A clinical study of failure in microchannel cooled diodes used in large laser systems Steven Jackel a,n , Avi Meir a , Zvi Horvitz a , Inon Moshe a , Yehoshua Shimoni a , Yaakov Lumer a , Revital Feldman a , Izhak Hershko b , Yotam Pekin a a Nonlinear Optics Group, Laser Department, Soreq NRC, 81800 Yavne, Israel b Nondestructive Testing Department, Soreq NRC, 81800 Yavne, Israel article info Article history: Received 2 March 2010 Received in revised form 15 September 2010 Accepted 22 September 2010 Available online 8 October 2010 Keywords: Diode lasers Microchannel coolers Corrosion abstract In this paper we investigate the source of failure in commercial, microchannel cooled CW diode bars placed in 12 bar horizontal arrays. The arrays were used to pump Nd:YAG rods in our 10 kW developmental laser. The laser was operated at low duty factor over a period of over 2 years. Experimental evidence indicated that the sudden, catastrophic failure was because of degraded cooling. We used optical microscopes, an X-ray microfocus imager, and a thermal neutron scattering camera to look inside the microcoolers. Our investigations revealed only one possible failure mechanism: cooling flow reduction because of delamination of the Au coating the walls of the microcoolers and the entrapment of Au flakes within the microchannel structures. We observed blisters in the microcoolers under working bars, and flake-like structures in the microcoolers under burnt-out bars (all taken from the laser). We observed no evidence of either massive blockages because of electrochemical deposits, or of corrosion/erosion in the microchannel walls. Integral operation times of the high flow-rate cooling system and of the diodes themselves were too short by one and two orders of magnitude, respectively, to explain the observed failures. Microchannel immersion times in the deionized water were, however, long enough to allow for corrosion of metals that may have been exposed through defects in the Au coatings. Three-dimensional heat flow simulations showed that blockage of multiple microchannels towards the edge of a bar can easily lead to catastrophic temperature increases. & 2010 Elsevier Ltd. All rights reserved. 1. Introduction Large diode pumped laser systems may contain hundreds or thousands of laser diode bars. The manufacturer lifetime testing of these diode bars is typically carried out on a continuous basis, with the bars being connected to the test setup (power supply and cooling system), being turned on, and then left to run nonstop. In all but heavy industrial applications, laser systems are used on a more sporadic basis, being turned on for periods of time ranging from seconds to hours, in order to facilitate laser or application development or to fulfill a specific process requirement, and are then left idle for equal or longer periods of time. The change in operating format and environment from that utilized by diode manufacturers may result in significantly different lifetime results. This fact was alluded to in a paper describing spectral wavelength shifts [1]. In that work, wavelength shifts were correlated to changes in the cooling characteristics of micro- channel cooled diode arrays. Such changes were postulated to be because of three possible effects: electrochemical deposition of material in the cooling channels, erosion and/or corrosion of the cooling channel walls, or pitting of the material underneath defects in the Au typically used to coat the microcoolers. (Beyond the inferences sited in that work, there is little published data from diode laser users.) The current generation of high power diode bars used in horizontal or vertical arrays is cooled with microchannel coolers. Microchannel coolers are designed to pass high speed water close to the surface being cooled. In order to perform this, the dimensions of the microchannels are made small. Typical dimensions are 100–300 mm high  300 mm wide. A typical 1 cm diode bar will be contacted to a cooler containing 14 micro- channels. If one considers that a CW bar contains 19 emitters, then there is an almost one to one dependence of emitter cooling on the performance of a sub-laying microchannel. Thus, if even one microchannel becomes blocked, then it can be expected that there will be changes in performance and lifetime of the emitter overlying it. An indication of degraded cooling is often a shift in wavelength. This is because emitter wavelength shifts with temperature at the rate of 0.3 nm/1C for 8xx nm diodes. Use of microchannel cooled diode arrays is complicated by the fact that deionized water must be used to prevent current leakage between adjacent bars [2–5]. Typical recommended water Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/optlastec Optics & Laser Technology 0030-3992/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.optlastec.2010.09.011 n Corresponding author. Tel.: + 972 50 629 2326. E-mail address: drstevenjackel@yahoo.com (S. Jackel). Optics & Laser Technology 43 (2011) 687–696