Home Search Collections Journals About Contact us My IOPscience A reassessment of intervalence band absorption in 1.6μm (GaIn)(AsP)/InP This content has been downloaded from IOPscience. Please scroll down to see the full text. 1987 Semicond. Sci. Technol. 2 761 (http://iopscience.iop.org/0268-1242/2/12/001) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 142.66.3.42 This content was downloaded on 08/09/2015 at 15:03 Please note that terms and conditions apply. Semicond. Sci. Technol. 2 ( 1 987) 761 -764.Printed in theUK A R Adams, K C Heasman and J Hilton Department of Physics, University of Surrey, Guildford, Surrey, UK Received 9 June 1987 paper w e re-evaluate the influence of intervalence band absorption on the temperature sensitivity of 1.6 pm (Galn)(AsP)/lnP lasers and emphasise its general importance in semiconductorlasers operating a t several times their threshold current. Abstract. In this 1. Introduction Thetemperature sensitivity of (GaIn)(AsP)/InPlasers fabricated to operate nearI .6 pm where opticalfibres have their minimum absorptioncontinues to be apractical problem. For example. if as is conventional. we write the threshold current density in the form Jth(TZ)=Jrh(TI) e x d ( T 2 - Tl)/TO] (1) then when temperatures TI and T , are near room temperature To is typically 50-70 K [ l ] . This compares with a value of To= 150 K [21 at lower temperatures and values of To= 200 K in GaAlAs/GaAs lasers up to about 350 K [31. Since this effect appears almost independent of the laser structure or the growth mechanism used. theoretical explanations in terms of thefundamental properties of the semiconductor band structure, such as Auger recombination [4] intervalence and band absorption [ 2 ] have been sought. This approach has been supported experimentallyby measurements using hydrostaticpressure [ 5 , 6 ] which showthatthe To problem improves as the direct band gapis increased with pressure andhas led to suggestions how thebandstructurefor long-wavelength lasers might be improved in strained layer systems 17, S]. However, recent careful calculations of the magnitude of Auger recombination rates and of intervalence band absorption have led authors to discount these mechanisms. Takeshima [91 considered both Auger recombination and IVBA and concluded that neither could explain theobservedtemperature sensitivity of lasers. Most recently, Childs and co-workers [IO] concentrated on IVBA calculating its magnitude using a pseudopotential band structure and matrix elements evaluated along 21 kspace directions through the Brillouin zone. They calculatedthatat1.6pmtheabsorption coefficient for IVBA, a= 39 c m - ' for 1 x lo18c m - ) holes atroom temperature, i.e. a= 39 x 10 - I 8 cm2. This is in very good agreement with the magnitude of a determined experimentally from laser characteristics [ 2 , 5 ] which require anabsorption coefficient in theactive region of about 100 cm-' at the threshold carrier density, which is 0268-1242/87/120761 + 0 4 $02.50 @ 1987 Ltd IOP Publishing typically NI,= 2 to 3 x 10l8c m - ) . However. Childs and co-workers concluded that the temperature sensitivity of I V B A ( I / a ) da/dT=0.45 96 K - ' at 300 K which they derived is insufficient to explain the observed temperature sensitivity of lasers. I n this paper we make use of the results of Childs and co-workers but show that when the rate of change of Nthwith temperature is takeninto account, I V B A can account almost entirely for the temperature sensitivity of 1.6 pm devices. Unlike Auger recombination. I V B A directly affects the quantum differential efficiency. vd. of lasers. Here we considerthe light output of devicesworking at several timestheirthreshold currentandshowthat it is the influence of IVBA on ?ld thataffects their temperature stability rather than thestability of the threshold current. 2. Temperature current density dependence of the threshold Childsandco-workersfoundthatthe intervalence band absorption coefficient depended approximately linearly on the hole density over the range considered. i.e. p = 7.5 x IO1' to 3 x I O l 8 cm -). Therefore, since an increase in a due to an increase in temperature requires an increase in threshold carrier density to overcome the extra loss. this in turn increases the absorption and the process iterates. If we write the I V B A absorption coefficient under threshold conditions as aac = aNlh (2) where we take Nth as the electron and hole densities in the active region in units of 10I8cm then -). d a-d a a c-NI,,dT dT dT + a-.dN,,, (3) If the absolute magnitude of a is large the second term of equation (3) can be very important. Indeed, wewill show that the value of To= 150 K expected for a lossless laser is 761 A R Adams et a/ reduced to 120 K when a= 39 x 10 - l 8 cm2 even when d a / d T is zero. The radiative currentdensity may be written as [2] 1 where We took the following typical values for the laser parameters; laserlength L = 300 x cm,thickness of the activeregion d = 0 . 2 x l o p 4cm, confinement factor r =0.636, reflectively of endfacetsR=0.42andloss external to the active region constant and equal to aex= 20 cm- l . From [2], it was assumed that the peak gain of the laser is linearly related to the carrier density so that we may write 20 40 IVBA,[X where A , = 1.75 x 10-l6 cm2 at 300 K and varies with temperatureas derived in [2]andthegainrequiredto reach transparency, ai,,= 200 cm - l at 300 K and was varied with temperature in such a way as togive a value of To= 150 K with aac=O[2]. The gain required to overcomelosses 80 60 (10”~ c m 2 ~ Figure 1. The effect of the magnitude and temperature dependence of intervalence band absorption on To the temperature sensitivity of the threshold current calculated at 300 K. a is the absorption coefficient for an injected hole density of 1 0l8cm-3. The values on the curves are ( I / a ) d a / d T ( % K - ’ ) . The arrow indicates the a value from [IO]. 1 These values resulted in NthN 2 x 10l8cm-3 near 300 K for a=39 x 10-18cm2. From equations (2), (6) and (7) we may write of which only a and its temperature dependence remain adjustable parameters. From Nth and a we determine a,, from (2) and hence vd from (5) and then J R from (4). Finally we introduce a loss current, J A , due to Auger recombination andwrite JR Jth= +J A (9) where J A = edA Nh : (10) A , theAuger where e is the electronic chargeand coefficient, is kept as an adjustable parameter. 3. R e s u l t s The results of the influence of intervalence band absorption on the temperature dependence of the thresholdcurrent of 1.6pm(GaIn)(AsP)/InPlasers is 762 Auger coefflclent , A c m 6 S” ) Figure 2. The variation of TOdue to a temperature independent Auger coefficient and intervalence band absorption with a temperature variation at 300 K of ( l / a )da/dT=0.45% K” [lo].The values on the curve are a ( I O ’ 8 cm’). lntervalence band absorption in 1.6 p m lasers clearly that a loss mechanism proportional to the carrier concentrationdoesnot need to be at all temperature sensitive to influence To. The full curve in figure 1 shows how To varieswith a if ( l / a ) ,da/dT=0.45 % K - I as given in [ 101. We see that when a = 3 9 c m - ' for 1 x 10I8 holes cm -3, the value determinedin [ 101 for 1.6 pm lasers, TOis decreased to 90 K. Increasing a to 60cm-', probably within theaccuracy of thecomplexabsorption calculations,decreases To to 70 K avalue observed experimentally. Although we can see that IVBA as calculated in [ 101 alone can count for most of the temperature sensitivity of 1.6pm lasers, since it causes N,, to change with the temperature, IVBA will also influence Auger recominationratethroughequation (10). To emphasise this point we haveconsidered a situation in which the Auger coefficient A is assumedto be independent of temperature. The results are shown in figure 2 where we 1 x lo1* holes, To is see that when a is 39 cm"for reduced from 90 to 73 K for a temperature independent Auger coefficient of 5 x cm6 S - l . loo/ 50 . 0 I 1 150 2b0 240 Temperature ,T (K) 1 300 Figure 3. Temperature dependence of the threshold current and differential quantum efficiency for 1.6p m (Galn)(AsP)/lnP lasers. Solid symbols, threshold current data [21; open symbols, differential quantum efficiency data [2]; full curves, a= 60 x 1 C'* cm2 broken curves a= 39 x 1O"* cm2 a t 3 0 0 K and varied with temperature as given in [ 101. summarised in figure 1. The threshold current was calculatedat290and320 K and To determined using equation (1). Figure 1 showsthevariation of To with intervalence band absorption a the absorption coefficient for 10l8 cm - 3 holes in the valence band. For a = 0, T = 150 K, the assumed value fora loss free laser. The various curves show the variation of To with a for different values of the temperature dependence of a. It is interesting to see how IVBA decreases To even whenda/dT=O,showing 0.9 0.8 I 300 I 310 Ternperature,T [ K ) I 320 1 Figure 4. ( a )The influence of IVBA on the light-current characteristic of 1.6pm (Galn) (AsP)/lnPlasers operating a t currents five times the threshold current.Full curve T = 2 9 0 K, aac= O ;broken curves T=320 K, cg, = O ;chain curve T= 3 2 0 K. a a c = 100 cm"; (To = 6 5 K, L = 300pm, R= 0.42, r =0.636, aex = 2 0 cm "1. ( b )The temperature dependence of the normalised light output at five times the threshold current as a function of theintervalence band absorption coefficient at3 0 0 K. The values on the curve are aac(cm"). 763 A R Adams et a/ So far we have only considered temperatures around roomtemperature, where lasersarenormallyoperated. However, when studyingthe T o problemauthorshave presented results over amuch wider range.Wehave thereforecalculated I,, and v d from150to320 K for comparison with experiment.Figure 3 showsthe results for a= 39 and 60 x 10 - l 8 cm2 at 300 K and varying with temperature as given graphically in [ 101. The experimental results are taken from [2]. The absolute value of qd for a=60 x 10-l8 c m 2 is in excellent agreement with experiment over the entire temperature range as might be expected. The calculated threshold current increases somewhat less quickly than is normally observed and may reflect the increasing influence of Auger recombination. Finally we would like to consider the influence of IVBA on lasers operating at several times their threshold current. This is becoming of increasing practical importance as the technologies of growthandfabricationimprove.The situation is illustrated diagrammatically in figure 4(a) which indicates how, at five times threshold, the absolute light outputdependsmore sensitively on v d than on Ith, Returningtoour lasermodel we plot in figure 4(b) the light output as a functionof temperature, normalised toits value at 290 K for a device held at a constant operating current five times thethresholdat290 K. Thecurves produced are for constant a,, so they are not the curves a laser would follow when heated and a,, increases. The top curve aac=O is derivedfor T 0 = 6 5K ,a e , = 2 0 c m - ’ assumingthat q d remainstemperatureindependentas would be expected tothe first approximation if Auger recombination alone is responsible for the T oproblem. We see that heating the laser from 290 to 320 K then reduces the light output by 14 % while including a very modest value of a,, = 20 cm - l reduces the output by double this and a more realistic value of a,,= 100 cm - l reduces the light output by 42 %. Clearly IVBA is particularly importantforlasersoperatingat several times their threshold current and may even need to be considered in GaAs devices. 764 4. Conclusion Takingthecalculated values forthemagnitudeand temperature dependence of intervalence band absorption (IVBA) in 1.6 pm (GaIn)(AsP)/InP lasers calculatedin [ lOJ we have shown thatit can accountfor most of the observed temperature sensitivity of the threshold current and quantum differential efficiency. Itwasalsoshownthat IVBA can strongly influence the Auger recombination rate at lasing threshold. Finally it wasshownthat IVBA is particularly important in devices operated at several times their threshold current. Acknowledgments One of us (KCH) wishes tothankSERCand Philips Research Laboratories, Redhill, for financial support. References [ 1 Horikoshi Y 1982 GaInAsPAllow Semiconductors ed. T P Pearsall (New York: Wiley) ch 15 [2] Asada M, Adams A R, Stubkjaer K E, Suematsu Y, Itaya Y and Arai S 198 1 I E E E J . Quantum Electron. JQE 17 61 1 [3] Kojima K, Noda S, Mitsunaga K, Kyuma K and Nakayama T 1985 Appl. Phq’s. Lett. 47 570 [4] Dutta N K and Nelson R J 1982 J . Appl. Phys. 53 74 [ 5 ] Patel D. Adams A R, Greene P D and Henshall G D 1982 Electron. Lett. 18 9 19 [6]MozerA,Hausser S and Pilkuhn M 1985 I E E E J . Quantum Electron. JQE 21 7 19 [ 7 ] Adams A R 1986 Electron. Lett. 22 249 G P [8] E P O’Reilly, K C Heasman,ARAdamsand Witchlow 1987 Superlatt. Microstruct. 3 99 [9] Takeshima M 1984 J . Appl. Phys. 56 69 1 [ l o ] Childs G N, Brand S andAbramRA1986 Semicond. Sci. Techno/. 1 1 16