Although the local resistivity of semiconducting silicon in its standard crystalline form can be changed by many orders of magnitude by doping with elements, superconductivity has so far never been achieved. Hybrid devices combining silicon's semiconducting properties and superconductivity have therefore remained largely underdeveloped. Here we report that superconductivity can be induced when boron is locally introduced into silicon at concentrations above its equilibrium solubility. For sufficiently high boron doping (typically 100 p.p.m.) silicon becomes metallic. We find that at a higher boron concentration of several per cent, achieved by gas immersion laser doping, silicon becomes superconducting. Electrical resistivity and magnetic susceptibility measurements show that boron-doped silicon (Si:B) made in this way is a superconductor below a transition temperature T(c) approximately 0.35 K, with a critical field of about 0.4 T. Ab initio calculations, corroborated by Raman measurements, strongly suggest that doping is substitutional. The calculated electron-phonon coupling strength is found to be consistent with a conventional phonon-mediated coupling mechanism. Our findings will facilitate the fabrication of new silicon-based superconducting nanostructures and mesoscopic devices with high-quality interfaces.

No abstract

We study within a first-principle approach the band structure, vibrational modes and electronphonon coupling in boron, aluminum and phosphorus doped silicon in the diamond phase. Our results provide evidences that the recently discovered superconducting transition in boron doped cubic silicon can be explained within a standard phonon-mediated mechanism. The importance of lattice compression and dopant related stretching modes are emphasized. We find that TC can be increased by one order of magnitude by adopting aluminum doping instead of boron. PACS numbers: 74.62.Bf, 71.15.Mb, 74.20.Fg, 74.25.Kc The experimental discovery by Bustarret and coworkers [1] of a superconducting (SC) transition in heavily boron-doped silicon in the diamond phase (labeled c-Si in what follows) concluded a long history of research on such a transition in silicon-based systems and in doped semiconductors in general.[2] Until recently, a SC behavior in silicon had only been observed in high-pressure metallic phases such as the hexagonal or β-tin structures [3] or low-pressure cage-like doped clathrate phases. [4,5] This latter structure, which becomes superconducting upon heavy doping, bears much similarities with the diamond phase, as it is a semiconducting sp 3 network, but with a band gap in the visible range. [6] Besides doped sp 3 silicon clathrates, superconductivity in c-Si was recently made likely by the discovery of a SC transition in highly boron-doped carbon diamond.[7] Tunnelling spectroscopy,[8] reflectivity measurements [9] and first-principles studies within the density functional theory (DFT), performed in the virtual crystal approximation (VCA) [10,11,12] or a supercell approach, [13,14,15] strongly suggested that the transition was phonon mediated. Further, the study of the SC origin in Si-clathrates and carbon diamond led theorists to predict the SC transition in doped c-Si within a crude rigid-band model for doping [5] and a more accurate VCA treatment [10] with emphasis on boron doping.In this paper, we study by means of ab initio simulations within a supercell approach the electronic, vibrational and electron-phonon coupling properties of boron, aluminum p-doped and phosphorus n-doped c-Si. We find that a standard phonon-mediated BCS approach can account for the experimental transition temperature observed at high boron content. We predict further that aluminum doping would allow to increase T C by one order of magnitude, an effect ascribed in particular to the negative effect of lattice compression on T C in the case of boron doping.The calculations are performed within a planewave pseudopotential implementation [16] of the DFT using the PBE functional [17] for exchange and correlation. Ultrasoft pseudopotentials are used with a 20 Ry (160 Ry) cutoff for the expansion of the wavefunctions (charge density), increased to 25 Ry (200 Ry) in the case of borondoping. A (2x2x2) supercell containing 16 atoms is built with one Si atom replaced by an impurity, leading to a ∼6.25% doping concentration, in the range of the e...