Phys. Status Solidi A 208, No. 7, 1590–1592 (2011) / DOI 10.1002/pssa.201000948 p s sa

statu

s

soli

di

www.pss-a.comph

ysi

ca

applications and materials science

Investigation of long wavelengthgreen InGaN lasers on c-plane GaN upto 529nm continuous wave operation

Jens Muller*, Uwe Strauß, Teresa Lermer, Georg Bruderl, Christoph Eichler, Adrian Avramescu,and Stephan Lutgen

OSRAM Opto Semiconductors GmbH, Leibnizstraße 4, 93055 Regensburg, Germany

Received 8 September 2010, revised 25 October 2010, accepted 30 October 2010

Published online 30 May 2011

Keywords gain, green InGaN laser, variable mirror coatings, variable stripe length

* Corresponding author: e-mail jens.mueller@osram-os.com, Phone: þ49 151 1625 5520, Fax: þ49 (941) 850 3344

We present analysis of the experimentally determined gain

coefficients g0 for green InGaN lasers with pulsed wavelengths

from 495 to 526 nm. Variable facet coating as well as variable

cavity length method are used to deduce the dependency of g0

on wavelength. From the experimental data we found that better

laser performance at shorter wavelengths correlates with higher

material gain. Therefore, lasers at 504 nm reach wall plug

efficiencies (WPE) up to 5.3% at 50 mW cw output power,

while lasers with 525 and 529 nm show a WPE up to 3 and 2.3%,

respectively. The WPE for longer wavelengths is limited due to

thermal rollover in cw operation.

� 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction InGaN laser diodes (LDs) were wellknown for multimedia data storage applications for manyyears. Few years ago, interest in blue InGaN-based LDs grewstrongly because mobile flying spot laser projection wascoming up. This application requires 450 nm emission athigh output powers and good wall plug efficiencies (WPE)for bright images and long battery lifetimes.

Up to now, projection systems use green laser lightsources based on frequency-doubled infrared lasers.However, future systems may use direct green emittingInGaN lasers with wavelengths beyond 515–530 nm. A firstbreakthrough of these long-wavelength InGaN lasers wasachieved in 2009 when two groups presented lasing at about500 nm for the first time [1, 2]. In the adjacent months,progress was very fast and pulsed operations at 515 nm weredemonstrated on c-plane substrates [3, 4]. Operation at520 nm (continuous wave) and 531 nm (pulsed) was shownby using a {20�21} semipolar crystal orientation [5],however, OSRAM’s polar design recently achieved InGaNlasers at 524 nm (cw) and 531.7 nm (pulsed) [6].

These results of the research groups mentioned beforewere realized within about 1 year due to continousimprovements in epitaxial growth of Indium-rich quantumwells, needed for true green wavelengths. Beside longwavelengths, thresholds below 100 mA and slope

efficiencies of 0.3–0.4 W/A are necessary for output powersof 50 mW. However, performance of cyan and true greenlasers still depends on emission wavelength [7]. Therefore,lasers on c-plane GaN with pulsed wavelengths from 495to 526 nm are analyzed more in detail in this work. Facetcoatings as well as cavity lenghts are varied to deducethe laser and material parameters, respectively. From theexperimental data we found that better laser performance atshorter wavelengths correlates with a higher material gaincoefficient.

2 Determination of gain coefficient Within thegain model for quantum well lasers, the gain g increaseslogarithmically with the carrier density n [8], i.e.

g ¼ g0lnð nntrÞ; (1)

where ntr represents the carrier density at transparency andg0 is the gain coefficient.

Thus, the laser threshold current density jth can bederived as

jth ¼ ntrqdhit

expðaiþamGg0

Þ; (2)

with the transparency carrier density ntr, the carrier lifetimet, the elemental charge q, the thickness d of the quantum

� 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Phys. Status Solidi A 208, No. 7 (2011) 1591

Original

Paper

Figure 1 (online color at: www.pss-a.com) Determination of gaincoefficientg0 byvariable facetcoatingmethod.Fit ismadeaccordingto the inserted equation.

Figure 2 (online color at: www.pss-a.com) Gain coefficientg0 versus wavelength. The black solid line is a phenomenologicalfit.

wells, the confinement factor G, the internal losses ai and themirror losses am which depends on the mirror reflectivitiesR1 and R2 and the cavity length L:

www

am ¼ 12L lnð 1

R2R2Þ: (3)

Figure 3 (online color at: www.pss-a.com) Pulsed median thresh-oldcurrents forgreenInGaNlasersshowing lower thresholdcurrentsfor shorter wavelengths. The black solid line was obtained by fittingthe gain coefficient g0(l) of Fig. 2.

By varying the facet reflectivities R1 and R2, am can bechanged resulting in a linear dependence between thedifference in mirror losses and the logarithm of thresholdcurrents ratio as shown in Fig. 1.

The slope directly gives the product of the confinementfactor and the gain coefficient. The confinement factor can becalculated by a 1D waveguide simulation tool and one getsthe gain coefficient g0. Since other parameters, e.g., theinjection efficiency are eliminated by the method of variablefacet coating, it is well suited to investigate the gain of longwavelength InGaN lasers.

To confirm the results of the variable facet coatingmethod, g0 was also derived by the method of variable stripelength and by Hakki–Paoli measurements although directcomparison with the last one is difficult due to the need ofestimating the carrier density. Therefore, the method ofHakki–Paoli was only used to confirm the order of magnitudefor g0.

3 Wavelength dependency of g0 andconsequences for threshold currents The results forg0 obtained by the three methods are shown in Fig. 2 forwavelengths from 495 to 526 nm. Below wavelengths ofabout 510 nm, an almost constant value for g0 of about1000 cm�1 was found. For longer wavelengths, g0 stronglydecreases resulting in a value of 370 cm�1 for a pulsedwavelength of 526 nm.

For further analysis, the g0 versus wavelength measure-ments were fitted (solid line) to get an analytical expressionfor g0(l).

.pss-a.com

Based on this fit, the calculated wavelength dependentlaser thresholds are shown in Fig. 3 as a solid line togetherwith the measured thresholds (red open cycles). Thecalculation is based on Eq. (2). For the internal losses, avalue of ai¼ 8 cm�1 was derived from Hakki–Paolimeasurements [9]. The threshold measurements presentmedian currents of about 100 LDs per wavelength. Laserswith pulsed wavelengths below 500 nm show thresholdcurrents of about 50 mA. For longer wavelengths at 526 nm,the median threshold increases up to 140 mA.

As can be seen in Fig. 3, the increase in threshold currentfor longer wavelengths can be well described by thecalculation by only assuming a decrease in g0 withwavelength. Since g0 links the gain to the carrier densityaccording to Eq. (1), a smaller gain coefficient means thatless carriers participate in the stimulated emission process ofthe predominating lasing transition. This could be explainedby a smaller electron–hole wavefunction overlap due to

� 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1592 J. Muller et al.: Long wavelength green InGaN lasers on c-plane GaNp

hys

ica ssp st

atu

s

solid

i a

Figure 4 (online color at: www.pss-a.com) PI and WPE of LDswith 504, 525, and 528.8 nm in cw operation.

Figure 5 (online color at: www.pss-a.com) cw spectrum of a528.8 nm InGaN laser at 45 mW. The corresponding WPE is 2.3%.

stronger piezo fields for higher Indium concentrations.However, for wavelengths from 500 to 530 nm, the Indiumconcentration in the quantum wells varies only by about 3%(see Ref. [6]) and can, therefore, not explain the observedstrong decrease in g0. Additionally, piezo fields are stronglyscreened by the high carrier densities corresponding tothreshold currents above 50 mA.

Instead, the decrease in g0 for longer wavelengths can beexplained with higher Indium fluctuations which lead tobroader gain spectra and lower peak gain values [10, 11].These fluctuations dramatically increase for higher Indiumconcentrations (see Refs. [11] and [12]) and, therefore, yieldhigher threshold currents for longer wavelengths.

This increase in threshold current is even morepronounced for cw operation due to thermal self-heating.Figure 4 shows the cw output power for two lasers withwavelengths of 504 and 525 nm up to 50 mW. Additionally, athird InGaN laser with 528.8 nm cw output shows over40 mW output power. It is (to the best of our knowledge) the

� 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

longest cw wavelength ever reported for an InGaN LD. Itsspectrum is shown in Fig. 5.

The maximal WPE of the 528.8 nm laser is 2.3%,while the 525 and 504 nm lasers have WPEs of about 3%and of over 5%, respectively. The WPE for longerwavelength is limited due to thermal roll over. This resultsfrom higher operation currents due to lower gain coefficientsand higher voltages for lasers with more Indium in thequantum wells.

4 Conclusion We analyzed the gain coefficient g0 ofInGaN LDs in the spectral range from 496 to 526 nm byvariable stripe length and variable facet coating method. Wefound that higher threshold currents for longer wavelengthscorrelate with increased compositional fluctuations forhigher indium concentrations. Hence, smaller thresholdcurrents of true green InGaN lasers were achieved byimprovements in material homogeneity yielding lasingabove 528 nm in cw operation.

Acknowledgements The authors thank J. Wagner(Fraunhofer IAF, Freiburg) for valuable comments and discussions.Part of this work was funded by the German Federal Ministry forEducation and Research (BMBF) under Contract No. 13N9373.

References

[1] K. Okamoto, J. Kashiwagi, T. Tanaka, and M. Kubota, Appl.Phys. Lett. 94, 071105 (2009).

[2] D. Queren, A. Avramescu, G. Bruderl, A. Breidenassel, M.Schillgalies, S. Lutgen, and U. Strauss, Appl. Phys. Lett. 94,081119 (2009).

[3] T. Miyoshi, S. Masui, T. Okada, T. Yanamoto, T. Kozaki, S.Nagahama, and T. Mukai, Appl. Phys. Express 2, 062201(2009).

[4] A. Avramescu, T. Lermer, J. Muller, S. Tautz, D. Queren, S.Lutgen, and U. Strauss, Appl. Phys. Lett. 95, 071103 (2009).

[5] Y. Yoshizumi, M. Adachi, Y. Enya, T. Kyono, S. Tokuyama,T. Sumitomo, K. Akita, T. Ikegami, M. Ueno, K. Katayama,and T. Nakamura, Appl. Phys. Express 2, 082101 (2009).

[6] A. Avramescu, T. Lermer, J. Muller, C. Eichler, G. Bruderl,M. Sabathil, S. Lutgen, and U. Strauss, Appl. Phys. Express 3,061003 (2010).

[7] S. Lutgen, A. Avramescu, T. Lermer, M. Schillgalies, D.Queren, J. Muller, D. Dini, A. Breidenassel, and U. Strauss,Proc. SPIE 7616, 76160G (2010).

[8] L. A. Coldren and S. W. Corzine, Diode Lasers and PhotonicIntegrated Circuits (Wiley, New York, 1995).

[9] W. G. Scheibenzuber, U. T. Schwarz, T. Lermer, S. Lutgen,and U. Strauss, Appl. Phys. Lett. 97, 021102 (2010).

[10] K. Kojima, U. T. Schwarz, M. Funato, Y. Kawakami,S. Nagahama, and T. Mukai, Opt. Express 15(12), 7730–7736 (2007).

[11] T. Lermer, A. Gomez-Iglesias, M. Sabathil, J. Muller,S. Lutgen, U. Strauss, B. Pasenow, J. Hader, J. V. Moloney,S. W. Koch, W. Scheibenzuber, and U. T. Schwarz, Appl.Phys. Lett. 98, 021115 (2011).

[12] U. Strauß, A. Avramescu, T. Lermer, D. Queren, A. Gomez-Iglesias, C. Eichler, J. Muller, G. Bruderl, and S. Lutgen,Status Solidi B 248(3), 652–657 (2011).

www.pss-a.com