Plan of the Thesis

Since nano metals are known as conductors that realize metallic conductivity at low tem- peratures, they potentially can provide a novel platform to explore quantum interference phenomena of coherent electrons, at small dimensions and across dimensionalities. Nano metals are known for many thin films, including bothnanometallic[11, 12, 13] andfilamen- tary resistance memory [56, 57, 58] (see Appendix A-B for a brief introduction of resistance memories, known as ReRAM, made from these two types of thin films.) But it is not known whether quantum phenomena for coherent electrons exist in them or not. This is an intrigu- ing question because nano metals are made of materials that are strong insulators when in the bulk form, and such strong insulators have never been thought to be possible candidates for observing coherent electrons. This question will be settled in this thesis, which seeks to establish quantum electronic interference phenomena in thin-film nano matals that are of particular interest to resistance-switching memory.

To provide the simplest ReRAM for a comprehensive study of these quantum electronic phenomena, I have developed an amorphous Si device that exhibits non-volatile resistance switching. It switches in the same way as the so-called nanometallic RRAM, which are metal-doped amorphous nitrides and oxides. This developmental effort is documented in Appendix A, which also explains why O and N doping can enable switching. Experimental

results in Chapter 2 unequivocally confirm that 3D EEI dominates the QCC in Si nano metals, causing a positive MR that is not saturated even though the diffusion length is cutoff by sample thickness. The result also allows the determination of diffusivity, which at 3×10−2 cm2/sec is about one to two orders smaller than conventionally seen in mesoscopic metals. Such lower diffusivity can be directly equated to a large effective mass for electrons, which is indicative of strong electron phonon interaction. The subject of electron-phonon interaction is of central importance to resistance switching and the stabilization of the insulator state in nanometallic ReRAM, and it is a subject that has been much explored in the past in our group. But further experimental evidence in the form of effective mass is provided here for the first time, through QCC and MR measurements, as will become clear throughout Chapter 2-4.

Amorphous Si has generally rather weak MR because of the small dimensions of the ma- terials, which prevent large loops to form to allow a large WL/WAL-MR effect. However, as shown in Eq. (1.14)-(1.16), low diffusivity can enhance QCC of both the WL and WAL origin in 3D, and this is used advantageously in Chapter 3 that studies three nanometallic oxide/nitride glasses, doped with Pt atoms. This is because doping allows tunable diffu- sivity (the lower the Pt concentration, the lower the D), reaching as low as 10−4 cm2/sec suggesting an effective mass as large as 100me. It also exhibits WAL because of the strong

spin orbit interaction of Pt. As mentioned previously, WAL and WL have opposite signs, and the spin-orbit interaction can play the role of imparting a “contrast” agent that will flip the sign of the QCC at a certain crossover temperature when scattering by Pt takes over other inelastic scattering. This adds new features to the QCC and with them it is possible to fully unravel all the interactions (WL, WAL, and EEI) in the Si3N4:Pt thin films.

Unlike the case of Si and nanometallic Si3N4:Pt in which quantum interference and con- ductivity are distinctly 3D, in filamentary ReRAM made of amorphous HfO2 and Al2O3, the conducting path is 1D [59, 60, 61], some part of it made of loops tilted at a certain angle relative to the film normal. Evidence of such loops is presented in Chapter 4 in the

form of remarkable Aharonov-Bohm oscillations under a magnetic field, originating from the quantum interference of coherent electrons in a single loop. The development of these filamentary ReRAM also takes advantage of electron-phonon interaction, as evidenced by the pressure-induced insulator to metal transition documented in Appendix B. But fila- ments in these ReRAM can be made by another method that is well known in the ReRAM field: applying a voltage to cause dielectric breakdown. Chapter 4 compares the QCC and Aharonov-Bohm oscillations from filaments and loops formed by both hydrostatic pressure and electrical voltage, and found different forming methods have remarkable effects on al- tering the dimensionality of and imparting spin-orbit interactions to QCC. In particular, voltage creates more random field due to the reduction of cations, distributed in the loop or in a 3D mesh, which impart strong spin-orbit interaction and cusp-like MR at small magnetic fields. Using these methods, we are able to measure loops of a size of 6.5 nm in both HfO2 and Al2O3, which is the smallest loop ever detected inside a solid.

This research has generated unprecedentedly detailed insight to the operation of ReRAM. It also for the first time demonstrates QCC and coherent electron interference in a nanoscale volume∼(10 nm)3. Remarkably, such observations were made in materials that are usually known, in their bulk form, as random insulators, whose conducting electrons have never been exposed so clearly before. A further outcome of this research is the development of a new metrology tool that enables resistance probing at the length scale of 10 nm, where standard four-point probe is so far impractical. These results are important for ReRAM technology and for understanding electron transport in nanodevices.

CHAPTER 2 : Electron Interference and Metal-Insulator Transitions Mediated by