Phase change superlattices and thin film effects : MBE-growth and characterization
Kaminski, Marvin; Wuttig, Matthias (Thesis advisor); Waser, Rainer (Thesis advisor)
Dissertation / PhD Thesis
Dissertation, RWTH Aachen University, 2020
With the introduction of the Optane memory in 2017, Intel and Micron made the first commercial use of the electrical properties of Phase Change Materials (PCM). PCMs can be switched between a high conductive crystalline state and a low conductive amorphous state by joule heating. These two states can be employed to store information. The unique properties of this material class are assumed to originate from their unconventional bonding mechanism, recently coined metavalent bonding. Metavalent bonding is characterized by electrons, which are on the verge of delocalization. In this thesis, the consequences of the tendency of the electrons towards delocalization are analyzed. Therefore three different effects of thin films are investigated in detail. The first effect is a spontaneous transition from an amorphous to a crystalline state at a critical film thickness in the range of a few atomic bilayers in PCM and metals at a constant temperature. After the transition, the crystalline film has an epitaxial relation to the substrate. During the transition the film crystallizes fully, there is no amorphous buffer layer. It is assumed that in thin films the electrons are so confined that it is not possible to develop the tendency needed for delocalization as employed in metavalent bonds. But this tendency is essential for metavalent bonding. Consequently, the thin film is bonded covalently. In literature, this transition is seen only for some combinations of substrates and PCMs. Moreover, a systematic analysis is lacking. In this work, it is shown that four out of five PCM undergo the transition on the Si(111)-7x7 surface. The only exception is Sb2Te3 due to its layered structure. In addition, a transition is demonstrated here for the first time in Ge2Sb2Te5 on Si(111)-7x7. One crucial but in literature not discussed factor for this transition is the growth mode. The transition is observed using Reflection High-Energy Electron Diffraction (RHEED). However, it can be shown that the needed lack of reflexes in RHEED occurs only during layer-by-layer growth. In addition it is shown that passivating the silicon surface with Sb forming the Si(111)-(sqrt(3) x sqrt(3)) R30° - Sb, prevents the transition in all cases studied and the film grows crystalline from the beginning. Moreover, the critical thickness of the transition depends on the substrate temperature. This dependency has not been reported so far. It is shown that the relation is based most likely on kinetic effects. The second effect which has been investigated concerns the in-plane lattice parameter at the start of the growth. The Si(111)-(sqrt(3) x sqrt(3)) R30° - Sb surface termination imprints a smaller lattice parameter on films like GeTe(111) or Ge2Sb2Te5 (111). Nevertheless, with RHEED a systematically too large lattice parameter has been found in all materials analyzed except Sb2Te3. The larger lattice parameter is assumed to be connected to the confined dimensions, influencing the bonding. Also, DFT calculations of free-standing films suggest a larger lattice parameter in thin films as compared to the bulk even without a substrate. As a third and last effect, the thin films are combined into multilayers. In literature, an unusual strain transfer is reported in multilayers of GeTe/Sb2Te3 or GeTe/Bi2Te3 stacks. The strain transfer is assumed to originate from the layered structure of, for example, Sb2Te3 and Bi2Te3. For the first time, GeTe and SnTe are combined into multilayers. Both materials have no layered structure in comparison to Sb2Te3 and Bi2Te3. Therefore, the strain transfer is of interest and is monitored via the in-plane lattice constant. Besides, the GeTe/SnTe superlattices are analyzed to determine the intermixing at different growth temperatures. Here, a rare combination of RHEED, X-Ray Diffraction (XRD), Transmission Electron Microscopy (TEM), and Atom Probe Tomography (APT) give insights into the structure of the stacks. The origin of the breaking of the symmetry in the growth of SnTe (001) on Si(111) is analyzed using Angular RHEED (ARHEED). Taking the crystal structure of Si(111) into account, three equal rotational domains of the SnTe should occur. In the past, it was possible to use the miscut of the substrate surface and the silicon lattice to break this symmetry. In this case, two out of the three rotational domains are suppressed resulting in an almost single crystalline film. With ARHEED it is possible to determine the critical thickness of the suppression of the two domains in the range of 1.5 to 2 nm. In addition, the surface sensitivity of ARHEED is used to determine a previously unknown rotational domain at ±2° at small film thicknesses of SnTe. These domains vanish at roughly 4 nm. The calculations of the intersections of the two superimposed grids of Si and SnTe also confirm the occurrence of the rotation domainal at ±2°. This discovery underscores the ability of ARHEED to measure rotational domains at film thicknesses where other techniques, such as XRD, fail. The analysis of the three thin film effects and the discovery of the critical thickness for the suppression of two out of three rotational domains helps improving the understanding of metavalent bonding in phase change materials.
- DOI: 10.18154/RWTH-2020-06480
- RWTH PUBLICATIONS: RWTH-2020-06480