(G03 Best Paper Award Winner) Analysis of Sn Behavior During Ni/GeSn Solid-State Reaction by Correlated X-ray Diffraction, Atomic Force Microscopy, and Ex-situ/In-situ Transmission Electron Microscopy

Abstract

GeSn material has been envisioned as an alternative to the complex integration of III-V lasers on Silicon (Si) for photonic applications [1]. GeSn heterostructures with 16 at.% of Sn were demonstrated [2]. Increasing the Sn at.% is of high interest in these devices. Additionally, high quality contacts are needed to achieve electrical pumping, for which Nickel intermetallics were proposed. Indeed, Ni(GeSn) exhibit low temperatures formation and low resistances values. Currently, it is not entirely known how Sn incorporates in Ni-GeSn phases. Sn might: (i) be soluble in the phase, (ii) segregate, and/or (iii) form Ni-Sn alloys [3,4]. The Sn content is otherwise expected to impact different aspects of the phase sequence [5], as well as morphological and electrical properties.We propose a comprehensive study of the Ni-GeSn intermetallics focusing on Sn segregation and its impact on phase stability. 60 nm-thick Ge and Ge1-xSnx layers with 6, 10 and 15 at.% of Sn, were epitaxially grown on Ge-buffered Si (100) substrates. Surface preparation was performed with 1% diluted HF. Then, 10 nm of Ni were deposited by physical vapor deposition and capped with 7 nm of TiN. The Ni-Ge1-xSnx phase sequence and crystalline evolution were monitored by in-situ X-ray diffraction (XRD). Cross-section transmission electron microscopy (XTEM) measurements coupled with energy-dispersive X-ray spectrometry (EDS) and electron energy-loss spectroscopy (EELS) atomic maps were acquired to follow Sn distribution and segregation. Figure 1 shows the phase sequence of the Ni-Ge0.90Sn0.10 system. The growth was sequential. The first phase appearing was the Ni5(GeSn)3 phase. It was followed by the growth of the Ni(GeSn) phase [5 – 7]. EDS profiles confirmed that Sn was incorporated in both the Ni-rich and Ni(GeSn) phases. Similar phase sequences were obtained for all systems, from a straightforward comparison of in-situ XRD patterns. The threshold temperature above which the Ni-rich phase disappeared (and the less resistive Ni(GeSn) phase appeared) increased with Sn content: from 180 °C up to 235 °C and 245 °C for 0, 6 and 10 at.% of Sn, respectively (Figure 2). A dependence of the phase transition on the Sn content in the layers was evidenced. This trend was not observed for the Ni-Ge0.85Sn0.15 system due to Sn segregation at lower temperatures. XTEM analyses coupled with EDS and EELS (Figure 3) enabled us to track Sn segregation. Flat and continuous layers were obtained from the as-deposited state up to 350 °C. At 350 °C, Sn segregation started. It happened first around grain boundaries (GB) and then near the surface. At 400 °C, the Ni(GeSn) layers lost their continuity due to grain boundary grooving and agglomeration (confirmed by atomic force microscopy (AFM)). Grains were mainly composed of Ni and Ge atoms. Sn was predominantly localized at GB and close to the surface. At 550 °C, Sn coming from the intermetallic and from the GeSn thick layers underneath, completely migrated towards the surface. Figure 4 presents a Sn behavior schematic summary. Sn distribution and segregation impacted the Ni-Ge(Sn) phase sequences. Due to the role GB have on atomic transport, Sn segregation in those highly energetic areas might have reduced atoms mobility. Indeed, GB normally offer preferential paths, with enhanced atomic mobility [8]. GBs modified/saturated by Sn, can reduce the diffusion rate for phase formation. We thus believe that Sn acts as a “stuffed barrier”, hampering two-way atomic traffic. The higher the Sn content, the greater the hampering effect is, which slowed down Ni diffusion and delayed phase formation.We showed that Sn segregation in Ni-GeSn layers resulted in higher thermal budgets needed to obtain the less resistive phase (Ni(GeSn)) and have proper electrical contacts. Those thermal budgets could be detrimental for layers’ stability. In-situ TEM-EDS, AFM and sheet resistance results will also be presented, together with alternative solutions to delay Sn segregation. Acknowledgements: We would like to thank Virginie Loup for the wet surface treatment. This work was supported by the French National Research Agency (ANR) under the “Investissements d’avenir” programs: ANR-10-AIRT-0005 (IRT-NANOELEC), ANR-10-EQPX-0030 (EQUIPEX-FDSOI-11) and by the CEA DSM-DRT Phare project “Photonics”.