LONG WAVELENGTH Pbsnte LASERS with CW OPERATION ABOVE 77K
Total Page:16
File Type:pdf, Size:1020Kb
LONG WAVELENGTH PbSnTe LASERS WITH CW OPERATION ABOVE 77K Koji Shinohara, Mitsuo Yoshikawa, Michiharu Ito and Ryuiti Ueda Kansai Research Center Fujitsu Laboratories Ltd.,JAPAN SUMMARY Lead tin telluride diode lasers with emission wavelengths6 to of9 pm easily operate continuously at temperatures above 77K. These lasers have the Pb Sn Te/Pbl-xSnxTE/Pb Sn Te/PbTe (substrate), (x > y) double 1-Y Y 1-Y Y heterostructure. To prepare this structure LPE,by the growth temperature must be below 600°C to suppress diffusion of the tin during the epitaxial growth. In this paper, the mechanism by which the heterojunction boundaries become irregular is cleared anda new LPEmethod which prevents the irregularity is explained. The lasers prepared from the wafers grown by the new method have demonstratedCW operation at wavelengths longer than10 urn above liquid nitrogen temperature. INTRODUCTION Lead tin telluride diode lasers are well known as excellent radiation sources in the wavelengths longer than6 pm"). The diode lasers should be of double heterostructures to reduce the threshold current densities or to operate at temperatures above77K(2). Lasers with wavelengths from 6 to 10 ~IUoperate easily above 77K and they in are practical use. On the otherhand, it is very difficult for the lasers to operateCW on with wavelengths over10 urn above liquid nitrogen temperature. The reason for this is the difficulty of preparing flat heterojunctions for the lasers with wavelengths over 10 pm. The heterojunction boundaries become irregular if the junctions are made by the usualLPE method. A new LPEmethod has been developed which overcomes the problem of irregular heterojunction boundaries. By the new method, lasers capable of CW operation have been made at wavelengths longer than10 pm at temperatures above 77K. 45 EXPERIMENTAL Highly reliable lasers with mesa strata structure have been developed (3) (4). The structure ofthe laser is shown in Figure 1. The heterojunctions have been prepared by the liquid phase epitaxial growth method usinggraphite sliding boat. To confinethe photons and carriers inthe activelayer, the proportion of SnTe which is denoted as -X inthe active layer is higherthan Y which is theproportion of two confininglayers. The insulating films have been made by the anodic oxidation (4). The wavelengthof the lasers is determinedby the composition of the activeregion and the temperature. To make thewavelength longer than 10 um at 77K, thevalue of X mustbe more than 0.2, whilefor wavelengths of 6 to 10 pm X is less than 0.2. When thedouble heterojunctions are formedby the usual LPE technique, theshapes of the heterojunction boundaries vary obviously whether the com- position of the active layer is over 0.2 or not. Two photomicrographsof etched cross sections of double heterowafers are shown inFigure 2. Both are grownby the same LPE method. One is anexample ofthe wafers for the 6 to 10 vm lasers, andthe others is an example of the wafersfor over 10 pm. Heterojunctionscontain many dislocations,which is due to the misfit of the lattice constant. The misfitdislocations between the substrate and the firstlayer spread in the direction of thecrystal growth. This is dueto the mutualself diffusion of Pb andSn(5). The growthtemperature of the secondand the third layers mustbe low enough tosuppress mutual diffusion. Therefore,the active region is grown at a temperatureof 600°C. A photomicrographfor the long wavelength laser shows thatthe hetero- junctionboundaries between the first andsecond layers and between the second and thirdlayers are irregular. When theheterojunction boundaries are irregular,the scattering loss that results becomes so seriousthat lasers made fromwafers shown inFigure 2A do notoperate even at 20K. But lasers withwavelengths from 6 to 10 vm, preparedfrom the wafer with flat hetero- junctions shown in Figure 2B, give good CW operationabove 77K. To confirmthe cause of irregularity in wafers for long wavelength lasers, anotherexperiment was performed. The result is shown inFigure 3. The first layer of Y = 0.13 was grown all overthe surface of the PbTe substrate. Afterremoving the solution for the first layer, the solution for the second layer was kept on just half of the first epitaxial layer for 1 minute isothermally at 6OO0C. Then thesolution was removed. The surfacewhere the secondsolution was kept on is uneven,while the remaining half of the surface is smooth.The cross sectional view shows that at some places,the meltingback occurs as was expected,and moreover, epitaxial growth occurs nearthe melted back area. From themeasurement of the lattice constant by the X-ray diffraction technique,the composition of the precipitated layer was determinedto be x = 0.25. From thisexperiment, the cause of irregularity is found tobe due not onlyto the melting back but also the precipitation. Of course,the saturated solutions were used for the epitaxial growth, but even with these saturated solutions, both melting back and precipitation occurred. The mechanismof the melting back and precipitation using saturated solutionscan be explained by the phase diagram. The ternaryphase diagram of Pb, Sn and Te was investigatedin detail by Harris, et al. (6) The temperatureof the epitaxial growth of the second and third layers is below 600"C, so theregion below the 600°C isothermalliquidus line must beconsidered. Note that the isocompositional solidus lines curve sharply wherethe composition is over x = 0.2. First, the irregularity ofthe heterojunction boundary between the first andsecond epitaxial layers is considered. The ternaryphase diagram which is relatedto the first and second layers is shown inFigure 4. When thefirst layer is already grown andthe solution forthe second layer makes contactwith the first layer, the first layer is in equilibriumwith the solution whosecomposition is denotedby the point B where Y = 0.13solidus intercepts the 600°C liquidus. So, thefirst layer is notin equilibrium with the solution at thepoint A where x = 0.25 solidusintercepts the 600°C liquidus,which is thecomposition of the second solution. The firstcomposition and this second composition are so different that a disequilibriumemerges instantaneously when thesecond solution is heldon the first epitaxial layer. To restoreequilibrium between the solid andthe liquid, the Composition of thesecond solution changes from the point A towardthe point B alongthe 600°C isothermalliquidus line. Inorder to change the composition of this solution, the molecular inter- changebetween the solid and the liquid of different composition is required. By thelocal melting back of the first layer of Y = 0.13,the composition ofthe solution moves fromthe point A towardthe point C which is the compositionof the first layer. This compositional change is indicated by the vector a. On theother hand, by the precipitation of x = 0.25, the composi- tion of thesolution moves fromthe point A againstthe point D which is thecomposition of the second layer. This compositional change due to the precipitation is indicatedby the vector b. When themelting back and precipitationoccur simultaneously, the motive force of thecompositional change along the 600°C isothermalliquidus emerges. This is thereason why the interface ofthe first andsecond layers becomes irregular. 47 Themechanism of the irregularity between the second and the third layers is just the reverse mechanismof the aforementioned irregularity at the inter- faceof the first and second layer; that is, thecomposition of the third solution moves towardthe composition of the second solution along the 6OOOC liquidus line. All this explains why the heterojunctions become irregular when double heterowafers are producedby the usual LPE method. However, a way hasbeen found to produce flat heterojunction boundaries on doubleheterowafers for the 6 to10 pm lasers. The reasonthis can be done can be understood from the phase diagram. The isocompositionallines are straightbelow x = 0.20. So, thecompositional differenceof the solutions between the first layer and the second layer is not so large,the result being that serious disequilibrium does not occur. Of course,the solution composition for the second layer moves towardthat of the first solution, but the motive force is so small thatthe boundaries do not become irregular. From theaforementioned analysis, it is seenthat to obtain flat hetero- junctionboundaries for the long wavelength laser, it is necessaryto produce precipitationonly, while suppressing the melting back. So, supercooled solutionsfor the growth of the second and third layers were used in order to allow no time forcompositional change. The temperature profile for the usual LPE growthmethod is shown in Figure 5. The first layer is grown slowlyto assure the quality of the epi- taxiallayer, then the second and third layers are grown continuously at the cooling rate of 1°C perminute. The modifiedtemperature profile in order to achieve the supercooled condition is shown inFigure 6. The beginninghalf of the first layer is grown underthe usual conditions, but in the middle point of the first layer growth,the cooling rate is changed to as much as 3°C per minute. While the last half of the first layer is growing,the solutions for the second and third layers are sufficiently supercooled prior to making contact with the substrate. Then thesecond and the third layers are grown continuously. The cooling rate of 3OC perminute is sufficient to keep the solutions supercooled. A photomicrographof the cross section grownby the new LPE method is shown inFigure