From evidence of strong lightâ•fimatter coupling ... - Wiley Online Library

PACS 42.55.Sa, 71.36.+c, 78.55.Cr, 78.66.Fd, 81.15.Hi. We present both experimental and theoretical results which outline our development of the molecular.
696KB taille 0 téléchargements 32 vues
phys. stat. sol. (b) 244, No. 6, 1882 – 1886 (2007) / DOI 10.1002/pssb.200674922

From evidence of strong light–matter coupling to polariton emission in GaN microcavities I. R. Sellers*, **, 1, F. Semond1, M. Zamfirescu2, 4, F. Stokker-Cheregi2, P. Disseix3, M. Leroux1, J. Leymarie3, M. Gurioli2, A. Vinattieri2, F. Réveret3, G. Malpuech3, A. Vasson3, and J. Massies1 1 2 3 4

CRHEA-CNRS, Rue Bernard Gregory, Parc Sophia Antipolis, 06560 Valbonne, France LENS, Dipartimento di Fisica, Universita di Firenze, 50019 Sesto Fiorentino, Italy LASMEA, Université Blaise Pascal, Clermont Ferrand II, Les Cézeaux, 63177 Aubière Cedex, France National Institute for Lasers, Plasma and Radiation Physics, P.O. Box MG-36, 077125 Bucharest, Romania

Received 30 October 2006, revised 8 December 2006, accepted 11 December 2006 Published online 3 May 2007 PACS 42.55.Sa, 71.36.+ c, 78.55.Cr, 78.66.Fd, 81.15.Hi We present both experimental and theoretical results which outline our development of the molecular beam epitaxy of GaN microcavities on (111) silicon. In particular we show that although in this material system the strong-light matter coupling regime can be observed at 300 K even with relatively low quality factor structures (Q = 60) in reflectivity measurements, it is necessary to increase the Q-factor by at least a factor of two to observe strong coupling in the emission. For an optimized microcavity structure (Q = 160), polaritonic emission is observed at 300 K, with the origin of the broadened luminescence features confirmed by co-incident reflectivity measurements. © 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction The physics of semiconductor microcavities operating in the strong light–mater coupling regime has progressed rapidly in the last decade [1]. This regime occurs when the light–matter coupling is stronger than the decay rates of excitons and the cavity mode [2]. Under such conditions, new quasi particles are created that are both of photonic and excitonic nature, called cavity polaritons [1, 2]. In arsenide or telluride based microcavities, the thermal increase of the excitonic decay rate often prohibits the strong coupling regime to pertain at room temperature. In this respect it was quickly recognised that the Nitride system was a good candidate for microcavity applications, since the large exciton binding energy and optical oscillator strengths of these materials offer the possibility to observe strong coupling at 300 K [3]. In this paper we present the optical characterization of two bulk λ/2-GaN microcavities grown by molecular beam epitaxy on epitaxial AlN/Al0.2Ga0.8N distributed Bragg reflectors (DBRs) grown directly on silicon (111). In sample A, the active layer is grown on a 7 pairs AlN/Al0.2Ga0.8N DBR, and the microcavity structure was completed by the deposition of a 10 nm transparent Al mirror. In sample B, in order to improve the cavity finesse, the number of AlN/Al0.2Ga0.8N pairs was increased to 10 and the upper mirror was an 8 period SiN/SiO2 dielectric DBR. The complete growth technique of (0001) oriented nitrides on silicon is described elsewhere [4].

* **

Corresponding author: e-mail: [email protected] Current address: Department of Physics, SUNY at Buffalo, Fronczak Hall, Buffalo, NY 14260, USA

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Original Paper

phys. stat. sol. (b) 244, No. 6 (2007)

1883

Figure 1(a) shows the reflectivity spectra of the microcavity sample A as a function of the incidence angle at 5 K. Also shown in Fig. 1(a) is the photoluminescence (PL) spectrum at 5 K measured prior to the deposition of the aluminium film. The PL emission peaks at 3.51 eV. This is high in comparison to conventional µm-thick GaN on silicon, emitting near 3.46 eV at low temperature [4], and results from the large biaxial compressive strain within the thin λ/2 GaN layer (~66 nm) induced by the AlN/Al0.2Ga0.8 DBR upon which it is grown. Regarding the angle-resolved reflectivity at 5 K, we note that at the lower incidence angle three dips are observed on the spectrum. At ~3.463 eV is the optical mode, negatively detuned with respect to the A- and B-free excitonic features at 3.521 eV and 3.533 (±0.002 eV) respectively. The observation of strong and well separated A- and B-free excitons in the reflectivity spectra is further evidence of the presence of a large biaxial compression of the GaN film, since it acts to increase the splitting of the valence bands [5]. As the angle of incidence is increased, there is a clear anti-crossing with a resonance between the A and B excitons and the cavity mode at 30° with a global Rabi splitting Ω of 50 meV. For energies larger than resonance, the cavity mode strongly broadens, which we attribute to strong band-to-band absorption in GaN. Transfer matrix simulations [6, 7] of the angle resolved reflectivity spectra are shown in Fig. 1(b). The simulations include both A and B excitons separated by an energy of 15 ± 2 meV consistent with the experimental splitting. The experimental results are very well reproduced with this analysis. The Rabi-splitting is well reproduced (Ω ∼ 50 meV) and the A- and B-excitons display a clear anticrossing

(a) Experimental

60°

60°

EEXC

50°

EEXC

Reflectivity

Reflectivity

A

AB

EPH

EPH 50°

EPH

EPH

(b) Simulation

(a) Experimental

(b) Simulation

B

30°

5° 5°

A EPH 3.4

5° B PL

3.5

3.6

Energy (eV)

3.7

A B

5° EPH

EPH 3.4

3.5

3.6

3.7

Energy (eV)

Fig. 1 (a) Experimental angle resolved reflectivity of sample A at 5 K. (b) Transfer matrix calculations of these reflectivity spectra. The dotted lines are a guides for the eye. The absence of the central dotted line is to preserve the clarity of the central polariton branch. EPH indicates the position energy of the photonic mode far from the resonance. www.pss-b.com

3.4

EEXC

EEXC EPH

3.5 3.6 3.7 3.4 3.5 3.6 Energy (eV) Energy (eV)

3.7

Fig. 2 (a) Experimental angle resolved reflectivity of sample A at 300 K. (b) Transfer matrix simulations of the reflectivity spectra at 300 K. EPH and EEXC indicate respectively the position energy of the photonic mode and the excitonic mode far from the resonance.

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1884

I. R. Sellers et al.: From evidence of strong light – matter coupling to polariton emission

with the photonic mode. It is also evident that the position of the central polariton branch is limited within the energy range determined by the A- and B-exciton energies. The angle dependent reflectivity of sample A at room temperature is shown in Fig. 2(a). At 5° the photon mode is again negatively detuned from the excitonic features, which can no longer be resolved independently due to the effects of thermal broadening. As the angle of incidence is tuned, a clear anticrossing is again observed with Ω ∼ 50 meV. Figure 2(b) shows the transfer matrix simulations of the 300 K reflectivity spectra, with again very good agreement with the experimental results. Although both excitons are included in the simulations at 300K the resolution of the individual excitons is screened by their thermal broadening resulting in a single broadened excitonic feature in the reflectivity spectra. Despite the good reflectivity properties of sample A (with an upper aluminium mirror) no evidence for strong-coupling was observed in its luminescence spectra. A transfer matrix analysis indicates this is due to the low Q-factor of this structure (Q = 60) and that a factor of at least 2 improvement [8] is necessary to observe strong coupling in the emission. This motivated the growth of sample B, with an improved epitaxial nitride bottom mirror and an 8 periods SiN/SiO2 dielectric upper mirror, resulting in a Q-factor of ∼160. Figure 3(a) and 3(b) show respectively the position dependent reflectivity and photoluminescence spectra of sample B at 10 K. In this case the photonic mode is tuned by using the variation in the thickness gradient created through an interruption of the substrate rotation during the growth of the GaN active layer. In Fig. 3(a) the photonic mode (labeled EPH) at 3.462 eV is negatively detuned with respect to the excitonic ones at ∼3.54 eV. The relative weakness of this last branch is related to the large negative detuning of the cavity at this position and the resulting low photon-like nature of the upper polariton. We note that the broadening of this excitonic dip is larger than in sample A (although the two excitons are apparent close to resonance). As the position on the sample is changed the photonic mode increases in energy consistent with a reduction in the active region thickness. As the detuning of the polariton modes is reduced the intensity of the exciton-like mode increases consistent with its increasingly photon-like nature. A clear anti-crossing is observed.

10K (b)

(a)

10K

Reflectivity

photoluminescence

n+1

EX

EPH 3.40

3.45 3.50

EPH E X1

1

3.55 3.60

Energy (eV)

3.40

3.45

3.50

EX2 3.55

3.60

Energy (eV)

Fig. 3 (a) Reflectivity and (b) photoluminescence spectra of sample B at 10 K as a function of position.

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.pss-b.com

Original Paper

phys. stat. sol. (b) 244, No. 6 (2007)

1885

50°

50°

Reflectivity

(b)



300K

(a)

photoluminescence

Figure 3(b) shows the photoluminescence spectra recorded at the same positions as reflectivity. In this case in the bottom spectrum (position 1) three features are visible with the photonic mode at 3.463 eV again negatively detuned with respect to two broad exciton-like features at 3.514 eV (EX1) and 3.539 eV (EX2). Although the A and B excitons are known to interact strongly with the photonic mode, in this case the energy separation of ~25 meV is much larger than the A- and B-separation in compressively strained GaN (∼12 meV) [5]. We tentatively attribute the EX1 PL feature to the combined contribution of localised neutral donor bound excitons (which dominate the GaN PL at low temperature) and of A-excitons, and the shoulder at 3.539 eV (EX2) to B excitons. Since the localised excitons have a much lower density than that of the free excitons strong light–matter coupling involving these states is unlikely. This analysis is consistent with the absence of the EX1 feature in the reflectivity spectra, which are closely related to the absorption and therefore a probe of the density of states. As the position on the sample is changed the excitonic and photonic modes in the PL spectra display again an anti-crossing, though due to thermalisation the population of the upper branch is weak. Figure 4(a) and (b) show respectively the angle-resolved reflectivity and photoluminescence spectra of sample B at 300 K. On the 5° reflectivity spectra, the photonic mode at 3.444 eV is negatively detuned with respect to the excitonic mode at 3.499 eV. As the angle is increased an anti-crossing is observed with a resonance at 35°. The angle-resolved PL spectra are shown in Fig. 4(b). Despite the thermal broadening of the transitions an anti-crossing of the modes is again observed. Since the luminescence of GaN is rather complex with contributions from localised, free excitonic and phonon related emission [9] such simultaneous measurement of the reflectivity and photoluminescence is essential, and confirms the origin of the features in the photoluminescence here to that of polaritonic modes in the strong-coupling regime. This is illustrated more clearly in Fig. 4(c), which shows the energy position of the polariton

5° 300K

3.4

3.5

3.6

Energy (eV)

3.58

Energy (eV)

3.56

3.4

3.5

Energy (eV)

3.6

Reflectivity Photoluminescence

(c)

3.54 3.52

photon

3.50

B-exciton

3.48

A-exciton

3.46 3.44 3.42

10

20

30

40

Fig. 4 (a) Angle resolved reflectivity and (b) PL spectra at 300 K of sample B. (c) Energy as a function of angle for both reflectivity (open squares) and photoluminescence (closed circles). Also shown are the uncoupled A- and B-excitonic (dotted lines) and photonic modes (dashed line). The solid lines represent the numerical dispersion relation using a (3 × 3) matrix analysis.

50

Angle (°) www.pss-b.com

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1886

I. R. Sellers et al.: From evidence of strong light – matter coupling to polariton emission

modes in both reflectivity and photoluminescence. Also shown in bold are the results of a (3 × 3) matrix calculation considering the system as the interaction between A -and B-excitons of constant energy and an angle dependent photonic mode. Despite the limited resolution of the room temperature emission, the resulting eigenvalues reproduce the experimental data rather well with a Rabi splitting of 31 ± 6 meV. The C-exciton is not included in this analysis since it is a much higher energy and has a considerably weaker oscillator strength at near normal incidence than A- and B-excitons in such highly biaxially compressed GaN [5]. Recently relevant confirmations of the mixing between excitons and photons on sample B have been reported by M. Gurioli et al. [10]. Through an analysis of the spectral linewidth and of the photoluminescence lifetime they highlight the polaritonic character of the sample B emission. In summary we have shown that although it is possible to observe strong light–matter coupling at room-temperature for relatively low finesse GaN microcavities in reflectivity measurements, it is necessary to improve that cavity finesse by at least a factor of 2 to observe polariton emission at 300 K. Acknowledgements The authors would like to acknowledge funding from the EU projects CLERMONT 2 (MRTN-CT-2003-503577) and STIMSCAT (STREP contract 517769).

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]

M. S. Skolnick, T. A. Fisher, and D. M. Whittaker, Semicond. Sci. Technol. 13, 645 (1998). V. Savona, L. C. Andreani, P. Schwendimann, and A. Quattropani, Solid State Commun. 93, 733 (1995). A. Kavokin and B. Gil, Appl. Phys. Lett. 72, 2880 (1998). F. Semond, B. Damilano, S. Vézian, N. Grandjean, M. Leroux, and J. Massies, phys. stat. sol. (b) 216, 101 (2001). B. Gil and O. Briot, Phys. Rev. B 55, 2530 (1997). N. Ollier, F. Natali, D. Byrne, P. Disseix, M. Mihailovic, A. Vasson, J. Leymarie, F. Semond, and J. Massies, Jpn. J. Appl. Phys. 44, 4902 (2005). I. R. Sellers, F. Semond, M. Leroux, J. Massies, P. Disseix, A. L. Henneghien, J. Leymarie, and A. Vasson, Phys. Rev. B 73, 033304 (2006). I. R. Sellers, F. Semond, M. Leroux, J. Massies, P. Disseix, G. Malpuech, J. Leymarie, A. L. Henneghien, and A. Vasson, Mater. Res. Soc. Symp. Proc. 892, 0892-FF20-04.1 (2006). M. Leroux, N. Grandjean, B. Beaumont, G. Nataf, F. Semond, J. Massies, and P. Gibart, J. Appl. Phys. 86, 3721 (1999). M. Gurioli et al., submitted.

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.pss-b.com