304 OPTICS LETTERS / Vol. 29, No. 3 / February 1, 2004

Jet-type, water-cooled heat sink that yields 255-Wcontinuous-wave laser output

at 808 nm from a 1-cm laser diode bar

Hirofumi Miyajima, Hirofumi Kan, Takeshi Kanzaki, and Shin-ichi Furuta

Central Research Laboratory, Hamamatsu Photonics K.K., 5000 Hirakuchi, Hamakita, Shizuoka 434-8601, Japan

Masanobu Yamanaka, Yasukazu Izawa, and Sadao Nakai

Institute of Laser Engineering, Osaka University, 2-6 Yamada-oka, Suita, Osaka 565-0871, Japan

Received September 22, 2003

A newly designed jet-type, water-cooled heat sink (the funryu heat sink, meaning fountain f low in Japanese)yielded 255-W cw laser output at 808 nm from a 1-cm bar made from InGaAsP�InGaP quantum-well activelayers with a 67% fill factor [70 quantum-well laser diode (LD) array along the 1-cm bar]. A funryu heatsink measuring 1.1 mm in thickness gave the LD 0.25 ±C�W thermal resistance, one of the lowest valuesachieved with a 1-cm LD bar. Over a short period of operation, the device reached a maximum cw power of255 W. To the best of our knowledge, this is the highest power ever achieved in 808-nm LD operation. Inthe future, the funryu heat sink may be capable of 80-W cw operation over an extended lifetime of severalthousand hours. © 2004 Optical Society of America

OCIS codes: 140.2010, 140.2020, 140.3290, 140.5960.

The output power of high-power laser diodes (LDs)and LD bars has increased remarkably in recentyears, and the use of these devices has substantiallyimproved the properties of LD-pumped solid-statelasers. High-power LDs are rapidly being introducedinto diverse fields such as laser processing, medicaltreatment, printing, and display devices, as well assolid-state laser pumping sources. The applicabilityof these LDs depends greatly on their electricity-to-light conversion eff iciency (hereafter simply referredto as efficiency). The efficiency of LDs generallyranges from 40% to 50%, with the remainder of theelectricity as heat. Since this generated heat de-grades a LD’s optical characteristics and dramaticallyshortens its lifetime, the heat sink is considered a keycomponent for realizing high-power LD bar operationwith an adequate operation lifetime.

The microchannel heat sink is well known as a heatsink for high-power LDs (especially LD bars). Mi-crochannel heat sinks for LD bars have been designedby use of Si etching technology by a group from theLawrence Livermore National Laboratory1,2 (LLNL)by use of Cu technology, as well as by researchersin Europe.3,4 The microchannel heat sink can havelow thermal resistance because of its large surfacearea for cooling water and its thin temperatureboundary layer. At the same time, however, themicrochannel heat sink has several disadvantages,such as a complex structure, high fabrication costs,and problems such as clogging by dirt particles andthe need for high water pressure during operation.As an alternative, our group proposed a “funryu” (jet)water-cooled heat sink5 and built the first prototypeof the device. The simple structure of the heat sinkmakes it possible to achieve low thermal resistanceat much lower thickness than with a microchannelheat sink, and thus the funryu is suitable for massproduction. This Letter f irst discusses the designand thermal characteristics of the funryu heat sink

0146-9592/04/030304-03$15.00/0

and then reports application results for high-power LDbar operation with 255-W cw output power at 808 nm,a level suitable for Nd:YAG laser pumping and directLD processing.

The thermal resistance of the funryu was experi-mentally evaluated by measurement of the tempera-ture increase with unit electricity consumption [±C�W].Because increases in the active-layer temperature maydegrade the output characteristics and reliability ofLDs, the temperature of the active layer must be keptbelow 50 ±C.2 When the efficiency of a LD is assumedto be 50%, the same power with optical output is gen-erated as heat. In the case of a LD bar with an op-tical output of 40 W, the required thermal resistanceis calculated to be 20�40 � 0.5 ±C�W or less. Simi-larly, when a LD bar with an optical output of 200 Wis assumed, the resistance must be 0.1 ±C�W or less.Therefore, our final target for the thermal resistanceof the funryu is at least 0.1 ±C�W.

Figure 1 shows a diagram and a schematic of thefunryu. The device is composed of a water inlet,the funryu heat-exchanging part, and a water outlet.The heat sink was fabricated by diffusion welding of

Fig. 1. Diagram and schematic of the funryu water-cooledheat sink.

© 2004 Optical Society of America

February 1, 2004 / Vol. 29, No. 3 / OPTICS LETTERS 305

Cu sheets with chemical etching, laser cutting, and adiamond milling process. Cool water f lows in fromthe bottom side, whereupon heat removal is performedeffectively by turbulent f low from small holes, andthe cool water is drained from the outlet. Thermalresistance is decreased by thinning of the temperatureboundary layer at the impingement target plate. Wehave also improved cooling efficiency by modifyingthe shapes of the holes, rearranging the hole positionsto facilitate the smooth drainage of hot water, reduc-ing the holes to less than 300 mm in diameter, andincreasing the surface roughness of the impingementtarget plate.

We attempted to use the funryu heat sink to gen-erate sufficiently high power from two types of 1-cmLD bar consisting of 30 and 70 broad-area emitters.Tensile-strained InGaAsP�InGaP quantum-well struc-tures were used for the bars’ active layers. Filling fac-tors (the ratio of the emitting width to the total barlength is 1 cm) of 40% and 67% were thus adopted.The LD bars were mounted junction down on the heatsinks by use of an indium solder to achieve betterthermal conductance to the LD active regions. Fig-ure 2 shows the wavelength distribution of the LD barwith 30 LDs operated in pulse (duty ratio of 1%) andcw (40-W) modes. The center wavelengths from the30 individual LDs were almost uniform across the 1-cmbar in both the pulsed and the cw operation modes.This result indicates a uniform temperature distribu-tion on the heat sink and uniform soldering along the1-cm bar. The spectral shift that was due to the tem-perature change was measured at 0.22 nm�±C.

Figure 3 shows the thermal resistance changesfor several LD bars as a function of the f low rateof the cooling water. The fill factors and cavitylengths of the evaluated LD bars used here were 67%and 1500 mm, respectively. Thermal resistance Rth(±C�W) of the LD bar module was estimated fromthe experimental values. We also estimated thethermal resistance by numerical analysis. A simplerepresentation of the heat-sink thermal resistance isgiven by6,7

Rth � Rjunc-Cu 1 Rcv 1 Rf . (1)

The resistance through the junction with the Cu(Rjunc-Cu) is estimated to be 0.11 ±C�W. Rcv , the ther-mal resistance component that is due to convectionheat transfer, is given by

Rcv �d

NuKS, (2)

where S is the total surface area of the heat-exchanging part, K is the thermal conductivity ofthe cooling water, and d is the hole diameter. In thecase of the turbulent jet,8 the Nusselt number (Nu) isexpressed as

Nu � 1.51Re0.44Pr

0.4�L�d�20.11, (3)

where Re is the Reynolds number, Pr is the Prandtlnumber, and L is the distance between the hole and thetarget plate. Thermal resistance component Rf arises

from the caloric increase of the water and is given by

Rf � DTf�q , (4)

where DTf is the temperature rise of the cooling wa-ter and q is the heat f lux. Figure 3 shows the re-sult of the calculation as a solid curve. The calculated

Fig. 2. Output center wavelength distribution versus theemitter position of the LD bar.

Fig. 3. Thermal resistance as a function of the f low rateof the cooling water.

Fig. 4. I–L characteristics of a LD bar on the funryu.

306 OPTICS LETTERS / Vol. 29, No. 3 / February 1, 2004

curve also (S is the f itting parameter) indicates thetendency of our experimental results. Here the valueof S, 33 3 1026 m2, was used. The expected value was27.5 3 1026 m2, so we believe that the model is rea-sonable. We adopted a water f low rate of 0.8 l�minfor evaluation of the characteristics of the LD bars,as increases in f low above the 0.8-l�min range resultin only minor reductions in the thermal resistance.Thus, the most favorable data recorded were thermalresistance below 0.25 ±C�W. With the simple calcula-tion discussed above, the funryu heat sink can be usedfor cw operation of a LD bar at 80 W with a thermalresistance of 0.25 ±C�W.

Figure 4 shows input current versus output light(I–L) characteristics of a LD bar mounted on the fun-ryu. The laser wavelength was 808 nm. The fill fac-tor and the cavity length of the evaluated LD bar were67% and 1500 mm, respectively. The inlet tempera-ture was 10 ±C, and the f low rate was 0.8 l�min. Athermal detector (Gentec, PS-1K) was used to mea-sure the cw power as a function of driving currentwhile the current was increased in steps of 2 A. Togive the detector time to reach thermal equilibriumwe kept the driving current constant for 10 s at eachmeasured point. An output power of 255-W cw wasachieved at a drive current of 296 A, and maximumefficiency of 49.1% was achieved at 167-W cw opera-tion. The junction temperature was estimated to be100 ±C (22-nm wavelength shift from pulsed operation:22 nm�0.22 nm�±C � 100 ±C) for 255-W cw operation.

Table 1 compares various types of heat sink for LDbars designed at different laboratories. As discussedabove, our apparatus produced a maximum cw powerof 255 W. This is the highest power ever achieved in808-nm LD operation to our knowledge, and it provesthe outstanding cooling performance of the funryuheat sink. In separate experiments under way toestimate the operational lifetime of LD bars on theheat sink in the 50-W cw mode, the device has beenin operation for over 2300 h and the LDs are stillrunning well. In the future we will be increasing thepower level in lifetime testing.

In conclusion, we have developed a novel funryuwater-cooled heat sink with thermal resistance con-firmed to be as low as 0.25 ±C�W for a 1-cm LD bar.

Table 1. Comparison of Various Types of Heat Sink for LD Bars

GeneralPhysics Institute

Parameter LLNLa Fraunhoferb OPCc (Porous Metal)d Fujitsue Funryu

Flow rate (l�min) 2.4 0.5 — — 0.2 0.8Rth (±C�W) 0.32 0.29 0.23 0.19 f 0.4 0.25Max. output power (W) 149 267 240 .50 .40 255Power at 50% efficiency (W) 100 150 240 .50 — 167g

Wavelength (nm) 808 980 — 795 808 808aRef. 2.bRefs. 3, 4, and 7.cOpto Power Corporation (Ref. 9).dRef. 10.eRefs. 11 and 12.fDesigned value.g49% efficiency.

In prototype testing we attained cw operation of 255 Wwith a maximum eff iciency of 49.1% at 808 nm. Thefunryu heat sink can feasibly attain 80-W cw opera-tion at 50% efficiency over long-life operation exceed-ing several thousand hours.

The authors thank T. Hiruma of HamamatsuPhotonics K.K. for his continuous financial supportand encouragement for this joint research betweenCentral Research Laboratory, Hamamatsu PhotonicsK.K., and the Institute of Laser Engineering, OsakaUniversity. We are also grateful to Y. Takiguchi ofHamamatsu Photonics K.K. for his helpful commentsand discussions. H. Miyajima’s e-mail address ismiyajima@crl.hpk.co.jp.

References

1. R. Beach, W. J. Benett, B. L. Freitas, D. Mundinger,B. J. Comaskey, R. W. Solarz, and M. A. Emanuel,IEEE J. Quantum Electron. 28, 966 (1992).

2. J. A. Skidmore, B. L. Freitas, J. Crawford, J. Satariano,E. Utterback, L. Dimercurio, K. Cutter, and S. Sutton,Appl. Phys. Lett. 77, 10 (2000).

3. F. Daiminger, F. Dorsch, and D. Lorenzen, Proc. SPIE3682, 13 (1998).

4. J. Jandeleit, N. Wiedmann, A. Ostlender, W.Brandenburg, P. Loosen, and R. Poprawe, Proc.SPIE 3945, 270 (2000).

5. S. Nakai, M. Yamanaka, H. Miyajima, T. Kanzaki, andH. Kan, “Heat sink,” Japanese patent 8-139479 pending(November 8, 1994).

6. G. A. Evans and J. M. Hammer, eds., Surface EmittingSemiconductor Laser and Arrays (Academic, San Diego,Calif., 1993).

7. R. Diehl, ed., High-Power Diode Lasers (Springer-Verlag, Berlin, 1999).

8. C.-L. Tien, Annu. Rev. Heat Transfer 6, 225 (1995).9. H. G. Treusch, J. Harrison, B. Morris, J. Powers, D.

Brown, and J. Martin, Proc. SPIE 3945, 23 (2000).10. V. V. Apollonov, S. I. Derzhavin, V. A. Filonenko, V. V.

Kuzminov, D. A. Mashkovsky, A. M. Prokhorov, andV. N. Timoshkin, Proc. SPIE 3889, 71 (2000).

11. Y. Nishikawa and H. Takigawa, Proc. SPIE 3628, 71(1991).

12. H. Takigawa and Y. Nishikawa, in Digest of Conferenceon Lasers and Electro-Optics/Pacif ic Rim (Optical So-ciety of America, Washington, D.C., 1997), paper TuH2.