A vital issue in the establishment of an ultralow anti-reflection coating process is detailed knowledge of the "substrate index" to which the coating must be carefully matched. A laser is a three-dimensional structure, where the refractive index of the waveguide is a complex function of its precise elemental composition and its geometry. Clearly, any coating design must include a healthy tolerance for fluctuations of these parameters. Nonetheless, despite all of these difficulties, facet reflectivities of <10-4 have been achieved at Helia by methodical calibration and appropriate monitoring approaches.

In addition to the obvious problems in fabricating such coatings, there remain considerable difficulties in the assessment of such small reflectivities of these diode facets. There are a number of methods of characterising such coatings, including analysis of threshold current change pre- and post-coating. However, it is widely regarded that the most reliable method of reflectivity characterisation is by examination of the longitudinal mode fringes - so-called “gain ripple”. Such methods use a measurement of the emission spectra to ascertain the coated facet reflectivity at the lasing wavelength, given knowledge of the uncoated facet’s reflectivity and the output parameters prior to coating. An example of the gain ripple change evident from a laser diode before and after coating is shown in Figure 3. One can see that in an uncoated forward-biased diode, the emission exhibits clear Fabry-Perot modulation caused by the two uncoated facet reflectivities (typically ~ 30%). In the AR-coated diode, there appears to be a greatly reduced gain ripple. In this example shown in Figure Y, the facet reflectivity is ~10-5. Methods which use variations of this approach, such as Kaminow (see I.P. Kaminow et al., IEEE Journal of Quantum Electron., vol 19, p493) are widely regarded as being reliable for reflectivity levels well below 10-4.

Helia engineers have spent many years perfecting facet coating using the IAD technique - with the major technological challenge being the transfer of such processes to high volume production. The most efficient method of coating laser diodes is the coat them in bar form, ie prior to cleaving into individual laser diode chips. Each bar is an array of, typically, 20-50 chips in total. An important parameter in the coating of laser diode bars is the level of non-uniformity evident along the length of a bar (typically 8-16mm). In order to produce anti-reflection coatings on bars to 10-4, it is necessary to have coating thickness uniformity to better than ~0.1% - this itself can be a challenge and requires careful consideration of the source, ion gun and laser geometries. This requires appropriate methods of holding the bars inside the coatings chamber, taking into account the possible thermal distribution caused by the evaporant source material. There are many other important factors, including coating overspill onto the top of the bar - a critical parameter as electrical contacts will be made later when the devices are in chip form. Other forms of overspill can be even more serious, such as overspill onto the back facet. Helia’s engineering team has developed sophisticated tooling methods to eradicate these problems.