06 Mar 2006
nLight is confident that no fundamental barriers stand in the way of squeezing 1 kW out of a single 1 cm diode laser bar. James Tyrrell speaks with Paul Crump to find out why.
Diode lasers may be more efficient and more powerful than ever before, but according to nLight there is a lot of mileage left in the technology. The message from the US firm at Photonics West's busy LASE conference was that there are no fundamental barriers to achieving peak powers of 1 kW per 1 cm diode laser bar. Commercially, this means that devices could operate reliably at 300 W per bar, more than double today's figure of around 125 W per bar.
"Peak optical power from single 1 cm diode laser bars is advancing rapidly across all commercial wavelengths," Paul Crump, nLight's director of device technology, told the audience. "Critical improvements have been a reduction in operating voltage and a reduction in packaging thermal resistance, together with advances in facet passivation." As discussed in detail below, increased power density offers big benefits to the customer and will help to open up new applications, for example, in direct material processing.
To identify just how far the technology could go, Crump and his colleagues devised a thermal model of a diode laser based on known characteristics of laser bar material and packaging. The prediction showed that peak powers of 1 kW per 1 cm diode laser bar were well within reach.
Back in the lab, the group benchmarked its current diode laser technology. Engineers obtained peak continuous-wave powers of 400 W at wavelengths from 800 to 980 nm by running single 1 cm diode laser bars (with a 3 mm cavity length and 80% fill factor) at close to 500 A. Mounted junction-down on copper micro-coolers using indium solder, the devices were operated with a cooler flow rate of 0.5 l/min and tested at 5 °C. Crump expects that these bars will become commercially available with a reliable output power of at least 150 W per bar, as these devices are operated typically at 30-50% of their peak power.
With a peak power of 700 W per bar as a realistic near-term goal, Crump explained that nLight has grouped its efforts around key device parameters, one of which is operating voltage. At a fundamental level, the minimum possible operating voltage for a diode laser is set by the band-gap of the quantum well. This is the minimum bias to "turn on" the diode. In reality, the applied voltage has to be much higher because charge carriers must travel through many layers within the device before reaching the quanum well. Junctions between dissimilar materials, so-called heterobarriers, are a major hurdle to designers looking to reduce operating voltage.
Back in 2004, when OLE spoke with nLight's Jason Farmer, now the firm's chief technology officer, one concept was to inject charge carriers laterally using filled trenches and approach the active region from the side. The plan was to use these deep grooves to bypass the heterobarriers, not a bad idea when you consider that semiconductor lasers can feature 10 or more dissimilar layers.
However, as Crump told OLE this month, side-injection has resistance issues of its own. "By laterally injecting into the quantum well area, you are trying to flow charge across a very small contact area," he explained. "Unfortunately, the associated increase in electrical resistance was just too high."
Tackling the voltage drop problem layer-by-layer proved to be more fruitful. Through increased understanding of the growth process and control of gases in the reactor chamber, nLight has improved the deposition and electrical contacting between layers within the device. This approach turned out to be more successful than the radical side-injection concept, driving down operating voltage and pushing up device efficiency.
To date, nLight has engineered 73% efficient 100 W (975 nm) bars on copper micro-channel coolers, with peak performing single emitters reaching 76%.
At the same time, the company has been busy optimizing heat-sink performance. Through a combination of operating and design parameters including water flow-rate, channel cross-section and location, the team has pushed thermal resistance down to figures of less than 0.2 K/W.
Put simply, this means that the temperature of the device will increase by just 0.2 K for every watt of excess heat left in the device. For example, a 73% efficient 100 W bar emits 100 W of optical power and leaves 37 W of "waste" heat in the device. This leads to an increase in temperature of <8 °C. Going back to nLight's original analysis, 1 kW devices would require a heat-sink thermal resistance of 0.14 K/W and Crump is encouraged by current progress.
The firm has tested single emitters at very low temperatures, reaching a power conversion efficiency as high as 85% at a temperature of just -50 °C.
However, despite the benefits of better thermal management and lower operating voltage, higher performance does have its price. At high output powers, catastrophic optical mirror damage (COMD) is a real threat to reliable operation. COMD occurs at the surface of the laser chip facets, which act as mirrors. Failure can be triggered by facet oxidation or by manufacturing defects that absorb light and act as a "hotspot". In response, diode makers protect their lasers with a passivation layer deposited over the raw facets which suppresses COMD.
With its vertically integrated approach to device fabrication (see "About nLight"), nLight can perform this critical and commercially sensitive process in-house. Crump was reluctant to go into detail, but explained that passivation can offer protection up to an optical power density of at least 140 mW/μm. Crump and his colleagues put the process to the test using a 50 μm stripe, 1.5 mm cavity, 800 nm diode laser and measured peak power at 10 °C heat-sink temperature. The treated device was bonded junction-down on an industry standard c-mount and rolled over thermally at 7 W.
In comparison, and according to the firm's analysis, a 1 kW diode laser bar would deliver an optical power density of 126 mW/μm. This figure is below the rollover limit shown in the lab and builds on a feeling of strong progress within the industry, especially in areas such as power conversion efficiency.
Efficiency has become the metric of the moment for the diode laser community thanks to DARPA's SHEDs initiative. Launched in autumn 2003, the goal of the US Department of Defense's super-high-efficiency diode sources (SHEDs) programme is to create diode lasers that are 80% efficient at converting electricity into light. Due to deliver in September this year, the 36 month programme targets 480 W diode stacks operating at 50 °C.
With its focus on battlefield laser weapons, the military is keen to drive down the waste energy released from diodes to allow smaller and more manoeuvrable refrigeration units. However, as all diode makers involved in the project are aware, the knock-on effect of more efficient diode lasers will reach way beyond defence applications.
"The SHEDs programme is one of those win-win situations where the US military gets access to new technology and the commercial sector benefits from more affordable and more powerful devices," said Crump. "It ultimately means that you can have more single diode products - handheld versions for photodynamic therapy, acne treatment, tattoo removal or even laser surgery, for example."
It is not just the reduction in size that will benefit consumers. Bringing more high-power devices on to the market means shorter treatment times for medical uses or improved throughput in production applications. Pushing up the power per bar reduces the cost per watt and stimulates new markets. Looking at commercial devices, the price per watt of diode lasers has come down from around $2000 (€1675) in the 1980s to less than $30 today.
As Crump goes on to explain, diode lasers are now being considered for automotive and even TV applications. Diode lasers offer a compact source of near-infrared emission and suit automotive night-vision. As OLE discovered last year, car maker DaimlerChrysler has come up with a prototype night-vision system based on a high-power 808 nm laser diode and a CCD camera. Devices could also find their way into vehicle ranging and intelligent parking systems.
What's more, the benefits of SHEDs technology can be migrated broadly to devices operating at other wavelengths, helping nLight achieve high bar powers from 635 to19xx nm and enabling many other applications. For example, longer wavelength versions available up to 19xx nm can be used for eye-safe communication and atmospheric sensing.
Crump explained that by tackling parameters such as passivation, thermal resistance and operating voltage separately, nLight's engineers are well on the road to that magic figure of 1 kW per 1 cm bar. The challenge now is putting all efforts into the one device and getting the processes to run in parallel.
This is the final step, and in the run up to the SHEDs deadline, nLight is sounding confident. In fact, back in January last year it said that the goal of 80% by September 2006 was "clearly achievable". "We are putting our money where our mouth is," said Crump. "Some of our first SHEDs devices were released at Photonics West this year with an output power of 80 W and with conversion efficiency guaranteed at >65%."