02 Oct 2008
The high-power diode laser market is seeing new applications emerge thanks to higher output powers and new emission wavelengths. Jörg Neukum of DILAS looks at recent progress.
These are exciting times for the high-power diode laser (HPDL) market. Driven by improvements in the technology, HPDLs are finding a growing number of uses and are emitting over an ever-wider range of wavelengths. Today, HPLDs are commonplace in applications spanning from welding of plastics in automotive and medical device manufacturing and direct imaging of printing plates through to medical treatments, and as pump sources for diode-pumped solid-state lasers for industrial and scientific applications. This article looks at the available wavelengths, improvements in the technology and applications.
Red laser diodes date back to 1962, when the first semiconductor laser was fabricated. Today, the largest application in terms of units is optical storage, such as CD players (785 nm) and DVDs (640 nm). Such low-power emitters are also used in laser pointers and printers.
Driven by the needs of photodynamic therapy, laser diode bars are now available with several watts of optical output power at 632, 635 and 652 nm. Improvements in the epitaxial growing processes of InGaAlP materials as well as in the optical coupling methods have led to the availability of 2–5 W fibre-coupled single-bar modules (see figure 1). With applications such as optical displays on the horizon, improvements in the beam profile and lifetime of red laser diodes are necessary.
New applications continue to emerge in the longer wavelength range between 1.0 and 2.2 µm (see table). For medical applications where the beam quality is not essential, up to 40 W at 1.06 µm can be generated by single laser diode bars based on GaAs substrates.
When moving to longer wavelengths, the use of GaAs becomes problematic because of the increasing mismatch between the lattice constant of the substrate and the epitaxial layers. This mismatch induces compressive strain, which in turn decreases the gain of the semiconductor laser diode. Therefore, for wavelengths between 1380 and 1550 nm, InP wafers are required. Such laser diode bars are only capable of emitting lower power levels, for example 20 W at 1470 nm compared with 60 W at 976 nm for conduction-cooled laser diode bars.
When emission wavelengths in the 1850–2200 nm range are required, the material system switches again from InP to GaSb-based structures, which have been shown to operate at power levels of 10–20 W. For a lot of these newer wavelengths, work is ongoing to improve the materials, processes and structures, and even better results can be expected in the near future.
The development of laser diodes in the standard wavelength range between 800 and 980 nm has been driven by the need for higher powers (up to 60–80 W conduction cooled) and longer lifetimes (>10,000 to 20,000 h). Beside higher optical output powers, there is a need to further improve the slow-axis divergence (see box). This becomes especially important for fibre-coupled laser diode modules.
Whereas the fast-axis divergence is nearly diffraction limited and can be collimated to >3 mrad using aspheric cylindrical lenses, a typical slow-axis beam para-meter product is around 400 mm mrad, far beyond the diffraction limit. Improvements to the slow-axis divergence can only be made at the epitaxial level by modifying the waveguide and cladding layers, or by lithographic optimizations of the lateral structure of the laser diode array.
Improving the technology
Increasing the optical power level has an impact on the mounting and cooling techniques applied as well as on the chip and its coatings. For example, the materials processing market requires on/off modulation of the laser, which leads to a repetitive heat load on the assembly and repetitive expansion and shrinkage of the laser diode chip. This causes thermo-mechanical stress due to a mismatch of the thermal expansion co-efficients of the chip, solder and submount.
One solution is to keep the temperature difference between the on and off states small, essentially driving the laser diode between full current and just below threshold current. For moderate current and power levels, a large part of the electrical input energy is turned into waste heat and leads to a large rise in temperature and thermal expansion. Starting just below threshold current (a so-called soft pulse) creates a "dc"-waste heat which does not contribute to the thermo-mechanical stress and a small amount of "pulsed-waste heat". This minimizes thermal expansion and reduces the risk of thermo-mechanical stress.
This soft-pulse method is only a temporary solution. At very high powers and operating currents (operating currents being 10 times higher than the threshold current), the same problems with thermo-mechanical stress can be expected.
The answer to this problem can be either to improve the efficiency of high-power laser diodes thus reducing the temperature increase or to use submounts and solder materials with coefficients of thermal expansion (CTE) that match that of the chip. Using CuW submounts that have the same CTE as the GaAs chip can reduce the thermal stress by a significant amount. In conjunction with wire-bonding for the n-contact, this has led to optimized mounting processes and has resulted in diode laser bars emitting in excess of 100 W continuous-wave (see figure 2).
As soon as the high-power chip is integrated with CTE-matched mounting technology, the standard front facet coatings may no longer be sufficient to withstand the large optical power levels produced by the chip. To increase the limit for catastrophic optical mirror damage (COMD or COD), the following improvements can be made: (i) the use of so-called un-pumped mirrors. This refers to a shaped electrical contact on the p-side of the chip to prevent electrical current flow close to the front facet and reduce the heat in that region; (ii) introducing non-absorbing regions close to the front facet mirror by generating a larger bandgap in the semiconductor in this region (iii) vacuum cleaving of laser bars to prevent oxidation of the front facets and appliance of epitaxial-grown front facet mirrors to reduce defects in the coating layer and (iv) use of thick waveguides to reduce the intensity at the output mirrors.
Several programmes are in place worldwide to address these improvements and translate them into reliable manufacturing processes so we can expect even better electro-optical performances in the future.
On one hand, the developments described above have been driven by the needs of the application but on the other, technical improvements have pushed HPDLs into new markets. Some examples are:
• Welding of transparent plastics at 1940 nm: visibly transparent plastics absorb light greater than 1.7 µm. This is ideal for medical device manufacturing, as transparent polymers can be welded without using additives or absorbing intermediate layers.
• Photodynamic therapy (PDT): with HPDL modules available in the range of 632, 635 and 652 nm, it is no longer necessary to couple dozens of single emitters together, which results in more compact devices with smaller fibre diameters.
• Other medical applications will benefit from higher powers, smaller fibre diameters and new wavelengths. For example, the water absorption of human tissue at 1940 nm is about three orders of magnitude larger than at 980 nm.
• Defence: high-efficiency, high-power CW laser bars are being developed for use in mobile multikilowatt diode-pumped solid-state lasers for defence applications.
• Fusion facilities: laser-based fusion facilities have driven the development of very high-peak power diode laser bars in excess of 300 W. Many of the outlined technical improvements will be incorporated to achieve the goals set by the different consortiums working in this field. Several hundred thousand bars are expected to be consumed to build such facilities.
• For more information, please see www.dilas.com or e-mail j.neukum.
• This article originally appeared in the October 2008 issue of Optics & Laser Europe magazine.
View pdf of article
|© 2023 SPIE Europe||