Monolithic and heterogeneous integration of advanced materials & devices on silicon

Resume : The continuum modeling of the growth and processing of thin-films is crucial to tailor material properties and predict the outcome of novel experiments accounting for realistic time and length scales. However, most of the theoretical approaches relying on continuum descriptions have important limitations when dealing with complex, three-dimensional (3D) evolutions and topological changes. The Phase-Field (PF) approach [1] naturally handles such a degree of complexity. In this presentation, the PF modeling of 3D crystalline structures and their morphological evolution is reviewed [2]. The extensions of the standard approach in order to account for detailed features of materials and processing such as surface faceting [3], flux-shielding effects during growth [4] and strain relaxation [5], are discussed. Special attention is devoted to show applications of the considered theoretical framework to understand and predict the results of technology-relevant processes. In particular, PF simulations are shown to enlighten the mechanisms at play during the annealing of patterned, single-crystal silicon films on insulator [6] and Ge/Si nano-strips [7]. References: [1] B. Li et al. Commun. Comput. Phys. 6, 433 (2009). [2] R. Bergamaschini et al., Adv. Phys. X 1, 331 (2016). [3] M. Salvalaglio et al. Cryst. Growth Des. 15 2787 (2015). [4] M. Salvalaglio et al. Phys. Rev. B 94 235432 (2016). [5] M. Albani et al. Phys. Rev. B 94, 075303 (2016) [6] M. Naffouti et al., Sci. Adv. 3 eaao1472 (2017). [7] M. Salvalaglio et al, Appl. Phys. Lett. (2018).

Resume : X-ray diffraction is a routine tool to reveal the mismatch and the degree of relaxation in heteroepitaxial systems. Only positions of x-ray diffraction peaks are needed for this purpose. The x-ray diffraction profiles are less commonly used but provide much more detailed information on the arrangement of the misfit dislocation arrays and other defects in relaxed epitaxial systems. The intention of the talk is to overview general theory of the x-ray diffraction from crystals with dislocations, analytical and numerical methods for analysis of the intensity distributions, and present various examples of detailed studies of the dislocation arrangements at different stages of relaxation of the silicon-based heterostructures by means of x-ray diffraction. The 60° and the edge (Lomer type) dislocations are distinguished at low dislocation densities by characteristic diffraction profiles and at large dislocation densities by comparing the FWHMs of different reflections. At the early stages of relaxation, the bunches of misfit dislocations result in a characteristic diffraction pattern. In the case of large dislocation densities at late stages of relaxation, the FWHMs of the diffraction profiles are notably smaller than expected for uncorrelated dislocations, which points out to their high positional correlations. Multilayers with different types of dislocations at different interfaces, for example the Ge-rich SiGe virtual substrates, can be characterized in detail.

Resume : Cross-hatch patterns are observed at the free surface of several heteroepitaxial systems undergoing strain relaxation via dislocation injection. Despite being known since half a century, such patterns are still poorly understood. While known to be linked to the underlying dislocation array, different mechanism controlling their formation were proposed. More importantly, such patterns display some degree of periodicity but typical wavelengths cannot be explained by simple parameters such as the dislocation-dislocation average distance. In this work we present a synergic theoretical and experimental effort aimed at building a deeper understanding of cross-hatch pattern formation and evolution. By focusing our attention on low-misfit Ge0.06Si0.94 /Si(001) films, we extensively characterize the dislocation distribution by combining state-of-the-art characterization techniques and dedicated continuum simulations. By exploiting recent developments in fast-scanning X-rays microscopy, we indeed record detailed “tilting-angle maps”, as caused by the dislocation distribution in the film and in the substrate. Combining such results with a large-scale TEM analysis and with Monte Carlo simulations, we are able to infer position and Burgers vector of each individual dislocation spanning 40µm in a cross-cut of the sample. Using the continuum model of Ref., accounting for local strain-dependent surface diffusion, we then simulate the evolution of the film roughness with time.

Resume : The formation of edge misfit dislocations (MDs) in a GeSi/Si(001) film is an important and long standing issue. The challenge is to find a mechanism accounting for the presence of these dislocations at the interface since they are not mobile and cannot nucleate at the surface and glide towards the interface. Experimental measurements can hardly detect the nucleation and early stages of growth because of the short time scale involved. Here we present the first semi-quantitative atomistic calculation of the formation of edge dislocations in such films. We use a global optimization method and density functional theory calculations, combined with computations using potential energy functions to identify the best mechanisms [1]. We show that mechanisms previously suggested are relevant only for a low film strain (i.e. for a Ge poor film) and we propose a new mechanism which accounts for the formation of edge dislocations at high film strain. There, a 60 deg. MD nucleates as a ``split'' half-loop with two branches gliding on different planes. One branch belongs to the glide plane of a complementary 60 deg. MD and therefore strongly favors the formation of the complementary MD which immediately reacts with the first MD to form an edge MD. The calculations are carried out using a stabilized version of the Nudged Elastic Band method. [1] E. Maras, L. Pizzagalli, T. Ala-Nissila and H. Jonsson, Sci. Rep. (7), 11966 (2017).

Resume : Quantum dots (QDs) – nanosized crystals – can not only offer low threshold current density but also provide reduced temperature sensitivity. In addition, QD structures have attracted increasing attention for the active element of III-V light emitting sources on Si substrate due to their enhanced tolerance to defects. First, we describe the first QD laser realized on Ge substrates. A long-wavelength InAs/GaAs QD structure was then fabricated on the high-quality GaAs buffer layer. Lasing at a wavelength of 1,305nm with a low threshold current density of 55.2A/cm2 was observed under continuous-wave current drive at room temperature. Based on techniques developed for growing III-V QD laser on Ge, we then demonstrated the first room temperature cw operation of III-V QD laser diodes monolithically grown on a Ge-on-Si substrate. Room temperature lasing at a wavelength of 1.28μm has been achieved with threshold current densities of 163A/cm2 and 64.3A/cm2 under cw and pulsed conditions for ridge-waveguide lasers with as cleaved facets, respectively. To bypass the need for the Ge buffer layer, we demonstrate cw InAs/GaAs QD lasers directly grown on Si substrates with a low threshold current density of 62.5A/cm2, a room-temperature output power exceeding 105mW and operation up to 120°C. Over 3,100h of continuous-wave operating data have been collected, giving an extrapolated mean time to failure of over 100,158h. More recently, we demonstrated the first silicon-based QD DFB laser array exhibited threshold currents as low as 12 mA, SMSRs as high as 50 dB, and a wavelength coverage of 100 nm with a precise channel spacing of 20±0.2 nm.

Resume : Molecular beams of Ga and As4 are implemented in an aberration-corrected transmission electron microscope. GaAs nanowires are grown in situ from Au catalyst particles. Real-time observation gives access to their morphological and structural parameters while growing and the formation of atomic planes at the catalyst-nanowire interface can be examined. We use various conditions which can result in solid or liquid catalyst particle. Surprisingly, the two cases lead to comparable growth rates (about 0.3 ml/s). For liquid catalyst, the contact angle of the droplet evolves rapidly with the V/III vapor flux ratio. At contact angles around 120°, the atomic plane stacking switches from hexagonal to cubic, with a transition region of mixed crystal phases. In agreement with recently reported results, but using a different growth technique and higher growth rates, we observe that the formation mechanisms of the two crystal phases differ singularly. Namely, hexagonal monolayers grow by slow and continuous step flow on a flat nanowire top facet; cubic monolayers appear incrementally and concomitantly with a truncation of the nanowire top facet. At low temperature, lateral growth on the sidewall facets is observed. It proceeds by step flow of one or more lateral monolayers along the nanowire axis direction. The steps are momentarily stopped and can accumulate at stacking faults present in the nanowire core. The underlying mechanisms will be discussed.

Resume : We report the heteroepitaxial growth at low temperatures (T = 450°C) of superflat In0.5Ga0.5N single crystalline thin film on Si (111) substrates using Plasma Assisted Molecular Beam Epitaxy. The influence of III/V ratio on the morphology and composition of grown layers was investigated. As expected, from previous growth studies performed at higher temperatures, the growth of In0.5Ga0.5N grown in N-rich conditions produces a very rough surface while in Metal-rich conditions a very smooth surface (RMS roughness ~1 ML) is observed. However, on metal rich side we observe the formation of metal droplets which strongly affect the growth dynamics of the InGaN epilayer. The droplets inhibit the adatom incorporation dynamics, thus making the growth rate a decreasing function of the metal flux impinging on the surface as soon as the metal dose exceeds the critical amount required for the nucleation of droplets. We explain this phenomenon via a rate equation model that considers droplet effects on the adatom incorporation dynamics and the relevant role played by the vapor-liquid-solid (VLS) growth mode that takes place under the droplets under Nitrogen flux. We also find an inverse relation between the surface coverage by metal droplets and Indium content in the grown epilayer. This effect can also be attributed to the leading role, in the metal rich conditions, of the VLS growth mode. There is therefore a very narrow window, just at the transition between Nitrogen and metal rich conditions, where is possible, avoiding the formation of droplets while still growing in metal rich conditions, to obtain high quality, composition controlled, and superflat In0.5Ga0.5N single crystal on Silicon substrates.

Resume : III-V semiconductors materials are well known for their direct bandgap allowing light emission in a widely tunable spectral range, going from UV to IR which is most attractive for optical applications. However, they exhibit a large lattice mismatch and a highly contrasting thermal expansion coefficient and a polarity difference with silicon. This makes integration on silicon difficult regarding not having too many defects in the layers. The methods usually demonstrated rely on extremely thick buffers superior in size to 2µm. Here we demonstrate on 300mm standard silicon wafers the feasibility of thin buffers on top of which a thin diode structure is grown, the total thickness being slimmer than 2µm. The diodes are demonstrated for 650nm, 970nm and 1.55µm emissions on nominal silicon and GaAs wafers. They exhibit similar electrical characteristics on both hetero and homoepitaxy. For 2x3mm2 diodes one observes threshold currents of the order of 0.5V and leakage currents of 10-9A with light emission starting at 30mA. The structure of these diodes is also integrated in patterned 300mm SOI wafers using the aspect ratio trapping technique and compared to the blanket diodes in terms of pholuminescence measurements. This demonstration provides indications pointing to the feasibility of a low cost CMOS-compatible platform for III-V material integration for optical applications. ACKNOWLEDGMENTS: This work was supported by the French government managed by ANR under the Investissements d’avenir economic stimulus package, with reference IRT Nanoelec ANR-10-IRT-05 and ANR-15-IDEX-02

Resume : Majorana bound states have been shown to be good candidate for the development of a topological quantum computer. These quasiparticles can be engineered in a hybrid system consisting of a high spin coupling III-V nanomaterial interfaced with a s-wave superconductor, such as Al. On the materials side, most progress in the field has relied until now on high quality vapor liquid solid growth of free-standing III-V nanowires functionalized at a later stage by ex-situ deposited Al. Advanced proposed device designs require a more scalable approach enabling easy formation of complex connected wire networks. Here I will report on the progress of our team in using selective area epitaxy in a molecular beam epitaxy (MBE) reactor to grow both III-V semiconductor nanowire networks and a high quality epitaxial s-wave metal superconductor. First, the general physics context and current state of the art will be introduced, with highlight of the challenges and advantages of the selective area epitaxy/MBE approach with respect to other growth techniques, geometries and growth mechanisms. Then we will show how to map the selectivity window enabling successful growth of the III-V semiconductor nanowires. Then, fundamental knowledge grounded in growth understanding and crystallography will be applied to obtain reproducible, high yield advanced high spin-orbit III-V nanowire/superconductor networks. The materials properties and quality are characterized by scanning and transmission electr

Resume : Majorana bound states have been shown to be good candidate for the development of a topological quantum computer. These quasiparticles can be engineered in a hybrid system consisting of a high spin coupling III-V nanomaterial interfaced with a s-wave superconductor, such as Al. On the materials side, most progress in the field has relied until now on high quality vapor liquid solid growth of free-standing III-V nanowires functionalized at a later stage by ex-situ deposited Al. Advanced proposed device designs require a more scalable approach enabling easy formation of complex connected wire networks. Here I will report on the progress of our team in using selective area epitaxy in a molecular beam epitaxy (MBE) reactor to grow both III-V semiconductor nanowire networks and a high quality epitaxial s-wave metal superconductor. First, the general physics context and current state of the art will be introduced, with highlight of the challenges and advantages of the selective area epitaxy/MBE approach with respect to other growth techniques, geometries and growth mechanisms. Then we will show how to map the selectivity window enabling successful growth of the III-V semiconductor nanowires. Then, fundamental knowledge grounded in growth understanding and crystallography will be applied to obtain reproducible, high yield advanced high spin-orbit III-V nanowire/superconductor networks. The materials properties and quality are characterized by scanning and transmission electr

Resume : Intriguing phenomena such as tunable ferromagnetism [1] and topological transition [2] have been theoretically predicted in III-VI layered semiconductors. However, experimental studies on spin related transport is very limited in these systems. Magnetoconductance (MC) at low temperature was measured to investigate spin-related transport affected by spin-orbit interaction (SOI) in III-VI n-type layered GaSe films. Results reveal that MC shows weak antilocalization (WAL). Its temperature and gate voltage dependences reveal that the dominant spin relaxation is governed by the D?yakonov-Perel? mechanism associated with the Rashba SOI. The estimated Rashba SOI strength in GaSe is much stronger than that of III-V compound GaAs quantum wells, although the energy gap and spin split-off band in GaSe closely resemble those in GaAs. The angle dependence of WAL amplitude in the in-plane magnetic field direction is almost isotropic. This isotropy indicates that the strength of the Dresselhaus SOI is negligible compared with the Rashba SOI strength. The SOI effect in n-GaSe thin films differs greatly from those of III-V compound semiconductors [3]. [1] T. Cao, Z. Li, and S. Louie, Phys. Rev. Lett. 114, 23602 (2015). [2] Z. Zhu, Y. Cheng, and U. Schwingenschloegl, Phys. Rev. Lett. 108,266805 (2012). [3] S. Takasuna, J. Shogai, M. Kohda, Y. Oyama, and J. Nitta, Phys. Rev. B 96, 161303 (R) (2017).

Resume : Top-down exfoliation from bulk transition metal dichalcogenides like MoS2 usually leads to micron-sized flakes. We have recently developed an alternative growth method of two-dimensional transition metal diselenides (TMDS) from multilayers down to a single layer based on the Van der Waals (VdW) epitaxy. In the VdW epitaxy, the TMDS is grown either on a passivated surface with a very low density of dangling bonds (it is then called quasi-VdW epitaxy) or on a layered VdW substrate. The basic concept of this growth method relies on the very weak interaction between the epilayer and the substrate in order to largely release the constraint of lattice matching. Therefore it leads to the formation of fully relaxed TMDS layers. Moreover, this technique allows for the growth of uniform layers over centimeter scale surfaces making it compatible with the development of a large scale 2D electronics based on these materials. In this presentation, we will present our recent results on three different topics relying on the VdW epitaxy: (i) the epitaxy of multi and monolayers of WSe2 on a mica substrate either non-intentionally doped, p-type doped (with Nb) or magnetically doped (with Mn), (ii) the full characterization of the layers using Raman and photoemission spectroscopies, x-ray diffraction and SQUID magnetometry and (iii) their transfer onto a SiO2/Si substrate to study magnetotransport properties. In particular, we find a ferromagnetic order in Mn-doped WSe2 layers at low temperature adding a new member to the recently discovered 2D ferromagnets family [1,2,3]. References: [1] C. Gong et al., Nature 546, 265 (2017). [2] B. Huang et al., Nature 546, 270 (2017). [3] M. Bonilla et al., Nature Nanotech. 13, 289 (2018).

Resume : Spin-charge interconversion (SCI) phenomena have attracted a large interest in nowadays spintronics. Here, we investigate SCI in ultrathin Bi films epitaxially grown on Ge(111) as a function of the Bi thickness t. We use x-ray diffraction and scanning tunneling microscopy to obtain a clear picture of the morphology and crystallography of the system. Through spin- and angle-resolved photoemission we show that spin-polarized surface states crossing the Fermi level are present. Then, we directly probe the charge-to-spin conversion by detecting with magneto-optical Kerr effect the electrically-induced spin accumulation in Bi, and the spin-to-charge conversion by generating a spin current in the system with either optical or ferromagnetic resonance driven spin injection. We recover large SCI signals in the thickness range between 1 and 3 nm, characterized by the presence of small Bi nanocrystals, whereas the conversion efficiency drastically decreases as t increases, when the Bi islands start to percolate. Since bulk SCI is small, the conversion is mainly related to the Rashba-Edelstein effect associated with electron transport in the spin-polarized surface states. We explain the observed thickness dependence of SCI by reminding that the Bi conductivity can be strongly affected by quantum confinement effects. In the high confinement conditions realized in the Bi nanoislands obtained between 1 and 3 nm, the bulk resistivity is high enough to electrically disentangle the upper and lower Bi surfaces, which otherwise would give rise to opposite contributions to SCI conversion that tends to cancel out. Our results indicate that quantum size effects might be exploited as a tool to tune SCI and investigate a very rich spin-physics.

Resume : We show room-temperature spin transport in both n-type and p-type germanium (Ge). First, we detect pure spin current transport in n-Ge using four-terminal nonlocal magnetoresistance measurements in lateral spin-valve (LSV) devices with Heusler-alloy/Ge Schottky-tunnel contacts [1-3], where the values of the resistance area product (RA) are less than 0.5 k?µm2. Next, using the same LSV devices, we demonstrate the detection of two-terminal local spin transport up to room temperature. Finally, we introduce methods for detecting spin transport even for p-Ge by using vertically stacked Ge/Heusler-alloy structures with Schottky-tunnel contacts [4,5]. K.H appreciates good collaborative research with Prof. K. Sawano, Prof. V. Lazarov, and the colleagues of our group in Osaka University. This work was partially supported by JSPS/MEXT KAKENHI (No. 16H02333, No. 17H06120, No. 17H06832, and No. 26103003). [1] Y. Fujita et al., Phys. Rev. Appl. 8, 014007 (2017). [2] M. Yamada et al., Phys. Rev. B 95, 161304(R) (2017). [3] M. Yamada et al., Appl. Phys. Exp. 10, 093001 (2017). [4] M. Kawano et al., Appl. Phys. Lett. 109, 022406 (2016). [5] M. Kawano et al., Phys. Rev. Mater. 1, 034604 (2017).

Resume : The maturity of Si-based technology provides compelling motivation to integrate the spin functionality in novel Si-based devices. Recently, we demonstrated the creation of a giant spin accumulation in Si up to room temperature in nonlocal (NL) devices with Fe/MgO tunnel contacts [1]. In the latter study, spin transport was studied in the low bias regime, where it can be described as purely diffusive using the standard theory for spin injection and diffusion. In the high bias regime, however, the presence of an electric field (E) in the Si channel causes spin drift that can strongly affect spin transport and spin accumulation voltages in Si [2,3]. Here, we study the spin transport in NL devices with a heavily-doped n-type Si channel in the presence of E in the high bias regime and quantify the effect of spin drift on the NL spin signals. The epitaxial Fe/MgO magnetic tunnel contacts were deposited by molecular beam epitaxy on a SOI substrate having a 70 nm-thick n-type Si(001) channel with a phosphorous concentration of 2.4 × 10^19 cm-3. The NL structures consist of two central ferromagnetic electrodes (FM1 and FM2) separated by a spacing d and two outer non-magnetic Au/Ti reference contacts. To study the spin-drift effect in the NL geometry, we use two different measurement configurations called NL1 and NL2, in which the direction of the electric field E is reversed, while the detector remains unbiased. To vary the strength of the electric field, we vary the current in the Si channel by either changing the bias across the injector, or by keeping the tunnel current density constant and changing the size of the injector tunnel contact. We first measure the NL spin-valve and Hanle signals in the low bias regime and extract a spin lifetime of 15 ns, a spin-diffusion length of 1.7 m, and a tunnel spin polarization of Fe/MgO of 42 % at 60 K. We then perform NL spin-valve measurements using both NL1 and NL2 schemes and different injector sizes at various biases. We demonstrate a clear modulation of the spin signal and the spin-transport length as a function of E. Using the spin-drift-diffusion model, we establish a simple and reliable method to accurately quantify spin drift and obtain the relevant spin-transport parameters as a function of the electric field or the bias current. Notably, we are able to enhance the spin-transport length by a factor of 3 and the magnitude of the spin signal up to 250 % at 60 K with a moderate electric field under the nonlocal detector of ~ -300 V/cm for spin injection condition. These findings show that spin drift is beneficial for spintronic purposes as it provides enhanced spin-transport length compared to the spin-diffusive regime. References: [1] A. Spiesser et al., Phys. Rev. Appl. 8, 064023 (2017), [2] Z.G. Yu and M. E. Flatte, Phys. Rev. B 66, 235302 (2002), [3] M. Kameno et al., Appl. Phys. Lett. 104, 092409 (2014).

Resume : SiGe alloys have been utilized in the field of Si-CMOS technologies. However, there is almost no information on spin transport properties in SiGe alloys. In this study, we demonstrate reliable spin transport (i.e., pure spin current transport) in SiGe-based lateral spin-valve (LSV) devices with Schottky-tunnel contacts. By using molecular beam epitaxy (MBE), an n-Si0.1Ge0.9 spin-transport layer (n ~ 5x10^18 cm^-3) was grown on a Ge/Si(111) virtual substrate. Then, we grew a Co2FeAl0.5Si0.5 (CFAS) ferromagnetic layer on the Si0.1Ge0.9 layer as a spin injector and detector. Finally, we fabricated LSV devices with n-Si0.1Ge0.9 spin-transport channel. Using four-terminal nonlocal measurements, we observed evident nonlocal magnetoresistance (?RNL) curves and Hanle-effect curves at 50 K, indicating reliable pure spin current transport in a SiGe alloy. We also obtained a spin diffusion length (?SiGe) of ~ 0.5 µm at 50 K by analyzing d dependence of ?RNL and Hanle-effect curves [1]. Because of the Ge-rich SiGe alloy (Si0.1Ge0.9), the value of ?SiGe was consistent with that of pure Ge at low temperatures [2]. We also observed two-terminal magnetoresistance in the SiGe-based LSV devices for the first time. This work was partially supported by JSPS/MEXT KAKENHI (No. 16H02333, No. 17H06120, No. 17H06832, and No. 26103003). [1] T. Naito et al., Appl. Phys. Express 11, 053006 (2018). [2] M. Yamada et al., Phys. Rev. B 95, 161304(R) (2017).

Resume : The rising interest of solid state community toward two-dimensional (2D) materials is driven by the capability of ab initio simulations to predict structural and electronic properties of these mono-layered compounds which have potential application in new 2D electronic devices ranging from molecular sensors to ultimate-scaled spintronic junctions. On the basis of first principles calculations we proposed and investigated a novel 2D material with honeycomb structure composed of SiGe random alloy with different Ge concentration, focusing our study on the structural, elastic, electronic and magnetic properties. We found that both with and without H passivation of dangling bonds, the lattice parameter of 2D-SiGe random alloy varies according to Vegard's law as a function of the concentration, making this material suitable as "seed ribbon" (i.e. the 2D analogous of conventional 3D substrates) in 2D lateral growth, therefore allowing the tuning of electronic properties of silicene and germanene (or of their H-passivized counterpart) in lateral 2D heterostructures, recently proposed in Ref.[1]. In particular, for partial H passivation, we found that this compound presents half-metallic properties with fully spin polarization at the Fermi energy, thus making this new 2D material a promising candidate for spin injection in 2D junctions and heterostructures. Further, for its tunable lattice and electronic properties, the H-passivized 2D-SiGe random alloy provides a excellent template also as 2D substrate for "more traditional" vertical 2D heterostructures, paving the way toward a ultra-scaled 2D electronics. [1] A. Debernardi, L. Marchetti, Ab initio simulations of pseudomorphic silicene and germanene bidimensional heterostructures, Phys. Rev. B., 93, 245426 (2016).

Resume : Recent experimental investigations have confirmed the possibility to synthesize and exploit polytypism in group IV nanowires. Indeed, while the crystal structure of Si and Ge nanowires (NWs) at standard conditions usually takes a well-defined cubic-diamond phase (as for their bulk counterparts), in the last few years several experimental observations of NWs exhibiting other phases - i.e. the hexagonal-diamond one - have been reported [1-2]. Driven by this promising evidence, here I will discuss recent first-principles calculations of the electronic and optical properties of hexagonal-diamond and cubic-diamond Si and Ge NWs as well as their homojunctions [3-4]. I will outline how a change in the crystal phase can strongly modify the electronic structure and optical response of the NW inducing novel and fascinating properties. Furthermore, I will show that, in the case of homojunctions, playing on crystal phase, size and length of the junction is an efficient tool to modulate band offsets and electron-hole separations. [1] S. Assali et al., Nano Lett. 15, 8062-8069 (2015) [2] J. Tang et al., Nanoscale, 9, 8113-8118 (2017) [3] M. Amato et al., Nano Lett. 16, 5694-5700 (2016) [4] T. Kaewmaraya et al., J. Phys. Chem. C 121, 5820-5828 (2017)

Resume : Recent experimental investigations have confirmed the possibility to synthesize and exploit polytypism in group IV nanowires. Indeed, while the crystal structure of Si and Ge nanowires (NWs) at standard conditions usually takes a well-defined cubic-diamond phase (as for their bulk counterparts), in the last few years several experimental observations of NWs exhibiting other phases - i.e. the hexagonal-diamond one - have been reported [1-2]. Driven by this promising evidence, here I will discuss recent first-principles calculations of the electronic and optical properties of hexagonal-diamond and cubic-diamond Si and Ge NWs as well as their homojunctions [3-4]. I will outline how a change in the crystal phase can strongly modify the electronic structure and optical response of the NW inducing novel and fascinating properties. Furthermore, I will show that, in the case of homojunctions, playing on crystal phase, size and length of the junction is an efficient tool to modulate band offsets and electron-hole separations. [1] S. Assali et al., Nano Lett. 15, 8062-8069 (2015) [2] J. Tang et al., Nanoscale, 9, 8113-8118 (2017) [3] M. Amato et al., Nano Lett. 16, 5694-5700 (2016) [4] T. Kaewmaraya et al., J. Phys. Chem. C 121, 5820-5828 (2017)

Resume : For extension of silicon photo detection range to mid-infrared region, several materials are proposed. Among them, the group IV alloy of germanium-tin material has shown promising characteristics extending the detection range to long wavelengths. Therefore we selected for study 400-nm thick Ge0.96Sn0.04 layers on silicon. The main material parameters important for photo-detectors are carrier lifetime, diffusion coefficient and diffusion length. We used contactless optical pump-probe techniques as differential transmittivity, differential reflectivity and light induced transient grating for that parameters determination. The determined free carrier lifetimes by differential transmittivity were in 20-30 ns range and weakly depended on excitation, whereas differential reflectivity decays provided fast 0.5 ns decays attributed to Auger and surface recombination with ~ 1500 cm/s velocity. The layer irradiation by 1064 nm laser for Sn concentration redistribution due to thermogradient effect as verified by XPS, Raman and reflectivity measurements did not induce appreciable changes of the recombination parameters. The determined diffusion length in range of few um proves GeSn material being suitable for effective infrared photodetectors. We acknowledge E. Kasper, K. Lyutovich from University of Stuttgart for provision of the samples. The work was supported as part of the Program on Mutual Funds for Scientific Cooperation of Lithuania and Latvia with Taiwan, project No. P-LLT18-6.

Resume : In the GAA geometry, horizontally-stacked Si or Ge nanowire or nanosheet channels are formed by selective etching of sacrificial SiGe layers epitaxially grown in SiGe/Si or SiGe/Ge multi-stacks, respectively. Owing to very aggressive target dimensions (e.g. sub-10 nm nanowire channel diameter), high etch selectivity and excellent process controls are mandatory. This sets stringent requirements on the epitaxial stacks (thicknesses and composition control, sharpness of interfaces) and on the etch process itself (high selectivity, limited channel material consumption). While processes for the selective etch of SiGe versus Si are available, the release of Ge channels remains more challenging. This step is usually done in advanced wet chemistries and is sensitive to both strain in and oxidation of Ge. For this reason, it requires a great precision in its adjustment, with the risk of experiencing process variabilities and yield issues. Furthermore, released channels should immediately be passivated to avoid Ge oxidation, which can cause severe material loss in the subsequent gate fabrication schemes. We propose an alternative for the preparation of suspended Ge horizontal nanowires. Our proposal relies on the combination of epitaxially grown GeSn/Ge stacks and a low-temperature vapor-phase GeSn etch process using Cl2. After an initial process screening on blanket layers, we demonstrate on patterned test structures that Cl2 etching enables high etch selectivity of n-GeSn towards Ge with high etch rates and throughput. In addition, the process to reveal the nanowires can be combined with an in situ passivation of newly exposed Ge surfaces, which provides perspectives for improved reliability and dimensions control.

Resume : To improve the performance of electronic and optical devices, extensive research efforts have recently focused on the effect of incorporating Sn into Ge. In the present work, we investigate how Sn composition x (0 ≤ x ≤ 0.12), film thickness t (50 ≤ t ≤ 300 nm), and deposition temperature T_d (50 ≤ T_d ≤ 200 °C) of the Ge_1−xSn_x precursor on glass affects subsequent solid-phase crystallization. Upon incorporating 3.2% Sn, which is slightly above the solubility limit of Sn in Ge, the crystal grain size increases and the grain-boundary barrier decreases, which increases the hole mobility from 80 to 250 cm²/V s. By increasing t from 100 nm to 200 nm, the hole mobility reaches 370 cm²/V s despite the grain becoming smaller. This suggests that the higher hole mobility with the thicker GeSn layer arises from the reduction of the GeSn/glass interface scattering. Furthermore, at T_d = 125 °C, the hole mobility reaches 410 cm²/V s, which is tentatively attributed to the formation of a dense amorphous GeSn precursor. This is the highest hole mobility for semiconductor thin films on insulators formed below 500 °C. These results thus demonstrate the usefulness of Sn doping of polycrystalline Ge and the importance of temperature while incorporating Sn. These findings make it possible to fabricate advanced Ge-based devices including high-speed thin-film transistors.

Resume : GeSn nanowires can be applied to metal-oxide semiconductor field-effect transistors (MOSFETs) and optoelectronics devices due to their high carrier mobility and the potential to achieve a direct band structure. Vapor-liquid-solid (VLS) growth with Au catalyst is widely employed to grow highly crystalline Ge nanowires [1]. Here, in this study, Au/Sn alloy was used as both Sn source and catalyst to guide Sn incorporation during the VLS growth of Ge nanowire. Sn nanoparticles and Au films with different thickness were deposited onto Si (111) substrate by a thermal evaporator. Then, those samples were loaded into a chemical vapor deposition (CVD) chamber to perform VLS growth of nanowires. Here, GeH4 was used as precursor gas under different growth temperature. Results of X-ray diffraction and Raman measurements proved that GeSn nanowires can be successfully grown with various conditions. Here, we could clearly say that higher initial Sn content in catalysts result in higher Sn incorporation and higher crystallinity of GeSn nanowires. Sample grown with Au 1nm/Sn 10 nm catalysts (360 °C, 20 min) showed the highest Sn content (5 at.%), which was estimated by Energy-dispersive X-ray (EDX) spectroscopy and confirmed by Raman measurement. [1] Fukata, N. et al. ACS Nano 4, 3807 (2010)

Resume : The E centers (dopant-vacancy pairs) play significant role in charge carrier compensation in semiconductors. In order to gain insight in the evolution of Pn-V clusters during epitaxial growth of in-situ phosphorus doped Ge, positron annihilation spectroscopy was performed on epitaxial Ge layers grown by chemical vapor deposition with different concentrations of Phosphorus (~1x1018-1x1020 cm-3). The increase (decrease) in S (W) line shape parameter with P concentration refers to an increased concentration of defects (vacancies) in the epi layers. The occurrence of Pn-V clusters leads to dopant deactivation in Ge. We corroborate our experimental results with electronic structure calculations for Pn-V clusters, corresponding positron states and annihilation characteristics were modeled as well. The fingerprint for the existence of Pn-V clusters in Ge:P layers is evident from their reduced contribution towards electron-momentum distribution in coincidence Doppler spectra. At high momenta (p > 1.4 a.u.), the Doppler spectra in Ge is dominated mainly by positron annihilation with 3d core electrons. The presence of P atoms around the vacancy instead of Ge atoms reduce the effective electron density due to missing 3d shell. This is turn leads to decreased positron annihilation probability with core electrons. The continuous decrease in intensity in high momentum region of Doppler spectra substantiates the evidence that the number of P atoms around the vacancy increases with P concentration.

Resume : Phase change materials (PCM) are a technologically important materials class. There are two mature main application areas for PCM. The first deals with their use in optical discs for data storage. The second application area is in electronics and is known as phase change random access memory (PRAM). These memory elements exploit the resistance contrast between the amorphous and crystalline phases. The prototype phase change material Ge2Sb2Te5 displays a cubic and a trigonal crystalline phase. In our group we pioneered the epitaxial growth of PCM using molecular beam epitaxy (MBE) and studied the ordering of the crystalline phases. Such materials are lamellar and display a pronounced bond hierarchy, featuring strong bonds within quasi 2D building blocks, while weak bonds link adjacent blocks. These weak bonds are frequently referred to as van der Waals (vdW) bonds. The weak interlayer interaction ? which causes the 2D nature of these materials ? is both a blessing and a curse. On the one hand, it allows the growth of heterostructures and superlattices of dissimilar 2D materials without epitaxial guidance (vdW epitaxy). Yet, it also creates adverse side-effects such as poor adhesion and wetting. More importantly, the weak coupling impedes strain engineering. Clearly, in the limit of zero coupling across vdW gaps, it should be impossible to introduce any strain in the growing 2D film. Yet, if enough coupling prevails across these gaps, strain engineering should be possible, too. The engineering of strain is an elegant concept to tailor physical properties without changing composition. Here I will discuss two examples of strain engineering in PCMs. The first method is to induce strain at the interface in the vdW epitaxy of GeTe-Sb2Te3 alloys by the employment of vicinal surfaces. The epitaxial layer is tilted along the growth direction with respect to the substrate, due to the out-of-plane lattice mismatch which introduces strain at the step edges, whereas the in-plane component is weakly bonded to the surface. Within the second method I will present evidence that p-bonded GeTe/Sb2Te3 superlattices (SLs), V2VI3 as well as their alloys devise a gap of non-pure vdW nature between the two chalcogenide atoms. Moreover, the larger coupling across the gap allows the tuning and engineering of strain. Most importantly, SLs of this class of materials develop a tunable distribution of in-plane lattice constants. Such a distribution of lattice constants within one solid has no precedent and is clearly beyond reach in classical 3D coupled solids.

Resume : Photonics has been the backbone for long range data communication since many years. More recently, the need for bandwidth drove the development towards a tighter integration of electro-optical systems, and silicon photonics became the baseline technology. As a key asset for the future of information and communication technology, silicon photonic integrated circuits will greatly benefit from the integration of novel materials. They should enhance performance or create functionalities not available in the standard materials set used in photonic foundries. Several oxide materials have very interesting optical properties and can nowadays be integrated with silicon while maintaining their superior optical characteristics. Polar materials such as ferroelectric oxides fall into this category. Such materials are extremely relevant because they are already in use as discrete components for long range communication, e.g. using LiNbO3 for electro-optical modulators. However, unless one can integrate such materials into silicon photonics, modulators will be limited to the use of the plasma dispersion in silicon. In this contribution, we show a path to integrate various BaTiO3 electro-optic devices monolithically with a silicon photonic platform.

Resume : The crystallite fraction of nano-crystalline Si:H (nc-Si:H) films deposited by means of hot-wire chemical vapor deposition (HWCVD) is investigated. The effect of Phosphorus and Boron as dopants on the film structure with varying levels of additional in-situ H2 gas are compared to un-doped control runs. All films were produced in a large scale inline coater using an array of 10 0.5 mm Tungsten wires with 50 mm spacing, kept at 2100 °C, with a filament to substrate spacing of 75 mm. The level of dopant gas flow relative to the Silane (SiH4) process gas flow was varied from 0.1 % to 1.0 % for n-doping (PH3), and from 0.01 % to 0.67 % for p-doping (B2H6). The crystallinity in the films was investigated via Raman spectroscopy, X-ray diffraction (XRD) and SEM imaging. The dopant concentration is found to have a strong effect on the hydrogen gas flow required for film crystallization. The increase of n-type doping corresponded to a shift of the onset of crystallinity by 500 sccm of H2 flow. Whereas the increase in p-type doping shifted the onset of crystallinity in the films, such that no crystallinity was measured in the 0.67 % B2H6 films within the working range of the device (5050 sccm H2 maximum flow). The final reachable value of crystallinity in the Silicon films also shows a strong dependency on the type and the level of doping. With an increase of doping corresponding to a strong decrease in maximum reachable crystallite fraction. The different behavior of P and B dopants may be explained by a size effect (misfit of covalent radii) or by a chemical effect.

Resume : Higher manganese silicide (HMS) have a complex band structure with multiple valleys close to the conduction and valence band edges. It has been recognized as the most promising p-type thermoelectric material for harnessing waste-heat in the mid-temperature range, owing to its earth-abundant, environmentally friendly nature and thermal stability of its constituent elements. It is well known that HMS exist as several incommensurable phases such as Mn4Si7, Mn11Si19, Mn15Si26, Mn27Si47, and all of these compounds are Nowotny chimney phases exhibiting tetragonal crystal structure. There are many methods to improve thermoelectric performance through compositional optimization by doping or substitution. In this research, we have investigated the effect of Sn doping on the microstructure and thermoelectric properties of polycrystalline HMS samples, which were prepared by arc melting and followed by spark plasma sintering (SPS). The electrical conductivity is slightly decreased then increase afterwards implying point defect was introduced by Sn doping. The Seebeck coefficient is increased as a result of further decreased hole concentration with 0.1 at% Sn doping. The abundant phonon scattering results in appreciably reduced thermal conductivity of 2.1 Wm-1K-1 at 748 K, which primarily contributes to the enhancement in ZT. The maximum ZT of Sn-doped HMS is 0.275 at 748 K. Key words: Higher manganese silicide, Thermoelectric properties, Tin, and Incommensurate phase