Optofluidic integration of a photonic crystal nanolaser

We demonstrate a new type of photonic crystal nanolaser incorporated into a microfluidic chip, which is fabricated by multilayer soft lithography. Experimentally, room-temperature continuous-wave lasing operation was achieved by integrating a photonic crystal nanocavity with a microfluidic unit, in which the flow medium both enhances the rate of heat removal and modulates the refractive index contrast. Furthermore, using the proposed system, dynamic modulation of the resonance wavelength and far-field radiation pattern can be achieved by introducing a bottom reflector across which various fluids with different refractive indices are forced to flow. In particular, by maintaining a gap between the reflector and the cavity equal to the emission wavelength, highly efficient unidirectional emission can be obtained. The proposed nanolasers are ideal platforms for highfidelity biological and chemical detection tools in micro-total-analytical or lab-on-a-chip systems. © 2008 Optical Society of America OCIS codes: (140.5960) Semiconductor lasers; (140.3600) Lasers, tunable; (140.3945) Microcavities; (220.4000) Microstructure fabrication; (230.5298) Photonic crystals; (280.4788) Optical sensing and sensors References and links 1. H.-G. Park, S.-H. Kim, S.-H. Kwon, Y.-G. Ju, J.-K. Yang, J.-H. Baek, S.-B. Kim, and Y.-H. Lee, “Electrically driven single-cell photonic crystal laser,” Science 305, 1444–1447 (2004). 2. J. M. Gérard and B. Gayral, “Toward high-efficiency quantum-dot single-photon sources,” Proc. SPIE 5361, 88–95 (2004). 3. D. Englund, D. Fattal, E. Waks, G. Solomon, B. Zhang, T. Nakaoka, Y. Arakawa, Y. Yamamoto, and J. Vučković, “Controlling the spontaneous emission rate of single quantum dots in a two-dimensional photonic crystal,” Phys. Rev. Lett. 95, 013904 (2005). 4. K. Nozaki, S. Kita, and T. Baba, “Room temperature continuous wave operation and controlled spontaneous emission in ultrasmall photonic crystal nanolaser,” Opt. Express 15, 7506–7514 (2007). 5. K. Srinivasan and O. Painter, “Momentum space design of high-Q photonic crystal optical cavities,” Opt. Express 10, 670–684 (2002). 6. J.-K. Hwang, H.-Y. Ryu, D.-S. Song, I.-Y. Han, H.-W. Song, H.-K. Park, Y.-H. Lee, and D.-H. Jang, “Roomtemperature triangular-lattice two-dimensional photonic band gap lasers operating at 1.54 μm,” Appl. Phys. Lett. 76, 2082 (2000). #93365 $15.00 USD Received 3 Mar 2008; revised 18 Apr 2008; accepted 18 Apr 2008; published 23 Apr 2008 (C) 2008 OSA 28 April 2008 / Vol. 16, No. 9 / OPTICS EXPRESS 6515 7. D. C. Duffy, J. C. McDonald, O. J. A. Schueller, and G. M. Whitesides, “Rapid prototyping of microfluidic systems in poly(dimethylsiloxane),” Anal. Chem. 70, 4974–4984 (1998). 8. M. Adams, M. Loncar, A. Scherer, and Y. Qiu, “Microfluidic integration of porous photonic crystal nanolasers for chemical sensing,” IEEE J. Sel. Areas Commun. 23, 1348–1354 (2005). 9. S.-H. Kim, S.-K. Kim, and Y.-H. Lee, “Vertical beaming of wavelength-scale photonic crystal resonators,” Phys. Rev. B 73, 235117 (2006). 10. S.-H. Kim, S.-K. Lee, Y.-H. Lee, and S.-M. Yang, “Microfluidic channel with built-in photonic crystal nanolaser,” Proc. SPIE 6645, 66451K (2007). 11. M. Nomura, S. Iwamoto, K. Watanabe, N. Kumagai, Y. Nakata, S. Ishida, and Y. Arakawa, “Room temperature continuous-wave lasing in photonic crystal nanocavity,” Opt. Express 14, 6308–6315 (2006). 12. Y. Akahane, T. Asano, B. S. Song, and S. Noda, “High-Q photonic nanocavity in a two-dimensional photonic crystal,” Nature (London) 425, 944–947 (2003). 13. M. Lonćar, A. Scherer, and Y. Qiu, “Photonic crystal laser sources for chemical detection,” Appl. Phys. Lett. 82, 648–650 (2003). 14. S. G. Johnson, S. Fan, P. R. Villeneuve, J. D. Joannopoulos, and L. A. Kolodziejski, “Guided modes in photonic crystal slabs,” Phys. Rev. B 60, 5751–5758 (1999). 15. O. Painter, R. K. Lee, A. Yariv, A. Scherer, J. D. O’Brien, P. D. Dapkus, and I. Kim, “Two-dimensional photonic band-gap defect mode laser,” Science 284, 1819–1821 (1999). 16. H.-Y. Ryu, M. Notomi, and Y.-H. Lee, “High-quality-factor and small-mode-volume hexapole modes in photonic-crystal-slab nanocavities,” Appl. Phys. Lett. 83, 4294–4296 (2003). 17. H.-G. Park, J.-K. Hwang, J. Huh, H.-Y. Ryu, Y.-H. Lee, and J.-S. Kim, “Nondegenerate monopole-mode twodimensional photonic band gap laser,” Appl. Phys. Lett. 79, 3032–3034 (2001). 18. C. Seassal, C. Monat, J. Mouette, E. Touraille, B. B. Bakir, H. T. Hattori, J. L. Leclercq, X. Letartre, P. RojoRomeo, and P. Viktorovitch, “Inp bonded membrane photonics components and circuits: toward 2.5 dimensional micro-nano-photonics,” IEEE J. Sel. Top. Quantum Electron. 11, 395 (2005). 19. B. Ben Bakir, C. Seassal, X. Letartre, P. Viktorovitch, M. Zussy, L. Di Cioccio, and J. M. Fedeli, “Surfaceemitting microlaser combining two-dimensional photonic crystal membrane and vertical bragg mirror,” Appl. Phys. Lett. 88, 081113 (2006). 20. J. Z. Chen, Z. Liu, Y. S. Rumala, D. L. Sivco, and C. F. Gmachl, “Direct liquid cooling of room-temperature operated quantum cascade lasers,” Electron. Lett. 42, 534–535 (2006). 21. H.-Y. Ryu, H.-G. Park, and Y.-H. Lee, “Two-dimensional photonic crystal semiconductor lasers: Computational design, fabrication, and characterization,” J. Sel. Top. Quantum Electron. 8, 891–908 (2002). 22. A. Taflove and S. C. Hagness, Computational Electrodynamics: the finite-difference time-domain method (Artech House, Norwood, MA, 2000), 2nd ed. 23. E. Chow, A. Grot, L. W. Mirkarimi, M. Sigalas, and G. Girolami, “Ultracompact biochemical sensor built with two-dimensional photonic crystal microcavity,” Opt. Lett. 29, 1093–1095 (2004). 24. J. Homola, S. S. Yee, and G. Gauglitz, “Surface plasmon resonance sensors: review,” Sens. Actuators B 54, 3–15 (1999). 25. S.-K. Lee, G.-R. Yi, and S.-M. Yang, “High-speed fabrication of patterned colloidal photonic structures in centrifugal microfluidic chips,” Lab. Chip 6, 1171–1177 (2006). 26. D. Erickson, T. Rockwood, T. Emery, A. Scherer, and D. Psaltis, “Nanofluidic tuning of photonic crystal circuits,” Opt. Lett. 31, 59–61 (2006). 27. E. A. Hinds, in Cavity Quantum Electrodynamics (Academic Press, Inc, Orlando, 1994). Edited by P. R. Berman. 28. M. L. Povinelli, S. G. Johnson, J. D. Joannopoulos, and J. B. Pendry, “Toward photonic-crystal metamaterials: Creating magnetic emitters in photonic crystals,” Appl. Phys. Lett. 82, 1069–1071 (2003). 29. M.-K. Kim, J.-K. Yang, Y.-H. Lee, and I.-K. Hwang, “Influence of etching slope on two-dimensional photonic crystal slab resonators,” J. Kor. Phys. Soc. 50, 1027–1031 (2007). 30. M. Forsberg, D. Pasquariello, M. Camacho, and D. Bergman, “InP and Si metal-oxide semiconductor structures fabricated using oxygen plasma assisted wafer bonding,” J. Electron. Mater. 32, 111–116 (2003). 31. S.-H. Kim and Y.-H. Lee, “Symmetry relations of two-dimensional photonic crystal cavity modes,” IEEE J. Quantum Electron. 39, 1081–1085 (2003). 32. K. Inoshita and T. Baba, “Fabrication of GaInAsP/InP photonic crystal lasers by ICP etching and control of resonant mode in point and line composite defects,” IEEE J. Sel. Top. Quantum Electron. 9, 1347–1354 (2003). 33. H. Altug, D. Englund, and J. Vučković, “Ultra-fast photonic crystal nanolasers,” Nat. Phys. 2, 484–488 (2006).


Introduction
Active photonic crystal (PhC) light emitters based on wavelength-scale cavities have been of particular interest to laser physics and quantum information researchers due to their potential applications in efficient single photon sources and ultra-low-threshold lasers.[1,2,3,4] However, highly divergent far-field emission inherent to the wavelength-scale small nature means that these systems suffer from poor vertical out-coupling efficiency.[5] Moreover, the poor thermal characteristics associated with air-suspended membrane structures degrade the performance of nanolasers.[6] Here we show that all those drawbacks can be simultaneously overcome by utilizing simple microfluidics technology based on soft lithography with poly(dimethylsiloxane) (PDMS).[7,8] Recently, Kim et al. proposed the use of a bottom reflector to take advantage of interference effects in order to control the far-field pattern of a PhC nanocavity.[9] They showed that, by matching the gap size with the emission wavelength, over 80 % of the photons generated inside a PhC nanocavity could be funneled into a small divergence angle of ±30 • .In the present study, a PhC nanocavity with a bottom reflector is incorporated in a soft lithographically featured microfluidic device.In the proposed system, a fluid is forced to flow in the vicinity of the PhC cavity, thereby modulating the effective optical thickness of the gap, and hence allowing dynamic control of the far-field pattern.[10] The large ranges of index modulation that are made possible using fluid flow enable real-time tunability and reconfigurability.
Room-temperature continuous-wave (RT-CW) operation is highly favorable in the application of low-threshold nanolasers.Recently, Nomura et al. [11] and Nozaki et al. [4] demonstrated RT-CW lasing operation from a 3-L cavity [12] and a point-shift nanocavity, respectively, with both systems exhibiting greatly reduced laser threshold and pump power.Here we show that continuous flow of a fluid (water in the present experiments) in the vicinity of the PhC cavity improves the thermal diffusion, enabling RT-CW operation to be achieved.The wavelength tunability induced by the refractive index variation of the fluid is also of practical importance for achieving perfect spectral overlap between the single quantum dot and the cavity mode.The same principle can also be applied to high-fidelity refractive index sensors [13] that can detect a specific chemical and biological species.

Microfluidic integration of PhC nanocavities
Two-dimensional (2-D) PhC slab [14] structures have been widely investigated as platforms for wavelength-scale optical cavities.Photonic engineers can generate various resonant modes to meet specific needs by intentionally introducing a structural defect into an otherwise perfectly periodic PhC.[15] The nanocavity design employed in the present study is the deformed hexapole mode [16] cavity shown in Fig. 1(c).After removing one hole at the center, the six nearest neighbor holes are modified (modified radii are denoted by mr).[17] Then, the radii of two air holes facing each other are enlarged by p.As reported in our previous publication, [9] this secondary modification enables well directed vertical beaming together with linearly polarized emission, in which the direction of the polarization is determined by the positioning of the two air holes.
To increase the thermal conductivity, we used a high refractive index InP (n InP = 3.167) membrane that was wafer-bonded onto a SiO 2 (n silica = 1.445) substrate.[18] According to a previous report, the maximum duty cycle that can be obtained from this wafer structure is ∼ 20 % at RT. [19] Another advantage of using a silica bonded wafer is that optical pumping through the relatively thin glass side (∼ 500 μm) is possible due to the very wide transmission window of silica, ranging from visible to near-infrared.Therefore, a very thick PDMS (∼ 1 cm) microfluidic channel can be easily attached to the InP side, as shown in Fig. 1(a).
A schematic diagram of the proposed structure is shown in Fig. 1(b).Optical pumping and collection of the PhC nanolaser are performed by an optical microscope objective lens positioned on top of the glass substrate.The fluid flowing under the PhC membrane is used as a coolant [20] or to change a nearby effective refractive index.As depicted in Fig. 1(d), if a sufficiently thick gold layer were deposited inside the microfluidic channel, highly diverging far-field radiation could be replaced by quite good directional beaming owing to the far-field interference effect.[9] 2-D PhC patterns are defined by electron-beam lithography followed by dry-etching processes such as Ar ion milling and chemically-assisted ion-beam etching (CAIBE).[21] The lattice constant (a) is chosen to vary in the range of 480-530 nm to match the photoluminescence peak (around 1.55 μm) of 4 InAsP quantum wells, which are embedded in the middle of the slab.A detailed description of the epitaxial structure can be found in our previous publication.[18] The microfluidic channel can be brought into the vicinity of the PhC nanocavity by using conventional soft-lithography technology.There is no severe constraint on the width of the fluidic channel; a channel width greater than that of the whole PhC nanocavity (typically, ∼ 10 μm) is sufficient.However, the height of the channel should be on the order of the emission wavelength (a few μm) in order to utilize the aforementioned interference effect to control the far-field emission pattern.

Effect of background refractive index
In the proposed system, the fluid flowing close to the PhC nanocavity causes the background refractive index to increase.Here, we investigate the effect of the fluid in relation to this refractive index change.Figure 2(c) shows the variations in the cavity Q factor and resonant wavelength (λ ) as a function of refractive index, as determined from simulations using the finite-difference time-domain (FDTD) method.[22] For simplicity, the fluid is assumed to completely fill both sides of the slab, as depicted in Fig. 2(b).We choose two representative modes, the dipole mode and the hexapole mode, whose electric-field intensity distributions are displayed in Fig. 2(a).Generally, the total radiated power (1/Q tot ) is decomposed into a vertical contribution (1/Q vert ) and an in-plane contribution (1/Q horz ).However, because the in-plane loss (1/Q horz ) can be arbitrarily reduced by increasing the number of PhC layers surrounding the cavity, the vertical contribution 1/Q vert is typically referred to as the inherent optical loss of the resonant mode.Note that the Q factor data presented in Fig. 2(c) correspond to Q tot calculated using a sufficiently large horizontal computational domain (16a × 16a).
Firstly, the Q factor decreases drastically as the background refractive index (n bg ) increases.This is because the size of the light-cone [k 2 x + k 2 y = (n bg ω/c) 2 ] of the background material is effectively enlarged by a factor of n bg , and an expanded light-cone allows more plane waves to be coupled into the background-propagating modes.[5] Secondly, the resonant wavelength increases with n bg due to the effective increase of the cavity length.[13] When the change of n bg is moderate, the observed behavior is quite well described by the linear relation λ = κ n bg , where κ is a proportionality factor.For the dipole mode (the hexapole mode), κ is estimated to be ∼ 152 (247) nm/RIU (refractive index units).The larger κ of the hexapole mode can be explained by its electric-field intensity distribution.As shown in Fig. 2(a), the electric-field maxima lie inside the six nearest air-holes, [16] which increases the optical overlap with the Fig. 3. Calculated far-field radiation patterns for the two representative modes (Fig. 2), the hexapole mode and the dipole mode, in which the photonic crystal nanocavity is assumed to be immersed in a liquid of refractive index n bg .All the far-field data (x, y) in this article are represented using a simple mapping defined by x = sin θ cos φ and y = sin θ sin φ (the radius of the plot corresponds to θ ).Here, it is assumed that the far-field patterns are measured inside the liquid.
background material and henceforth the larger wavelength tuning.Note that Lonćar et al., using an optimized dipole cavity design with a central air-hole, obtained a κ value of 266.[13] Therefore, by utilizing this wavelength tuning property, one may achieve: 1) a high sensitivity biochemical sensor based on the refractive index change, as originally suggested by Lonćar et al. [13,23] and 2) a fluidically tunable photonic crystal nanolaser.For both applications, achieving a high Q factor is highly desirable for stable laser operation.Once the device is lased, an extremely narrow emission linewidth (∼ 0.1 nm) compared to other competing strategies [24,25] will enable a much higher refractive index sensitivity ( n ∼ 0.001).[13] However, another critical issue remains to be addressed, namely the strong diffractive far-field emission from wavelength-scale small nanocavities.This topic will be discussed in the following section.

Unidirectional beaming from the hexapole mode
To calculate far-field radiation patterns, we have used the FDTD method and the near-to farfield transformation algorithm.[9] Figure 3 shows the calculated far-field radiation patterns for the two representative modes discussed in the previous section, the hexapole and dipole modes.Again, the PhC slab is assumed to be completely surrounded by the background material.First, let us examine the results for n bg = 1.0 (air).Both resonant modes seem to have favorable directional patterns in spite of the wavelength-scale cavity size.This finding emphasizes the role of the secondary perturbation in controlling the directionality of the hexapole mode.Specifically, the original hexapole mode with perfect six-fold symmetry cannot show such directional emission; [16,9] however, when this perfect symmetry is broken by slightly increasing the size of the two air holes, both linear polarization and directionality are obtained.
As n bg increases, however, undesired side lobes appear along the horizontal direction.The situation for the dipole mode is more serious.Even when n bg reaches 1.3 (a value similar to the refractive index of water), the directionality of the dipole mode is totally degraded.Remember that only planewave components inside a light-cone [k 2 x + k 2 y ≤ (n bg ω/c) 2 ] can couple with propagating modes.Thus, the expanded light-cone size in the fluid turns the originally evanescent, non-propagating components into propagating components, which eventually manifest along the horizontal direction.
Unidirectional emission can be achieved by depositing a gold layer at the bottom of the PDMS channel, as described in Fig. 1(d) and Fig. 4(a).A detailed discussion of the physical origin of the directional emission can be found in our previous publication.[9] Here, we explain the key phenomena in a qualitative manner.The role of the gold mirror can be easily understood in terms of far-field interference.The downward-emitted photons from the PhC nanocavity are redirected in an upward direction by the reflector, and eventually form an interference pattern with the originally upward-emitted photons.If multiple reflections between the PhC slab and the gold mirror are neglected (this is valid when the slab thickness (φ ) satisfies the so called 'slab resonance condition', φ = π), then simple two-beam interference is obtained; when the gap size (d) satisfies 2nd = mλ , where m is an integer, we obtain the constructive vertical interference.Whereas, at an angle sufficiently distant from the vertical point (θ = 0), destructive interference occurs, which can suppress the unwanted side lobes.In general, however, the PhC slab thickness does not satisfy the slab resonance condition and the residual reflections cannot be neglected.Even in such general cases, we previously showed that almost the same condition as 2nd ∼ mλ holds for good vertical beaming.[9] Figure 4(b) shows the calculated far-field radiation patterns for the deformed hexapole mode as a function of the gap size d.In each pattern, the effective gap size, n(water) × d, is represented in wavelength units.To model the gold, the auxiliary differential equations technique was employed in the FDTD method, in which a single-pole Drude model was adopted.[22] In contrast to the far-field radiation patterns shown in Fig. 3, fairly good unidirectional emission is achieved with an effective gap size of 1λ .Unwanted side lobes are now suppressed by destructive interference, while vertical radiation is enhanced by constructive interference.In terms of the real distance, the gap size should be 1.17 μm (=λ /n = 1.557 μm / 1.33) with the pre-assumed structural parameters.A channel height of 1.17 μm is large enough for actual fabrication and the fluid flow is still governed by the microhydrodynamics.[26] In some cases, however, a slightly larger gap size may be preferable.Here, we present results for two additional cases, namely gaps of 2λ and 3λ (see Fig. 4(c)).The results show that the directional beaming condition is preserved for these multiple-wavelength gap sizes; for 2λ (3λ ) gap case, ∼ 36 % (∼ 31 %) of the total emission power can be collected within the numerical aperture of 0.4 (θ ≤ 23.6 • ) in the glass.The present findings thus establish that use of a bottom reflector is essential to achieving high-efficiency wavelength-small PhC nanolasers.
Finally, we investigated the variation in the Q factor as a function of the effective gap size.In Fig. 4(d), one can see modulation of the Q factor with a period of approximately λ /2, which manifests the above-mentioned far-field interference effect.Given that the Q factor represents the lifetime of the resonant mode (τ = Q/ω), the present results indicate that by varying the gap size, inherent radiation properties of the PhC resonant mode such as the radiation lifetime (τ) and the radiation pattern can be modified.This behavior is analogous to that observed for a dipole antenna in the vicinity of an ideal plane mirror.[27] This similarity may provide an interesting perspective for PhC nanocavities, whereby the resonant mode is considered as an artificial emitter whose emission characteristics are equivalently described by its electric multipoles and magnetic multipoles.[9,28] It is interesting to find that the Q factor can be more enhanced compared to the Q factor in the absence of the gold mirror.However, the maximum Q factor (∼ 1900) is still unexpectedly small compared to the result in Fig. 2(c); a Q factor of ∼ 2800 is expected when n bg is 1.33.The small value of the Q factor can be attributed to the broken vertical symmetry with respect to the PhC slab, [14] which causes TE-TM coupling loss through the horizontal direction of the PhC slab.[29] The use of a slightly thicker PhC slab or a slightly higher refractive index slab will diminish this effect, leading to a higher Q factor.For example, when the thickness and the refractive index of the PhC slab are assumed to be 250 nm and 3.4 (this corresponds to the case of replacing InP with InGaAsP), respectively, the Q factor can be as high as 3700 in the absence of the reflector and it can be in excess of by optimizing the gap size.

Fabrication
As summarized in Fig 1(a), the overall fabrication process consists of two main steps: 1) formation of PhC nanocavities in the InP slab and 2) bonding of a PDMS microfluidic channel onto the PhC pattern.First, we spun PMMA of molecular weight 950K (2%, dissolved in chlorobenzene) at 2500 rpm onto the InP surface.The thickness of the resulting PMMA layer was ∼ 150 nm.The PMMA-coated substrate was then baked in an oven at 160 • C for 3 hours.Then, we performed electron-beam lithography with a modified scanning electron microscope (SEM) to define the PhC nanocavity pattern.The e-beam exposed parts of the PMMA were selectively removed by a chemical mixture of methanol:ethoxy-ethanol = 7:3.The resulting perforated PMMA layer was used as a mask in the following dry-etching process.PhC hole patterns were transferred by Ar ion milling and CAIBE (Ar/Cl 2 ), with the latter process proceeding until the underlying silica was completely revealed.The remaining PMMA on the InP slab was removed by O 2 plasma treatment, thus completing formation of the PhC nanocavities.An SEM image taken after this first fabrication step is shown in Fig. 5(d).
A typical lateral size of the InP/silica wafer used in the present work is about 5 mm × 5 mm.We used two layers of a PDMS mold in order to separate the fluid inlet/outlet from the PhC patterns.Master patterns for the two PDMS microfluidic channels were fabricated using conventional photolithography.A negative photoresist (SU-8 2, MircoChem) was spun onto a 4 inch-diameter silicon wafer.The resulting thickness of the patterned photoresist is controlled to be ∼2 μm.The width of the microfluidic channel is designed to be 100 μm, which is sufficiently wide to cover the PhC nanocavity pattern (typically, 10 μm × 10 μm).Using these master patterns, PDMS (PDMS 184-A and B, Dow Corning) layers were fabricated by the following conventional soft lithography process.PDMS 184-A (monomer) and 184-B were mixed with a ratio of 10:1.Then, the PDMS molds were cured at 70 • C for more than 3 hours.[25] As shown in the optical microscope image [Fig.5(c)], the first PDMS layer containing a winding microfluidic channel was directly bonded onto the InP slab.In this process, the entire microfluidic channel should fit into the 5 mm × 5 mm InP structure and the PhC patterns should be aligned into the 100 μm-wide channel [see inset of Fig. 5(c)].Finally, the second PMDS layer containing the fluid inlet/outlet was bonded onto the first PDMS layer.
Before the first PDMS layer was bonded to the other two materials (the InP surface and the second PDMS layer), it was subjected to oxygen plasma treatment on both sides.Then, the two treated sides were brought into conformal contact with the InP surface and the second PDMS layer and irreversible bonding occurred.It should be noted that this method, which has been widely used for PDMS-PDMS bonding, is also effective for PDMS-InP bonding.[30] The final fabricated microfluidic chip is shown in Fig. 5(a).

Experimental results
We performed photoluminescence (PL) measurements as water was flowed through the microfluidic channel.As shown in Fig. 5(b), a 20× infrared (IR) objective lens was used to focus a pump laser (λ pump = 980 nm) on one of the nanocavities with a spot diameter of ∼ 4μm.The same objective lens was used to collect the emitted laser light from the sample.Because of the presence of the silica substrate, the duty cycle could be increased up to 20 % even when the channel was filled with air.However, when water was flowed through the channel at a typical flow rate of 5 μl/h, CW lasing operation could be achieved.Continuous flowing of water may prevent the nanolaser structure from overheating.Unless otherwise stated, the following PL data were measured at RT under pulsed pumping at a repetition rate of 1 MHz with a 10 % duty cycle.
In Fig. 6(a), we compare two PL spectra, one with air and the other with flowing water in the channel.The strongest resonant peak shows a +19 nm wavelength shift when the PhC nanocav- ity is immersed in water.As will be discussed below, by using the contour FDTD simulation, we confirmed that the strongest resonant peak originates from the deformed hexapole mode.The two minor peaks observed at around 1540 nm (in air) are the two degeneracy-split quadrupole modes.[31] Figure 6(b) summarizes the measured wavelength tuning characteristics for the three different PhC nanocavity modes mentioned above.C1 and C2 denote PhC nanocavities with different air-hole sizes.We found that each distinctive resonant mode has its own characteristic peak wavelength shift: the two quadrupole modes are red-shifted by ∼15 nm, while the hexapole mode shows a peak wavelength shift of +19 nm.This is because each mode has a distinct modal distribution, and the peak shift is mainly determined by the degree of modal overlap with the background material.However, the observed peak shift is smaller than expected.In Fig. 6(c), the C1 nanocavity data are compared with the corresponding contour FDTD simulation results.Here, the numerical structural input data directly obtained from the SEM images were used in order to compensate for any fabrication imperfections.[1] We found that, if the microfluidic channel was filled with air, the errors in the absolute values associated with the peak wavelength were within 1 %, which corresponds to ∼ 15 nm in wavelength units.If we assume that water completely backfills the air-holes of the PhC structure, as depicted in Fig. 6(d), then the estimated wavelength shift should be over 40 nm for the hexapole mode.As mentioned before, the peak shift can be considered to be an inherent characteristic of each resonant mode.However, the simulation result (∼ 40 nm) is obviously far from the measured value of ∼ 20 nm.We attribute this discrepancy to the possibility that the water may not have completely filled the air-holes, leaving air-pockets, as depicted in Fig. 6(d).This hypothesis is reasonable since the InP surface is rather hydrophobic.To confirm this scenario, we conducted contour FDTD simulations including air pockets.These simulations predicted a 20 nm peak shift, which exactly coincides with the experimental data.
From the perspective of wavelength tuning, the low wettability of the PhC slab is undesirable for sensor applications.This problem can be solved by introducing a specific surfactant such as cetyltrimethylammonium (CTAB) into the microfluidic channel.[26] However, the hydrophobic InP surface may have its own applications.Firstly, the PhC slab structure that leaves air-holes unfilled with fluid may enable a higher Q factor and stronger optical confinement in the horizontal direction, due to the larger refractive index difference.Secondly, in some applications where precise wavelength tuning is crucial (e.g., spectral overlap between the cavity mode and a single quantum dot), [3] moderate refractive index sensitivity would be preferred.Finally, the presence of the air pockets may make it possible to realize a unique microfluidic delivery system in solid semiconductor materials, which can be formed by wet-chemical-etching of an underlying sacrificial layer through PhC air-holes.Water may flow inside the undercut region without leakage through the PhC air-holes.Such a micro-plumbing system could be adopted in electrically-driven PhC nanolasers.[1] In Fig. 7, we demonstrate, for the first time, RT-CW lasing operation of the PhC single cell cavity, which is enabled by a more than 100-fold increase in thermal conductance compared to that for the air suspended membrane structure.[4,32] (We have measured the power dependence of the peak wavelength to be ∼ 0.1 nm/mW in the above threshold.)A detailed discussion of the water-cooled RT-CW operation will be given in a forthcoming publication.Here, we present only some of the key results.As confirmed by the contour FDTD simulation, the laser mode originates from the deformed hexapole mode [see inset of Fig. 7 (a)].The hexapole mode shows a redshift of ∼ 23 nm in water, which is similar to the above results.Again, we believe that the air-holes are not filled with water.In Fig. 7(b), the laser threshold (P th ) of ∼ 1.7 mW is slightly higher than those of previous reports on RT pulsed lasing operation [17,19,21]; this discrepancy can be attributed, in part, to the lower Q factor of our system.Note that the pump power was measured in front of the sample; hence, P th represents the real irradiated pump power.
RT-CW operation is of great interest in contexts where stability of the laser wavelength is required, such as high-sensitivity biochemical sensors [13] and efforts to more reliably estimate spontaneous emission behavior below P th , [4].Furthermore, our CW nanolaser does not show wavelength chirping phenomena caused by thermal effects.RT-CW operation is also well suited to achieving practical high-speed modulation of the PhC nanolaser.[33]

Conclusion
Highly divergent far-field emission and poor thermal characteristics have been daunting problems limiting the utility of PhC nanocavities in thin dielectric membranes.Here, we have shown that both of these weaknesses can be simultaneously solved by using simple microfluidics technology.We have demonstrated that, by flowing water near the laser structure, RT-CW laser operation can be achieved.Furthermore, through FDTD simulations, we have shown that fairly good unidirectional beaming can be obtained by introducing a gold reflector at the bottom of the microfluidic channel.The ability to tune the wavelength by varying the refractive index of the fluid is of critical importance to applications such as 1) sensitive biochemical detection of femto-liter small volume analytes and 2) precise tuning of the resonant wavelength for perfect spectral overlap with a single quantum dot.In addition, the proposed structure is compatible with PDMS based 'lab-on-a-chip' systems.We believe that such reconfigurability and the high heat capacity of the fluid will enable application of the proposed system in the fields of photonics, optoelectronics, and quantum optics.

Fig. 1 .
Fig. 1.(a) Simple description of the fabrication process.(b) Schematic of the proposed photonic crystal nanolaser integrated with a microfluidic channel, where optical pumping and collection can be performed through the thin glass side with the assistance of a long working distance infra-red objective lens.(c) The deformed hexapole mode cavity.Here, two air-holes facing each other are enlarged by p to generate well-directed vertical emission.(d) ∼100 nm thick gold film deposited at the bottom of the microfluidic channel can be used to enhance the directivity of the laser emission.

Fig. 2 .
Fig. 2. (a) FDTD simulation of near-field (|E| 2 ) distributions of the dipole mode and the hexapole mode.The thickness of the InP slab and the lattice constant (a) were chosen to be 200 nm and 500 nm, respectively.Other structural parameters [See Fig. 1(c)] are as follows: r = 0.35a, mr = 0.25a, and p = 0.05a.(b) A photonic crystal slab nanocavity completely immersed in a liquid with refractive index n bg .(c) FDTD simulation of cavity quality factors and resonant wavelengths of the dipole mode and the hexapole mode, where the background refractive index was varied from 1.0 to 1.4.

Fig. 4 .
Fig. 4. FDTD simulation of far-field emission from a photonic crystal nanocavity with a bottom reflector.Here, it is assumed that the far-field patterns are measured inside the glass.The InP slab thickness and the lattice constant of the photonic crystal (a) are assumed to be 200 nm and 530 nm, respectively.Other structural parameters [see Fig. 1(c)] are as follows: r = 0.35a, mr = 0.25a, and p = 0.05a (a) The microfluidic channel is filled with water, and the height of the channel (d) is adjustable.(b) Calculated far-field emission patterns from the modified hexapole mode.In each pattern, the effective gap size, n(water) × d, is normalized by the wavelength (λ ).A fraction of angle integrated power within the numerical aperture (NA) of 0.4 in the glass, η = [ θ ≤(NA=0.4)(dP/dΩ)dΩ]/[ θ ≤90 • (dP/dΩ)dΩ], is also presented.(c) Far-field emission patterns from structures with gaps equal to multiples of the wavelength, such as 2λ and 3λ .(d) Cavity Q factor as a function of the effective gap size, n(water) × d.A horizontal dotted line represents the Q factor in the absence of the gold mirror.

Fig. 5 .
Fig. 5. (a) A sample after completion of the fabrication process.(b) A photo taken during the photoluminescence measurement.A long-working distance objective lens is positioned on the glass substrate side, facilitating optical pumping and collection of the emitted laser light.(c) Optical microscope image showing that the photonic crystal nanocavities are well aligned with a winding microfluidic channel of width ∼ 100 μm.(d) Scanning electron microscope image of the fabricated photonic crystal nanocavity.The measured hole-tohole distance (lattice constant) is ∼ 530 nm.

Fig. 6 .
Fig. 6.(a) Measured photoluminescence spectra for systems with air or flowing water inside the microfluidic channel.(b) Wavelength tuning characteristics of two PhC nanocavities (C1 and C2) with different air-hole sizes.(c) and (d) Comparison between the measurement and the FDTD simulation for the C1 cavity.The observed wavelength shift agrees well with the simulation result obtained assuming that the water does not fill the air-holes of the photonic crystal slab.

Fig. 7 .
Fig. 7. Room-temperature continuous-wave lasing under the constant flow of water at a rate of rate of 5 μl/h.(a) Measured wavelength tuning characteristics as water flows inside the microfluidic channel.The peak wavelength shows a redshift of ∼ 23 nm.The inset shows the contour FDTD simulation of the laser mode (Deformed hexapole).(b) Light-in versus light-out (L − L) curve, in which the pump power (Light-in) was measured in front of the sample (on top of the glass substrate).