structure of the film. FIB milling is based on a system very similar to SEM; but, instead of using focused electrons, it uses a beam of focused ions with a high enough energy to etch the film from the substrate. Though this method is relatively slow it has a resolution of 10’s of nm [Figure 2] but carries with it the danger of ion contamination[14].

Fig.2 SEM images of (a) 1m x 5m x 50nm wire created using the FIB method, (b) and (c) show a comparison of edge roughness on 1m wide wires using FIB milling and wet-etch photolithography respectively

E-beam and laser writing are similar to traditional wide beam photoresistive lithography in their etching procedure but do not require a mask during resist development which increases the flexibility to quickly create novel geometries without the need to fabricate new masks while creating structures of 10-100 nm. However the exposure times can be considerable longer due to the need to finely scan the active beam[15, 16].

The most widely used and fastest method to create confined structures is photolithography [Figure 3]. In this method, a film is grown on a lattice matched or slightly mismatched substrate. A layer of photoresist is applied to the film. This layer is reactive to UV light. A metallic mask having the desired geometry is placed on top of the photoresist layer and exposed to a UV light source. This creates a patterned photoresist layer on the film. The structure is then placed in an etching solution. The etchant removes the exposed film while leaving the film under the photoresist layer. The remaining photoresist layer is then removed using a solvent and leaves the confined structure without damage. This method is only capable of creating structures larger than about 500 nm and limits the geometry to the predefined parameters of the mask used. To create new geometries, a new mask must be made.

Fig.3 Diagram of etching procedure steps and optical image of two confined etched from single thin film having constriction wires of 10m x 50m x 50nm and 1m x 5m x 50nm with gold contact pads for 4-contact transport.

The methods discussed each have strengths and weaknesses that must be matched to the material system under consideration. For the study of very long range interactions in a material such as PS manganites, photolithography is a very attractive and useful method; however as correlation lengths and organization levels of interest shrink to nanometer scales, FIB milling and e-beam lithography become the only methods capable of allowing for consecutive scaling transport measurements from a single sample.

4.2.2. Cryogenic Four-Probe STM:

Measuring electron transport behaviors in nanostructured materials presents a significant challenge to experimental physicists. Conventional transport electrodes and probes are very invasive; namely, they change what we are trying to measure. For a large conductor, the probes only represent a minor perturbation. But for a small conductor, especially at the nanoscale, the probes can very well be the dominant source of scattering. Therefore, using weakly coupled scanning tunneling probes to detect transport phenomena around individual scatterers appears to be the best approach [17]. The advantage of using the four-point method is that it separates the current supplying electrodes from the voltage-probing electrodes and thus can eliminate the parasitic resistance introduced by the probes. Furthermore, for observing quantum behavior of electrons, it is often necessary for transport measurements to be carried out in the nanometer scale at cryogenic temperatures so that electronic excitation processes can be controlled and studied in such nanosystems.

Fig.4 (a) Photo of ORNL quadraprobe STM system: A) a four-probe STM, B) an SEM column C) a liquid helium Dewar, D) an electron energy analyzer for SAM, E) a probe load-lock chamber, F) an MBE chamber, G) a sample load-lock chamber, and H) three air legs for vibration suppression. (b) Photograph of four STM scanners and a sample holder on the cryostat stage. (c) SEM image showing four STM probes located on a single CNT. Adapted from [18, 19]

Recently, several groups have developed four-probe scanning tunneling microscopes (STM) that can perform four-point electrical measurements with probe spacing down to the microscopic scale [18, 20-24]. With only a few exceptions [18, 23], these systems provide a cryogenic environment only for the sample stage, while the STM probes stay at an unregulated higher temperature. Transport experiments in these systems would involve hot electrons injected from probes into samples at much lower temperature. Such a temperature gradient between the probe and the sample severely impacts the stability and thus the reliability of the experiments. Moreover, injected hot electrons will take a long time to reach thermal equilibrium with the sample (lattice temperature) usually through an electron-phonon coupling that is very weak at low temperatures, imposing an additional level of complexity to the interpretation of the experiments. The temperature management in these multiple-probe STM systems is a serious challenge.

The cryogenic four-probe STM represented in Figure 4 directly addresses the demand of nanotransport research, which allows for in situ sample preparation and treatment, nanoscale precision of sample and probe positioning, together with the four-point transport probing potential. The cryogenic multiprobe STM system has integrated multiple functional modules in a single system to provide capabilities of material synthesis, high resolution microscopy, chemical analysis, cryogenic temperature control both for sample and probes, atomic resolution imaging and spectroscopy, four independently controllable STM probes, and local electrical transport probing. In the cryogenic four-probe STM, both sample and probes are kept at the same low temperature. Each of the four probes works independently as a cryogenic STM [25], and together, they provide the capability of four-point electrical transport measurement for nanostructured materials [18, 19, 26].

4.3 Results and Discussions

4.3.1. Percolative Mott Transition in Sr3(Ru1-xMnx)2O7:

The layered Sr3Ru2O7 belongs to the Ruddlesden-Popper series with a stacking of two layers of corner- sharing RuO6 octahedra separated by cation oxide rock-salt layers [27]. The lamellar structure of the ruthenate makes the crystal amenable to creating a fresh ab plane by cleaving in-situ in an ultra-high vacuum (UHV) system. This offers a controlled way to investigate electronic phases using multiple surface analytical techniques. In general, 4d TMOs (e.g., ruthenates) have weaker correlation effects than 3d TMOs (e.g., manganites, cuprates, and cobaltates) as the radial extension of 4d wave functions is significantly larger than that of transition-metal (TM)-3d orbitals [28]. On the other hand, the extended wave functions of the 4d TM enable an interesting competition between local and itinerant physics. Doping the 4d host with a dilute 3d TM is thus extremely effective in tuning valence, spin, and orbital characteristics and, in turn, the macroscopic physical properties as manifested in Sr3(Ru1-xMnx)2O7[28, 29]. In the undoped compound of Sr3Ru2O7, a field- induced quantum phase transition was discovered [30, 31]. With the substitution of Ru by a few percent of Mn, a metal-to-insulator transition (MIT) and the emergence of a Mott antiferromagnetic (AF) ground state were revealed [28, 29]. The MIT is believed to be driven by the onset of AF correlations with a clear phase boundary between the paramagnetic (PM) metal and AF insulator regions at Tc ~ 80 K for 10% Mn substitution [28, 29].

To study the Mott MIT in ruthenates, we substituted the Ru site with 20% Mn and grew single crystals of Sr3(Ru0.8Mn0.2)2O7 using the floating zone technique [32]. Direct visualization of phase separation was carried out on cleaved single-crystal surfaces in a cryogenic four-probe scanning tunneling microscopy (STM) system [18]. An UHV (pressure <1.0 × 10-10 Torr) field emission scanning electron microscopy (SEM), scanning tunneling spectroscopy (STS), and four-probe STM were used to probe electronic properties on surfaces in situ.

Fig.5 Electronic phase separations displayed in the cleaved surfaces of Sr3(Ru1-xMnx)2O7. (a) Bubble- and (b) Stripe-shaped domain structures revealed by SEM for x=0.20 at 81 K. (c) No domain structure observed at 300 K for x=0.20 sample or in an undoped sample (x=0). Adapted from [33]

Figure 5 shows the striking domain structures in a Sr3(Ru0.8Mn0.2)2O7 surface at low temperatures revealed by SEM operated in the secondary electron emission mode. At 81 K, micrometer-scale dark bubble- and stripe-like (Figure 5a and 5b) domains are clearly seen in an otherwise bright matrix of the cleaved surface. However there is no domain structure at room temperature (Figure 5c) or for an undoped sample.

Contrast in the field-emission SEM image reflects the difference in the secondary electron (SE) yield which, in turn, is determined by topography, chemical composition, and conductance of the sample [34, 35]. In particular, the differences in conductance between the neighboring areas affect the work function, stopping power, and electron mean free path, all intimately associated with the SE emission process. In fact, there is a growing interest in utilizing the conductance contrast of SEM to map electron density distributions, particularly in semiconductors [36]. Across the metal-insulator transition (MIT) point of Sr3(Ru1-xMnx)2O7, topographical and chemical contrasts are not expected to show appreciable change. However, the formation of an energy gap near the MIT point modifies the electron band structures relative to the corresponding metallic phase. As the energy distribution of the SE is narrow and peaked at very low energy, generally in the range of 2–5 eV [34, 35], even a small shift in band structure can generate a remarkable change in SE yield and a clear contrast at the boundary of the insulator and the metal phases.

The temperature (T)-dependent evolutions of stripe domains have been displayed in Figure 6. When warming up the sample from 50 K, the stripe domains start to melt gradually with T and disappear completely at 150 K. Upon cooling down to 50 K again, the stripes reappear. In comparing to the domain pattern obtained in the first thermal cycle, the new pattern shows a few new remarkable regions — a new stripe developed, a stripe disappeared in the second thermal cycle, and some stripes changed locations. Based on theoretical simulations, the chemical disorder introduced by the impurity ions of different radii can form large- scale phase separations [37, 38]. The inhomogeneity of a chemical origin would occur at more or less the same location although the shape may be different in different thermal cycles. On the other hand, pure electronic interactions can also induce inhomogeneity of charge density distribution; however, this inhomogeneity would have totally random distribution at the nanometer scale [3, 38]. Thus, neither electronic interactions nor chemical inhomogeneity can be used to fully explain the above observations.

Fig.6 T-dependent phase percolations in a Sr3(Ru0.8Mn0.2)2O7. (a-e), Domain images measured at various T for the same sample location. (f), Superimposed view of images (a) and (e), showing in blue and purple respectively. Adapted from [33]

To investigate the origins of the bright/dark domain contrast in SEM images, we have performed STM/STS measurements on cleaved surfaces. STS spectra were taken on individual surface domains, which probes the integrated density of electronic states near the Fermi energy, EF. The derivative of I(V) data (Figure 7a) reveals an energy gap near EF in the dark domains, in contrast to a finite density of states in the bright regions. Therefore, the SEM contrast observed here reflects the electronic state difference between dark and bright regions that correspond to insulating and metallic phases, respectively. The Mott transition nature is reflected in the abrupt opening of the energy gap measured by STS and shown in Figure 7b. At the microscopic scale, the onset of the MIT (namely Tc) varies from location to location, consistent with the observation of the continuous transition behavior in domain percolation and the bulk MIT.