John Marsh chief technology officer, Intense says integration is vital in reducing costs
The laser industry is starting a slow process of evolution, from a collection of small specialised companies offering niche products to a mature industry led by a few large players. As the industry becomes mature, it is becoming volume-driven and photonic integration is one aspect seen as vital in reducing costs. However, it is also an industry where standards are slow to emerge, so any integration technology is presented with considerable challenges. Integration technologies need to be versatile and offer rapid turnaround from concept to prototype samples and through to production volumes. Photonic integrated circuits need to be built from standard basic building blocks, yet also accommodate customisation. Although volumes might appear large by the historical standards of the photonics industry, they would be seen as small in the electronics industry, emphasising the need to control development and production costs.
John Marsh believes integration is the key
Photonic integration, particularly monolithic integration, has been a research topic for around 20 years. Historic research has been driven by the demands of the telecommunications industry, particularly for deployment in the access network. Most technologies involve the use of selective etching and masked growth steps during wafer production, generally with several stages of epitaxial growth. It is clear most of such projects have simply been too technically ambitious – despite considerable worldwide efforts, multiple etch and regrowth steps have been unsuccessful in producing the satisfactory yields that provide a low-cost solution for mass markets.
There has also been over-ambition in integrating electronic with photonic devices. Because transistors can be made in the same semiconductor materials as optoelectronic components, it is not necessarily a good idea to do so. The first question that should be asked of electronic/photonic integration is what real or potential need is being met. The physical structures of transistors and optoelectronic components are quite different, so bringing them together is likely to compromise performance heavily. In terms of functionality, the benefits of integrating driver transistors with lasers are small. Arguably the only area where functionality is enhanced is in receivers, where the integration of a photodiode in the first stage of amplification can enhance sensitivity and electrical bandwidth. The net result is that, until recently, the only integrated component in widespread use was the electro-absorption modulated laser; this is now being joined by a variation – the tunable laser and the tunable laser integrated with a modulator. These are deployed in the backbone telecom network in which, despite enormous pressures on component manufacturers, data aggregation across many users can still support significant costs.
Where photonic integration is becoming successful is in making multiple copies of the identical or very similar components, particularly laser arrays and photodiode arrays. This mirrors the development of silicon integrated circuits in which many copies of a single component – the transistor – are integrated. In telecoms, laser arrays are now found in certain designs of tunable source. A further level of integration is appearing in the long-haul network, with photonic integrated circuits containing either arrays of single wavelength lasers or arrays of modulators along with multiplexers. However, it is outside telecoms where laser arrays are making their greatest impact.
An innovative technology development called Quantum Well Intermixing (QWI) has been very successful in driving integration technology for photonic integrated circuits. QWI is a fundamentally different integration technology from the older regrowth or selective growth processes as it takes place completely post growth. This allows device processing to be separated completely from epitaxy; in turn this brings about significant economies as separate specialist foundries can be utilised for epitaxy and processing.
In lasers QWI is used to integrate passive regions at the facets of emitters, arrays, and bars giving excellent high power performance, enhanced reliability and superior manufacturing yield. For a photonic integrated circuit containing 100 components to be viable, the individual component yield must be >99 per cent. QWI is also effective in integrating multiple semiconductor components into complex, monolithic photonic integrated systems.
By separating epitaxy and processing, QWI allows for the rapid and cost effective development of versatile optical systems. As an example, multi-element laser chips containing around 100 singlemode, high-power emitters can be packaged with drive ASICs and micro-optics in small form-factor modules to form the building blocks of complex optical systems. These systems are changing the dynamics of existing products that utilise lasers and are also creating whole new market segments for photonic systems. The technology is now being deployed in a wide range of applications that require the precise delivery of optical energy.
As the cost of photonic integrated circuits falls, new markets are opening up. QWI technology provides an exciting alternative to the older epitaxy-oriented approaches to photonic integration, separating epitaxy from wafer processing and delivering the yield and reliability that make integration viable. Precise delivery of light and energy is required in a host of industrial and consumer products, and it is certain that scalable, volume-driven integration technologies such as QWI will revolutionise the photonics industry.