Continuous Wave Blue Lasing in III-Nitride Nanobeam Cavity on Silicon

Letter pubs.acs.org/NanoLett

Continuous Wave Blue Lasing in III-Nitride Nanobeam Cavity on Silicon Noelia Vico Triviño, Raphael̈ Butté,* Jean-François Carlin, and Nicolas Grandjean* Institute of Condensed Matter Physics, École Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland S Supporting Information *

ABSTRACT: III−V photonics on silicon is an active and promising research area. Here, we demonstrate room-temperature (RT) lasing in short-wavelength III-nitride photonic crystal nanobeam cavities grown on silicon featuring a single InGaN quantum well (QW). In the low-absorption QW region, high quality factors in excess of 104 are measured, while RT blue lasing under continuous-wave optical pumping is reported in the high-absorption wavelength range, hence the high QW gain region. Lasing characteristics are well accounted for by the large spontaneous emission coupling factor (β > 0.8) inherent to the nanobeam geometry and the large InGaN QW material gain. Our work illustrates the high potential of III-nitrides on silicon for the realization of low power nanophotonic devices with a reduced footprint that would be of prime interest for fundamental light−matter interaction studies and a variety of lab-on-a-chip applications including biophotonics. KEYWORDS: III-nitrides, lasing, nanobeam, silicon integration, photonic crystal, short-wavelength

H

but also via the demonstration of GaN nanolasers relying on the nanowire geometry.18 Such wide bandgap materials offer tunable emission, spanning the UV to near IR spectral region, leading to a high level of design flexibility and hence potentially allowing for the integration of devices operating at short wavelengths as well as in the telecommunication window on the same platform. Furthermore, the large second-order nonlinear susceptibility, stemming from the wurtzite structure, of such materials is crucial for the development of on-chip nonlinear optics and frequency conversion components.19 Beyond their unique optical properties, III-nitrides possess additional features such as high thermal robustness, mechanical hardness, chemical inertness, and biocompatibility.20 Despite the aforementioned advantages and the well-known PhC performance the development of III−N based PhC structures has long been hindered due to several processing issues until the report of a relatively high-Q (∼2400) AlN PhC membrane in 2007.21 Since then, the demonstration of active PhC slabs exhibiting Q-factors up to several thousand has been reported both at UV and visible wavelengths.22−25 In the IR, a Q-factor as high as ∼146 000 has been measured in passive AlN nanobeam structures19 and 34 000 in passive GaN 2D-PhC membranes grown on Si,26 further confirming the promising and rapidly evolving capabilities of this semiconductor family. Lasing has also been obtained in III-nitride band-edge and PhC

igh quality (Q) factors combined with ultrasmall mode volumes (V) are essential features of photonic crystal (PhC) nanocavities. The resulting extremely large Q/V ratios can enhance spontaneous emission thanks to the Purcell effect,1 hence enabling a wide range of applications including cavity quantum electrodynamics (CQED), compact light sources for optical interconnects,2 optical switches3 and memories,4 high sensitivity biosensors,5 and ultrafast low threshold lasers6,7 hinting toward thresholdless lasing.8,9 One-dimensional PhC nanobeam cavities have recently emerged as very promising candidates10−14 that can further enhance the Q/V figure-of-merit with theoretical Q-factors exceeding 106even with low refractive index materials15 and V values close to the diffraction limit ∼(λ/2n)3.12 Besides, their small-footprint and simple shape make them very attractive for optomechanics,16 high integration density, lowpower consumption photonic circuits, or needle-like biosensors.17 So far most of the reported PhC lasers operate in the infrared (IR)-telecom spectral range. However, a shorter wavelength emission is required in numerous applications including optical manipulation, optical trapping, fluorescence-based biosensing, and optogenetics. In this respect, III−nitride (III−N) semiconductor compounds (GaN, AlN, InN, and their ternary alloys) proved extremely successful and versatile for the realization of efficient short wavelength light-emitters. This is exemplified by the fast growing solid-state lighting market based on III−N light-emitting diodes (LEDs) as well as the Blu-ray technology which relies on InGaN-based laser diodes © XXXX American Chemical Society

Received: November 18, 2014 Revised: January 8, 2015

A

DOI: 10.1021/nl504432d Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

nanocavities under pulsed optical excitation,27,28 and room temperature (RT) pulsed blue-violet electrically driven PhC surface emitting lasers have even been realized.29 Those achievements clearly illustrate the growing maturity of the III−N compound semiconductor family and its inherent potential for a wide variety of applications going beyond white LEDs and visible edge-emitting laser diodes. In this work we report RT continuous wave (cw) III−N based blue lasing in a nanobeam cavity in which the gain is provided by a single InGaN/GaN quantum well (QW). This was made possible by taking advantage of epitaxial layers grown on silicon, which were subsequently processed using a single lithographic step and conventional dry etching techniques. The structural and material quality of these devices are confirmed by experimental Q-factors exceeding 14 000. This demonstrates the potential of III-nitrides grown on silicon for integrated photonics and opens new avenues for biophotonics and CQED that would fully benefit from the excellent optoelectronic properties of GaN and related alloys. Here the technological challenge with the fabrication of GaN-based PhCs essentially arises from (i) the mechanical hardness of III−N with dry etching eventually leading to rough and nonvertical sidewalls, (ii) extended structural defects in heteroepitaxial layers due to the large lattice-mismatch with standard substrates, e.g., 17% for GaN grown on Si (111) leading to a dislocation density >109 cm−2, and (iii) the lack of a straightforward way to release membranes. So far III−nitride PhC fabrication techniques have mostly included conformal growth on prepatterned Si substrates,22 layer-transfer bonding from SiC to Si,23 and selective thermal decomposition of GaN.25 Sputtering deposition of AlN-on-insulator for passive devices in the IR has also been used.19 The present nanobeam structure is based on the design proposed in ref 15. It consists of two PhC mirrors with a 7-hole taper cavity whose schematic diagram is displayed in Figure 1a. In this article all of the reported results were obtained on structures that were designed and optimized to emit at wavelengths ranging from the violet to the blue-green spectral

range. The reason behind this will be discussed hereafter. The theoretical Q factor, whatever the structure we consider in this work, is on the order of 106, as estimated by three-dimensional (3D) finite-difference time-domain (FDTD) simulations (see Supporting Information Methods). The mode volume of the first cavity mode (Ey profile shown in Figure 1b) amounts to ∼1.38 (λ/n)3. The fabrication procedure is similar to that described in our previous report.30 It is based on the growth on Si (111) of an AlN buffer layer and a GaN layer with an embedded 3 nm thick In0.15Ga0.85N QW by metal organic vapor phase epitaxy. The total thickness is ∼310 nm. The lithography was performed by means of e-beam using a SiO2 hard mask. Dry etching techniques were used for both III-nitride etching and silicon substrate undercut to release the PhC membrane (relevant details can be found in ref 30). The roughness of postprocessed back and top sides of membranes fabricated with the same fabrication process were shown by atomic force microscopy to be lower than 1.5 nm over a 500 × 500 nm2 area,30 with a peakto-valley height lower than 2.5 nm, which further confirms the high etching selectivity between III−N and Si as well as the advantage and simplicity of this fabrication technique. Figure 1c shows a top scanning electron microscope (SEM) image of one of the fabricated structures. The latter are characterized by an aspect ratio about 1:5, smooth hole sidewalls with a deviation from verticality lower than 5° and a regular PhC pattern. The photonic structures were optically pumped at RT with a cw frequency-doubled Ar+ laser (λ = 244 nm) using a diffraction-limited microphotoluminescence (μPL) spectroscopy setup equipped with a piezoelectric stage and camera imaging (see Supporting Information Methods). This experimental configuration allowed us confirming the strong spatial localization of the resonant modes, whose PL signal quickly vanishes when moving away from the central nanobeam region either along (x) or perpendicularly (y) to the nanobeam axis. The collected light is preferentially polarized along the y-axis evidencing the transverse electric nature of the cavity mode (see Supporting Information, Figure S1). To assess the structural and material quality, Q-factor measurements were carried out on a nanobeam structure whose cavity mode wavelength lies in the low energy tail of the QW. Indeed, the large Stokes shift,31 which strongly suppresses QW absorption in this spectral range, makes the present sample well-suited for the determination of the actual Q-factor, i.e., that of the nanobeam unaffected by the QW absorption-induced optical losses. Otherwise at shorter wavelengths the measured Q value can decrease by up to 1 order of magnitude as shown in microdisks with similar active layers.32 Since this region is inappropriate for the occurrence of gain, it also avoids any line width narrowing induced by lasing that could result in an overestimation of the Q value. Thus, line widths on the order of 60 pm (Q > 8000) were systematically measured. The narrowest line width (∼36 pm), measured at ∼507 nm (corresponding to a Q value >14 000, see Figure 2), is presently limited by the low light intensity issued from the sample in the weak excitation regime (see Supporting Information, Discussion S1 and Figure S2). It clearly demonstrates that no intrinsic limitations inherent to the III−N material system itself are at play. In Figure 3 the emission spectra measured on three different nanobeams with cavity mode in the violet, blue, and blue-green spectral range are displayed together with the RT PL of the bare InGaN QW for comparison. Notice that the active

Figure 1. (a) Schematic drawing of the nanobeam cavity membrane (not drawn to scale). The taper and cavity length are intentionally magnified for clarity. (b) Electric field (Ey) intensity profile of the fundamental cavity mode obtained by 3D-FDTD simulations. (c) Top-view SEM image of a typical nanobeam structure. B

DOI: 10.1021/nl504432d Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

semiconductors. We also briefly emphasize that the emission properties of the single QW were not altered by processing, i.e., that etching defects related to the nanobeam sidewalls are not responsible for an enhancement in the nonradiative recombination rate due to surface recombinations. This can be readily seen in Figure 3b where the blue-green emitting nanobeam cavity exhibits a strict linear increase in the integrated PL intensity as a function of input power density as expected from emission governed by radiative recombination, before any Auger recombination effect takes place. As a result, the present QW behaves similarly to InP QW heterostructures which are known to be much less sensitive to surface nonradiative recombinations than their GaAs counterparts.13 This is likely due to the role played by disorder as recently highlighted by Woolf and co-workers for microdisk structures containing an InGaN active region.33 In contrast to the nanobeam structure whose cavity mode is centered at ∼480 nm (Figure 3b) for which the input−output light curve (Lin−Lout) curve exhibits a linear slope going through the origin that cannot be accounted for by standard laser rate equations in a system characterized by nonradiative recombinations;7,8 the Lin−Lout curves of the two cavity modes centered at 460 and 445 nm are characterized by a change in the slope of the emitted intensity as shown in Figure 3c and d, respectively. Such a feature is naturally compatible with lasing and is well-reproduced via laser rate equation modeling as demonstrated hereafter, which is also supported by the overlap of those cavity modes with the high energy part of the QW emission band that is characterized by a high material gain, as previously reported in GaN microdisks with similar embedded

Figure 2. Spectrum of a nonlasing mode centered at ∼507 nm exhibiting a fwhm as low as 36 pm corresponding to a Q-factor >14 000 acquired at RT under cw excitation with a HeCd laser. The red curve corresponds to a Lorentzian fit to the data.

medium emission line width (Figure 3a) is much larger than that of the cavity modes (∼21 nm vs less than 0.27 nm, Figure 3b−d) and that such a value is typical of InGaN/GaN QWs and is not overdegraded due to the growth on silicon substrate. As far as the active medium is concerned, it is worth recalling that the material gain in InGaN/GaN QWs is extremely large, several thousand inverse centimeters, which is inherited from the large joint-density of states characteristic of wide bandgap

Figure 3. (a) Bare InGaN QW PL emission measured at RT. The dots indicate the three cavity mode positions shown in b−d. The GaN luminescence is also observed at 369 nm. (b) Linear plot of the experimental Lin−Lout curve of the nanobeam cavity mode emitting at 480 nm (nonlasing). Inset: corresponding spectrum acquired at ∼7 kW/cm2. In c and d, curves and spectra analogous to b corresponding to the cw lasing modes at 460 and 445 nm, respectively, are displayed. The additional inset shown in c corresponds to the far-field radiation patterns measured below and above threshold. The input power density at which the patterns were acquired is identified via the picture number, i.e., ∼710 W/cm2 (position 1), ∼2100 W/cm2 (pos. 2), and ∼6400 W/cm2 (pos. 3). C

DOI: 10.1021/nl504432d Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

InGaN QWs.32 The power dependence of the mode profile taken with the optical microscope for the structure emitting at 460 nm is shown as insets in Figure 3c. The observation of a structured far-field pattern above threshold is an additional support in favor of lasing in our structure.12 For most PhC cavities, the usual unambiguous assessment of lasing, i.e., coherence, relies on second-order correlation measurements.7 Here we could circumvent the need for such correlation measurement through the investigation of the dependence of the Lin−Lout curve of nanobeam structures emitting at three different wavelengths that behave differently. It allows us discriminating with a high level of confidence between lasing and non lasing features. Regarding the power dependence of the full width at half-maximum (fwhm) of the mode emitting at 460 nm no significant reduction above threshold has been ascertained (Supporting Information, Figure S3). Such a limited line narrowing is a usual signature of nanobeam lasers and has been previously ascribed in similar GaAs-based cavities to heating effect.11 The lasing threshold can be determined by extrapolating the linear fit above threshold of the Lin−Lout curve. Thus, for the mode emitting at 460 nm (445 nm), it was found to be ∼740 W/cm2 (∼1.65 kW/cm2). The fwhm (Q factor) evaluated just below threshold amounts to 0.27 nm (1700) and 0.18 nm (2600) for the lasing modes centered at 445 and 460 nm, respectively. We believe that the actual Q value of these nanobeam structures is essentially limited by thermo-optic effects that induce a redshift and subsequent peak broadening (Supporting Information, Figure S2 and Supporting Information, Discussion S1). We wish to point out that the nature of the active region plays a crucial role in order to correctly describe the emission regime of nanocavities. Contrary to quantum dots, which represent a model two-level solid state emitter, QWs, due to their continuum of states and their large homogeneous broadening at RT, exhibit a gain bandwidth much larger than the cavity mode line width, which prevents any enhanced spontaneous emission.8,34 This latter aspect now being clarified, the lasing threshold will hence be governed by the properties of the active region, namely gain, but also by the optical quality of the resonator as in a conventional laser diode. The β value, which represents the ratio of the spontaneous emission rate into the lasing mode to the spontaneous emission rate for all the optical modes, is a fundamental parameter for achieving low threshold lasing. Thus, a β value ∼1 is a necessary but not a sufficient condition for thresholdless lasing.8,9 In this respect it is worth mentioning that another advantage of nanobeam structures is that, typically, they do not exhibit mode degeneracy, which is crucial for achieving high β values. A usual way to determine this parameter is obtained via fitting of the Lin−Lout curve with standard laser rate equations35 (see Supporting Information, Methods). As shown in Figure 4 experimental data obtained on the cavity mode centered at 460 nm are in very good agreement with the curve deduced from rate equations when β > 0.8 and the gain coefficient g0 ∼ 3.5 × 104 cm−1. We can notice that a deviation from a straight line in a β = 1 laser is naturally expected for structures where the nonradiative recombination lifetime remains finite,7,36 hence altering the very notion of thresholdless lasing. The analogous fit corresponding to the lasing mode emitting at 445 nm is displayed in Supporting Information, Figure S4. In this latter case the best fit leads to β > 0.6. This lower value compared with that shown in Figure 4 is attributed to the presence of additional competitive modes,

Figure 4. Logarithmic plot of the experimental Lin−Lout curve of the nanobeam cavity laser emitting at 460 nm acquired at RT under cw Ar+ laser optical excitation together with theoretical curves deduced from rate equations for various β values supporting β > 0.8.

predicted by FDTD simulations, overlapping with the QW emission band as observed in the inset of Figure 3d (see also Supporting Information, Discussion S3). To further support the previous analysis we have quantitatively compared the features of the two lasing modes centered at 445 and 460 nm. The differential slope efficiency of the violet nanobeam structure is larger likely due to the larger material gain.35 Knowing that the lasing threshold power density (Pthr) is inversely proportional to the β × Q product,9 we can thus estimate the ratio between the expected threshold power density values for those two modes. The factor of ∼2 which is obtained is in good agreement with the ratio derived from the experimental threshold values obtained via linear fit extrapolation (∼740 W/cm2 and ∼1.65 kW/cm2) (Supporting Information, Discussion S4). Before concluding we emphasize that the lowest absolute threshold power density value (∼740 W/cm2) corresponds to a total power of ∼2.3 μW, an extremely small power by III-nitride standards. Note however that the present nanobeam laser threshold could be further reduced by using multiple QWs, which would also increase the modal gain. It would open the door to energy-efficient short wavelength cw coherent light sources operating with a submicrowatt threshold pumping power. In summary, low-threshold cw lasing has been demonstrated at RT in single InGaN QW small mode-volume nanobeam cavities grown on silicon whose operation window lies in the blue spectral range. Additionally, experimental Q factors larger than 14 000 have been measured. In this regard, the combination of mature InGaN-based heterostructures, exhibiting high optical gain, and high quality factor/small modevolume photonic cavities paves the way toward geometries suitable not only for performing fundamental studies on lightmatter interaction occurring at the nanoscale, but also for biophotonics, optomechanics or the realization of compact coherent light sources. It also offers promising prospects for the integration of wide bandgap semiconductors with the silicon technology platform.

■

ASSOCIATED CONTENT

S Supporting Information *

Methods; supporting figures (S1: polarization dependence of a nanobeam cavity mode, S2: RT μPL spectra taken at various excitation power densities, S3: cavity mode line width power D

DOI: 10.1021/nl504432d Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

dependence, S4: rate equation fit of the mode at 445 nm); supporting discussion (S1: heating effects, S2: determination of transparency and threshold carrier densities, S3: impact of competitive cavity modes on the lasing threshold, S4: estimation of the threshold power density). This material is available free of charge via the Internet at http://pubs.acs.org.

■

(11) Gong, Y.; Ellis, B.; Shambat, G.; Sarmiento, T.; Harris, J. S.; Vučković, J. Opt. Express 2010, 18, 8781−8789 DOI: 10.1364/ OE.18.008781. (12) Zhang, Y.; Khan, M.; Huang, Y.; Ryou, J.; Deotare, P.; Dupuis, R.; Lončar, M. Appl. Phys. Lett. 2010, 97, 051104 DOI: 10.1063/ 1.3475397. (13) Fegadolli, W. S.; Kim, S.-H.; Postigo, P. A.; Scherer, A. Opt. Lett. 2013, 38, 4656−4658 DOI: 10.1364/OL.38.004656. (14) Jeong, K.-Y.; No, Y.-S.; Hwang, Y.; Kim, K. S.; Seo, M.-K.; Park, H.-G.; Lee, Y.-H. Nat. Commun. 2013, 4, 2822 DOI: 10.1038/ ncomms3822. (15) McCutcheon, M. W.; Lončar, M. Opt. Express 2008, 16, 19136− 19145 DOI: 10.1364/OE.16.019136. (16) Eichenfield, M.; Chan, J.; Camacho, R. M.; Vahala, K. J.; Painter, O. Nature 2009, 462, 78−82 DOI: 10.1038/nature08524. (17) Shambat, G.; Kothapalli, S.-R.; Provine, J.; Sarmiento, T.; Harris, J.; Gambhir, S. S.; Vučković, J. Nano Lett. 2013, 13, 4999−5005 DOI: 10.1021/nl304602d. (18) Qian, F.; Li, Y.; Gradečak, S.; Park, H.-G.; Dong, Y.; Ding, Y.; Wang, Z. L.; Lieber, C. M. Nat. Mater. 2008, 7, 701−706 DOI: 10.1038/nmat2253. (19) Pernice, W. H. P.; Xiong, C.; Schuck, C.; Tang, H. X. Appl. Phys. Lett. 2012, 100, 091105 DOI: 10.1063/1.3690888. (20) Stutzmann, M.; Garrido, J. A.; Eickhoff, M.; Brandt, M. S. Phys. Status Solidi A 2006, 203, 3424−3437 DOI: 10.1002/pssa.200622512. (21) Arita, M.; Ishida, S.; Kako, S.; Iwamoto, S.; Arakawa, Y. Appl. Phys. Lett. 2007, 91, 051106 DOI: 10.1063/1.2757596. (22) Néel, D.; Sergent, S.; Mexis, M.; Sam-Giao, D.; Guillet, T.; Brimont, C.; Bretagnon, T.; Semond, F.; Gayral, B.; David, S.; et al. Appl. Phys. Lett. 2011, 98, 261106 DOI: 10.1063/1.3605592. (23) Sergent, S.; Arita, M.; Kako, S.; Tanabe, K.; Iwamoto, S.; Arakawa, Y. Appl. Phys. Lett. 2012, 101, 101106 DOI: 10.1063/ 1.4751336. (24) Vico Triviño, N.; Rossbach, G.; Dharanipathy, U.; Levrat, J.; Castiglia, A.; Carlin, J.-F.; Atlasov, K. A.; Butté, R.; Houdré, R.; Grandjean, N. Appl. Phys. Lett. 2012, 100, 071103 DOI: 10.1063/ 1.3684630. (25) Arita, M.; Kako, S.; Iwamoto, S.; Arakawa, Y. Appl. Phys. Express 2012, 5, 126502 DOI: 10.1143/APEX.5.126502. (26) Roland, I.; Zeng, Y.; Han, Z.; Checoury, X.; Blin, C.; Kurdi, M. E.; Ghrib, A.; Sauvage, S.; Gayral, B.; Brimont, C.; et al. Appl. Phys. Lett. 2014, 105, 011104 DOI: 10.1063/1.4887065. (27) Lai, C.-F.; Yu, P.; Wang, T.-C.; Kuo, H.-C.; Lu, T.-C.; Wang, S.C.; Lee, C.-K. Appl. Phys. Lett. 2007, 91, 041101 DOI: 10.1063/ 1.2759467. (28) Kim, D.-U.; Kim, S.; Lee, J.; Jeon, S.-R.; Jeon, H. IEEE Photon. T ec hn o l . L et t . 2 0 1 1 , 2 3, 1 4 54 − 1 45 6 D O I: 1 0 . 11 0 9/ LPT.2011.2162944. (29) Matsubara, H.; Yoshimoto, S.; Saito, H.; Jianglin, Y.; Tanaka, Y.; Noda, S. Science 2008, 319, 445−447 DOI: 10.1126/science.1150413. (30) Vico Triviño, N.; Dharanipathy, U.; Carlin, J.-F.; Diao, Z.; Houdré, R.; Grandjean, N. Appl. Phys. Lett. 2013, 102, 081120 DOI: 10.1063/1.4793759. (31) Berkowicz, E.; Gershoni, D.; Bahir, G.; Lakin, E.; Shilo, D.; Zolotoyabko, E.; Abare, A. C.; Denbaars, S. P.; Coldren, L. A. Phys. Rev. B 2000, 61, 10994−11008 DOI: 10.1103/PhysRevB.61.10994. (32) Simeonov, D.; Feltin, E.; Altoukhov, A.; Castiglia, A.; Carlin, J.F.; Butté, R.; Grandjean, N. Appl. Phys. Lett. 2008, 92, 171102 DOI: 10.1063/1.2917452. (33) Woolf, A.; Puchtler, T.; Aharonovich, I.; Zhu, T.; Niu, N.; Wang, D.; Oliver, R.; Hu, E. L. Proc. Natl. Acad. Sci. U.S.A. 2014, 111, 14042−14046 DOI: 10.1073/pnas.1415464111. (34) Auffèves, A.; Gerace, D.; Gérard, J.-M.; Santos, M. F.; Andreani, L. C.; Poizat, J.-P. Phys. Rev. B 2010, 81, 245419 DOI: 10.1103/ PhysRevB.81.245419. (35) Coldren, L. A.; Corzine, S. W. Diode Lasers and Photonic Integrated Circuits; Wiley: New York, 1995. (36) Björk, G.; Karlsson, A.; Yamamoto, Y. Phys. Rev. A 1994, 50, 1675−1680 DOI: 10.1103/PhysRevA.50.1675.

AUTHOR INFORMATION

Corresponding Authors

*R.B., e-mail: raphael.butte@epfl.ch. *N.G., e-mail: nicolas.grandjean@epfl.ch. Author Contributions

N.V.T. performed the sample fabrication, the measurements, the data processing, and the 3D-FDTD simulations. N.V.T., R.B., and N.G. performed the data analysis. R.B. handled the framework of the rate equation fit and wrote the corresponding Supporting Information. J.-F.C. grew the sample. N.V.T. and R.B. wrote the paper. N.G. supervised the whole project. All authors contributed to numerous discussions and revised the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to thank M. Glauser and G. Rossbach for their assistance in optical measurements, R. Houdré for his critical reading of the manuscript and A. Castiglia for fruitful discussions. This work was supported by the NCCR Quantum Photonics, research instrument of the Swiss National Science Foundation (SNSF) and by the SNSF (Grants Nos 200020113542 and 200020-150202).

■

ABBREVIATIONS RT, room temperature; QW, quantum well; Q-factor, quality factor; V, mode volume; PhC, photonic crystal; CQED, cavity quantum electrodynamics; IR, infrared; III−N, III−nitrides; LED, light-emitting diode; cw, continuous-wave; 1(2)(3)D, one(two)(three) dimensional; SEM, scanning electron microscope; μPL, microphotoluminescence; Lin−Lout, light-in lightout; fwhm, full width at half-maximum

■

REFERENCES

(1) Purcell, E. Phys. Rev. 1946, 69, 681 DOI: 10.1103/PhysRev.69.674.2. (2) Miller, D. A. B. Proc. IEEE 2009, 97, 1166−1185 DOI: 10.1109/ JPROC.2009.2014298. (3) Tanabe, T.; Notomi, M.; Mitsugi, S.; Shinya, A.; Kuramochi, E. Appl. Phys. Lett. 2005, 87, 151112 DOI: 10.1063/1.2089185. (4) Kuramochi, E.; Nozaki, K.; Shinya, A.; Takeda, K.; Sato, T.; Matsuo, S.; Taniyama, H.; Sumikura, H.; Notomi, M. Nat. Photonics 2014, 8, 474−481 DOI: 10.1038/nphoton.2014.93. (5) Falco, A. D.; O’Faolain, L.; Krauss, T. F. Appl. Phys. Lett. 2009, 94, 063503 DOI: 10.1063/1.3079671. (6) Painter, O.; Lee, R. K.; Scherer, A.; Yariv, A.; O’Brien, J. D.; Dapkus, P. D.; Kim, I. Science 1999, 284, 1819−1821 DOI: 10.1126/ science.284.5421.1819. (7) Nomura, M.; Iwamoto, S.; Kumagai, N.; Arakawa, Y. Phys. Rev. B 2007, 75, 195313 DOI: 10.1103/PhysRevB.75.195313. (8) Björk, G.; Yamamoto, Y. IEEE J. Quantum Electron. 1991, 27, 2386−2396 DOI: 10.1109/3.100877. (9) Yamamoto, Y.; Machida, S.; Björk, G. Phys. Rev. A 1991, 44, 657−668 DOI: 10.1103/PhysRevA.44.657. (10) Notomi, M.; Kuramochi, E.; Taniyama, H. Opt. Express 2008, 16, 11095−11102 DOI: 10.1364/OE.16.011095. E

DOI: 10.1021/nl504432d Nano Lett. XXXX, XXX, XXX−XXX