Towers of power

Stephen Mounsey looks at the ways in which laser diodes are evolving, and the new markets opening up as a result

Laser diodes have come a long way. Since their relatively recent introduction, high-power laser diodes in particular have been applied to a wide range of applications, be it pumping solid state lasers, pumping fibre lasers, or used directly in materials processing. Much development has gone into increasing the power and beam quality of laser diodes over these years, but how much further can the technology be pushed, and is there even a need to push it?

A laser diode is, in simple terms, a combination of an LED with a resonating cavity, the latter consisting of two parallel facets on the surface of the emitting chip. The light generated is reflected from these facets, travelling through the cavity several times and stimulating further emissions as it propagates.

For many low-power applications such as telecommunications, laser pointing, and optical data retrieval, a single laser diode emitter provides sufficient power. In higher-power applications, laser diode bars are used. These bars are monolithic arrays of between 50 and 100 laser diode emitters – typically about 70 of them. The resulting diode bar will be bright, i.e. it will have high beam quality.

Power struggle

Since their introduction into commercial markets, the power of laser diodes, and of laser systems powered by diodes, has increased steadily. At the upper end, laser bars are available at powers up to 300W, but as Joerg Neukum, director of sales and marketing for laser diode manufacturer Dilas, says, few applications demand more power than this. Specialist projects such as the HiPER inertial confinement nuclear fusion experiment (currently at an early planning stage) will require large numbers of diodes, as lamp technology will not be able to provide the efficiency required. HiPER’s predecessor, the National Ignition Facility (NIF) at Lawrence Livermore National Laboratories, USA, is due to begin tests in the next few months. The beam lines of the NIF use lamp-pumped lasers, but this technology only achieves a wall-plug efficiency of around one per cent. Outside of nuclear fusion, applications exist in high-energy materials testing. The applications require petawatt-scale lasers, short pulse durations, and high peak energy lasers. These requirements combine to necessitate pumping diodes capable of very high peak power.

However, these large-scale high-energy experiments are rather few and far between, and so the market for laser diodes has shifted away from increasing output power towards other goals, such as increasing beam quality and efficiency while reducing cost of ownership. When discussing high-power laser diodes for pumping larger YAG-based lasers, Neukum states: ‘It’s fair to say that for high-energy laser applications, people are looking for high peak-power pulsed laser diodes. In industrial lasers, however, there are limitations on how much power you can actually pump into a YAG rod per centimetre length. The only [industrial] applications for which very high CW power diodes are required is in direct-diode materials processing, such as metal welding, brazing, and cladding, but these are currently small markets compared to the overall diode market.’

Focusing on applications

Direct diode systems are those in which the beam produced by the diode is used on the workpiece, rather than for pumping another gain medium. While material processing applications currently represent a small portion of the diode market, the capabilities of direct diode lasers have become more advanced over the last decade, as beam quality has increased. These increased capabilities have allowed direct diode solutions to offer an energy-efficient alternative to other laser technologies.

In the early days of high-power diode lasers, poor beam quality limited the systems to those applications that do not require particularly high precision, such as brazing, heat treating and soldering. Since then, diode lasers have been delivering progressively higher power in a higher quality beam. Due to another relatively recent shift in the predominant market preference, the beam is now almost always fibre-coupled for ease of delivery to the workpiece. The market for diode lasers has accordingly opened up into welding, initially allowing conduction welding regimes, and more latterly keyhole welding regimes as achievable brightness continues to increase.

A 300W passively cooled diode stack produced by Dilas.

Gary Broadhead, director for industrial systems at Laser Lines, explains that the drive towards better beam quality has outlasted the drive towards higher power: ‘10kW lasers have been available for years and years, and there are not many people driving them [at that rated power] and saying “we need more than 10kW”’. That said, high powers have been a key factor in increasing beam quality: ‘The industry standard is the 1cm bar. Ten years ago, we might have gotten 10W out of that bar, but now get 150W out of the same cross-section,’ says Broadhead. The improvements have, for the most part, been made by the people who design and grow the wafers. Careful design has meant that more power can be emitted without damaging the facet. Alongside these developments, advancements in microoptical technology have allowed the beams to be efficiently collimated and combined into delivery fibres.

While the industry will continue to strive to improve the beam quality of diode lasers, Broadhead believes that the demand may reach an asymptote: ‘There’s an economic limitation in one sense: as the diode manufacturers try to push the beam quality, they’re going to see competition from people who already make very high beam quality lasers – typically this is going to be a fibre laser or a disk laser. It might be more effective for a fibre laser manufacturer to make a “poor” beam-quality laser, rather than trying to push and push the diode technology.’ Very high beam quality is important in some cutting applications, and in the emerging technology of remote-welding. ‘Why chase cutting applications when the fibre and disk lasers can do it perfectly well already?’ asks Broadhead. ‘I think that what’s interesting about diode lasers, however, is that they continue to offer the best efficiency.’


Regardless of whether or not the perception was justified, the lifetime of laser diodes was perceived to be something of a drawback when the technology first emerged, but advancements in cooling and in materials science have allowed manufacturers to supply diode bars with warranties of up to five years.

Of the two cooling regimes – passive and active, the latter is considered to have a shorter life span. ‘It all comes down to how well you engineer these micro channels,’ explains Broadhead; ‘they’re very small bore, and so you also need to have a very high quality of cooling water with absolutely no particulates in it. Because the channels are so small-bore, you have to have quite a high flow rate (you haven’t got much water in there). If you don’t engineer the channels correctly, they can end up corroding like a river bank. A lot can go wrong down the line if you don’t engineer [active cooling] correctly, and so some manufacturers have pulled back and only offer passively cooled bars.’

The crossover point at which passive cooling is no longer sufficient and below which active cooling is prohibitively complicated (both in terms of reliability and cost) is, according to Broadhead, at about one kilowatt of output power, assuming that the application requires high brightness and therefore requires emitters to be closely packed.

Diode manufacturers are typically very conservative in their stated estimates of their products’ lifespan. Bob Murphy, director of business development at diode producer Opnext, states that the company’s qualification process sees a large sample of laser diodes driven at their full rated power and their maximum temperature for 2,000 hours. The criteria Opnext specifies is that no components of the sample group should fail in this time. Murphy explains that this does not mean that the life of the diode is 2,000 hours, but it does mean that the company can confidently guarantee that the diodes will last at least this long without failure. Murphy believes that diode lifespan is important for the company’s reputation, and that giving lifetime figures that are always met is more important than giving longer lifetimes. ‘The customers believe in our products, because we’re very conservative in our estimations,’ he says.

Dilas, for example, recently released a 150W bar with a warranty, despite the fact that, according to Neukum, the company has already demonstrated the bar at 250W in its labs. ‘If we specify lifespans for industrial applications, we are conservative. De-rating always helps, and it allows us to build in a safety margin. That safety margin is then slowly and conservatively reduced as the device proves itself,’ he says. ‘We have feedback from the field stating that water-cooled stacks have lasted way beyond 27,000 hours of performance, but a user could damage the stack within minutes if it is operated incorrectly.’

Broadhead believes that fibre laser manufacturers have been very good at promoting the fibre lasers as having a very high reliability, but customers must remember that a fibre laser is pumped by laser diodes. ‘There’s no reason to think that putting a fibre in front of those diodes makes them magically more reliable. If there is a perception that diode lasers are not as reliable, that’s clearly a misconception,’ he says. Some manufacturers now are offering five-year warranties on the diodes themselves.

Where next...?

According to Neukum, Dilas has recently focused its efforts on refinements in beam shaping, which, he says, always relies upon micro optics. With respect to the output power of the next generation of laser diodes, Neukum believes that ‘there are some people who claim the cost per Watt of laser diode power will continue to decline. On the other hand, that cost can only decline if for the same effort in assembly and material costs we get more power from the chip. At the moment, a lot of the power increase is done by increasing [the size of the] gain medium, so we get laser bars with longer resonators... going from 1mm up to about 4mm bars.’ Neukum thinks that this trend will lead to a trade-off between three attributes of a laser diode: ‘The cost of a chip scales with the length of the resonator, and so it will be interesting to see whether or not there will be a compromise reached in terms of high yield, high output power, and long lifetime. Customers look for longer lifetime components at lower prices.’

When it comes to increasing the reliability further, most developments revolve around reducing the intensity of the beam at the output facet of the optical cavity. Neukum explains that increasing the dimensions of the waveguide achieves this goal, and that lower intensity at the facet reduces the risk of damage there. Cooling, he says, can also be improved by way of increased efficiency (often a product of improved materials science) and by way of careful, stable mounting on an effective heat sink.

When it comes to the future application-driven demands of the market, Murphy explains that Opnext, which primarily markets lower-powered diodes than the other companies mentioned here, is investing heavily in blue and green diodes. The company’s aim is to produce red, green and blue diodes, in order to address the market for highly portable and compact video projectors. Any diodes developed for this purpose will also have to be highly efficient, in order to be viable in a battery-powered device. ‘We want to develop diodes that emit at “true” blue, and not at the 405nm, which is currently available,’ explains Murphy. The emerging generation of laser televisions either uses pseudo-blues produced by 405nm emitting diodes, which does not allow the display to re-produce the millions of colours that a conventional TV can, or otherwise makes use of bulkier and less-efficient DPSS solutions, which are able to achieve the true-blues as 440-450nm, albeit with a wall-plug efficiency of approximately seven per cent. ‘405nm is more of a violet than a blue. True green is also under development. Once we get there [to true red, green and blue], I think that the market will be limitless; it’s laser televisions, it’s laser displays, it’s automotive, it’s handheld, and we think that there could be millions of devices sold every year in these markets.’