Innovative stacking techniques increase the output power and brightness of diode laser bars for materials-processing applications.
Semiconductor lasers are being used in a growing number of materials-processing tasks, including plastic and metal welding, marking, hole drilling, soldering, heat treating, and adhesive curing. Whether for direct use or pumping applications, ongoing research is focused on delivering higher laser output power, without sacrificing operating lifetime or beam quality. Higher power increases process throughput and extends the range of materials that can be processed by the lasers. Work on increasing semiconductor laser power currently focuses on three areas of the technology: device materials, heat dissipation technology, and beam-delivery optics.
The basic building block of a high-power semiconductor laser is the laser bar, a monolithic, linear array of laser edge emitters. A typical bar is about 10*1*0.1 (mm3) and contains 20 to 70 individual emitters. With a specified operating lifetime of more than 10,000 hours, bars typically supply 40 to 50 W of continuous-wave (CW) output at wavelengths from 800 nm to 980 nm.
Individual bars can be combined into stacks to produce higher total output power. This is usually done by stacking the arrays vertically, so that the rows of emitters are very close together (less than 2 mm). In principle, there is almost no limitation on how many bars can be combined in this manner. Typical commercial products consist of stacks of six or 12 individual laser bars and offer CW output powers as high as 600 W. Custom products have gone as high as stacks of 36 bars that reliably deliver 1.8 kW CW.
For most applications, the high-power output generated by a diode bar is most useful if it appears to emanate from a relatively small source area. This is quantified by the parameter known as brightness, which is defined as the total power emitted by the source per unit solid angle, per unit area. The higher the source brightness, the more easily and efficiently it can be concentrated into a small focused line, spot, or area. In other words, a higher brightness source can deliver greater power density at the work surface.
The brightness of a laser bar is limited by its fill factor and output divergence. Fill factor is the ratio of the individual emitter length to the spacing between emitters-essentially the percentage of the bar’s surface that emits light (see figure 10.1). Typical laser bar fill factor values range from 40% to 75%. For laser stacks, fill factor is also defined as the width of the output line from each bar divided by bar-to-bar spacing (called the pitch).
|Figure 10.1. The fill factor for a laser bar is the individual emitter length divided by the emitter spacing. For a bar stack, fill factor is the beam height divided by the beam spacing|
The output of each emitter in a laser bar emanates from an area that is very long and thin. The output in the axis of the smaller dimension, called the fast axis, diverges rapidly due to diffraction (typically 30° to 40° FWHM). However, because the source size is small in this dimension, the M2 is low (typically around 1.5), and thus the beam can be collimated using simple cylindrical optics. In the longer emitter dimension, known as the slow axis, the divergence is much lower, usually about 10° FWHM. However, the larger source size results in a higher M2 (even exceeding 1000), which limits the ability to collimate and/or refocus the light in this dimension.
There is a basic tradeoff in laser bars between output power and lifetime: The higher the power, the shorter the lifetime. The long-term performance of a high-power laser bar is limited by the laser temperature, the power density at the emitting facets, and the internal stress induced by bonding the laser to the heatsink. Laser temperature depends on drive current, which in turn depends on operating mode (CW, quasi-CW, or pulsed operation, average power, peak power, pulse width, repetition rate, duty cycle, etc.), and on the type of thermal dissipation (heat sinking and cooling) used.
Reducing thermally induced stress can be achieved by a combination of efficient cooling and innovative device architecture. High-power laser stacks can be thermally stabilized by microchannel cooling–running water through small channels in the heatsink attached to each laser bar. Ideally, the heatsink microchannel cooling ports are stacked, which enables all the water inlets and outlets to operate in parallel.
Our group has focused on reducing the thermally induced stress between the laser bar and heatsink. This stress occurs because the heatsink bonding process is performed at high temperature. As the assembly cools, the difference in the thermal expansion coefficient between the laser and heatsink produces mechanical stress that degrades device reliability. This problem can be mitigated by using a heatsink whose thermal expansion is more closely matched to the laser bar. We have made significant progress in this area by using a heatsink composed of a sandwich of materials, instead of a single material. Test data indicate that this technique should soon enable the production of laser bars that deliver more than 70 W of power with lifetimes of more than 10,000 hours.
Our microchannel cooling architecture uses a ceramic spacer to accurately position the individual bars and a spring foil to provide electrical contact from the n-side of one bar to the p-side of the next. With this construction, a stack having a 1.8-mm pitch can be produced with a pitch tolerance of less than 0.02 mm. The importance of this is that tight mechanical tolerances facilitate the accurate positioning of the fast axis collimation lenses, which have focal lengths of less than 1 mm.
With a tight enough pitch, it is possible to use optical methods to increase laser brightness, in some cases virtually doubling the source brightness. In one approach, we begin by assembling two stacks side by side, but with a vertical offset between the two corresponding to half the stack pitch. The bars in each stack are collimated in the fast axis using a lens with a focal length short enough to produce a fill factor of less than 50%. The light from each bar in the stack on the left then passes through a plane parallel plate, whose height is half the stack pitch. Each of these plates is rotated so that the emergent light is shifted to the right by a distance that is half the spacing (center-to-center separation) between the stacks. Likewise, light from the bars on the right is shifted to the left the same distance by another set of plates. The result is that the output of the two stacks is interleaved, nearly doubling the source brightness. We have successfully used this approach to combine the output of two 15-bar stacks of 50 W CW lasers to yield a total power of 1.4 kW. Most importantly, the combined beam possesses the same M2 as the individual laser stacks.
Laser brightness can again be doubled, without any degradation in M2, by using a dichroic prism to combine the output of lasers having different wavelengths. We have demonstrated this experimentally with a 12-bar stack. The top and bottom three bars of this stack are 808-nm lasers, and the middle six have output at 940 nm. A dichroic prism deflects the two sets of 808-nm beams so that they emerged coincident with the 940-nm beams. The output of the dichroic prism, which consists of six dual-wavelength beams with a pitch of 1.8 mm, is then combined with the output from an identical system using the interleaving stack just described. There is nothing that prevents this technique from being scaled to even larger stacks.
Semiconductor lasers offer many desirable characteristics for industrial applications, including compact size, high reliability, long lifetime, and electrical efficiency. Improvements in laser bar output power and stacking technology will deliver even higher total power and higher brightness–critical advantages that can open up a new range of applications in materials processing.
By Franck Leibreich and Hans-Georg Treusch, Spectra-Physics Semiconductor Lasers,
SPIE, OEMagazine, September 2001