Laser bars (arrays of multiple parallel emitters on a semiconductor substrate) are on the verge of becoming a consumer product. According to estimates, within two to three years, more than a million laser bars will be needed to meet the demands of industry, for example, in materials processing and medical applications. If that happens, it will lead to a significant drop in price for high-power diode lasers. Interest has also been growing in using thermal-expansion-matched microchannel heat sinks with very high cooling performance to increase the reliability of high-power diode-laser bars.1 One new approach to producing heat sinks uses micrometal injection molding (μ-MIM). Unlike conventional heat sinks, which are made of copper, these devices combine tungsten—a material with a very low coefficient of thermal expansion (CTE) and moderate thermal conductivity—with copper, which has moderate CTE and high thermal conductivity (>400W/mK). Manufacturing heat sinks by μ-MIM enables economical mass production of complex microscale near-net-shape parts. Especially in runs of over 10,000, the μ-MIM process reduces the cost per part considerably (up to 70%). μ-MIM offers several other significant advantages, such as the possibility of producing structurally and geometrically complex composite parts with desired mechanical and thermal properties. The copper-tungsten heat sinks have a gallium arsenide (GaAs)-matched CTE, combined with high thermal conductivity. Yet another advantage of μ-MIM is that the needed green bodies—molded components—of the heat sink can be joined together in a cosintering process.
In the field of expansion-matched heat sinks, high power means a thermal load of more than 600W/m2 underneath the diode-laser bar. The continuously increasing output power is achieved by a similarly increasing resonator length and, consequently, the expanding footprint area of the laser bar. This ensures that thermal loads do not increase as wall-plug efficiency simultaneously rises to over 55%. Effective cooling requires water-cooled active sinks.
We fabricated heat sinks using a material composition of tungsten-copper 80/20% by weight. This composition shows a CTE of 8.8 parts per million (ppm)/K, slightly above the required 6.5ppm/K of GaAs.2,3 With μ-MIM, it is possible to create the required flow-channel structures. Powders with an average diameter of only 5μm are used for metal-powder injection molding. These powders allow high contour accuracy and good mixing of both constituents. Densities of more than 98% are achieved following sintering. The sprue material is recycled and reused, which makes μ-MIM environmentally friendly (see Figure 1).
Figure 1. Three ‘green body’ components of an injection-molded heat sink.
In cosintering, individual green bodies are positioned one on top of the other and joined inside a kiln at high temperatures (>800°C), with no need for additional joining steps. For our purposes, the joining material is copper. As an alternative to μ-MIM, we used silver diffusion soldering. Here, presintered heat sinks are silver coated (which constitutes one additional processing step), braced together, and then also joined together inside a kiln at temperatures >500°C (see Figure 2). As is generally the case with injection molding, it can be assumed that the relatively high cost of tools for manufacturing heat sinks using μ-MIM will drop with an increasing number of units. To first approximation, the price per heat sink should be under 20Euros (~$27) for a total of 10,000 manufactured units.
Figure 2. Sintered heat sink with a mounted laser bar and a copper n-contact sheet.
Test results have shown that thermal design of the heat sinks with a minimal pressure drop ensures a high flow rate. The aim is thermal resistance <0.5K/W. The units we manufactured using the second joining process show no significant change in flow rate after 1000h of extended testing. Only after 2400h during a long-term test under full load with a constant pressure drop of 3bar did we encounter a leak, located in the joining zone between the sintered parts. Prior to the long-term test, we had identified thin areas of pure copper under a microscope, and these turned out to be the culprit. We plan to modify the heat-sink design to prevent formation of pure copper areas in the water-cooled section of the device.
In a follow-up project, we will investigate the possibility of manufacturing a heat sink consisting of only two parts using μ-MIM. We are especially interested in knowing whether relocating the joining zone away from the water-cooled area can improve the sink's lifespan. Cosintering will still be the joining process of choice. In summary, the advantages of the μ-MIM process are its amenability to complex geometries, its cost-effectiveness, and the opportunity to recycle excess material back into the production process. The advantages of μ-MIM-manufactured heat sinks themselves are very-high-performance cooling and a high life expectancy with respect to erosion and corrosion.
The authors would like to thank the AiF (German Federation of Industrial Research Associations) for funding the development of active, cooled heat sinks by μ-MIM.
Michael Leers, Erik Liermann
Fraunhofer Institute for Laser Technology (ILT)
Michael Leers is currently head of the packaging group. He received his diploma in mechanical engineering from the RWTH Aachen University of Technology (Germany) in 2002.
Erik Liermann has been a scientist at the Fraunhofer ILT since 2009.
Fraunhofer Institute for Manufacturing and Advanced Materials
Philipp Imgrund leads the Shape Forming and Functional Materials Group. He specializes in powder technology and μ-MIM.