Superluminescent diodes have an eye for diagnosis

Healthcare costs are in the spotlight. The US is wrestling with the proposed introduction of universal access and many other countries are looking at the financial burden of their burgeoning state-run systems.

Reducing the cost of diagnosis and improving treatment efficiency are areas of major focus. One of the largest increases in costs in recent years has been the use of diagnostic imaging tools such as ultrasound, computed tomography (CT) and magnetic resonance imaging (MRI), access to which can cost several thousand dollars per examination. Diagnostic imaging based on relatively low cost photonics could be the answer in some applications.

Whilst not a direct replacement for tools like MRI, near surface photonic imaging techniques can have an important place within certain diagnostic schemes. Even for internal imaging where the lack of light penetration can be an issue, there is much invention to be gained using optical probes to bring the source light into the region to be interrogated. Furthermore, if the applications are intended to be used for diagnosis in vitro rather than in vivo, these restrictions are considerably relaxed, since then specimens can be prepared in which penetration depth is not an issue.

Non-contact near surface imaging of the skin or eye using Optical Coherence Tomography (OCT) is an example of a technique that has become popular both in the lab and in the clinic. It has gained considerable favour for its potential to provide two and three-dimensional real-time images of biological tissues with micron-like resolution. The technique can be described as an optical analog to ultrasound in which light is used instead to probe the variation of reflected light as a function of depth. Combined with areal scanning it is relatively easy to obtain useful cross sectional images of skin and tissue.

The basis of OCT is low-coherence interferometry in which a broadband source illuminates a Michelson interferometer. The system is very easy to implement in fibre and the ideal sources for such applications are the edge-emitting superluminescent diodes. The axial resolution of such systems is inversely proportional to the bandwidth of the source, which explains the strong desire for broadband sources.

Aside from fibre optic test equipment, the edge-emitting superluminescent diode has served relatively few applications over the years. But this device may well have found its niche as a broadband source in the near infrared (800-1300nm), the wavelength range most suited to the OCT application. Companies such as Exalos, Superlum, and Inphenix have emerged as suppliers of sources for the emerging OCT supply chain. Yet these products are generally based on InP or GaAs-based quantum well technology, and realising a significantly wide broadband output from this class of device is very challenging. However following the lead from researchers in EPFL, who were later to found Exalos, we became interested in exploiting the properties of quantum dots for such applications and have been working on these devices within the European FP6 project ‘Nano-UB sources’ together with OCT technology partners.

Indium Gallium Arsenide quantum dots (QD’s) have many useful properties for this application. They are naturally broadband, exhibiting a statistical variation in size and shape within a typical ensemble. In addition, under increasing carrier injection the QD ground state can saturate in favour of emission from a first, or subsequent excited states. Combining just the ground and first excited state (GS and ES) emissions can give a broadband output of typically around 100nm.

But there are two complicating factors. Firstly efforts have to be introduced to reduce the dip between emission states otherwise a multiple peaked image results in multiple or ‘ghost’ OCT images. This is done by introducing a variation into some of the quantum dot layers in the structure to progressively shift (or ‘chirp’) the emission. In the MBE growth that is carried out at the University of Sheffield, this is achieved by varying a structural parameter, usually the position of the QD within a surrounding quantum well.

The second problem with quantum dot structures is the relatively low gain. Gain values of 20-40cm-1 are typical of multi-quantum dot active layers whilst a single InGaAs/InP quantum well can offer 50-100cm-1. Despite this, it is relatively easy to obtain low threshold and high power output from QD lasers because the losses can also be very low.

However the same is not true of superluminescent diodes which operate on single-pass amplification of spontaneous emission, rather than multi-pass feedback to induce stimulated emission. A quantum dot superluminscent diode is therefore a long device (typically 4-6mm) in comparison to a laser diode (1-2mm) and it needs to operate under higher injection and therefore higher thermal loading.

Project partners Alcatel-Thales III-V lab and Optocap Ltd have fabricated and packaged superluminescent diodes targeting two wavelengths, one an established market segment around 1300nm and one at the shorter wavelength of 1050nm which is more suitable for ophthalmology.

Output powers (ex-facet) of up to 160mW and bandwidths of 115nm have been observed, although it is hard to obtain high power and broad bandwidth at the same time. For the more commercially viable 1050nm devices, a more typical result would be the measurement of 130mW ex-facet and 65mW ex-fibre packaged from a device exhibiting ~70nm bandwidth. Such a device was exhibited by Alcatel-Thales III-V lab at the Munich CLEOEurope meeting in 2009. The QD technology is very competitive against any other alternatives at this less developed wavelength.

Development of compact fibre-optic OCT diagnostic tools based on III-V quantum dot superluminescent diode sources has been undertaken by colleagues at the technical University of Denmark and the University of Cardiff. Two major markets exist within dermatology and ophthalmology, for which high power and broad bandwidth sources can give unprecedented enhancement in the ability to resolve the structural morphology of human tissue.

In one such example, a 3D OCT image of normal human retina, made up of 512 individual cross-sectional scans is shown right, together with that of a single scan below. The image shows the fovea (to the right), peripheral parts of the optic disc (to the left) and has good depth penetration into the retina as a consequence of the 1060nm centre wavelength of the source.

So what next for OCT? The medical advances are clear: in the hands of qualified medical clinicians, the technique is already playing a role in early-state cancer diagnosis in skin and in the monitoring of ocular diseases that are the world’s leading causes of blindness. But the technology does not stand still and neither do the source requirements. Attention is now turning to high speed OCT scanning performed by a tuneable laser source rather than the movement of an interferometer mirror. Low cost, widely sweeping sources need to be developed.

Furthermore the benefits of multispectral imaging are now being realised by imaging with a linear array or even full 2D cameras. So as the application moves on, perhaps the simple superluminescent diode has had its day already. However broadband gain materials, such as quantum dots will play an important role in future devices for this application.