Synthesis and Modulation-Doping of VO2 (001) Epitaxial Thin Film Heterostructures

Abstract

Correlated electron materials (CEMs) host a rich variety of condensed matter phases. Vanadium dioxide (VO2) is a prototypical CEM that undergoes a temperature-driven metal-to-insulator transition (MIT) at ~340 K accompanied by a symmetry-lowering transition in the crystal structure. However, external control of MIT in VO2 – especially without inducing lattice strain in the parent phases - has been a long-standing challenge. All the previous approaches, for example, the control of MIT in VO2 by elemental doping, strain, oxygen vacancy creation, and hydrogenation affect the lattice parameters of VO2. In these cases, simultaneous lattice strain and carrier concentration change make it challenging and, sometimes, impossible to disentangle the role of carrier concentration changes from the role of lattice strain. On the other hand, the electric field-driven control of MIT could, in principle, enable conductivity modulation without causing lattice strain. However, it is limited by either the lack of sufficient field-induced electron density in the VO2 channel when utilizing solid-state gating or by the oxygen deficiency-induced lattice distortion in VO2 when using ionic liquid gating. We propose a VO2-based modulation-doped heterostructure in which carrier densities can potentially be modulated by remotely injecting carriers into VO2, which, in principle, enables us to control the metal-insulator transition temperature (TMIT) of VO2 without introducing lattice strain. In Chapter 1, we introduce a comprehensive overview of the area of correlated electron materials and provide a broad explanation of metal-insulator transitions (MIT) phenomena in transition metal oxides. Specifically, we discuss Mott-Hubbard transition and Peierls transitions in the context of MIT in VO2. We also discuss crystal field theory (CFT) which explains the energy landscape near the fermi level depending on the electrostatic interaction between metal ions and the surrounding ligands in transition metal oxides. We further introduce vanadium dioxides as a prototypical example of CEMs and describe the electrical and structural phase transition that occurs at the metal-insulator phase transition temperature. Toward the end of this chapter, we provide a brief discussion of the state-of-the-art studies of MIT behavior in VO2. We also discuss the successes and challenges of controlling MIT by external stimulus including elemental doping, oxygen vacancy creation, hydrogenation, lattice strain, and application of electric field. We conclude that even though modulation of TMIT was successfully achieved in these approaches, it is always accompanied by microscopic changes to the crystal lattices. In Chapter 2, we provide a detailed demonstration of the experimental techniques and specifically the important experimental methodologies which were utilized in this dissertation. First, we introduce the Pulsed Laser Deposition (PLD) system, the preferred thin film deposition technique for the growth of oxide thin films, which was employed for the growth of all the thin films used in these studies. Next, we discuss the fundamental working principles along with utilized measurement parameters of several characterization techniques which are crucial to measuring the physical properties of VO2 thin films. We discuss in-situ Reflection High Energy Electron Diffraction (RHEED), Atomic Force Microscopy (AFM), High-Resolution X-ray Diffraction (HR-XRD), Reciprocal Space Mapping (RSM), Scanning Transmission Electron Microscopy (STEM), Energy Dispersive X-ray Spectroscopy (EDS), Temperature-dependent electrical transport measurement, Temperature-dependent Hall measurement, X-ray Photoelectron Spectroscopy (XPS), and Wedge bonding. Finally, we present details of an in-house built low-temperature electrical transport measuring system (Dipstick), covering details of instrumentation, calibrations, and measurements. In Chapter 3, we first introduce an overview of thin film growth mechanisms in general. Further, we comprehensively discuss the optimization conditions to accomplish atomically sharp and single-crystalline VO2 thin film growth on TiO2 (001) substrate using the PLD technique. By combining RHEED, AFM, asymmetrical RSM, and HR-XRD, we confirm the epitaxial growth of VO2 films. Further, the films are atomically flat, single crystalline, and highly oriented to the TiO2 (001) substrates. With the aid of temperature-dependent electrical transport and HR-XRD, we show that VO2 films undergo ~3.5 orders of magnitude change in resistance along with a concomitant structural phase transition across the MIT. Temperature-dependent Hall measurements suggest that such changes in resistance are predominantly contributed by the equivalent change in carrier densities whereas carrier mobility changes little, in comparison. Furthermore, temperature-dependent hard X-ray photoelectron spectroscopy (HAXPES) measurements of the V2p core level and valence band reveal the signature of the non-local screening as well as the shift in the density of states (DOS) across MIT. Finally, we demonstrate that the TMIT of VO2 (001) films remain almost identical at ~295 K throughout the thickness range of 9.5 nm to 1.5 nm. This finding is very crucial since in the following chapters we will demonstrate how this TMIT could be modulated by varying the thickness of VO2 using modulation-doped heterostructures. In Chapter 4, we demonstrate the control of MIT by varying the carrier concentration in VO2 utilizing modulation-doped heterostructures where Nb:TiO2 layer was used as a dopant layer and the single crystalline VO2 (001) film as a channel layer. Here we show, the carrier densities can be modulated in two pathways – either by varying the dopant densities in the dopant layer (i.e. concentration of Nb in Nb:TiO2) for a constant thickness of VO2 or by varying the thickness of the VO2 (001) layer while maintaining a fixed dopant density in the dopant layer. Using a combination of temperature-dependent transport and hall measurements, we present that a systematic reduction in TMIT up to 40 K was achieved. A remarkable feature of this work is the deposition of the dopant layer at room temperature which ensures two important criteria for achieving high-quality modulation-doped heterostructures. Firstly, the low kinetics of the deposited dopant ions at room temperature prevents the dopant ions (Nb5+) from diffusing into VO2 and preventing the formation of NbxV1-xO2, which could lower the TMIT. Secondly, room temperature deposition of dopant layers also minimizes the migration of oxygen ions from VO2 to Nb:TiO2 layer which could result in oxygen vacancy formation in VO2 along with the induced strain and lowered TMIT. Both the transport and XPS data are not consistent with oxygen vacancy formation in VO2. We further show that, for all the heterostructures, in-plane lattice parameters of VO2 remain unchanged. However, interestingly, we found an asymmetrical shape of VO2 (002) peaks in the XRD spectra, which were not observed in pristine VO2 films. With the help of XRD simulations and time-evolved RHEED images during the growth of Nb:TiO2 dopant layers, we attributed the presence of this asymmetric shape to the formation of a few atomic layers of crystalline Nb:TiO2 layer on VO2 film even for room temperature growth. However, for such thin films, this makes it challenging to distinguish and exactly quantify whether these asymmetric natures in the spectra are due to out-of-plane strain in VO2 (if any) or due to the formation of Nb:TiO2 crystalline layers. Since both VO2 and Nb:TiO2 have a similar rutile crystal structure at room temperature, it's energetically more favorable (due to epitaxial stabilization) to grow a few crystalline atomic layers of Nb:TiO2 on VO2 even for deposition at room temperature. To address this, we propose the use of an amorphous spacer layer which we discuss in the next chapter. In Chapter 5, we designed an upgraded version of the modulation-doped heterostructure that contains LaAlO3 (LAO) as a spacer layer between the VO2 channel and the dopant layer. Furthermore, to avoid the interdiffusion of higher valence metallic dopants such as (Nb5+) from the Nb-doped TiO2 dopant layer to the VO2, we use oxygen-deficient TiO2-x layer as a dopant layer instead of the Nb-doped TiO2 dopant layer. The crystal structure of LAO is different from that of rutile VO2 which enhances the possibility of the LAO layer being amorphous for growth at room temperature. Besides, it is an insulator with a high band gap of 5.6 eV. In addition, because of low oxygen vacancy-diffusivity in LAO, it prevents the oxygen migration from the VO2 to TiO2-x dopant layer. Here, we synthesize and demonstrate that these modulation-doped heterostructures closely emulate a textbook example of filling control in a correlated electron insulator. Using a combination of charge transport, Hall measurements, and structural characterization, we demonstrate that the insulating state can be doped to achieve carrier densities greater than 5x1021 cm-3 without inducing any measurable structural changes. With the aid of Hall measurements, we find that the TMIT is strongly correlated with carrier concentration and decreases continuously (by ~65 K) with increasing carrier concentration. A remarkable feature of this study is the possibility of bulk metallization in modulation-doped VO2. A sharp MIT is observed for heterostructures with VO2 thicknesses as high as 9.5 nm, which are much higher than the Thomas Fermi screening length of ~1 nm. Additionally, by performing temperature-dependent XRD across the MIT we validate that the structural phase transition is still accompanied by the electrical phase transition. Further, we also discuss the role of spacer layer thickness where we found about 2 nm thickness of LAO is optimal for both allowing significant tunneling current from the dopant to the VO2 and preventing oxygen migration across these two layers. In Chapter 6, we further extend our studies as presented in the previous chapter. By performing bulk-sensitive HAXPES measurements, we show the presence of two characteristics peaks, V2p3/2, and V2p1/2, at ~515.8 eV and ~523.1 eV, respectively, for the 7.5 nm VO2 film in the insulating phase. Importantly, a systematic shift of the main component of the V2p3/2 peak to higher binding energies is observed for the VO2 heterostructures, with the highest increase in binding energy (~250 meV) observed for the heterostructure with the thinnest VO2 layer (1.5 nm) suggesting the signature of band bending of an electron-doped VO2 in the insulating phase. However, these V2p spectra show two remarkable features. The first one is the gradual increase in lower binding energy shoulder around ~514.5 eV in the insulating phase with the increase in carrier densities in VO2 suggesting that the additional charge transferred to the VO2 channel layer enables non-local screening previously observed only in the metallic phase of VO2. The second is the emergence of an additional peak with a larger binding energy at around 517.5 eV. This higher binding cannot be a trivial higher valence peak, such as V5+. This is because a more oxidative state for vanadium ions is inconsistent with electron doping implied by Hall measurements. Furthermore, the intensity of this additional higher binding energy shoulder is directly correlated with carrier density. Based on the LDA+DMFT Anderson impurity model calculation of an electron-doped VO2, we tentatively assign the 517.5 eV peak to a satellite peak induced by electron doping. Remarkably, across the MIT, HAXPES spectra show a clear change of V3d spectral weight at the fermi level even for the heterostructure with the thinnest VO2 layer (1.5 nm) and the highest carrier density suggesting that the insulating state is robust even at doping concentrations as high as ~0.2 e-/vanadium. In Chapter 7, we conclude by pointing out the future direction of this study in general and more specifically application of modulation-doped heterostructures which is a powerful technique for achieving high carrier densities, close to those possible with elemental doping. Since our approach does not need any epitaxially matched spacer and dopant layers, it expands the library of materials that can be utilized to explore ‘pure’ electronic effects in correlated oxides and related systems.