Complex, expensive serial integration of the whole processing subsystem is giving way to less costly, lower-risk parallel integration of multiple functions.
John Marsh Intense Photonics
Integration is widely viewed as a critical technology for progress in optical components, but there is debate about the best way forward. Should it be serial integration of a complete processing subsystem on a single chip, or parallel integration; putting multiple iterations of similar or identical functions in a single package?
Given the current economic climate - with next-generation equipment designs on hold and a lot of component inventory still in the system - the parallel approach is viewed by many as a good low-risk strategy in the shorter term.
QWI allows a semiconductor's energy bandgap to be varied and lasers to be monolithically integrated with amplifiers, filters, switches and other passive devices on a single chip.
Different techniques include laser-induced, implantation-induced, impurity-induced and - in Intense Photonics' case - impurity-free vacancy disordering.
In the latter, optical absorption from electrically active dopants is avoided. Two dielectric caps are deposited on the surface of the semiconductor: one creates vacancies on the group-III lattice site; the other has no effect on the underlying semiconductor.The vacancies diffuse through the semiconductor as a result of heating. Individual atoms hop from one lattice site to another, intermixing the quantum well with the adjacent barrier material and hence widening its bandgap (Fig. 1). The wider-bandgap structure can be used to implement further functions, such as passive waveguides in a laser array (Fig. 2).
A key advantage of QWI over both alternative monolithic integration techniques, such as re-growth, and hybrid integration is that active elements are in alignment with their passive counterparts, so there is negligible reflection at the active/passive interface.
One of the first applications to be tackled by Intense Photonics is pumping waveguide amplifier arrays, for lower-cost products for markets such as metropolitan area networks, having developed the IP0980 monolithic multi-device array of gallium arsenide-based 980nm pump lasers for EDFA-type applications.
Designed to simplify systems integration, the array may be specified with a range of application-specific parameters to suit individual systems. These include the number of sources (up to 10), device pitch and interface angles (to simplify connection to arrayed waveguide gratings in a planar lightwave circuit), power selections of up to 250mW (to suit different pump applications), and n- or p-side down mounting.
Although the device is basically a straightforward replacement for existing pump lasers, integration techniques have been used not only to increase laser count but also to optimise manufacturability, yield and reliability.
In addition to the laser sources, the device integrates two types of passive element: "non-absorbing mirrors" (NAMs) and waveguides.The NAMs combat hot spots, tackling a common laser failure mode: catastrophic optical damage caused by energy absorption at the laser facet, which can be aggravated by micro-positioning inaccuracies of the active chip onto the sub-mount.
The passive waveguide brings several benefits. No heat is generated in the passive region. This further relaxes manufacturing tolerances, because the passive sections can overhang the heat-sink without impacting reliability. This waveguide feature additionally supports the option of p-side-down mounting if required by the application - a technique which optimises heat transfer by reducing the distance between the active region and the heat-sink.
Compared with the silicon industry, III-V semiconductor material processing is not as advanced, and yields are somewhat lower. Monolithic integration processes for III-V materials rely on re-growth - with yield reductions accumulating at each stage.
In comparison, I believe Intense Photonics' QWI integration approach enables multi-function devices to be produced in a single processing stage while establishing new levels of manufacturing yield and reliability for photonics ICs. It will help the industry in its attempts to break out from the telecom backbone to reach the metro and desktop/ fibre-to-the-home markets.
John Marsh
Chief Research Officer
Intense Photonics
[email protected]
John Marsh is Chief Research Officer with Intense Photonics www.intensephotonics.com