Printed Large-Area Single-Mode Photonic Crystal Bandedge Surface-Emitting Lasers on Silicon

We report here an optically pumped hybrid III-V/Si photoic crystal surface emitting laser (PCSEL), consisting of a heterogeneously integrated III-V InGaAsP quantum well heterostructure gain medium, printed on a patterned defect-free Si photonic crystal (PC) bandedge cavity. Single mode lasing was achieved for a large area laser, with a side-mode suppression ratio of 28 dB, for lasing operation temperature ~200 K. Two types of lasers were demonstrated operating at different temperatures. Detailed modal analysis reveals the lasing mode matches with the estimated lasing gain threshold conditions. Our demonstration promises a hybrid laser sources on Si towards three-dimensional (3D) integrated Si photonics for on-chip wavelength-division multiplex (3D WDM) systems for a wide range of volume photonic/electronic applications in computing, communication, sensing, imaging, etc.

very narrow linewidth and high side-mode suppression ratio (SMSR). It is worth noting that currently the lasing performance is largely limited by the transfer printing process. With further optimization of the transfer printing process, and precise control of the interface quality, higher performance lasers are feasible.

Cavity design and fabrication
As shown schematically in Fig. 1(a), the lasing cavity consists of a transferred III-V InGaAsP multi-quantum well (MQW) heterostructure active region, printed on a single-layer Si-PC cavity. Also shown in Fig. 1(a) is the simulated optical magnetic H-field distribution profile for the designed lasing cavity mode. In such cavities, patterned PC structure was designed to be placed very close to the unpatterned QW structure to provide strong in-plane distributed feedback. The lightwaves propagating in various 2D directions are coupled with one another, and a 2D standing wave or 2D cavity mode is constructed over a broad area 21 . The in-plane propagating waves are also diffracted toward the vertical direction by the PC structure itself due to the first-order Bragg diffraction 19 .
Si-PC cavity was first designed based on a silicon-on-insulator (SOI) substrate, with a top Si layer thickness (t 2 ) of 230 nm and a buffered oxide (BOX) layer thickness (t 3 ) of 400 nm. Total thicknesses of the top InGaAsP MQW heterostructure (t 1 ) and the bottom Si-PC cavity (t 2 ) vary for two device designs operating at different temperatures, as shown in the structure table in the inset of Fig. 1(a), denoted by PCSEL-I and PCSEL-II, respectively. The lattice constant for the square lattice air hole Si-PC lattice is chosen to be a = 480 nm. To match InGaAsP MQW emission peak locations at different operation temperatures, the air hole radii are r = 0.2a and r = 0.15a for PCSEL-I and PCSEL-II, respectively.
The whole InGaAsP MQW structure was grown on (001) n-InP substrate by metal-organic chemical vapor deposition (MOCVD), consisting of top and bottom InGaAs contact layer, top and bottom InP cladding layers, and eight pairs of strain-compensated In 0.76 Ga 0.24 As 0.83 P 0.17 /In 0.485 Ga 0.515 As 0.83 P 0.17 with the center emission wavelength of 1,527 nm at room temperature 27 . A InGaAs/InP sacrificial bilayer was grown first on InP substrate for selective wet-etching and substrate removal. Total thickness (t 1 ) is 427 nm for device PCSEL-I and 297 nm for device PCSEL-II. For the PCSEL-II device, the bottom InP cladding layer were selectively etched away for stronger evanescent coupling between the MQW and the Si PC cavity. Accordingly, for PCSEL-I, the Si-PC layer was first thinned down from 230 nm to 120 nm based on thermal oxidation and selective wet etching processes.
The 2D 0.5 × 0.5 mm 2 square lattice PC pattern was fabricated based on the standard e-beam lithography (EBL) patterning technique and plasma dry-etching processes. Substrate removal process is used here for the release of top large area InGaAsP MQW structure from the native InP substrate with black wax as the protection layer on top. InP substrate was removed by immersing the sample in 20%HCl + 80%H 3 PO 4 for about two hours. The active region was then lifted off as a membrane by selective wet etching to release the bottom InGaAs/InP sacrificial bi-layers and the InGaAs n-contact layer. For PCSEL-II, the bottom InP n-cladding layer was also selectively etched away. The released InGaAsP MQW membrane is finally printed onto the patterned Si PC cavity. Wax was removed with trichloroethylene (TCE) and IPA solutions. At the end, a 150 × 150 μ m 2 active MQW square area is defined by

Device characterizations
Devices are mounted inside a cryostat and characterized with a monochrometer-based micro-photoluminescence (μ -PL) set up (see Methods section for more details). We report here two groups of PCSEL on Si devices (PCSEL-I and -II) operating at T = 25 K and T = 180 K, respectively. The L-L plot (light output for different input light pump powers) and the corresponding spectral linewidths are shown in Fig. 2(a) for PCSEL-I operating at T = 25 K. The threshold pump power is ~6 mW, or 0.1 kW/cm 2 . The measured spectral linewidths are reduced from ~12 nm (below threshold) to 0.7 nm (above threshold). Shown in Fig. 2(b) is the measured lasing spectrum above threshold. Single mode lasing was observed at around 1424 nm with a narrow linewidth of 0.7 nm, which is essentially limited by the measurement set-up. The side mode suppression ratio (SMSR) reaches ~22 dB for the pump power P = 40 mW. Also shown in Fig. 2b inset is a measured far-field image with collimated circular single-mode output. Additionally the lasing output shows one dominant polarization, as shown in Fig. 2(c).
The measured results for PCSEL-II designed for the T = 180 K operation are presented in Fig. 2(d-f). A pumping threshold of 25 mW was observed with the linewidths reduced from ~10 nm to 0.6 nm. The measured single mode lasing spectrum and far-filed image are displayed in Fig. 2(e) where one can see a SMSR of 28 dB at a pumping power of 50 mW. Correspondingly, the output is also polarized.
Temperature-dependent characteristics are also investigated for both devices. Shown in Fig. 3(a,b) are the measured L-L plots at different operation temperatures for PCSEL-I and PCSEL-II devices. For PCSEL-I, the design was optimized for device operation at T = 25 K. Single mode lasing operation was obtained up to T = 140 K. However, threshold increases with the increase of the operation temperature. For PCSEL-II, on the other hand, the design was optimized for device operation around T ~ 180 K. Lasing thresholds increase with the decrease or increase of the temperature, due to the detuning of the cavity resonance and MQW emission peak. The temperature-dependent lasing characteristics are summarized in Fig. 3(c,d). For PCSEL-I, with the increase of temperature (T) from 25 K to 140 K, the lasing peak red-shifted from 1421.8 nm to 1426.55 nm, with lasing linewidth increased slightly from 0.6 nm to 0.95 nm. The lasing peak shifts at a rate of ~0.041 nm/K. The speed of cavity mode shift is much lower than the QW PL peak shift rate at 0.43 nm/K, as reported earlier 26 . It is also lower than that of the conventional VCSEL (~0.1 nm/K) 28 . The SMSR was also reduced with the increase of the temperature, with the lasing threshold increased from 6 mW to 25 mW,. For PCSEL-II, with the change in temperature (T) from 25 K to 220 K, the lasing peak red-shifted from 1496.5 nm to 1507 nm, with lasing linewidth changing slightly around 0.6 nm. The lasing peak shifts at a rate of ~0.054 nm/K. Both SMSR and threshold degrade with the change in temperature away from the optimal operation temperature of 160-180 K. It is worth noting that an increase in the coupling between QW and Si-PC cavity by reducing the separation between the QW layer and Si-PC layer results in a reduced threshold and better lasing performance. With further optimization in the coupling design, it is feasible to achieve lasing at room temperature, with much reduced threshold and much higher output.

Discussions
To better understand the underlying physics of these two PCSEL devices, the experiment results are also correlated to theoretical and numerical analysis. First, the photonic bands are calculated from the Fano/guided resonances in the simulated reflection spectra. In Fig. 4(a,b), the eight bands are displayed in a small in-plane wave vector k // range from 0 to ~0.01 (2π /a) with blue-solid lines for the transverse electric (TE, s) and red-dashed lines for the transverse magnetic (TM, p) polarizations, for PCSEL-I and -II respectively. One can see that all of these bands are very flat at the edges close to Γ point, which indicates very small group velocity and longer light-matter interaction time for these bandedge modes. Figure 4(c,d) show the simulated reflection spectra (the incident angle θ = 0.2° is used here) and measured lasing spectra for PCSEL-I and -II, respectively. Broadband PL emission spectra are also shown as references at these two different operation temperatures. These measured lasing modes correlate well with the TM guided modes (modes A and C′ ) in the reflection spectra for these two devices, respectively.
In order to determine why the lasing occurs at TM mode A and C′ for PCSEL-I and -II, respectively, the gain threshold (g th = α/f c ) is investigated according to the field confinement factor (f c ) at MQW and the loss α (or quality factor, Q). Since the device is large enough with the size around 150 μ m × 150 μ m, which is more than 300a × 300a, it can be assumed that the edge scatting is not the dominating loss. Here only the radiation loss is considered for the infinite cavity with the relationship of α ≈ 2π/aQ 29 . The gain threshold g th values of the four band edge modes A, B, C, and D in PCSEL-I, A′ , B′ , C′ and D′ in PCSEL-II are estimated to be 0.92 cm −1 , 190.4 cm −1 , 6.96 cm −1 , 444.9 cm −1 , 1.61 cm −1 , 390.4 cm −1 , 3.79 cm −1 , and 536.2 cm −1 , respectively. Further information on the calculation can be found in the Supplementary Information (SI) Section. As we can see, the modes A and A′ have the lowest g th in PCSEL-I and -II, respectively, which are expected to be the lasing modes. In experiments, the mode A in PCSEL-I lases as expected, while in PCSEL-II, the lasing is mode C′ , instead of mode A′ . The reason can be attributed to the fabrication imperfection, such as III-V/Si-PC interface defect or unflatten at QW membrane edge. Since the modes A′ and C′ have similar g th , we observed lasing from mode C′ .
In addition, from the band structure, one can see that the mode separation is very large, due to the high refractive index contrast above and below the cavity, as compared to the thicker cavity (a few micrometers) 21 , which is highly desirable in order to achieve a single wavelength operation. This larger mode separation also indicates the stronger optical coupling in this kind of PCSEL cavities. And the coupling is further improved by putting a thicker Si-PC closer to the active region, as designed in PCSEL-II, due to higher optical confinement factor within Si-PC layer (see SI section). Comparing with PCSEL-I, the enlarged mode separation in PCSEL-II provides a stronger optical feedback, which results in a lower pump threshold, a narrower linewidth, and a higher SMSR as observed in experiments.

Summary.
In conclusion, employing membrane transfer and printing technique, hybrid Si-PC bandedge surface emitting lasers are demonstrated with optical pumping. With the careful cavity design and the control of the transfer process, single mode lasing was achieved with excellent linewidth ~6 Å and side mode suppression ratio of 28 dB. According to the numerical analysis, the experimental results agree well with the theoretical expectations. These results indicate that the combination of the transfer printing process and the photonic crystal bandedge cavity offers a simple and flexible method to realize single mode high power lasers for large-scale silicon-based photonic integration. Further optimization of the cavity design and improvement of the fabrication quality can lead to the realization of electrically pumped lasers operating under room temperature with potential 2D multi-wavelength arrays for on-chip multi-layer WDM systems with the desired spectral, spatial, and energy-efficiency.

Methods
Device design is based on the PCSEL principle by utilizing the bandedge modes in the vicinity of the second order Γ points 30 . These modes correspond to the Fano/guided resonance modes (peaks or dips) in the reflection or transmission spectra of the cold cavity. The cavity structure parameters are optimized by tuning these Fano resonance modes to match with the MQW emission peak. All of the simulations used the Fourier Modal Method with Stanford Stratified Structure Solver (S 4 ) software package 31 . In the simulation, a periodic boundary condition is used, corresponding to the cavity with in-plane infinite size.
Device characterization are carried out by mounting the devices inside a cryostat and characterized with a monochrometer-based micro-photoluminescence (μ -PL) set up. See the details in the third section of Supplementary Information (SI).