A simplified multiwavelength prototype of an axially symmetric diode laser device based on stacks madeof single emitters has been made, and the performance of the device has been demonstrated experimen-tally. The results verify that kilowatt-level light power can be focused into a circular spot with a 1�e2
The development of high-power edge-emitting diodelasers (DLs) has been rapid. Single emitters withcontinuous wave (cw) output power as high as 16 Wand with brightness as high as 6.6 � 1012 W m�2 sr�1
have been reported.1,2 Brightness is defined asB � P�A�, where P is the radiant power of the lightsource, A is the area of the emitting surface, and � isthe solid angle of the emitted field.
The performance of practical direct high-power DLdevices has not been sufficient for the most demand-ing materials processing applications such as metalcutting. One reason for this has been the fact that thebuilding block of direct DL devices has typically beena 10 mm wide laser bar containing some tens of emit-ters. If the fill factor (the width of one emitter dividedby one period of the bar) of a laser bar is f and thebrightness is Bb, the brightness of a single emitter inthe same bar is Be � Bb�f. For DL bars designed forcw operation, f is typically 0.2–0.5. This fundamentaladvantage of a single emitter compared with a bar isfurther increased by the fact that it is easier to cool asingle emitter, which facilitates higher power peremitter or, alternatively, more-reliable operation.
In a standard single broad area edge-emittinghigh-power DL the beam quality in the slow-axis di-
rection, i.e., parallel to the pn–junction plane, is typ-ically an order of magnitude poorer than in the fast-axis direction, i.e., normal to the pn-junction plane.Therefore, for a 10 mm wide DL bar made of suchemitters, the beam quality in the slow-axis directionis roughly 3 orders of magnitude poorer than in thefast-axis direction. Such a light source is naturally farfrom ideal and cannot be focused to a small spotwithout complex beam-shaping optics. However, ifsingle emitters are stacked one on top of another inthe fast-axis direction, the resultant beam exhibitssimilar beam quality in both slow- and fast-axis di-rections.3 Such a stack is therefore an excellent build-ing block for high-beam-quality high-power DLdevices.
To increase the power to a kilowatt level, the lightemitted from several stacks made of single emittersneeds to be combined into a single beam. Axial sym-metry has been proposed4 to provide an efficient ge-ometry for this purpose.
In this paper we report what is to our knowledgethe first experimental demonstration of a simplifiedaxially symmetric high-power DL prototype designbased on single emitter stacks. In addition, the prac-tical limits of the performance are studied.
2. Diode Laser Stacks
Two different stack structures with single emitterswere made. The macrochannel stacks were madewith customized macrochannel coolers and sub-mounts, while the microchannel stack consisted ofcommercial microchannel coolers and DLs on sub-mounts.
The authors are with the Optics Laboratory, Institute ofPhysics, Tampere University of Technology, P. O. Box 692, 33101Tampere, Finland. E. Lassila’s e-mail address is [email protected].
Received 16 November 2005; revised 21 December 2005; ac-cepted 6 January 2006; posted 10 January 2006 (Doc. ID 66072).
The DLs used in macrochannel stacks were mountedp-side down on customized CuW submounts and Cucoolers with AuSn solder.5 Figure 1 shows one emit-ter after mounting. The stripe width of these emitterswas 150 �m, and the emission wavelengths were 803,931, and 965 nm. The commercial DLs on submountsused in the microchannel stack had a 90 �m stripewidth, and they emitted at 960 nm. Each DL wascharacterized in cw mode after mounting on a cooler.Figure 2 presents typical P–I (power–current) andU–I (voltage–current) curves.
B. Macrochannel Structure
Easier cooling of a single emitter compared with thatof a wide bar makes it possible to use macrochannelcooling (even conductive cooling is sufficient in somecases) instead of microchannel cooling, which hasbeen somewhat unreliable in practice. The cross sec-tion of the cooling channels in the customized coolerswas 2.5 mm2. According to heat transfer calcula-tions,5 the maximum temperature increase per ab-sorbed watt of heat was 9 K. Figure 3 shows one 10layer macrochannel stack with fast-axis collimation(FAC) microlenses. The stack was mounted on a car-rier. The layer thickness was approximately 1.1 mm.The DLs were electrically connected in series, whilethe cooling channels were connected in parallel.
FAC was carried out for six 10 layer macrochannelstacks (i.e., 60 microlenses) by use of a custom mo-torized precision positioning system with six degreesof freedom.6 The height of the commercial micro-lenses was 1.0 mm. Table 1 lists the measured re-sults of the FAC.
According to Table 1, the divergence values ob-tained are close to the diffraction limit ��1 mrad�,but the angular error is relatively large. This angularerror was caused by the unfortunate and unpredict-able vertical movement of the stack during UV curingof the microlens attachment glue. This movementhad a magnitude of �1 �m that was probably due tothe combined effect of glue shrinkage, translationstage backlash, and possible torques in the system.Angular error can be completely removed by carefuldesign and use of high-quality equipment. It shouldbe noted that a certain value of angular error has amore severe effect than a corresponding value of ex-tra divergence, because angular error deflects alllight from the desired direction, whereas divergenceaffects mainly the wings of the light distribution.
C. Microchannel Structure
As an alternative approach, a five-layer stack wasmade of commercial microchannel coolers and singleemitters mounted p-side down on AlN submounts.The submounts were attached to the coolers withSnPb solder preforms. Figure 4 presents the micro-channel stack without FAC microlenses. The layerthickness in this construction was 2.2 mm. If desired,
the effective layer thickness can be decreased to1.1 mm, e.g., with a suitable step or stripe mirror.7–9
The optical properties of the collimated microchan-nel stack were studied in a focusing experiment. Thefocusing lens was an achromatic doublet with 30 mmfocal length. Because the height of the collimatedlight field was approximately 10 mm, the numericalaperture (NA) of the focused beam was 0.16. Figure 5shows the focus obtained. The 1�e2 diameter of thefocus was 100 �m in the fast-axis direction and90 �m in the slow-axis direction. The correspondingFWHM values were 60 and 40 �m, respectively. Be-cause the structure can provide approximately 50 Wof cw light power into the focus �1.1 mm layer thick-ness), the average power density in the 1�e2 focus canbe of the order of 7 kW�mm2 and the brightness of theorder of 9 � 1010 W m�2 sr�1.
3. Prototype
A prototype for an axially symmetric DL light sourcebased on the multiplexing of a large number ofstacked single emitters was designed according to thegeneral concept presented in Ref. 4. The concept sug-gests the arrangement of the laser stacks in multiplesectors around the optical axis. The prototype con-tained one sector pair consisting of two sectors oppo-site each other, each sector including threemacrochannel stacks. The design can accommodateseveral sector pairs. However, the main aim of the
present research was to demonstrate the high bright-ness achievable with this design and to analyze itssensitivity to inaccuracies in the positioning of theelements. For this purpose one sector pair was con-sidered sufficient, with the considerable expenditureand labor related to the manufacturing of the sectorsfor the first time taken into account.
The wavelengths of the stacks were 803, 931, and965 nm. Figure 6 shows the assembly. The position-ing of the stacks, prisms, and dichroic mirrors is as in
Fig. 4. Microchannel stack.
Table 1. Results of FAC of Macrochannel Stacks
MeasurandDivergence
(mrad)aAngular error
(mrad)b
Average value 1.2 0.9Standard deviation 0.6 0.8
aHalf-angle of 1�e2 cone.bAngle between the optical axis and the true propagation direc-
Fig. 4(b) of Ref. 4. All stacks were on carriers, whichwere attached to the prototype in such a way that thelight was initially propagating radially toward theoptical axis. The beams originating from the leftmoststacks were turned 90° into the axial direction withprisms. The beams from the other stacks were turnedinto the axial direction by use of dichroic beam com-biners that reflected radial beams into axial beamsand transmitted axial beams. In this way the threeoriginally radial beams in each sector were combinedinto a single axial beam.
The two axial combined beams originating from thetwo sectors were collimated in the slow-axis directionby a standard cylindrical lens having a 100 mm focallength. The radial position of each stack could beadjusted to provide optimum collimation of the com-bined beam in the slow-axis direction. The use ofsingle emitters makes it possible to use slow-axiscollimation (SAC) lenses that can typically be 2 or-ders of magnitude larger in size than the stripe widthof the emitters. This results in low divergence in theslow-axis direction, which cannot be obtained in apractical way with traditional 10 mm wide bars.
After SAC the light field was focused with a com-mercial achromatic doublet having a 50 mm diameterand a 100 mm focal length (maximum NA, 0.24). Thedimensions of each of the combined beams on thefocusing lens were approximately 11 mm in the fast-axis direction and 13 mm in the slow-axis direction.The focus was recorded on a CCD array. Figure 7presents the focus obtained. The 1�e2 diameter of thefocus was 360 �m in the fast-axis direction and280 �m in the slow-axis direction. The correspondingFWHM values were 160 and 150 �m, respectively.
It is possible to expand the two-sector prototypedescribed above into a corresponding device witheight sectors, thus quadrupling the total light power.The focus of such a device would be formed by rotat-ing the focus shown in Fig. 7 by 45° three times. Theresultant focus would therefore resemble a circle witha 1�e2 diameter of 360 �m. Assuming 5 W outputpower per emitter and 90% efficiency of the optical
system, each sector provides approximately 135 W ofcw light power into the focus. This predicts an aver-age power density in excess of 10 kW�mm2 and abrightness of the order of 6 � 1010 W m�2 sr�1. Onecan further increase the power level by adding newwavelengths and (or) polarization multiplexing intothe structure.
In principle, the beam quality of the prototypeshould be better in the fast-axis direction than in theslow-axis direction. The fact that the experimentgives the opposite result is explained by the presenceof angular error that results from the deficiencies inthe FAC process. In searching the practical limitsof the prototype we were therefore interested in find-ing the size of the focus in the fast-axis directionwhen angular error was not present. We studied thisfocal size experimentally by moving a well-collimatedsingle emitter that had practically no angular error inthe fast-axis direction over the focusing lens with50 mm diameter and 100 mm focal length. The totalmovement was 48 mm, and the emitter was moved in6 mm steps. An image of the focus was recorded foreach emitter position, after which an average imagewas calculated from the individual images. The focusobtained is shown in Fig. 8. The 1�e2 diameter of thefocus was 200 �m in the fast-axis direction and240 �m in the slow-axis direction. The correspondingFWHM values were 120 and 140 �m. This exper-iment verifies that it is possible to concentratekilowatt-level light power into a focus with a 1�e2
diameter of 240 �m, a NA of 0.24, and a 100 mmfocal length. Thus elimination of the angular errorimproves the predicted average power density toapproximately 24 kW�mm2 and the predicted bright-ness to the order of 1 � 1011 W m�2 sr�1.
In principle, by using DLs with better beam qualityin the slow-axis direction (e.g., narrower stripe widthor tapered DLs), one can make the focus even smallerin the slow-axis direction than in the fast-axis direc-tion. This would result in a circular focus with a 1�e2
diameter of �200 �m for an eight-sector device. Fur-thermore, by utilizing polarization multiplexing it ispossible to increase the cw light power level in the
Fig. 7. Focus of prototype.Fig. 8. Practical limit of prototype’s focus.
focus to approximately 2 kW. This would result in anaverage power density of more than 60 kW�mm2 anda brightness of the order of 3 � 1011 W m�2 sr�1.
It is possible to increase the power level by increas-ing the number of sectors. This can be done severalways. First, SAC lenses with shorter focal lengths canbe used. This will, however, increase the size of thefocus in the slow-axis direction. Second, a larger fo-cusing lens can be used. This will increase the NA ofthe focused beam. Third, customized SAC lens stackscan be utilized, as explained in Ref. 4.
In addition to direct DL materials processing, theprinciple is interesting for high-power fiber laserpumping because of its small circular focus, highbeam quality, and free optical axis.
4. Summary
A two-sector prototype light source based on axialsymmetry and single-emitter DL stacks has beenmade, and its performance has been experimentallyverified. The 1�e2 diameter of the focus obtained at100 mm focal length was 360 �m in the fast-axis di-rection and 280 �m in the slow-axis direction. Thecorresponding FWHM values were 160 and 150 �m,respectively. Expanding the prototype device to in-clude eight sectors will result in a circular focus witha 1�e2 diameter of 360 �m. Assuming 1.1 kW cw lightpower in the focus, the average power density will bemore than 10 kW�mm2 and the brightness of the or-der of 6 � 1010 W m�2 sr�1.
It was also experimentally verified that with accu-rate fast-axis collimation the 1�e2 diameter of thefocus can be decreased to 200 �m in the fast-axisdirection. By using suitable DL emitters, one canmake the focus even smaller in the slow-axis direc-tion. The concept is scalable and enables high cwpower levels to be achieved. This demonstrationtherefore suggests that it is possible to focus approx-imately 2 kW of cw light power into a circular focuswith 1�e2 diameter of 200 �m at a 100 mm focallength and a 0.24 NA. This would result in an averagepower density of more than 60 kW�mm2 and a bright-ness of the order of 3 � 1011 W m�2 sr�1 in the focus.
As well as for direct DL materials processing ap-plications, the method is interesting for high-powerfiber laser pumping owing to its small circular focus,high beam quality, and free optical axis.
This work has been financially supported by theNational Technology Agency of Finland (Tekes),Cavitar Oy, Corelase Oy, Finn-Power Oy, and LiekkiOy. The helpful assistance of Petri Harkko, HannuSipponen, Tuomas Siekkinen, Jukka Karinen, andAntti Lepistö in manufacturing the prototype andcarrying out the measurements is gratefullyacknowledged.
References1. N. Pikhtin, S. Slipchenko, Z. Sokolova, A. Stankevich, D. Vino-
kurov, I. Tarasov, and Z. Alferov, “16 W continuous-wave outputpower from 100 �m-aperture laser with quantum well asym-metric heterostructure,” Electron. Lett. 40, 1413–1414 (2004).
2. M. Kelemen, J. Weber, G. Kaufel, G. Bihlmann, R. Moritz, M.Mikulla, and G. Weimann, “Tapered diode lasers at 976 nmwith 8 W nearly diffraction limited output power,” Electron.Lett. 41, 1011–1013 (2005).
3. T. Fan, A. Sanchez, and W. DeFeo, “Scalable, end-pumped,diode-laser-pumped laser,” Opt. Lett. 14, 1057–1059 (1989).
4. T. Alahautala, E. Lassila, and R. Hernberg, “High power den-sity beam from narrow diode-laser arrays in axial symmetry,”Appl. Opt. 43, 2760–2766 (2004).
5. S. Kuusiluoma, M. Hakamo, E. Lassila, P. Heino, R. Hernberg,and E. Ristolainen, “Assembly of high power diode laser emit-ters,” presented at the 2004 IEEE International Conference onSemiconductor Electronics, Kuala Lumpur, Malaysia, 7–9 De-cember 2004.
6. P. Harkko, “Diode laser light field collimation by a micro lens,”Master of Science thesis (Tampere University of Technology,2005); in Finnish.
7. Y. Liao, K. Du, S. Falter, J. Zhang, M. Quade, P. Loosen, and R.Poprawe, “Highly efficient diode-stack, end-pumped Nd:YAGslab laser with symmetrized beam quality,” Appl. Opt. 36,5872–5875 (1997).
8. A. König, “Double image micrometer,” U.S. patent 587,443 (3August 1897).
9. H. Ohashi, X. Gao, H. Okamoto, M. Takasaka, M. Saito, and K.Shinoda, “Enhancement of emitting power density with a beam-shaping technique for a high-power laser-diode array stack,”Opt. Eng. 43, 2206–2207 (2004).
