Quantum-dot semiconductor disk lasers

ARTICLE IN PRESS Journal of Crystal Growth 310 (2008) 5182–5186

Contents lists available at ScienceDirect

Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro

Quantum-dot semiconductor disk lasers T.D. Germann a,, A. Strittmatter a, U.W. Pohl a, D. Bimberg a, J. Rautiainen b, M. Guina b, O.G. Okhotnikov b a b

¨t Berlin, Germany ¨ r Festko ¨rperphysik, Technische Universita Institut fu Optoelectronics Research Centre, Tampere University of Technology, Finland

a r t i c l e in f o

a b s t r a c t

Available online 9 July 2008

We demonstrate quantum-dot (QD)-based, optically pumped semiconductor disk lasers (SDLs) for wavelengths ranging from 950 to 1210 nm. QDs grown either in the submonolayer (SML) or in the Stranski–Krastanow (SK) regime are employed as active layers of the SDLs which are based on two different design concepts. Output power of up to 1.4 W continuous wave (CW) is achieved with an InAs/ GaAs-SML SDL at 1040 nm. Up to 21 InGaAs SK-QD layers within a single SDL gain structure are used to realize the ground-state CW lasing with 0.3 W at 1210 nm. The SK-QD-based SDL shows temperature and pump-power stable emission. Threshold and differential efficiency do not depend on heat-sink temperature. & 2008 Elsevier B.V. All rights reserved.

PACS: 42.55.Px 78.67.Hc 78.67.Pt 81.07.Ta 42.60.v 81.15.Gh Keywords: A3. QD A3. Quantum dot A3. Submonolayer B3. SDL B3. Semiconductor disk laser B3. VECSEL

1. Introduction The concept of optically pumped semiconductor disk lasers (SDLs) enables established GaAs technology to access the visible wavelength range by efficient frequency up-conversion [1,2]. Lowcost, high-power lasers with excellent power scalability and outstanding beam quality [3–7] in a wide spectral range for continuous wave (CW) and short pulse emission [8,9] can be realized. SDLs are key elements for future applications such as low cost, full color video displays, and projection systems. To realize these systems with a maximum possible color gamut, wavelengths close to 1220, 1040, and 940 nm are required for red, green, and blue emission, respectively. Previously reported quantum well (QW)-based SDLs have already proved the advantages of this laser concept [1–6,8,9]. However, QWs emit narrow and temperature-dependent gain spectra, which lead to a strong temperature dependence and consequently to a strong pumppower-dependent emission wavelength [6]. Thus thermal management and careful adjustment of the cavity resonance become essential for all QW-based SDL gain structures. Additionally, the large lattice mismatch of InGaAs QWs limits the application at

wavelengths beyond 1150 nm for GaAs-based devices. Quantumdot (QD)-based gain structures combine a number of advantages for SDLs as compared to QWs like a broad and flat gain spectrum, higher temperature stability, low lasing threshold, and extended wavelength range already demonstrated for a wide range of applications [10,11]. The lower modal gain per QD layer as compared to QWs is counterbalanced by easier stacking of many QD layers without strain compensation. The optical pumping employed for SDLs ensures efficient excitation up to large layer numbers [12]. This paper deals with the realization of the first QD-based SDLs for wavelengths at 950, 1040, and 1210 nm. Submonolayer (SML) deposition of InAs/GaAs QDs as well as Stranski–Krastanow (SK) grown InGaAs QDs are employed as active media in such SDLs. CW operation at room temperature (RT) is achieved for all SDLs. The SML-QD-based SDLs show watt-level output power and reasonable slope efficiencies of up to 12% at wavelengths of 950 and 1040 nm, respectively. The SK-QD-based SDLs exhibit lower output power and slope efficiencies. In contrast, these devices feature temperature-stable output characteristics.

2. Gain structure design  Corresponding author at: Berlin Institute of Technology, Office EW 5-2,

Hardenbergstr. 36, 10623 Berlin, Germany. Tel.: +49 3031422060; +49 3031422596. E-mail address: [email protected] (T.D. Germann). 0022-0248/$ - see front matter & 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2008.07.004

fax:

All SDL gain-chip structures include a semiconductor Bragg mirror, a GaAs pump-light absorber with an embedded stack of active QD layers and a confinement window preventing diffusion

ARTICLE IN PRESS T.D. Germann et al. / Journal of Crystal Growth 310 (2008) 5182–5186

of the photo-excited charge carriers towards the surface. The design of the whole structures is based on calculations carried out by using the transfer-matrix method for normal surface incidence. A nominal reflectivity of the integrated distributed Bragg reflector (DBR) of 99.92% and 99.96% was reached by 35.5 and 37.5 Al0.98Ga0.02As/Al0.2Ga0.8As pairs of l/4 layers, respectively. For the active region two different approaches were used to achieve sufficient gain for a single gain-chip SDL configuration. The first one aims at increasing the modal gain per QD layer while the second aims at a higher total gain by increasing the number of active layers. Table 1 gives an overview on all realized SDLs. Design-A for SDLs emitting at 1040 and 950 nm targets at an increased gain per active layer. It uses a DBR with 35.5 pairs and an Al0.3Ga0.7As charge-carrier barrier with a 10 nm GaAs antioxidation cap. The 1040 nm structures include 13 active QD layers divided into two groups of three (2  3), two groups of two (2  2), and three single (3  1) QD layers. Each group is placed at an antinode of the standing optical wave. The 950 nm SDL was grown with 2  3 and 2  2 grouped layers resulting in a stack of 10 active layers. This nonlinear approach accounts for an exponentially decreasing pump-light intensity within the absorbing GaAs matrix and assures a more homogenous pumping of all active layers. The high gain needed per QD layer was realized by two discrete approaches. QDs grown in the SK growth mode were tuned and optimized for high luminescence at 1100 nm to use the first excited state transition of the QDs for the SDL gain at 1040 nm. Thus the gain generated per QD layer is theoretically up to threefold higher compared to ground-state emission at 1040 nm. Based on comparable structures SK-QD densities of up to 1011 cm2 can be deduced. Much higher densities can be assumed by SML QDs due to significantly increased PL intensities. For precise data SML structures are currently investigated by cross section scanning tunneling microscopy.

Table 1 Overview of all realized QD-based SDL devices presented within this work Type

Design-A

Design-A

Design-A

Design-B

QD layers Transition Wavelength (nm) CW output (mW)

10  SML Ground state 950 500

13  SML Ground state 1040 1400

13  SK Excited state 1040 300

21  SK Ground state 1210 300

5183

The SML growth mode represents thus an appealing alternative approach to achieve ultra-high density QDs with very high gain sufficient for high-power SDLs. In addition, the SML deposition extends the accessible wavelengths towards 900 nm. Therefore, this growth mode was used for SDLs with ground-state emission at 1040 and 950 nm. A comparison of both growth modes is shown in Fig. 1 for RT and operating temperature (366 K) by photoluminescence (PL) spectroscopy of test structures with three active layers. The SK structure shows a broad and flat spectrum compared to the SML luminescence resulting in a constant gain independent of the spectral position of the cavity resonance. Thus temperature changes do not cause a resonancegain detuning. Design-B attempts to use the QD ground-state emission for wavelengths larger than 1200 nm and includes a higher number of active QD layers within the gain structure, which are closely grouped around the antinodes of the optical standing wave. Seven groups of three symmetrically positioned SK-QD layers are placed at the antinodes of the standing optical wave, and a DBR with 37.5 pairs is used. From the characteristics of QD edge emitters based on similar QDs a transparency condition of about 10 A/cm2 can be deduced [13]. This condition is readily fulfilled for all 21 QD layers already at incident power density levels below 1 kW/cm2. Given a modal gain value of 5 cm1 per QD layer for the ground-state transition [14,15] modeling shows that lasing can be achieved with 21 QD layers using a highly reflecting (R499.8%) output coupling mirror. The topmost carrierdiffusion barrier is realized by a lattice-matched 65 nm thick In0.48Ga0.53P top layer which is transparent to the pump light and eliminates the requirement of an absorbing GaAs oxidationprotection layer. Defect-free QDs are essential for all designs as defect multiplication at successive QD layers is highly expected to occur upon stacking leading to degraded device performance. Furthermore, for distances below 40 nm between SK-QD layers, strain-induced structural coupling between individual SK-QD layers causes an increased inhomogeneous broadening upon stacking due to increased SK-QD sizes [16]. Requirements for uniform defect-free dots are met for a minimum spacer thickness of 45 nm as proven by inspection of the surface morphology and PL. The SML growth mode shows good results already for a minimum spacer dimension of only 20 nm. Thus a much closer positioning of the active layers around the antinodes is possible.

Fig. 1. PL test structures with threefold SML-QD or SK-QD stack measured at RT and at 360 K with an excitation density of 500 W/cm2. Vertical lines indicate the position and temperature shift of the sub-cavity resonance of the SDL gain chip.

ARTICLE IN PRESS 5184

T.D. Germann et al. / Journal of Crystal Growth 310 (2008) 5182–5186

3. Growth and characterization All devices are grown on Si-doped GaAs 2-in substrates using metalorganic vapor phase epitaxy (MOVPE). After de-oxidizing the wafers at 715 1C under arsenic atmosphere, a GaAs buffer layer, the DBR structure, and a GaAs layer as part of the sub-cavity prior to the first active layer are grown at 680 1C. Subsequently, the temperature is cycled between 500 1C for SK-QD growth or SML deposition and 595 1C for GaAs spacer growth. Devices with design-A are grown solely with alternative precursors and without arsine. Design-B is optimized to use tertiarybutylarsine (TBAs) only for the active layers grown at 500 1C and to use arsine for all other layers with growth temperatures of 595 or 680 1C. The In0.48Ga0.52P window layer is grown using tertiarybutylphosphine at 595 1C. The use of arsine results in a reduced carbon incorporation and thus in a significantly lowered background pdoping especially of the AlGaAs layers. This effect was checked by capacitance–voltage measurements showing a reduction for the DBR from p ¼ 5  1017 to 2  1016 cm3. Growth of the SML-QD layers is performed for every active layer by fivefold or 10-fold cycled alternate deposition of nominally 0.25 monolayer (ML) or 0.5 ML InAs and 2.3 ML GaAs for 950 or 1040 nm SDL, respectively. In contrast, QD growth in the SK growth regime consists of a single deposition of 2.7 ML In0.65Ga0.35As followed by a growth interruption of 180 s. For emission at 1100 nm, the SK QDs are directly overgrown with GaAs. In order to tune the SK-QD groundstate emission to 1210 nm, the QDs are overgrown by an

additional 3.9 nm thick In0.12Ga0.88As layer. Details of SK-QD growth are given in Ref. [17]. Growth runs are optically in situ monitored by reflectance anisotropy spectroscopy and reflectivity measurements for runto-run fingerprint control. DBR growth reflectance measurements are used to check growth rate and material compositions during the long SDL growth process of more than 15 h. Optical reflectance spectra recorded by an optical spectrum analyzer are used to adjust the DBR stop band and the cavity length. To characterize QD emission RT PL measurements are performed on PL test structures with excitation by a frequency-doubled Nd:YAG laser at 532 nm. The PL test structures comprise two AlGaAs diffusion barriers enclosing the QD layers in 300 nm GaAs matrix. To investigate the stacking properties of the active layers, PL test structures with up to six active layers are characterized. Due to improved growth conditions with low V/III ratios for cap and spacer layers [13] neither broadening nor shifting of the groundstate emission upon stacking is observed in the PL spectrum. To assess the characteristics of complete SDL structures, surface reflectivity and RT PL are measured. Normalized surface PL measurements and reflectivity of SML- and SK-based SDLs are shown in Fig. 2(a) and (b), respectively, showing the temperature dependence of the emission wavelength. The PL peak of the SML structure is completely detuned at RT and is shifted into resonance at operating temperature, whereas the SK structure emits already at RT in resonance and shows the same small temperature shift as the SDL resonance.

4. Device setup and operation

Fig. 2. Normalized surface-emitting PL at RT and at 366 K and measured reflectivity at RT of (a) the 1040 nm SML–SDL structure and (b) the 1040 nm SKQD SDL structure.

Setup and operation conditions are identical for all SDL. To assess SDL operation characteristics a natural diamond heat spreader is liquid-capillary bonded onto the surface of a 2.5  2.5 mm2 gain-mirror piece. The bonded wafer is mounted on a water-cooled copper heat sink kept at 15 1C unless otherwise declared. Optically pumping is provided by a fiber-coupled diode laser delivering CW power of up to 20 W at 808 nm for design-A and up to 50 W at 790 nm for design-B at an incident angle of 351 to the surface normal. Pump-spot size on the gain-chip surface and the size of the cavity mode are matched to a diameter of 180 mm. The external V-shaped cavity is realized by a focusing mirror with a radius of curvature of 200 mm, the gain chip itself, and an output-coupler (OC) mirror with reflectivities of 98%, 99%, or 99.8%. The use of TBAs for the design-A structures leads to a high carbon incorporation in all AlGaAs layers and consequently to an increased absorption and heat generation, resulting in limited lifetime of the SDLs. For design-B, TBAs is replaced by arsine for all layers grown at T ¼ 595 or 700 1C, leading to a DBR with low (p ¼ 2  1016 cm3) background doping. No degradation was observed for hours of operation at high pump-power level for the 1210 nm SDL. Nevertheless, the SML SDLs show output powers of 0.5 W at 950 nm and up to 1.4 W at 1040 nm, limited by the pump power of 20 W. For the SK-QD-based SDLs thermal rollover is observed leading to a maximum output power of 0.3 W for both concepts. The high power of the 1040 nm (with 98% OC) SML SDL matches with the highest slope efficiency of 12% (15% if optical interfaces are taken into account) of all devices and permits the use of a 98% OC. An M2 value of 1.1 demonstrates the nearly diffraction-limited emission of the 950 nm SDL. For all other devices a beam intensity profile was recorded by a charge-coupled device camera to verify the Gaussian beam profile. The emission wavelengths of both SML SDLs shifts with the gain spectra of the active layers (Fig. 2(a)). In contrast, the SK SDLs demonstrate temperature-stable emission spectra independent of the pump power and the heat-sink temperature. This is attributed to the

ARTICLE IN PRESS T.D. Germann et al. / Journal of Crystal Growth 310 (2008) 5182–5186

broader gain spectrum of the SK QDs as shown in Fig. 1. Emission spectra recorded at different heat-sink temperatures (for a medium pump-power level) shown in Fig. 3(c–e) demonstrate the constant emission wavelength. Pump-power-independent emission-wavelength stability is proved by the almost constant center wavelengths in the spectra recorded at different output powers; cf. Fig. 4(a–e). Use of the SK-QD ground-state emission at 1210 nm leads to a temperature-independent slope efficiency of 2% (Fig. 4(f)), in contrast GaAsSb-QW SDLs at 1220 nm reported by Gerster et al. [18]. which show a strong temperature dependence of laser characteristics A very low pump-power threshold density of 2–3 kW/cm2 is measured for all temperatures (Figs. 3(b) and 4(g)) which is twice lower than recently reported for InGaAsNQW-based SDLs [19]. Minimum threshold pump power is 0.48 W at 15 1C and an excellent center wavelength stability of 0.06 nm/K was measured. This is in agreement with an almost constant peak wavelength of the PL emission which shows a shift of only 2 nm and a broadening of 1 nm of full-width at half-maximum as the pump power is increased from 30 to 190 kW/cm2. The spectral

5185

shift value corresponds to the shift of the cavity resonance. In comparison, typical values of GaInNAs-QW-based SDLs show a shift of 0.3 nm/K [6]. A temperature-independent differential efficiency of 2% is found. The low value results from the limited modal gain in the structure requires the use of a 0.2% OC. Modeling suggests lasing with an improved OC of 2% for 30 grouped QD layers. In conclusion, we demonstrated QD-based CW lasing of QD SDLs at 1210, 1040, and 950 nm with two different QD growth concepts. High-power emission up to 1.4 W CW at 1040 nm and 0.5 W CW at 950 nm are realized by active layers grown in the SML regime. The SK growth mode leads to temperature-stable laser characteristics in terms of threshold, differential efficiency, and emission spectrum, with a fixed emission wavelength independent of the incident pump power and the heat-sink temperature due to the broad and flat gain spectrum of the SK QDs. QD ground-state operation of design-B resulted in a very low pump-power threshold of 2–3 kW/ cm2. The arsine-based growth process leads to stable operation without indications of degradation for the 1210 nm SDL.

Fig. 3. (a) Light output characteristics of all SML–SDL devices and (b) of both SK-QD devices (c–e) Laser spectra of the SK-QD SDL recorded at a medium pump-power level of 30 kW/cm2 with heat sink set to different temperatures.

Fig. 4. (a–e) Laser spectra recorded at different output power levels with heat-sink temperature set to 15 1C. (f) Slope efficiency of the 1210 nm SK-QD SDL and (g) threshold pump power both measured with copper heat sink set to different temperatures.

ARTICLE IN PRESS 5186

T.D. Germann et al. / Journal of Crystal Growth 310 (2008) 5182–5186

Acknowledgment The authors acknowledge the support from EU FP6 NATAL Contract no. 016769 and Sfb 787 funded by DFG. References [1] J. Lee, S. Lee, T. Kim, Y. Park, Appl. Phys. Lett. 89 (2006). [2] A. Harkonen, J. Rautiainen, M. Guina, J. Konttinen, P. Tuomisto, L. Orsila, M. Pessa, O.G. Okhotnikov, Opt. Express 15 (2007) 3224. [3] M. Kuznetsov, F. Hakimi, R. Sprague, A. Mooradian, IEEE J. Selected Topics Quantum Electron. 5 (1999) 561. [4] A.C. Tropper, H.D. Foreman, A. Garnache, K.G. Wilcox, S.H. Hoogland, J. Phys. D-Appl. Phys. 37 (2004) R75. [5] J.H. Lee, et al., IEEE Photonics Technol. Lett. 18 (2006) 2117. [6] J. Konttinen, A. Harkonen, P. Tuomisto, M. Guina, J. Rautiainen, M. Pessa, O. Okhotnikov, New J. Phys. 9 (2007). [7] T.D. Germann, A. Strittmatter, J. Pohl, U.W. Pohl, D. Bimberg, J. Rautiainen, M. Guina, O.G. Okhotnikov, Appl. Phys. Lett. 92 (2008). [8] P. Klopp, F. Saas, M. Zorn, M. Weyers, U. Griebner, Opt. Express 16 (2008) 5770.

[9] D. Lorenser, D.J.H.C. Maas, H.J. Unold, A.R. Bellancourt, B. Rudin, E. Gini, D. Ebling, U. Keller, IEEE J. Quantum Electron. 42 (2006) 838. [10] M. Grundmann, O. Stier, S. Bognar, C. Ribbat, F. Heinrichsdorff, D. Bimberg, Phys. Stat. Sol. a-Appl. Res. 178 (2000) 255. [11] D. Bimberg, Electron. Lett. 44 (2008) 168. [12] A. Strittmatter, T.D. Germann, J. Pohl, U.W. Pohl, D. Bimberg, J. Rautiainen, A. Guina, O.G. Okhotnikov, Electron. Lett. 44 (2008) 290. [13] T.D. Germann, A. Strittmatter, T. Kettler, K. Posilovic, U.W. Pohl, D. Bimberg, J. Crystal Growth 298 (2007) 591. [14] N. Kirstaedter, O.G. Schmidt, N.N. Ledentsov, D. Bimberg, V.M. Ustinov, A.Y. Egorov, A.E. Zhukov, M.V. Maximov, P.S. Kopev, Z.I. Alferov, et al., Appl. Phys. Lett. 69 (1996) 1226. [15] L.F. Lester, A. Stintz, H. Li, T.C. Newell, E.A. Pease, B.A. Fuchs, K.J. Malloy, IEEE Photonics Technol. Lett. 11 (1999) 931. [16] B. Lita, R.S. Goldman, J.D. Phillips, P.K. Bhattacharya, Appl. Phys. Lett. 74 (1999) 2824. [17] R.L. Sellin, I. Kaiander, D. Ouyang, T. Kettler, U.W. Pohl, D. Bimberg, N.D. Zakharov, P. Werner, Appl. Phys. Lett. 82 (2003) 841. [18] E. Gerster, I. Ecker, S. Lorch, C. Hahn, S. Menzel, P. Unger, J. Appl. Phys. 94 (2003) 7397. [19] W. Diehl, P. Brick, B. Kunert, S. Reinhard, K. Volz, W. Stolz, Appl. Phys. Lett. 91 (2007).