GaAs single quantum well laser diodes

Microelectronics Reliability 55 (2015) 62–65

Contents lists available at ScienceDirect

Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel

The effect of external stress on the properties of AlGaAs/GaAs single quantum well laser diodes Hui Zhu ⇑, Kun Liu, Cong Xiong, Shiwei Feng, Chunsheng Guo College of Electronic Information and Control Engineering, Beijing University of Technology, Beijing 100124, PR China

a r t i c l e

i n f o

Article history: Received 17 April 2014 Received in revised form 3 August 2014 Accepted 15 September 2014 Available online 18 November 2014 Keywords: AlGaAs/GaAs single quantum well laser diodes Uniform in-plane stress Energy band gap Wavelength

a b s t r a c t The change of spectrum of the AlGaAs/GaAs single quantum well laser diode is measured under the application of uniform uniaxial in-plane tensile and compressive stress. In the range of the tensile stress we apply (up to 597 MPa), the wavelength increases linearly at a rate of 5.3 nm GPa1. The energy band gap decreases with the tensile stress with the slope of 10 meV GPa1, which is close to the theoretical change of the heavy hole band edge with respect to the conduction band edge. There is a shorter wavelength peak existing on the spectrum as the tensile stress increases, suggesting a transition from the conduction band to a higher energy valence band. For the compressive stress (up to 516 MPa), the wavelength decreases with the stress, but it shows an abrupt reduction from 162 to 200 MPa. The threshold current also varies as a result of the change of the energy band structure. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Laser diode bars (LDBs) with AlGaAs active regions have a wide range of applications including solid-state laser pumping, materials processing, optical communications, and printing machines [1–3]. The device structures consist of very thin active layers epitaxially grown on appropriate substrates. There usually exist residual stresses in the thin film layer due to the thermal and lattice mismatches between the epilayer and substrate [4,5]. The stress can also be introduced from the packaging processes, as a result of the differences in the thermal expansion coefficients between the chip and heat sink materials [6–9]. Previous investigations have revealed that the performance, such as the wavelength, threshold, optical gain, and reliability of laser diodes are affected by the residual stresses [10,11]. Furthermore, the energy band structure of III–V semiconductor materials have been found to change with the lattice mismatch stresses, which result from the variation of the composition of the semiconductor materials [12–15]. The band gap increases when under compression and decreases under tension [12–17]. Therefore, for the AlGaAs/GaAs single quantum well laser diodes which are of wide application interest, it is desirable to study in detail the effect of external mechanical stress on the properties for a better understanding of the device performance, for interpretation of experimental results, and for optimization of the device design. However, it is difficult to ⇑ Corresponding author. Tel.: +86 10 67396955; fax: +86 10 67396539. E-mail address: [email protected] (H. Zhu). http://dx.doi.org/10.1016/j.microrel.2014.09.012 0026-2714/Ó 2014 Elsevier Ltd. All rights reserved.

apply a uniform stress either in-plane or out-of-plane on a laser chip, especially when the stress level is much larger than 1 MPa. Adams and colleagues [10] applied an in-plane stress on a unbounded InGaAsP/InP laser chips by supporting it on two diamonds apart while pushing a probe against the back side of the substrate. They observed the threshold current and wavelength increased under in-plane tensile stress and decreased under in-plane compressive stress. The same approach has also been used in other works [18]. However, the in-plane stress applied by such a method is not uniform across a sample and hence the measured results depend not only on the applied force but also the relative position of a device on the sample. In this work, we present the results on the change of spectrum of the unbounded 808-nm AlGaAs/GaAs laser chips under external mechanical stresses. By using a four point bending test rig, a uniform uniaxial in-plane stress (both tensile and compressive) can be applied to the lasers. The change of the energy band structure is also investigated based on the properties change. 2. Experiments The laser diodes studied in this work are grown by metalorganic chemical vapor deposition (MOCVD) on (1 0 0) GaAs substrate and consist of a single quantum well AlGaAs structure. The active region has a thickness of 9 nm which is made up of AlInGaAs. The output and reverse facets are asymmetrically coated with Al2O3 and Al2O3/Si. The chip dimensions are 130 lm (bar thickness)  900 lm (cavity length)  2765 lm (total lateral

H. Zhu et al. / Microelectronics Reliability 55 (2015) 62–65

width), sawed from a 2 in. wafer. The laser bar contains 5 emitters with a width of 500 lm. To avoid heating effect, the lasers are operated with 30 ls current pulses at a duty cycle of 0.3%. The uniform in-plane tensile stress is provided by a four-point bending test rig shown in Fig. 1. Because the sample is too small to be loaded, it is bonded on the center region of a sheet of spring steel which is 30 mm (length)  8 mm (width)  0.26 mm (thickness) using the indium solder. To make sure the bonding is tight enough, we use the uv-cured optical adhesive on the periphery of the sample. The steel sheet with the adhered sample is put between the two sets of center aligned upper and lower rollers. When the height of adjustable platform is increased, the sample would start to bend. The advantage of this method is that the inplane stress induced between the two lower rollers is intrinsically uniform. The lower roller stage is supported by a sensor placed on the platform, which is connected with a monitor to read the force loaded on the sample. The measurement accuracy can reach 0.01 N. When we apply a load on a sample using two sets of rollers as shown in Fig. 1, the bending moment is constant between the two lower rollers. Hence, the in-plane strain and stress at the sample top surface is uniform. The derivation of the in-plane stress from the loading force is described in the previous work [19]. The active region of most lasers is located closer to the top surface of the sample. Therefore, either tensile or compressive stress can be produced within the active region depending whether the laser chip is positioned with the active region up or down. In our experiment, the maximum stress level we apply is less than 600 MPa before the sample breaks up. We increase the stress gradually from 0 to 600 MPa (tensile or compressive), and then reduce it from 600 MPa to 0. To confirm the reliability of the results, we repeat the experiment for two samples.

where h is the Planck’s constant, c the speed of light, and k the wavelength. The linear slope of the energy band gap with respect to the external stress is about 10 meV GPa1 (or the linear slope of the energy band gap with respect to the strain is about 1 eV). It has been known that the stress defeats the symmetry of the semiconductor and results in removal of the degeneracy at the top of the valence band with a splitting between the heavy and light hole states. The energy difference between the bottom of the conduction band and the top of the degenerate valence band under the biaxial stress has been discussed and calculated in the literature [12–17]. For the uniaxial stress in our case, the corresponding energy difference is given by [20]

b DEhh ¼ aðexx þ eyy þ ezz Þ þ ðexx þ eyy  2ezz Þ 2 b DElh ¼ aðexx þ eyy þ ezz Þ  ðexx þ eyy  2ezz Þ 2

ð1Þ ð2Þ

where DEhh is the shift in the heavy hole valence band edge with respect to the conduction band edge, DElh is the shift in the light hole valence band edge with respect to the conduction band edge, exx, eyy, and ezz are the strain components, a and b are the hydrostatic deformation potential and the shear deformation potential, respectively. In the experiment, the uniaxial stress is applied along the x direction in the layer plane (ie. xy-plane); neither the stress along the growth axis (z-axis) nor the shear stresses are imposed on the sample. Therefore, the in-plane uniaxial stress are expressed as

rxx ¼ r ryy ¼ rzz ¼ 0 rxy ¼ ryz ¼ rzx ¼ 0

ð3Þ ð4Þ ð5Þ

where rij is the stress components. The strains due to the stress are given by

3. Results and discussion Fig. 2(a) and (b) shows the spectra measured at different levels of in-plane tensile and compressive stress, respectively. The change of the peak wavelength is plotted in Fig. 3(a), displaying a linear increasing relationship with the tensile stress within the stress level. The linear slope is about 5.3 nm GPa1. For the compressive stress (Fig. 3(b)), the wavelength decreases first linearly, but shows a rapid and big decrease at around 200 MPa. Afterwards, it changes following a similar trend as before. Moreover, when the stress is removed from the sample (represented by the open symbol in Fig. 3), the wavelength can return to their original values with no measurable hysteresis, suggesting that the in-plane stress we apply does not cause any plastic deformation in our sample. The peak wavelengths of the two samples show the changing characteristics under the external stress consistent with each other, although their magnitude deviates from each other which is due to the samples intrinsic difference. The corresponding energy band gap can be derived from the relationship Eg = hc/k,

Sample

Steel sheet

63

2 upper rollers 2 lower rollers

Force sensor Height adjustable platform

Fig. 1. The sketch of the four-point bending test rig.

exx ¼ e eyy –0 ezz –0 exy ¼ eyz ¼ ezx ¼ 0

ð6Þ ð7Þ ð8Þ

The strain and stress are related by

r ¼ C 11 e þ C 12 eyy þ C 12 ezz 0 ¼ C 12 e þ C 11 eyy þ C 12 ezz 0 ¼ C 12 e þ C 12 eyy þ C 11 ezz

ð9Þ ð10Þ ð11Þ

giving that

C 12 e C 11 þ C 12 ! 2C 212 C 11  e C 11 þ C 12

eyy ¼ ezz ¼

ð12Þ

r¼

ð13Þ

where Cij is the elastic stiffness coefficients. From Eqs. (1) and (2) we obtain

  b  DEhh ¼ a exx þ eyy þ ezz þ exx þ eyy  2ezz 2   C 11  C 12 ðC 11 þ 2C 12 Þ ¼ a þb e 2ðC 11 þ C 12 Þ C 11 þ C 12   b  DElh ¼ a exx þ eyy þ ezz  exx þ eyy  2ezz 2   C 11  C 12 ðC 11 þ 2C 12 Þ ¼ a b e 2ðC 11 þ C 12 Þ C 11 þ C 12

ð14Þ

ð15Þ

Using the numerical values for AlGaAs (a = 8.2 eV, b = 2 eV, C11 = 118.8 GPa, C12 = 53.8 GPa) [20,21], Eqs. (14) and (15) yield for AlGaAs, DEhh = 1.8e, DElh = 4.4e. It can be seen that the linear slope of the energy band gap with respect to the uniaxial tensile

64

H. Zhu et al. / Microelectronics Reliability 55 (2015) 62–65

1.2

Tensile stress (MPa)

(a)

1.0

0 140 265 370 475 597

0.8 0.6 0.4 0.2 0.0 804

808

Normalized intensity

Normalized intensity

1.2

1.0

Compressive stress (MPa)

(b)

0 -88 -162 -177 -212 -284 -516

0.8 0.6 0.4 0.2 0.0 800

812

804

808

812

Wavelength (Å)

Wavelength (Å)

Fig. 2. The change of the spectrum of the AlGaAs/GaAs laser chip under the (a) tensile stress and (b) compressive stress.

812

1.565 -1

1.555

5.3 nm GPa Sample 1

804 800

(b)

1.560

808

Eg (eV)

wavelength (nm)

(a)

Sample 2

1.550 1.545 1.540 1.535

796

Sample 2

-10 meV GPa

-1

Sample 1

1.530 792 -600

-400

-200

0

200

400

600

1.525 -600

-400

Stress (MPa)

-200

0

200

400

600

Stress (MPa)

Fig. 3. The change of the (a) wavelength and (b) energy band gap versus the external stress. The solid symbol represents the process to increase the magnitude of the stress. The open symbol represents the process to remove the stress.

0.88

Threshold current (A)

strain from the experimental results are close to the theory results of DEhh, which is also consistent with the previous simulation results [10,22,23]. It is reasonable to believe that the top of the heavy hole valence band and light hole valence band split due to the application of the in-plane tensile stress. The heavy hole band is at the lowest energy state and the change of band gap under tensile stress up to 597 MPa is due to the shift of heavy hole band edge with respect to the conduction band edge. Furthermore, as the tensile stress increases, there is a peak existing on the shorter wavelength side of the heavy hole peak, whose intensity increases gradually with the stress, as it is shown in Fig. 2(a). It is likely caused by a transition from the conduction band to a higher energy band. According to the energy difference between the low and high energy peak which is about 3.85–4.75 meV, the transition tends to be from the conduction band to the light hole subband [24]. For the compressive stress up to 162 MPa, the energy band gap increases as the increase of the magnitude of stress with the linear slope of about 10 meV GPa1, implying that it is also caused by the change of heavy hole band edge with respect to the conduction band edge. However, it displays a rapid increase under the stress from 162 to 200 MPa. Based on the quickly varying in energy under stress [24], the change is probably associated with the shift of light hole subband. Hence it can be derived that, as the compressive stress increases, the final state for the transition from the conduction band may change from the 1st heavy hole subband to the 1st light hole subband. However, because the light hole subband shifts faster than the heavy hole subband with stress, it can intersect the 2nd heavy hole subband at a certain stress level and then increase further. Consequently, the transition final state can return to the heavy hole subband and the change of energy band gap under stress returns to be caused by the shift of the heavy hole band edge with respect to the conduction band edge. This process is supported by the experimental results in Fig. 3(b), where the changing slope reduces to 10 meV GPa1 after 200 MPa. The change of the energy transition under external stress requires further investigation.

0.84 0.80 0.76 0.72 0.68 0.64 -600

-400

-200

0

200

400

600

Stress (MPa) Fig. 4. The change of the threshold current as a function of the external stress. The solid symbol represents the process to increase the magnitude of the stress. The open symbol represents the process to remove the stress.

Besides the wavelength, we also study the threshold current under the application of the external stress. It can be seen from Fig. 4 that the threshold current increases by 14% at 400 MPa which is close to the result from Adams and Cassidy [10,25]. It decreases under compressive stress, but displaying a fluctuation at around 200 MPa. The change of threshold current can also be explained by the change of the energy band structure under external stress. Because the quantum well barrier is decreased by the tensile stress, the carrier confinement can be reduced, and hence the threshold current is increased. Furthermore, the threshold current follows an opposite changing trend under compressive stress as a result of the increase of the quantum well barrier. 4. Conclusion In summary, we apply a uniform in-plane stress to the AlGaAs/ GaAs single quantum well laser chip using a four-point bending

H. Zhu et al. / Microelectronics Reliability 55 (2015) 62–65

test rig and measure the spectrum between 516 and 597 MPa. Our results demonstrate that the wavelength changes linearly (with the slope of 5.3 nm GPa1) and reversibly in the range of the stress we apply. The rate of the corresponding change of the energy band gap under tensile stress is about 10 meV GPa1. By comparison with the calculation result of the change of energy band gap with respect to the uniaxial stress, it can be derived that such change of the energy band gap is due to the shift of the heavy hole band edge with respect to the conduction band edge. Furthermore, there is a high energy peak existing on the spectrum whose intensity increases with the stress. It may be caused by the transition from the conduction band to the light hole energy subband. Besides the linear changing trend, the wavelength and energy band gap show an abrupt change from 162 to 200 MPa under the compressive stress. We attribute this to the change of the final state of the energy band gap from the heavy hole subband to the light hole subband. Finally, we find that the threshold current varies with the stress as a result of the change of the quantum well barrier. In order to minimize the stress effect, lattice matched substrates and expansion matched heat sinks are needed. Moreover, laser bars with smaller bar width (less emitter numbers) is expected to lower the external stress effect. In addition, a stress buffer layer coating can be considered to apply on the laser chip to mitigate the laser spectrum dependence on stress. Acknowledgements This work is supported by the Beijing Natural Science Foundation China under the Grant Number 4122005, by the Natural Science Foundation, China, under the Grant Number 61201046, and by the Research Fund for the Doctoral Program of Higher Education of China under Grant Number 20121103120019. References [1] Treusch HG, Ovtchinnikov A, He X, Kanskar M, Mott J, Yang S. High-brightness semiconductor laser source for materials processing: stacking, beam shaping, and bars. IEEE J Sel Top Quant Electron 2000;6:601–14. [2] Yang HD, Shih CT, Yang SM, Lee TD. High-power broad-area InGaNAs/GaAs quantum-well lasers in the 1200 nm range. Microelectron Reliab 2010;50: 722–5. [3] Bettiati MA. High optical strength GaAs-based laser structures. Microelectron Reliab 2013;53:1496–500. [4] Davydov VY, Averkiev NS, Goncharuk IN, Nelson DK, Nikitina IP. Raman and photoluminescence studies of biaxial strain in GaN epitaxial layers grown on 6H–SiC. J Appl Phys 1997;82:5097–102.

65

[5] Perry WG, Zheleva T, Bremser MD, Davis RF, Shan W, Song JJ. Correlation of biaxial strains, bound exciton energies, and defect microstructures in GaN films grown on AIN/6H-SiC(0 0 0 1) substrates. J Electron Mater 1997;26: 224–31. [6] Ma YY, Sudharsanam K, Pamidighantam R, Ishimura E. Design for enhancing reliability of LD chip in a 10 Gps optoelectronic package. IEEE Electron Packag Technol Conf 2002:164–9. [7] Xia R, Larkins EC, Harrison I, Dods SRA, Andrianov AV, Morgan J, et al. The effect of mounting-induced strain and defect on the properties of AlGaAs 808 nm LDs. Synth Met 2002;127:255–9. [8] Martin E, Landesman JP, Hirtz JP, Fily A. Microphotoluminescence mapping of packaging-induced stress distribution in high-power AlGaAs laser diodes. Appl Phys Lett 1999;75:2521–3. [9] Lisak D, Cassidy DT, Moore AH. Bonding stress and reliability of high power GaAs-based lasers. IEEE Trans Compon Packag Technol 2001;24:92–8. [10] Adams CS, Cassidy DT. Effects of stress on threshold, wavelength, and polarization of the output of InGaAsP semiconductor diode lasers. J Appl Phys 1988;64:6631–8. [11] Yang J, Cassidy DT. Strain measurement and estimation of photoelastic effects and strain-induced optical gain change in ridge waveguide lasers. J Appl Phys 1995;77:3382–7. [12] Kuo CP, Vong SK, Cohen RM, Stringfellow GB. Effect of mismatch strain on band gap in III–V semiconductors. J Appl Phys 1985;57:5428–32. [13] Asai H, Oe K. Energy band-gap shift with elastic strain in GaxIn1xP epitaxial layers on (0 0 1) GaAs substrates. J Appl Phys 1983;54:2052–6. [14] Khan N, Li J. Effects of compressive strain on optical properties of InxGa1x/GaN quantum wells. Appl Phys Lett 2006;89:151916-1–6-3. [15] Olsen GH, Nuese CJ, Smith RT. The effect of elastic strain on energy band gap and lattice parameter in III–V compounds. J Appl Phys 1978;49:5523–9. [16] Zhao DG, Xu SJ, Xie MH, Tong SY. Stress and its effect on optical properties of GaN epilayers grown on Si(1 1 1), 6H–SiC(0 0 0 1), and c-plane sapphire. Appl Phys Lett 2003;83:677–9. [17] Aumer ME, LeBoeuf SF, Moody BF, Bedair SM. Effects of tensile, compressive, and zero strain on localized states in AlInGaN/InGaN quantum-well structures. Appl Phys Lett 2002;80:3099–101. [18] Chang CT, Hsiao SK, Chang EY, Huang JC, Lee CT. Changes of electrical characteristics for AlGaN/GaN HEMTs under uniaxial tenaile strain. IEEE Electron Dev Lett 2009;30:213–5. [19] Zhu H, Chu DP, Fleck NA, Pane I, Huber JE, Natori E. Polarization change of PZTN ferroelectric thin films under uniform in-plane tensile stress. Integr Ferroelectr 2007;95:117–27. [20] Chao CY, Chuang SL. Spin-orbit-coupling effects on the valence-band structure of strained semiconductor quantum wells. Phys Rev B 1992;46:4110–22. [21] Zory Jr PS. Quantum well lasers. Sandiego: Academic press, INC.; 1993. [22] Jagannath C, Koteles ES, Lee J, Chen YJ, Elman BS, Chi JY. Uniaxial stress dependence of spatially confined excitons. Phys Rev B 1986;34:7027–30. [23] Gil B, Lefebvre P, Mathieu H. Reflectance spectroscopy on GaAs–Gao.sAlo.sAs single quantum wells under in-plane uniaxial stress at liquid-helium temperature. Phys Rev B 1988;38:1215–20. [24] Lee J, Jagannath C, Vassell MO, Koteles ES. Mixing of valence subbands in GaAs/ AlxGa1xAs multiple quantum wells by uniaxial stress. Phys Rev B 1988;37: 4164–70. [25] Eliseev PG, Sverdiov BN, Shokhudzhaev N. Reduction of the threshold current of InGaAsP/lnP heterolasers by unidirectional compression. Sov J Quantum Electron 1984;14:1120–1.