16 Mar 2016
1310 nm emitting device could be the key to silicon photonics, say University College London researchers.
A team of researchers in the UK has grown a reliable laser diode on a silicon semiconductor wafer, suggesting a significant breakthrough for photonic integration.
The University College London (UCL), Cardiff University and University of Sheffield partnership used molecular beam epitaxy (MBE) to grow a gallium arsenide (GaAs) quantum dot laser structure directly on the silicon wafer. It emits close to 1310 nm, one of the wavelengths widely used for optical communications.
And unlike previous attempts to produce similar devices, the laser also operates at high temperatures and with a low threshold current – the kind of properties needed for commercial deployment.
UCL’s Huiyun Liu told optics.org: “Although lasers monolithically grown on silicon were demonstrated previously, the lifetime of these lasers is very short because of the high defect density generated at the interface between III-V and silicon substrate.”
“The short lifetime limits the practical application of these silicon lasers. The significance of this work is that we demonstrate the first practical III-V laser monolithically and directly grown on silicon substrates. This could be the key for silicon photonics.”
According to the team’s Nature Photonics paper, published last week, they were able to suppress the impact of threading dislocations in the semiconductor structure by depositing a nucleation layer between the silicon base and the light-emitting material.
They also grew “strained-layer superlattices”, which act as dislocation filters and confine the dislocations to the part of the structure nearest the silicon wafer and furthest from the light-emitting part, and used a series of thermal annealing steps to minimize the dislocations.
Liu’s colleague Alwyn Seeds added: “We previously published lasers grown on germanium on silicon wafers, an approach taken up by other groups worldwide. The new approach does not require the use of germanium on silicon and so has much better compatibility with silicon photonics and silicon electronics. Other groups, for example Intel, currently base their approach on wafer bonding compound semiconductor material to silicon.”
Seeds also points out that accelerated lifetime tests suggest that the device should operate for more than 100,000 hours, which is unprecedented for a laser structure on silicon. “This is a laser epitaxially grown direct on silicon, with a threshold current comparable to lasers grown on native substrates, with high output power (>100 mW) demonstrated [and] high-temperature (120°C) operation," he said.
In their paper, the team adds that the laser performance should be even better once standard industry techniques like hard soldering the device to a heat-sink and facet coatings are incorporated.
Photonic integration potential
If the breakthrough could be replicated on a volume scale, it may come to represent a major breakthrough in the development of truly integrated photonic circuits.
While a large number of photonic devices and functions, like detection, amplification and modulation of light, can already be performed at the level of the silicon chip, producing a laser directly on silicon material has always been a major stumbling block.
Silicon itself does not emit light with the kind of efficiency that would be needed in commercial photonic integrated circuits (PICs), so developers have previously resorted to schemes such as unusual doping, fabricating porous silicon, and bonding together indium phosphide and silicon wafers after first producing device structures on each.
“Our results demonstrate that the large lattice mismatch between III-V materials and silicon will no longer be a fundamental hurdle for monolithic epitaxial growth of III-V photonic devices on silicon substrates,” the UK team states in its paper. “These results are a major advance towards reliable and cost-effective silicon-based photonic–electronic integration.”
The next steps for the team include demonstrating the technology on a larger wafer platform that is closer to the sizes used in high-volume silicon microelectronics, ultimately with a view to commercializing the approach.
Liu said: "The scale-up [of] wafer [diameter] should not be a problem as this kind of procedure is very mature. I think the next big challenge should be to incorporate these lasers into a silicon waveguide and hence other silicon photonics components."
”We are, of course, aware of the large commercial opportunity for this technology, which is protected by appropriate IP,” noted Seeds, with Liu adding that the potential pathways for commercialization were “under discussion”.