NAMBE 2023 Session ST-WeM2: Fundamentals of MBE Growth

The bandgap of germanium-tin (Ge_1-xSn_x) alloys can vary from ~0.8 eV into the far-IR as the Sn composition increases, and transition from indirect to direct bandgaps between 6 and 25 at% Sn depending on strain. Consequently they are being developed for a range of thermophotovoltaic (TPV), thermal imaging, and photonic data transmission applications. However, at useful Sn contents these alloys are metastable and require low epitaxial growth temperatures. Therefore they can only be grown to limited thickness, otherwise films may experience Sn segregation and defect formation. Growing below ideal temperatures for Ge can also induce defects, reducing electrical conductivity. Growth and thickness limitations complicate PV cell design and optimization. One mitigation strategy for non-optimal growth conditions is to deposit and maintain on the growth surface a thin coverage of an additional atomic species, which acts as a surfactant to modify adatom interactions. Surfactants change the surface energy of the growth front, which controls the mobility of adatoms and affects the rate of positional exchange between buried and surface layers. A surfactant that preferentially moves to the growth surface could increase the incorporation of Sn into deeper layers and reduce defects formed due to growth at low temperatures. Additionally, the filling of partial growth layers by surfactant atoms may reduce islanding and so improve surface roughness.

In this work, antimony was applied as an n-type dopant and surfactant during the growth of GeSn alloys to potentially improve the crystallinity of Ge and GeSn films grown by molecular beam epitaxy. Previous work has shown Sb suppresses roughness in SiGe film growth. Gallium was applied as a p-type dopant and surfactant, providing a comparison between dopant types and differing surfactant atomic sizes. Pairs of GeSn samples were deposited without and with the codeposition of Sb or Ga. The growth temperature, film thickness, and Sn content were varied to investigate the impact of each surfactant on Sn incorporation and critical thickness. Amount of Sb or Ga incorporation into the growing film depends on growth conditions, so this study also provides data on the retention of Sb and Ga as dopants during low-temperature GeSn growth. Sn content and film quality were characterized by X-ray diffraction, photoluminescence and Raman spectroscopy, spectroscopic ellipsometry, and optical microscopy. Electrical properties were examined by Hall effect measurement. Results will be discussed in terms of how surfactants have modified the kinetics of the growth process and considered with design of a GeSn based photodiodes and TPV cells in mind.

Recently it has been shown that structural disorder in 3D Topological Insulators (TIs) has considerable effects on the properties of the materials. The addition of magnetic ions like Mn breaks time reversal symmetry and opens a gap in the Dirac point.¹ This addition also changes the crystal structure from the typical quintuple layer (QLs) structure of non-magnetic TIs to a septuple layer (SLs) structure.² Furthermore, addition of a Mn flux during MBE growth results in self-assembled structures of mixed QLs and SLs.²

Previously we reported the MBE growth of self-assembled structures of (MnSb₂Te₄)_x(Sb₂Te₃)_1-x magnetic topological materials, and showed that their Curie temperature (T_C) is dependent on the composition x (or %SL). Samples with 0.7 < x < 0.85 exhibit very high T_C values. Additionally, it was observed that as the Mn beam equivalent pressure (BEP) ratio used was increased, there was a corresponding increase in T_C.Decreasing the growth rate further increased T_C to >100 K, the highest values reported for this material system. An understanding of how the changes in growth conditions lead to the T_C enhancement is not well-established.

Here we investigate the structural properties of the materials as they relate to the growth conditions, specifically growth rate (GR). Samples grown at slow GRs (0.4 – 0.6 nm/min) were compared to samples grown at fast GRs (0.8 – 1.0 nm/min). Samples were investigated by X-ray diffraction, Energy Dispersion X-Ray Spectroscopy (EDS), Hall effect and scanning transmission electron microscopy (STEM).

We found that for the same Mn BEP ratio, low GR yields similar composition x as fast GRs. On the other hand, EDS showed that for x > 0.7 there was increased intermixing between Sb and Mn in both the fast and the slow GR samples. However, the samples grown with slow GR showed much greater Mn and Sb intermixing, as well as Mn and Te intermixing, suggesting increased Mn incorporation at the slow GR. Cross sectional EDS studies reveal a high Mn content in the QLs, consistent with (Sb_1-yMn_y)₂Te₃ alloy formation. Hall effect measurements show that GR does not significantly affect the electrical doping in (MnSb₂Te₄)_x(Sb₂Te₃)_1-x supporting the proposal that a super-exchange magnetic mechanism is likely at play. Other techniques, such as magnetic force microscopy (MFM) are being explored to better understand the magnetic mechanisms leading to high T_C values. Our results provide insight as to how to achieve on-demand magnetic TIs with enhanced properties.

¹ Y. Tokura et al, Nature Reviews Physics1, 126 (2019)

² J.A. Hagmann et al, New Journal of Physics 19, 085002 (2017)

³ I. Levy et al, Crystal Growth & Design 22, 3007 (2022)

Bismuth ferrite (BiFeO3) is a lead-free magnetoelectric multiferroic showing antiferromagnetic order and a large spontaneous polarization at room temperature. This antiferromagnetic order in BiFeO3 is complex, where, as a consequence of the Dzyaloshinskii-Moriya Interaction (DMI), a small canting of the antiferromagnetic order forms a chiral spin cycloid in bulk samples. Understanding the interplay between the ferroelectric polarization and the spin cycloid, as well as its electric field manipulation, is of significant interest for antiferromagnetic spintronics and next generation computation. There is still much to learn about BiFeO3’s intrinsic antiferromagnetic structure in thin films, where epitaxial strain imposed by an underlying substrate can perturb and even destroy the spin cycloid. MBE provides an ideal pathway to create BiFeO3 with unparalleled structural perfection. As a model system, we have synthesized epitaxial thin films (2-100 nm) of BiFeO3 on (110) TbScO3 substrates via oxide molecular-beam epitaxy. In this work, we explore how epitaxial strain affects the formation and orientation of the spin texture in BiFeO3 using nitrogen-vacancy magnetometry (NV magnetometry). NV magnetometry uses a nitrogen vacancy implanted at the tip of a diamond cantilever which acts as a single-spin magnetometer to sensitively map nanoscale surface stray fields. By correlating the spin texture to ferroelectric information measured using piezoresponse force microscopy, we are able to explore coupling between the two order parameters. Electric-field-dependent measurements show that the spin cycloid can be manipulated by switching the ferroelectric polarization, which is of great interest in low dissipation magnonics and spintronics.

Complementary metal-oxide-semiconductor (CMOS) technology is an important part of today’s integrated circuit technology. CMOS replaced n-channel metal-oxide-semiconductor (NMOS) in the 1980s and with the 100 to 1000x power savings advantage it provides, enabled integrated circuits to grow from tens of thousands of NMOS transistors on a chip to tens of billions of CMOS transistors today. Today we stand at similar cross-roads for transistors made from oxide semiconductors. Only high-performance n-channel oxide transistors (and thus NMOS) exist for oxide transistors. If high performance p-channel oxide transistors could also be made, low-power CMOS would be possible in oxide systems and enable low-power transparent electronics. Although many p-type semiconducting oxides have been predicted based on their theorized electronic properties, few have been realized in experiment and those that have been achieved have much lower mobility than established n-type oxides like indium-gallium-zinc-oxide or In₂O₃. Unfortunately, the realization of p-type oxides is particularly difficult due to the localization of the oxygen 2p orbitals. One design criterion to realize p-type oxides is to create oxides with Sn²⁺, Pb²⁺, or Bi³⁺ cations that have dispersive filled s-orbitals that will hybridize with oxygen 2p orbitals to delocalize the hole states and enhance mobility. We show that suboxide MBE can be used to grow Ba₂BiTaO_6, a candidate transparent p-type oxide.

Suboxide MBE utilizes molecular beams of suboxides, where the incoming cation precursors are already in the desired oxidation states. Many oxides crystallize with a perovskite structure and offer flexibility in accommodating a variety of cations, resulting in a wide range of properties. Double perovskites, of which Ba₂BiTaO₆ is an example, offer additional flexibility and another degree of freedom to explore in the interplay of structure and properties. For Ba₂BiTaO₆, we take advantage of suboxide MBE and avoid using an electron beam to evaporate Ta by supplying a molecular beam of TaO₂ from a Ta₂O₅ source. Additionally, with the volatility of Bi at our growth temperatures, we grow in an adsorption-controlled regime and fine tune the flux of Ba and Ta cations to create the desired phase.