13 Jul 2005
A Swedish start-up has found an ingenious way to protect the surface of semiconductors and glass. Its Nitrel treatment could boost the performance of laser diodes, photodetectors and optical crystals, reports Oliver Graydon.
A small Swedish firm believes that it has found a way to fabricate semiconductor lasers that can be driven at much higher temperatures and output powers than ever before, without failing. If the claim of Stockholm-based Comlase turns out to be true, it could have dramatic consequences in fields ranging from data storage and telecoms, to fibre pumping and laser welding.
Until now, the maximum output power and temperature of operation of high-power laser diodes has been severely limited by the onset of a failure mechanism called catastrophic optical mirror damage (COMD). Put simply, COMD is irreversible thermal damage to the surface of the laser chip facets, which act as mirrors. The problem is often caused by facet oxidation or manufacturing defects that absorb light and act as a "hotspot". It is also the main reason behind the poor manufacturing yield of high-power lasers.
Today, edge-emitting laser diodes are made by cleaving a semiconductor wafer into individual chips. The cleaved ends are then covered with thin-film coatings - one end with a high-reflectivity (HR) coating, the other end, antireflection (AR) - to form the laser's rear mirror and front output coupler.
The problem is that, after cleaving, the exposed facets are highly chemically reactive because the bonds of the atoms in the surface layer have been physically torn apart, creating "dangling bonds". If these come into contact with even low concentrations of moisture or oxygen, they oxidize to create a light-absorbing hotspot that can lead to COMD.
Although the conventional thin-film coatings that are applied to the facets do offer some protection against moisture and oxygen, they are not a durable solution. For example, aluminium oxide, which is often used as a coating, is relatively permeable to water and is prone to moisture ingress over time.
To date, approaches to overcoming the problem have focused on depositing a passivation layer, such as silicon (in Bookham's E2 process), on top of the raw facets to protect them prior to depositing the thin-film reflection coatings. However, the process is only effective at certain wavelengths.
"Applying a thin layer of silicon works very well for lasers operating at 900-980 nm, but as soon as you go to lower wavelengths, you get negative effects," explained Alfred Feitisch, Comlase's CEO. "Our Nitrel passivation process is applicable to any semiconductor material - going all the way from GaN in the blue, to the infrared." The Comlase "Nitrel" process effectively tidies up all the dangling bonds on the raw facet surface so that they cannot oxidize and act as a catalyst for COMD. Conventional high-reflection and antireflection thin-film coatings are applied to the treated facets after.
"What Comlase does is atomically seal the surface by taking off the oxygen and substituting it with nitrogen atoms," said Feitisch. "We actually form surface nitrides with the dangling chemical bonds on the surface. This renders the surface chemically stable so that it cannot react."
The process is performed in a specially designed ultra-high vacuum chamber (reactor) that features two electron guns and an ion gun for bombarding the facets of the laser with a stream of nitrogen ions. The nitrogen ions polish the surface smooth, remove any oxides and then seal it with a nitride layer. The electron guns are then used to deposit the HR and AR thin-film coatings onto the rear and front facets by e-beam evaporation.
"The reactor has a carousel that puts a stack of 50 laser bars in front of the ion gun at a time. We then flip them over and do the reverse side. Then we lay down the HR and AR coatings," said Feitisch. "We are now building a new high-volume chamber that will have at least three electron guns and will be able to process 800 laser bars in a single run."
The benefits from the process are substantial, according to lifetime test data collected by Comlase. It says that multimode 805 nm AlInGaAs laser diodes fabricated with the Nitrel process showed no degradation in performance over a period of 9000 h when driven at a 60-80 W power level at a junction temperature of 90 °C. In comparison, untreated diodes driven under the same conditions degraded rapidly, with three-quarters of the batch failing before reaching 6000 h.
Tests with InGaAs diodes at the longer wavelength of 980 nm also showed great improvements in performance. "We've put them on life tests at 180 mW/μm [emitter width] and seen no degradation over thousands of hours," said Feitisch. "In short, what this means is that if you take a 100 μm-wide single-chip emitter, [it] would be a reliable 18 W laser chip, which is enormously important for pumping high-power fibre lasers."
As for applications for the process, Feitisch says that it would be ideal for making high-power infrared (IR) lasers for pumping solid-state lasers, such as the US military's mobile 100 kW laser weapon, which is currently under development. The ability to run diode lasers at much higher temperatures (90 °C, instead of room temperature) and higher output powers should dramatically simplify the cooling requirements for the pumps and minimize the number that are needed.
"In theory, you could run everything on a car-engine cooler, rather than using refrigeration equipment," explained Feitisch. "We have set up a US subsidiary in Delaware so that we can deal more easily with the US military and government."
Fibre lasers and thin-disc lasers could also benefit from diodes with improved reliability and power levels. Ultimately, the technology could result in bright, compact diode bars that are ideal for performing direct diode welding and surface treatment.
The improved reliability of the lasers could also be a big benefit for other mission-critical applications. "We are in talks with the European Space Agency to design and qualify a special laser bar for spaceborne applications and hope to have a project started next year," said Feitisch. "After all, if a laser fails on a satellite in space, it's a big problem."
Other potential applications include raising the output power of blue and red lasers used in optical storage industry. The result could be faster writing and reading of data onto CD-ROMS, DVD and BluRay discs. Comlase says that it has already received enquiries from Sony and Hitachi, who are both interested in the technology.
Although the company is initially focusing on applying its process to semiconductor lasers, Feitisch says that there is no reason why the Nitrel process couldn't be applied to semiconductor photodetectors, such as avalanche photodiodes (APDs) and pin diodes, to help reduce their noise. "These devices have dark current [noise] issues that our process should be able to eliminate," said Feitisch. "Nitride is an electrical insulator and can prevent electrons from flowing over the surface of these devices, which causes the dark current. We want to investigate this idea further."
In the case of APDs, the Nitrel process allegedly lowers the risk of voltage breakdown over the junction region, thereby reducing the likelihood of device failure.
And that's not all, Feitisch says that the process could potentially be used to protect glass components such as fibres and crystals. "For example, you could treat the end of optical fibres before laying down an antireflection coating, in order to get extremely good adhesion and remove any surface imperfections."
Another possibility is treating borate crystals, which are used for nonlinear optics, but are very sensitive to moisture (hygroscopic). By creating a thin, protective film of nitride on the surface of a crystal, it could be protected from water vapour.
Comlase has no shortage of possibilities for its technology, but as it has a staff of just 20, the firm is having a tough time deciding which applications to pursue first. "Before we explore other opportunities, we need to get some traction with laser manufacturers and secure revenue," said Feitisch. No firms have licensed the Nitrel process yet, but he says that it is currently being trialled by three parties and is optimistic about the outcome.