(Invited) Theory of Silicon-Based Light Emitting

Abstract

The indirect band gap preclude the silicon from making light-emitting devices, which is, however, a quest for exploration of large-scale integration of both electronic and photonic devices on a single microchip. Why silicon and germanium are both indirect band gap and most of group III-V and all group II-VI semiconductors are direct band gap is a long standing puzzle. In this talk, I will first present our recently developed unified theory, based on occupied d orbitals in cations, for explaining why diamond, Si, Ge, and Al-containing group III-V semiconductors have indirect band gap, and the remaining group III-V and II-VI semiconductors, except GaP, have direct band gap. This theory also unravels that it is impossible to make allotropes of Si and compounds consisting by Si mixed together with other elements high efficient direct bandgap light emitting. Interface and quantum confinement in low-dimensional Si nanostructures are two knobs to relax the momentum conservation law of optical transition. Our theory confirms that the light emission efficiency of Si quantum dots increases exponentially as reducing dot size, but its highest reachable efficiently is still two orders of magnitude smaller than that of direct bandgap InP and GaAs. Fortunately, we discovered that Si/Ge magic sequence superlattices exhibit orders more efficient at emission light than their existing counterpart records and approach more than 10% brightness of real direct gap materials, such as GaAs. Furthermore, regarding the direct bandgap is only 140 meV above the indirect bandgap in Germanium (Ge), it is well established that a tiny (1-2%) extensile strain could render an indirect-to-direct bandgap transition and makes it direct bandgap. However, there is no feasible way to create such extensive strain on Ge since the lattice parameter of Ge is even 4.3% larger than Si. We propose insert external atoms into interstitial sites of Ge, which grows on Si substrate through SiGe buffer layer, to induce an effective tensile uniaxial strain on Ge, a new pathway toward direct band gap Ge. Because our proposed scheme is highly compatible with CMOS fabrication process, our presented new pathway would be a solution for on-chip lasers.