We report on single rolled-up microtubes integrated with silicon-on-insulator waveguides. Microtubes with diameters of ~7 μm, wall thicknesses of ~250 nm, and lengths greater than 100 μm are fabricated by selectively releasing a coherently strained InGaAs/GaAs quantum dot layer from the handling GaAs substrate. The microtubes are then transferred from their host substrate to silicon-on-insulator waveguides by an optical fiber abrupt taper. The Q-factor of the waveguide coupled microtube is measured to be 1.5×105, the highest recorded for a semiconductor microtube cavity to date. The insertion loss and extinction ratio of the microtube are 1 dB and 34 dB respectively. By pumping the microtube with a 635 nm laser, the resonance wavelength is shifted by 0.7 nm. The integration of InGaAs/GaAs microtubes with silicon-on-insulator waveguides provides a simple, low loss, high extinction passive filter solution in the C+L band communication regime.
© 2011 OSA
The integration of III-V micro-cavities with silicon-on-insulator (SOI) waveguides is in high demand for next generation chip-level optical interconnects and silicon photonics in general. Traditional approaches based on the flip-chip integration  scheme are slow and expensive and restricted to relatively large Fabry-Perot micro-cavities. Other micro-resonators with much better optical properties, such as those based on photonic crystals, microdisks, and micropillars, are hard to be removed from their host substrates without damage due to the presence of large surface tension.
Recently, rolled-up InGaAs/GaAs microtubes have emerged as a promising alternative for realizing future high performance micro-resonator-based devices [2,3]. Strong coherent emission [2,3] and room temperature lasing  has been recently reported in these microtubes, demonstrating their excellent optical properties. These unique devices can be controllably released from the handling GaAs substrate by selectively etching an underlying sacrificial layer . It is also possible to do controlled transfer and exact positioning of a single InGaAs/GaAs microtube on a foreign substrate by utilizing transfer-printing  or an optical fiber abrupt taper . The remarkable optical properties together with the controllable transfer process make semiconductor microtubes ideally suited to providing an integrated III-V light source on the SOI photonics platform. However, the integration of InGaAs/GaAs microtubes with SOI waveguides has not been realized so far.
In this paper, we report on the integration of InGaAs/GaAs microtubes with SOI waveguides. The InGaAs/GaAs quantum dot microtubes are picked up using optical fiber abrupt tapers from the handling GaAs substrates. They are subsequently transferred, in a precisely controlled fashion, directly onto the SOI waveguide. Using the abrupt taper to eliminate the effects of surface tension and vibrations, we can measure the Q-factor of the microcavities in a transmission configuration as opposed to using photoluminescence. This lets us find the intrinsic cavity Q-factor without any extraneous broadening . The highest Q-factor we measured is 1.5×105, the highest recorded for a semiconductor microtube cavity up to date. Detailed studies also confirm that the resulting microtube optical cavities are relatively free of structural defects and exhibit strong mode confinement, thereby promising integrated micro-cavities with greatly simplified packaging.
2. Microtube fabrication, transfer and SOI waveguide fabrication
The microtubes were fabricated from a GaAs/InGaAs bilayer grown on a GaAs substrate, with a 50 nm AlAs sacrificial layer . The bottom In0.18Ga0.82As layer is 20 nm thick, compressively strained and capped with a 30 nm thick GaAs layer. There is one layer of InAs quantum dots in the GaAs layer. When the sacrificial layer is etched away using a concentrated HCl solution, the bilayer rolls upon itself along the  crystal axis to release its stress . The rolling process and the sample structure are sketched in Fig. 1(a) . In order to optimize the optical quality of the microtubes, we use a U-shaped mesa structure for the fabrication (Fig. 1(b)). This results in a microtube with three parts: a thinner free-standing region which is isolated from the substrate and two thicker “legs” which remain in contact (Fig. 1(c)). The process results in microtubes with diameters of ~7 μm, wall thicknesses of ~250 nm, and lengths larger than 100 μm. Surface corrugations patterned in the free-standing region provide mode confinement along the microtube axial direction [2,3], which has been clearly demonstrated in our experiments. When the tube is deposited on top of a waveguide sample, the presence of the legs will make sure that there is a gap between the free-standing part and the waveguide itself. This gap could, in principle, be tuned by carefully designing the tube geometry.
Optical fiber abrupt tapers are used to transfer InGaAs/GaAs microtubes from their host substrate to the SOI substrate. The abrupt tapers are made by a splicer machine and have a tip diameter of less than 1 µm, and a tapered length of less than 1 mm . One abrupt taper is mounted on a micro positioning stage with a 100 nm resolution. The tip of the abrupt taper is inserted 10 to 200 µm (depending on the required experiment) into the microtube. Then the stage is moved up by 5 µm steps to fully release the InGaAs/GaAs microtube from its host substrate. Figure 2(a) shows an optical microscope image of a microtube held by an abrupt taper. The tip of the taper is inserted into the InGaAs/GaAs microtube by 10 µm, which is less than 10% of its total length. Surface tension forces strongly attach the microtube to the abrupt taper to the point that the microtube doesn’t fall down even if the tapered fiber hangs upside-down. A striking demonstration of this strong microscale attachment is shown in Fig. 2(b), where an InGaAs/GaAs microtube is transferred onto the surface of a cleaved single-mode SMF-28 fiber. The axis of the fiber is vertically oriented, while the contact length between it and the microtube is around 10 µm. Even though the contact length is no more than 10% of its total length, the microtube stays level, with most of its body hanging in the air. This strong attachment force can potentially simplify the packaging of integrated microtube/SOI waveguide devices. For single microtube transfer, in contrast to the flip-chip method, this fiber abrupt taper aided method provides advantages such as faster transfer speed, visual feedback during the transfer, and high accuracy. Also, several tapered fibers could be manipulated in parallel. Coupled with precise control of the distance between the rolled-up microtubes by careful design of the lithography patterns used for defining the microtube understructure, this would allow for the simultaneous transfer of multiple microtubes.
The waveguides used in our studies were fabricated by colleagues at the University of Sherbrooke on SOI with a 260 nm thick silicon device layer and a 2 μm buried oxide layer, using e-beam lithography . This thickness of the Si layer ensures that the waveguides exhibit a single transverse slab mode operation . Once the pattern is defined using an e-beam writer and developed, it is etched all the way to the buried oxide layer using reactive ion etching. Grating couplers are then fabricated for coupling light in- and out- of the waveguide with SMF-28 optical fibers. The grating couplers are second order gratings with 20 periods of 600 nm period each, 50% duty cycle, and an etch depth of 90 nm . The waveguides were adiabatically tapered to 10 μm at the edges to accommodate grating couplers. The total area of the grating coupler region was thus 12×10 μm2. Several waveguides of widths varying from 1 to 2 μm were fabricated. A scanning electron microscope (SEM) micrograph of the facet of a waveguide and the grating coupler is shown in Fig. 3 .
3. Microtube/SOI waveguide coupling modeling and experiment
We studied the coupling properties of the waveguide using 2D-FDTD to quantify the mode coupling between the microtube resonators and SOI waveguides. Figure 4(a) shows the calculated field profile of a 260 nm-high SOI waveguide next to a ridge representing a transverse cut of the microtube, with a 50-nm air gap between them. The confinement factor in the silicon region is 59.4% and in the microtube region is 27.2%, indicating a relatively high modal overlap with the waveguide mode, and a high evanescent field coupling from the microtube to the waveguide. The high overlap is evident on the vertical cut of the field intensity as shown in Fig. 4(b), and the field is symmetrical along the horizontal axis as shown in Fig. 4(c).
Figure 5(a) illustrates the experimental setup for the integrated microtube with the SOI waveguide. An abrupt taper was used to transfer the microtube from its host substrate and load it onto the SOI waveguide. Two cleaved SMF-28 fibers were used to couple light in and out of the grating couplers. A broadband source (1520-1600 nm) or a tunable laser (1500-1580 nm) was used as the optical source, and an optical spectrum analyzer or a power meter was used as detector, respectively. Figure 5(b) shows an SEM image of a microtube mounted on the waveguide. The diameter of the microtube is 7 µm and the length is larger than 100 µm. There are at least 5 layers in the leg part, resulting in a thickness larger than 250 nm. During the transfer process, the microtube was held by the abrupt taper through the lower end leg. After the transfer there was no damage to the leg part, as shown in the left inset of Fig. 5(b). The right inset of Fig. 5(b) illustrates the corrugations along the axial direction, which provide the optical confinement in this direction.
To measure the Q-factor of the microtube cavity, the abrupt taper is inserted inside of the microtube for precise alignment. A broadband source and an optical spectrum analyzer are used as the source and detector in Fig. 5(a). The abrupt taper is moved down by 100 nm steps until a small ripple in the normalized transmission spectrum, as shown in scan 1 of Fig. 6(a) , can be seen. Due to the strong attracting force between the microtube and waveguide, and the flexibility of the abrupt taper, the spectrum then evolves very quickly from scan 1 to scan 3 via scan 2 as the microtubes moves even closer to the waveguide. At this point the microtube is resting on top of the substrate and the spectrum remains as shown in scan 3 for a long period; the abrupt taper needs to be moved up by 10 µm to release the microtube from the SOI waveguide. The free spectral range (FSR) of the dominant mode in scan 3 is 33 nm at 1550-nm wavelength. With a diameter of the free-standing part of the microtube equal to 7 µm (estimated from the image shown on Fig. 5(b)) the effective refractive index of the mode is calculated to be 3.4. This is very close to the effective refractive index of the InGaAs/GaAs bilayer, demonstrating the ring-like nature of the modes in the free-standing part of the microtube  and good mode confinement in the wall of the microtube.
In order to eliminate the undesired motion of the microtube due to the attractive force, and to get closer to the critical coupling point, the abrupt taper is fully inserted through the microtube. The tip of the abrupt taper is put down on the SOI wafer, as shown in Fig. 5(a), and plays the role of a fulcrum, stabilizing the microtube position and making it more robust against mechanical perturbations. Then, the separation between the microtube and the waveguide can be precisely controlled until there is a ripple in the transmission spectrum, similar to scan 1 in Fig. 6(a) (which then is much more stable). To get better resolution, a tunable laser with a 1 pm step and an optical power meter were used as the source and detector in Fig. 5(a). The measured 3-dB linewidth is 0.01 nm, as shown in Fig. 6(b), resulting in a Q-factor of 1.5×105. This is the highest Q-factor reported of a microtube cavity and this is also the first measurement of the cold-cavity Q-factor of a microtube in a transmission configuration. The separation between the microtube and waveguide under critical coupling is estimated to be between 50 and 75 nm.
By moving the microtube along its axial direction, various spectra are observed, as shown in Fig. 7(a) . It is clear that more modes are excited when the waveguide is aligned with the center of the free-standing part than when aligned close to the leg part. This is most likely due to the excitation of higher order axial modes when coupling farther away from the corrugation center. These higher order modes are also broader due to the presence of extra leakage outside of the corrugation area. These patterns are consistent with those reported in . The insertion loss due to the microtube is around 1 dB, and extinction ratios up to 34 dB are obtained. This is due to the low absorption loss at C+L band wavelength and good confinement of the mode in three dimensions. The variation of the resonator performance with input intensity is obtained by varying the tunable laser power from −13 to −1 dBm. The corresponding transmission spectra are shown in Fig. 7(b) and demonstrate that there is not enough power coupled into the resonator at these pump levels to induce nonlinear effects. This is due to the large insertion loss of the vertical coupler section.
Since there are InAs quantum dots embedded in the wall, the microtube can act as an active micro-cavity device at shorter wavelengths where the quantum dot absorption is significant. For instance, it can start lasing when pumped by a 635 nm laser . In our experiment, we use a 1 mW 635 nm laser as the pump laser through the abrupt taper holding the microtube. Due to the scattering light in the abrupt taper itself, only a fraction of the emitted light from the pump laser is eventually delivered to pump the microtube. The transmission spectra of the microtube in the presence and absence of pumping are shown in Fig. 8(a) . The only significant effect is that the resonance wavelength shifts linearly with the pump laser power to a longer wavelength, with a maximum observed shift of 0.7nm, as shown in inset of Fig. 8(a). There are two main mechanisms by which the pumping light can alter the microtube response: thermal effects and carrier generation. The former increases the refractive index of the microtube , causing resonances to shift to longer wavelengths, while carrier generation normally decreases it (shifting resonances to shorter wavelength). Therefore, the red shift of the resonance wavelength induced by the pump light is due to the thermal effect . We used this effect to turn the microtube into an optical modulator. We set the tunable laser wavelength to one of the resonance dips (1544.8 nm in particular), and modulated the 635 nm pump laser at 1 Hz (the low frequency was due to limitations in the speed of our power meter). Due to the wavelength shift in the presence of the pump, the transmission of the 1544.8 nm laser is modulated with an extinction ratio of 10 dB, as shown in Fig. 8(b). Another electrical-optical modulation scheme is under development, and is expected to deliver a high speed modulation (10 G/s) response.
Utilizing the low absorption in the C+L band, this kind of microtube can be used as a low loss and high extinction ratio filter integrated with a planar optoelectronic circuit. Due to the multiple axial modes supported, the free-spectral range is smaller compared to the other types of single mode microcavities like microring and microtoroid resonators. Microcavities with small FSRs are of great interest in realizing filters with continuous tuning over a wide wavelength range . These integrated InGaAs/GaAs microtubes with SOI waveguides can also be used as differential phase shift keying demodulators by optimizing the mode confinement , as active optical intensity or phase modulators [15,16], or as signal quality monitors .
In this paper, we have demonstrated integration between a single rolled-up InGaAs/GaAs quantum dot microtube and silicon-on-insulator waveguides. The microtube is fabricated using standard processes and transferred from its substrate using an optical fiber abrupt taper. We have performed the first transmission measurements of a microtube coupled to a waveguide obtaining a Q-factor of 1.5×105. When the microtube is coupled to the waveguide, the transmission spectrum shows an insertion loss of 1 dB and an extinction ratio of up to 34 dB for resonances in the C+L band. The spectra do not show nonlinear effects due to the near-infrared light used for measuring the transmission, even at high source power levels, but show a shift in resonance wavelengths when pumped with 635 nm light. Using this effect, we could transfer a 1 Hz modulated signal from 635 nm to 1544.8 nm with an extinction ratio of 10 dB. These microtube/SOI waveguide systems, thanks to their excellent optical properties and ease of manipulation, will find more applications in the area of optical communication and signal processing.
This work was supported by the Natural Science and Engineering Research Council of Canada.
References and links
1. J. Bruns, T. Mitze, M. Schnarrenberger, L. Zimmermann, K. Voigt, M. Krieg, J. Kreissl, K. Janiak, T. Hartwich, and K. Petermann, “SOI-based optical board technology,” AEU-Int. J. Electron. C. 61, 158–162 (2007). [CrossRef]
2. Ch. Strelow, H. Rehberg, C. M. Schultz, H. Welsch, Ch. Heyn, D. Heitmann, and T. Kipp, “Optical microcavities formed by semiconductor microtubes using a bottlelike geometry,” Phys. Rev. Lett. 101(12), 127403 (2008). [CrossRef]
5. S. Vicknesh, F. Li, and Z. T. Mi, “Optical microcavities on Si formed by self-assembled InGaAs/GaAs quantum dot microtubes,” Appl. Phys. Lett. 94(8), 081101 (2009). [CrossRef]
6. X. Li, “Strain induced semiconductor nanotubes: from formation process to device applications,” J. Phys. D Appl. Phys. 41(19), 193001 (2008). [CrossRef]
7. Z. Tian, F. Li, Z. T. Mi, and D. V. Plant, “Controlled transfer of single rolled-up InGaAs–GaAs quantum-dot microtube ring resonators using optical fiber abrupt tapers,” IEEE Photon. Technol. Lett. 22(5), 311–313 (2010). [CrossRef]
8. A. Meldrum, P. Bianucci, and F. Marsiglio, “Modification of ensemble emission rates and luminescence spectra for inhomogeneously broadened distributions of quantum dots coupled to optical microcavities,” Opt. Express 18(10), 10230–10246 (2010). [CrossRef]
9. A. V. Prinz, V. Y. Prinz, and V. A. Seleznev, “Semiconductor micro- and nanoneedles for microinjections and ink-jet printing,” Microelectron. Eng. 67–68, 782–788 (2003). [CrossRef]
10. V. Veerasubramanian, A. G. Kirk, G. Beaudin, A. Giguère, B. LeDrogoff, and V. Aimez, “Waveguide coupled drop filters on SOI using vertical sidewalled grating resonators”, 23rd Annual Meeting of the IEEE Photonics Society, 634–635 (2010).
11. G. T. Reed, Silicon Photonics: The State of the Art, (John Wiley & Sons, 2008).
12. D. Taillaert, W. Bogaerts, P. Bienstman, T. F. Krauss, P. Van Daele, I. Moerman, S. Verstuyft, K. De Mesel, and R. Baets, “An out-of-plane grating coupler for efficient butt-coupling between compact planar waveguides and single-mode fibers,” IEEE J. Quantum Electron. 38(7), 949–955 (2002). [CrossRef]
13. G. S. Murugan, J. S. Wilkinson, and M. N. Zervas, “Selective excitation of whispering gallery modes in a novel bottle microresonator,” Opt. Express 17(14), 11916–11925 (2009). [CrossRef]
14. S. Adachi, “Optical Properties” in Properties of group-IV, III–V and II–VI semiconductors, 241(John Wiley & Sons, 2005).
15. R. Kumar, L. Liu, G. Roelkens, E.-J. Geluk, T. de Vries, F. Karouta, P. Regreny, D. V. Thourhout, R. Baets, and G. Morthier, “10-GHz all-optical gate based on a III–V/SOI microdisk,” IEEE Photon. Technol. Lett. 22(13), 981–983 (2010). [CrossRef]
16. L. Zhang, J.-Y. Yang, M. Song, Y. Li, B. Zhang, R. G. Beausoleil, and A. E. Willner, “Microring-based modulation and demodulation of DPSK signal,” Opt. Express 15(18), 11564–11569 (2007). [CrossRef]
17. S.-W. Jeon, Y. H. Kim, B. H. Lee, M. A. Jung, and C.-S. Park, “OSNR monitoring technique based on cascaded long-period fiber grating with optically tunable phase shifter,” Opt. Express 16(25), 20603–20609 (2008). [CrossRef]