Ultra-low threshold polariton lasing in photonic crystal cavities
Stefano Azzini, Dario Gerace, Matteo Galli, Isabelle Sagnes, Rémy Braive, Aristide Lemaître, Jacqueline Bloch, Daniele Bajoni
aa r X i v : . [ c ond - m a t . o t h e r] A ug Ultra-low threshold polariton lasing in photonic crystal cavities
Stefano Azzini, Dario Gerace, Matteo Galli, Isabelle Sagnes, R´emyBraive, Aristide Lemaˆıtre, Jacqueline Bloch, and D. Bajoni ∗ Dipartimento di Fisica “A. Volta,” and UdR CNISM, via Bassi 6, 27100 Pavia, Italy CNRS-Laboratoire de Photonique et Nanostructures, Route de Nozay, 91460 Marcoussis, France Dipartimento di Elettronica and UdR CNISM, via Ferrata 1, 27100 Pavia, Italy (Dated: August 9, 2018)The authors show clear experimental evidence of lasing of exciton polaritons confined in L3photonic crystal cavities. The samples are based on an InP membrane in air containing five InAsPquantum wells. Polariton lasing is observed with thresholds as low as 120 nW, below the Motttransition, while conventional photon lasing is observed for a pumping power one to three orders ofmagnitude higher.
PACS numbers: 78.55.Et, 71.36.+c, 78.45.+h
Polariton lasing originates from the spontaneous for-mation of a coherent population of exciton-polaritons outof incoherent excitation [1]. Exciton-polaritons are thedressed states arising from the strong coupling of a pho-tonic mode in a semiconductor microcavity with exci-tons confined in an embedded quantum well (QW) [2].Polariton lasers act as coherent light sources very simi-lar to conventional lasers, the main difference being thatpolariton lasing occurs below the pumping rates neces-sary for population inversion: the formation mechanismof the coherent polariton state is stimulated relaxation ofpolaritons [3], as opposed to stimulated emission of pho-tons. As a result, the threshold for polariton lasing hasbeen predicted [4] and observed to be several orders ofmagnitude below the conventional photon lasing thresh-old in the same samples [5, 6], and recently reported upto room temperature in GaN based samples [7]. How-ever, the threshold powers for polariton lasing reportedup to date are larger than (or comparable to) the lowestthreshold reported for conventional lasers obtained withthe same materials [8]. This is mainly due to the inabilityto confine polaritons in volumes comparable to their op-tical wavelength: polariton “boxes” such as micropillarsand cavity corrugations have been reported with confine-ment volumes on the order of tens of µ m [9].To date, photonic nanocavities realized by point de-fects in photonic crystal (PC) slabs [10]can be fabricatedby top-down lithographic techniques [11], yielding un-precedented figures of merit in terms of quality factor(Q) over effective confinement volume (V eff ) [12]. Thetypical L3 cavity design [13], with three missing holesalong the Γ K direction in a triangular lattice, supportsdiffraction limited cavity modes V eff ≃ ( λ/n ) , allowingthe demonstration of basic cavity QED effects [14, 15]and ultra-low threshold lasing [16].Despite an ongoing research effort to observe the strongcoupling regime in PC structures with embedded QWs[17], the structures realized so far rely on the periodicmodulation of the evanescent tail in the photonic mode[18, 19]. This is mainly due to the introduction of fast P=300 nWP=180 nWP=120 nW
P=2 nW (x100)P=11 nW (x30)P=15 nW (x30) I n t en s i t y ( a . u . ) Energy (meV)
P=45 nW (x5) (a)
P=0.18 W (x30)P=0.3 W (x10)P=0.75 W (x3)P=1.5 W (x2)
Energy (meV)(b) 1 m
P=4.5 WP=9 WP=15 W
Energy (meV)
P=3 nW I n t en s i t y ( a . u . ) FIG. 1: (color online) (a) PL spectra measured on a samplewith a = 235 nm for increasing pump power P ; in the inset adetail of the spectrum for P = 3 nW is shown. (b) PL spectrameasured on the same sample at higher pumping; in the inseta SEM image of the cavity is reported. recombination channels for the excitons when pattern-ing GaAs based QWs, which hinders exciton coherenceto point of preventing strong coupling [18]. In this workwe choose InP-based materials for their negligible nonra-diative recombination issues, even after patterning, andwe report experimental evidence of polariton lasing in L3photonic crystal cavities.The cavities have been fabricated on an InP suspendedmembrane. A 230 nm-thick InP guiding layer was grownby molecular beam epitaxy on top of a 1.5 µ m thick In-GaAs sacrificial layer, on InP substrate. The InP guidinglayer contains five 8 nm-thick shallow InAsP QWs sepa-rated by 12 nm InP barriers at its center. The L3 cavities P (c)(b) Photon LasingThreshold
P P I n t eg r a t ed I n t en s i t y ( a . u . ) Pump Power (nW)
Polariton LasingThreshold (a) L i ne w i d t h ( m e V ) B l ue s h i ft ( m e V ) Pump Power (nW)
FIG. 2: (color online) (a) Integrated intensity of the modefrom PL spectra shown in Fig. 1 as a function of the pumppower. The continuous (black) line, the dotted (red) line andthe dashed (green) line are guides to the eye proportional to,respectively, the pump power P , P and P . (b) Linewidthand (c) blueshift the mode from PL spectra shown in Fig. 1. have been obtained via standard electron beam lithogra-phy followed by inductively coupled plasma dry etchingof the InP top layer. After the etching, the 1.5 µ m sac-rificial layer was selectively removed by wet etching toproduce air suspended membranes. The structural pa-rameters of the PCs were chosen to have the resonancecondition between the fundamental L3 cavity mode andthe QW s-wave exciton. By lithographic tuning, the lat-tice constant a was scanned between 230 and 250 nmevery 5 nm, while the ratio between the hole’s radiusand the lattice constant is kept fixed as r/a = 0 .
32. Theholes at specific positions around the cavity were slightlyvaried in size to maximize the out-of-plane emission fromthe cavity mode [20]. A SEM picture of the cavity regionis shown in the inset of Fig. 1(b). Photoluminescence(PL) experiments were carried out exciting the sampleswith a pulsed laser pump (10 ps pulse width) at λ =750nm focussed on a 500 nm spot through a high numericalaperture microscope objective, and the PL signal was se-lectively collected from the cavity using a confocal set-upthrough the same objective. The sample was kept at 10K in a cold finger cryostat.A typical PL spectrum from the cavities is shown in theinset of Fig. 1(a). Two main features can be highlighted:a broad resonance at 1360 meV, visible also outside thepatterned area, due to the emission from bare QW exci-tons, and a sharp resonance on the low energy side of theexciton transition. The quality factor of these polaritonresonances was, in all samples, between 3000 and 6000,corresponding to a lifetime of the order of a ps. Spec-tra taken from a sample with a = 235 nm are reportedin Figs. 1(a) and (b) for increasing pump power, P . ThePL emission shows a clearly nonlinear behavior: when the excitation power is increased above P ∼
100 nW an evi-dent blueshift and a super-linear increase of PL from thepolariton line can be observed in Fig. 1(a). Another simi-larly nonlinear threshold, accompanied by an even largerblueshift is observed for
P > µ W in Fig 1(b). Betweenthese two thresholds, the line significantly broadens.We summarize in Fig. 2 the behaviors of the integratedpeak intensity, its linewidth and blueshift, respectively,as a function of pump power. Both the first and the sec-ond thresholds are accompanied by a spectral narrowingof the emission, implying the increase of temporal coher-ence. Within the first threshold the emitted peak shiftsby about 1 meV, while a total shift of more than 5 meVis observed before the onset of the second threshold. Thepresence of both thresholds is an unambiguous (althoughindirect) evidence that the sample is in strong coupling atlow pumping powers, and that we are indeed in presenceof both polariton lasing (with threshold around P ∼ ∼
50 W/cm ),and conventional photon lasing (with threshold around P ∼ µ W, ∼ ).Unfortunately, it was not possible to observe anticross-ing between exciton and bare cavity mode. In fact, tem-perature cannot be used as a tuning parameter, as theInAsP QW exciton shifts by less than 1 nm between 4K and 70 K, while the cavity mode shifts by less than2 nm using thin film coating in the cryostat. Moreover,lithographic tuning is too coarse and the points too fewto be used for a reliable anticrossing plot. However, westress that the presence of two thresholds, separated bythe Mott transition, is a sufficient proof that the sample isin strong coupling for pumping powers below 1 µ W. Thisis also confirmed by the blueshift, which continues wellabove the first threshold. This is a clear indication thatthe sample is entering the weak coupling regime, and theemission resonance is shifting from the lower polariton tothe bare cavity mode [6]. Notice that just below polaritonlasing (between 30 nW and 100 nW) there is a quadraticincrease in the emission intensity: such a dependenceis the fingerprint that the dominating relaxation mecha-nism giving rise to polariton lasing is polariton-polaritonscattering, as predicted [4]. Notice also that the lasingthreshold in these samples is reduced by three orders ofmagnitude with respect to the existing literature [5–7],and is comparable to the lowest thresholds reported forquantum dots lasers [16] so far. The threshold for pho-ton lasing, on the contrary, occurs at powers consistentto those reported for other InP-based PC cavity lasers[21].We have observed polariton lasing in samples witha different lattice constant, and thus different exci-ton/cavity detuning, ∆ = E cav − E exc . The thresholdpower increases with increasing ∆, hence with the pho-tonic component of the polariton, as expected. Polari-ton lasing relies on polariton-polariton scattering, so it isstrongly dependent on the excitonic fraction. In Fig. 3 wereport PL spectra collected for increasing pump power ona sample with lattice constant a = 250 nm. In this case,∆ is too large and the exciton fraction is not enough toobtain polariton lasing: as it is shown in Fig. 3(c) theemitted intensity increases linearly, while the line broad-ens and blueshifts due to the progressive loss of strongcoupling. The crossover to weak coupling is observedaround P ∼ µ W as in all other samples. When thesample is in weak coupling the blueshift stops, and for
P > µ W conventional photon lasing sets in with asuper-linear increase of the emitted intensity. The factthat the first threshold is not observed far from the exci-ton resonance proves it is due to excitonic gain, and notto conventional gain due to band filling.At such large negative detunings, changes in refrac-tive index with pumping related to the exciton resonanceare negligible [22]. However effects due to the injectedelectron-hole pairs have to be taken into account follow-ing Ref. 23 and using InP parameters [24]. We obtainthat the bare cavity mode is blueshifted by ∼ E cav = 1329 . E exc =1360 meV, and measuring the lower polariton energy be-low threshold as E LP = 1328 . ~ Ω = q (2 E LP − E cav − E exc ) − ∆ ≃ . ∼ ~ Ω is consistent with what expectedfor five GaAs-based QWs in a comparable system [17].The correspondent detuning is ∆ ≃ −
10 meV for thesample of Figs. 1 and 2 (i.e. a = 235 nm), and ∆ ≃ − a = 250 nm).In conclusion the reduction of the modal volume withrespect to previously studied solutions for polariton con-finement leads to a reduction of more than three orders ofmagnitude in polariton lasing threshold. The ability toconfine polaritons in volumes comparable to cube of theirwavelength should also enable to observe effects relatedto the enhancement of their repulsion, such as polari-ton self-phase modulation [25] and ultimately polaritonblockade [26].This work was supported by CNISM funding throughthe INNESCO project PcPol, by MIUR funding throughthe FIRB “Futuro in Ricerca” project RBFR08XMVYand from the foundation Alma Mater Ticinensis. Weacknowledge L. C. Andreani and M. Patrini for fruitfuldiscussions and L. Ferrera for a first characterization ofthe sample. ∗ [email protected][1] A. Imamoglu, R. J. Ram, S. Pau, and Y. Yamamoto,Phys. Rev. A
100 1000 10000 100000 012310 100 1000 10000 10000010 L i ne w i d t h ( m e V ) Pump Power (nW) P I n t eg r a t ed I n t en s i t y ( a . u . ) Pump Power (nW) P3 P=50 W (x10)P=25 W (x10)P=40 W (x10)P=20 W (x10)P=15 W (x10)
P=10 W (x10)P=30 W (x10)
Energy (meV) (d)(c) (b)
P=4 W (x10)P=2 W (x4) I n t en s i t y ( a . u . ) Energy (meV)
P=1 W (x4)P=100 nWP=200 nWP=500 nW(a) B l ue s h i ft ( m e V ) Pump Power ( W)
FIG. 3: (a) and (b) PL spectra measured on a sample with a = 250 nm for increasing P; inset: blueshift extracted fromthe data. (c) Integrated intensity as a function of the pumppower. The continuous (black) line and the dashed (green)line are guides to the eye proportional to, respectively, thepump power P and P . (d) Linewidth from the data of panels(a) and (b)cond. Sci. Techn.
645 (1998).[3] G. Malpuech, A. Kavokin, A. Di Carlo, and J. J. Baum-berg, Phys. Rev. B , 085304 (2002).[5] J. Kasprzak, M. Richard, S. Kundermann, A. Baas, P.Jeambrun, J. M. J. Keeling, F. M. Marchetti, M. H. Szy-manska, R. Andr´e, J. L. Staehli, V. Savona, P. B. Lit-tlewood, B. Deveaud, and Le Si Dang, Nature , 409(2006).[6] D. Bajoni, P. Senellart, E. Wertz, I. Sagnes, A. Miard,A. Lemaˆıtre, and J. Bloch, Phys. Rev. Lett. , 047401(2008).[7] G. Christmann, R. Butte, E. Feltin, J. F. Carlin, and N.Grandjean, Appl. Phys. Lett. , 886 (1995).[9] J. Bloch, R. Planel, V. Thierry-Mieg, J. M. G´erard, D.Barrier, J. Y. Marzin, and E. Costard, Superlatt. Mi-crostruct. , 371 (1997); O. El Da¨ıf, T. Guillet, J. P.Brantut, R. Idrissi Kaitouni, J. L. Staehli, F. Morier-Genoud, and B. Deveaud, Appl. Phys. Lett. , 061105(2006).[10] J. D. Joannopoulos, S. G. Johnson, J. N. Winn, and R.D. Meade, Photonic Crystals: Molding the Flow of Light. (Princeton University Press, Princeton, 2008).[11] K. H. Lee, S. Guilet, G. Patriarche, I. Sagnes, and A.Talneau, J. Vac. Sci. Technol. B , 1326 (2008); F. Ro- manato et al. , Nanotechnology , 644 (2002); M. Be-lotti, M. Galli, D. Bajoni, L.C. Andreani, G. Guizzetti,D. Decanini, Y. Chen, Microelect. Eng. , 405 (2004).[12] B. S. Song, S. Noda, T. Asano and Y. Akahane, NatureMaterials , 207 (2005); S. Combri´e, A. De Rossi, Q. V.Tran, and H. Benisty, Opt. Letters , 1908 (2008).[13] Y. Akahane, T. Asano, B. S. Song, and S. Noda, Nature , 944 (2003).[14] D. Englund, D. Fattal, E. Waks, G. Solomon, B. Zhang,T. Nakaoka, Y. Arakawa, Y. Yamamoto, and J. Vuˇckovi´c,Phys. Rev. Lett. , 013904 (2005).[15] K. Hennessy, A. Badolato, M. Winger, D. Gerace,M. Atat¨ure, S. Gulde, S. F¨alt, E. Hu, and A. Imamoˇglu,Nature , 896 (2007).[16] S. Strauf, K. Hennessy, M. T. Rakher, Y.-S. Choi, A.Badolato, L. C. Andreani, E. L. Hu, P. M. Petroff, andD. Bouwmeester, Phys. Rev. Lett. , 127404 (2006).[17] D. Gerace and L.C. Andreani, Phys. Rev. B , 235325(2007).[18] D. Bajoni, D. Gerace, M. Galli, J. Bloch, R. Braive, I.Sagnes, A. Miard, A. Lemaˆıtre, M. Patrini, and L. C.Andreani , Phys. Rev. B , 201308(R) (2009). [19] S. Zanotto, G. Biasiol, R. Degli Innocenti, L. Sorba, andA. Tredicucci, Appl. Phys. Lett. , 231123 (2010).[20] N.-V.-Q. Tran, S. Combri´e, and A. De Rossi, Phys. Rev.B , 041101(R) (2009); S. L. Portalupi, M. Galli, C.Reardon, T. F. Krauss, L. O’Faolain, L. C. Andreani,and D. Gerace, Opt. Express , 16064 (2010).[21] L. J. Martinez, B. Al´en, I. Prieto, D. Fuster, L. Gonzalez,Y. Gonzlez, M. L. Dotor, and P. A. Postigo, Opt. Express , 14993 (2009).[22] D. Bajoni, P. Senellart, A. Lemaˆıtre, and J. Bloch, Phys.Rev. B , 201305(R) (2007).[23] I. Fushman, E. Waks, D. Englund, N. Stoltz, P. Petroff,and J. Vuˇckovi´c, Appl. Phys. Lett., , 091118 (2007).[24] ; K.Inoshita and T. Baba IEEE J. Sel. Top. Quant. , 1347(2003)[25] D. M. Whittaker and P. R. Eastham, Europhys. Lett.73