Direct-modulated waveguide-coupled microspiral disk lasers with spatially selective injection for on-chip optical interconnects.

Direct-modulated waveguide-coupled microspiral disk lasers with spatially selective injection for on-chip optical interconnects Yue-De Yang,1,2 Yu Zhang,1 Yong-Zhen Huang,2 and Andrew W. Poon1,* 1

Photonic Device Laboratory, Department of Electronic and Computer Engineering, The Hong Kong University of Science and Technology, Hong Kong 2 State Key Laboratory on Integrated Optoelectronics, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China * [email protected]

Abstract: We investigate direct-modulated waveguide-coupled microspiral disk lasers for on-chip optical interconnects. Microspiral resonators, with a rotationally asymmetric shape and a waveguide directly gapless coupled to the notch, offer a compact unidirectional-emission on-chip laser source. We employ spatially selective injection by means of a ring-shaped p-contact on top of the microdisk rim region to selectively inject current to the whispering-gallery-like modes and thus enhance the laser performance. Here we report room-temperature continuous-wave electrically injected AlGaInAs/InP waveguide-coupled microspiral disk lasers with a disk radius of 30 and 40 μm. For a 30μm microspiral disk laser gaplessly coupled with a 100μm-long passive waveguide that is directly connected to an on-chip AlGaInAs/InP photodiode, we estimate a laser output power of at least 200 μW upon a 70mA injection. We realize small-signal modulation with a 3dB bandwidth exceeding 10 GHz for the 30μm microspiral disk. We demonstrate an open eye diagram at 15 Gbit/s with a bias current of 90 mA at a stage temperature of 15°C. ©2014 Optical Society of America OCIS codes: (140.5960) Semiconductor lasers; (140.3945) Microcavities.

References and links 1.

D. A. B. Miller, “Device requirements for optical interconnects to silicon chips,” Proc. IEEE 97(7), 1166–1185 (2009). 2. G. Roelkens, L. Liu, D. Liang, R. Jones, A. Fang, B. Koch, and J. Bowers, “III-V/silicon photonics for on-chip and intra-chip optical interconnects,” Laser Photonics Rev. 4(6), 751–779 (2010). 3. L. He, Ş. K. Özdemir, and L. Yang, “Whispering gallery microcavity lasers,” Laser Photonics Rev. 7(1), 60–82 (2013). 4. S. L. McCall, A. F. J. Levi, R. E. Slusher, S. J. Pearton, and R. A. Logan, “Whispering-gallery mode microdisk lasers,” Appl. Phys. Lett. 60(3), 289–291 (1992). 5. S. Chang, N. B. Rex, R. K. Chang, G. Chong, and L. J. Guido, “Stimulated emission and lasing in whisperinggallery modes of GaN microdisk cavities,” Appl. Phys. Lett. 75(2), 166–168 (1999). 6. H. Cao, J. Y. Xu, W. H. Xiang, Y. Ma, S.-H. Chang, S. T. Ho, and G. S. Solomon, “Optically pumped InAs quantum dot microdisk lasers,” Appl. Phys. Lett. 76(24), 3519–3521 (2000). 7. T. Baba and D. Sano, “Low-threshold lasing and Purcell effect in microdisk lasers at room temperature,” IEEE J. Sel. Top. Quantum Electron. 9(5), 1340–1346 (2003). 8. A. C. Tamboli, E. D. Haberer, R. Sharma, K. H. Lee, S. Nakamura, and E. L. Hu, “Room temperature continouswave lasing in GaN/InGaN microdisks,” Nat. Photonics 1(1), 61–64 (2007). 9. J. Van Campenhout, P. Rojo Romeo, P. Regreny, C. Seassal, D. Van Thourhout, S. Verstuyft, L. Di Cioccio, J. M. Fedeli, C. Lagahe, and R. Baets, “Electrically pumped InP-based microdisk lasers integrated with a nanophotonic silicon-on-insulator waveguide circuit,” Opt. Express 15(11), 6744–6749 (2007). 10. V. N. Astratov, S. Yang, S. Lam, B. D. Jones, D. Sanvitto, D. M. Whittaker, A. M. Fox, M. S. Skolnick, A. Tahraoui, P. W. Fry, and M. Hopkinson, “Whispering gallery resonances in semiconductor micropillars,” Appl. Phys. Lett. 91(7), 071115 (2007). 11. P. Jaffrennou, J. Claudon, M. Bazin, N. S. Malik, S. Reitzenstein, L. Worschech, M. Kamp, A. Forchel, and J.M. Gérard, “Whispering gallery mode lasing in high quality GaAsAlAs pillar,” Appl. Phys. Lett. 96(7), 071103 (2010).

#200669 - $15.00 USD (C) 2014 OSA

Received 4 Nov 2013; revised 22 Dec 2013; accepted 23 Dec 2013; published 7 Jan 2014 13 January 2014 | Vol. 22, No. 1 | DOI:10.1364/OE.22.000824 | OPTICS EXPRESS 824

12. J. U. Nöockel, A. D. Stone, G. Chen, H. L. Grossman, and R. K. Chang, “Directional emission from asymmetric resonant cavities,” Opt. Lett. 21(19), 1609–1611 (1996). 13. J. U. Nöckel and A. D. Stone, “Ray and wave chaos in asymmetric resonant optical cavities,” Nature 385(6611), 45–47 (1997). 14. C. Gmachl, F. Capasso, E. E. Narimanov, J. U. Nöckel, A. D. Stone, J. Faist, D. L. Sivco, and A. Y. Cho, “Highpower directional emission from microlasers with chaotic resonators,” Science 280(5369), 1556–1564 (1998). 15. S.-B. Lee, J.-H. Lee, J.-S. Chang, H.-J. Moon, S. W. Kim, and K. An, “Observation of scarred modes in asymmetrically deformed microcylinder lasers,” Phys. Rev. Lett. 88(3), 033903 (2002). 16. S.-K. Kim, S.-H. Kim, G.-H. Kim, H.-G. Park, D.-J. Shin, and Y.-H. Lee, “Highly directional emission from few-micron-size elliptical microdisks,” Appl. Phys. Lett. 84(6), 861–863 (2004). 17. Y. Baryshnikov, P. Heider, W. Parz, and V. Zharnitsky, “Whispering gallery modes inside asymmetric resonant cavities,” Phys. Rev. Lett. 93(13), 133902 (2004). 18. T. Tanaka, M. Hentschel, T. Fukushima, and T. Harayama, “Classical phase space revealed by coherent light,” Phys. Rev. Lett. 98(3), 033902 (2007). 19. G. D. Chern, H. E. Tureci, A. D. Stone, R. K. Chang, M. Kneissl, and N. M. Johnson, “Unidirectional lasing from InGaN multiple-quantum-well spiral-shaped micropillars,” Appl. Phys. Lett. 83(9), 1710–1712 (2003). 20. M. Kneissl, M. Teepe, N. Miyashita, N. M. Johnson, G. D. Chern, and R. K. Chang, “Current-injection spiralshaped microcavity disk laser diodes with unidirectional emission,” Appl. Phys. Lett. 84(14), 2485–2487 (2004). 21. S. Y. Lee, S. Rim, J. W. Ryu, T. Y. Kwon, M. Choi, and C. M. Kim, “Quasiscarred resonances in a spiral-shaped microcavity,” Phys. Rev. Lett. 93(16), 164102 (2004). 22. R. Audet, M. A. Belkin, J. A. Fan, B. G. Lee, K. Lin, F. Capasso, E. E. Narimanov, D. Bour, S. Corzine, J. Zhu, and G. Höfler, “Single-mode laser action in quantum cascade lasers with spiral-shaped chaotic resonators,” Appl. Phys. Lett. 91(13), 131106 (2007). 23. C. M. Kim, J. Cho, J. Lee, S. Rim, S. H. Lee, K. R. Oh, and J. H. Kim, “Continuous wave operation of a spiralshaped microcavity laser,” Appl. Phys. Lett. 92(13), 131110 (2008). 24. J. Wiersig and M. Hentschel, “Combining directional light output and ultralow loss in deformed microdisks,” Phys. Rev. Lett. 100(3), 033901 (2008). 25. Y. D. Yang, S. J. Wang, and Y. Z. Huang, “Investigation of mode coupling in a microdisk resonator for realizing directional emission,” Opt. Express 17(25), 23010–23015 (2009). 26. Q. Song, L. Ge, B. Redding, and H. Cao, “Channeling chaotic rays into waveguides for efficient collection of microcavity emission,” Phys. Rev. Lett. 108(24), 243902 (2012). 27. M. W. Kim, C. H. Yi, and C. M. Kim, “Emission characteristics of a spiral-shaped microcavity laser with a waveguide,” J. Korean Phys. Soc. 60(5), 750–753 (2012). 28. Y. D. Yang, Y. Zhang, Y. Z. Huang, and A. W. Poon, “AlGaInAs/InP waveguide-coupled unidirectionalemission microspiral lasers for on-chip optical interconnects,” in CLEO (2013), paper CTh1G.2. 29. G. Bjork and Y. Yamamoto, “Analysis of semiconductor microcavity lasers using rate equations,” IEEE J. Quantum Electron. 27(11), 2386–2396 (1991). 30. A. Kapsalis, D. Syvridis, U. Troppenz, M. Hamacher, and H. Heidrich, “7Gb/s direct modulation of bertically coupled microring lasers,” in 2008 Conference on Optical Fiber Communication/National Fiber Optic Engineers Conference (2008), pp. 2917–2919. 31. L. Liu, G. Roelkens, J. Van Campenhout, J. Brouckaert, D. Van Thourhout, and R. Baets, “III-V/silicon-onInsulator nanophotonic cavities for optical network-on-chip,” J. Nanosci. Nanotechnol. 10(3), 1461–1472 (2010). 32. M. H. Mao and H. C. Chien, “Transient behaviors of current-injection quantum-dot microdisk lasers,” Opt. Express 20(3), 3302–3310 (2012). 33. X. M. Lv, Y. Z. Huang, L. X. Zou, H. Long, and Y. Du, “Optimization of direct modulation rate for circular microlasers by adjusting mode Q factor,” Laser Photonics Rev. 7(5), 818–829 (2013). 34. S. K. J. Luo, J. Y. Xu, H. Cao, Y. Ma, S. H. Chang, S. T. Ho, and G. S. Solomon, “Dynamics of GaAs/AlGaAs microdisk lasers,” Appl. Phys. Lett. 77(15), 2304–2306 (2000). 35. M. Tassaert, H. J. S. Dorren, G. Roelkens, and O. Raz, “Passive InP regenerator integrated on SOI for the support of broadband silicon modulators,” Opt. Express 20(10), 11383–11388 (2012). 36. L. Coldren and S. Corzine, Diode Lasers and Photonic Integrated Circuits (John Wiley, 1995). 37. L. Xie, J. Man, B. Wang, Y. Liu, X. Wang, H. Yuan, L. Zhao, H. Zhu, N. Zhu, and W. Wang, “24-GHz directly modulated DFB laser modules for analog applications,” IEEE Photon. Technol. Lett. 24(5), 407–409 (2012). 38. R. S. Tucker, J. M. Wiesenfeld, P. M. Downey, and J. E. Bowers, “Propagation delays and transition times in pulse-modulated semiconductor lasers,” Appl. Phys. Lett. 48(25), 1707–1709 (1986).

1. Introduction Optical interconnects has been regarded as a promising solution to address the bandwidth limitation and power consumption in the conventional on-chip electrical interconnects [1]. Semiconductor whispering-gallery-mode (WGM) microresonator lasers are potential compact, energy-efficient light sources for on-chip optical interconnects due to their small footprints, high cavity quality (Q) factors and planar geometry [2, 3]. Among WGM microresonator lasers, circular microresonator lasers with a circularly symmetric geometry have attracted the most research interest, given their high-Q WGMs [4–11]. However, due to the circular rotational symmetry, one major shortcoming for circular microresonator lasers is

#200669 - $15.00 USD (C) 2014 OSA


that their lasing emission is non-directional and homogeneous along the cavity rim. This is undesirable for the on-chip optical interconnects application. In order to realize directional or unidirectional lasing emission from WGM microresonator lasers, researchers in the past two decades have been studying various deformed-shape microcavities in order to properly break the shape symmetry, while preserving high-Q cavities with efficient directional or unidirectional output-coupling [12–18]. By carefully designing the cavity shape, researchers have obtained unidirectional lasing emission from various asymmetric microresonator lasers, including microspiral pillar or disk lasers and limaçon microdisk lasers [19–24]. Although these microcavity designs can produce unidirectional lasing emission with a small in-plane divergence angle, coupling the light to other on-chip devices still requires further light-coupling techniques that typically introduce additional insertion losses. Thus, waveguide-coupled microcavity lasers with unidirectional emission should be a potential solution [25, 26]. One candidate for such a waveguide-integrated onchip light source is the microspiral disk laser, which offers the key merit of direct gapless coupling from a characteristic notch that is only a fraction of the microdisk radius to an integrated waveguide [27, 28]. Another useful criterion for on-chip optical interconnect light sources is the possibility of high-speed direct-modulation without the need of an additional external optical modulator. The direct-modulation characteristics of semiconductor WGM microcavity lasers have long been theoretically studied [29]. Experimentally, researchers have recently demonstrated electrically injected direct-modulation in a 70μm-radius microring laser with a 7Gbit/s modulation [30], in a 7.5μm-diameter circular microdisk laser bonded on a silicon-oninsulator (SOI) waveguide with a 3dB bandwidth of 3.5 GHz [31] and in a quantum-dot circular microdisk laser with a 1Gbit/s modulation [32]. Most recently, one of us has demonstrated 12.5Gbit/s direct-modulation in a waveguide-coupled circular microdisk laser [33]. However, the direct-modulation speed for microdisk lasers can be limited by carrier diffusion over the non-uniform spatial field distribution of WGMs [34]. In this paper, we report room-temperature continuous-wave (cw) electrically injected waveguide-coupled microspiral disk lasers of 30 and 40μm radius with direct-modulation. We design ring-shaped p-contacts on top of the microspiral disk rim in order to maximize the spatial overlap between the injection current and the high-Q modes of the microspiral cavity to reduce the lasing threshold and increase the modulation bandwidth. Spatially selective electrical injection has been used to lower the lasing threshold of microspiral pillars of a relatively large radius [20]. We obtain for a 30μm-radius microspiral disk laser that is directly connected to an on-chip photodiode an estimated laser output power of at least 200 µW upon a 70mA injection. We compare the DC and direct-modulation performances between three different ring-shaped and disk injection designs. We demonstrate direct-modulation of the waveguide-coupled microspiral disk laser, with a small-signal-response 3dB bandwidth exceeding 10 GHz for a 30μm-radius microlaser. An open eye diagram at 15 Gbit/s with a bias current of 90 mA is demonstrated at a stage temperature of 15 °C. 2. Waveguide-coupling and spatially selective injection: designs and simulations Figure 1(a) illustrates the microspiral shape with a radius linearly varying with the azimuthal angle as follows: r = r0 (1 − εθ / 2π )

(1)

where r is the azimuthally varying radius of the microspiral, r0 is the radius at the azimuthal angle of 0°, θ is the azimuthal angle and ε is the deformation parameter. In our work, we have chosen relatively small ε values in order to obtain high-Q modes from the microspiral disk cavity. An output-coupled waveguide with a width w = ɛr0 is directly gapless coupled to the microspiral notch for unidirectional out-coupling. We choose the waveguide width w = 1.5 and 2 μm in our designs. The minimum waveguide width is limited by both our contactaligner photolithography and laser direct-write mask resolutions.

#200669 - $15.00 USD (C) 2014 OSA


Fig. 1. Top-view schematics of waveguide-coupled AlGaInAs microspiral disk lasers with (a) ring-shaped and (b) disk-shaped p-contacts. (c) Cross-sectional-view schematic of microspiral disk lasers with a ring-shaped p-contact.

We use a ring-shaped p-contact on top of the microdisk rim and an arc-shaped n-contact surrounding the microdisk on top of the substrate, as shown in Fig. 1(a). The patterned pcontact metal is used to define the injection area for preferentially injecting along the rim region of the microdisk for better current injection efficiency to the high-Q WG-like modes, to be shown in Fig. 3. The patterned n-contact on the same side of the wafer forces carriers flow towards the rim region of the microdisk. This helps the carriers spatially overlap with the high-Q optical modes, and also minimizes carrier diffusion to the microdisk center. As a control, we also design a disk-shaped p-contact injecting essentially the entire microdisk, as shown in Fig. 1(b). In all cases, the waveguide is not electrically injected, and thus it absorbs part of the laser output power. We use coplanar electrodes on the same side of the substrate for the ease of planar integration. Figure 1(c) schematically shows the cross-sectional view of the vertical p-i-n diode structure and the layer thicknesses of the ring-shaped injected microspiral disk lasers. We employ two-dimensional (2D) finite-difference time-domain (FDTD) method to simulate the cavity internal-field spectrum and the spatial field-amplitude distributions at resonances. We calculate the transverse-electric (TE) modes in a microspiral with r0 = 30 μm, ɛ = 0.05 and a 1.5μm-wide multimode waveguide butt-coupled to the microspiral notch, following our typical device designs. We assume the cavity has a constant effective refractive index of 3.2. The calculation only includes two sets of modes with the highest and secondhighest Q values. Figure 2(a) shows the calculated internal-field spectrum revealing the highest-Q mode with a Q ≈4 × 104 at 1549.2 nm and the second-highest-Q mode with a Q ≈1 × 104 at 1549.9 nm. The calculated free-spectral range (FSR) for both modes is about 4.3 nm, which is consistent with the microspiral circumference.

#200669 - $15.00 USD (C) 2014 OSA


Fig. 2. 2D-FDTD simulations of a 30μm-radius microspiral cavity with a 1.5μm-wide waveguide coupled to the microspiral notch. (a) Internal-field spectrum of the TE modes normalized to the peak intensity at 1549.2 nm. (b), (c) Internal-field amplitude distributions normalized to the peak amplitude inside the cavity for the resonance mode at (b) 1549.2 nm and (c) 1549.9 nm. Zoom-in: high-order transverse modes of the waveguide.

Figures 2(b) and 2(c) show the calculated distributions in the x-y plane of the magnetic field-amplitude component in the z-direction |Hz| of the modes at 1549.2 nm and 1549.9 nm. Both reveal a WG-like mode. We simulate the field-amplitude distributions with a clockwise excitation source in the rim region of the microdisk. Our simulation results suggest a high waveguide-output-coupling efficiency via the microspiral notch. We define the waveguide-output-coupling efficiency as the ratio of the energy flux through the output-waveguide to the total energy flux coupled out of the cavity. The calculated output-coupling efficiency exceeds 70% for the simulated WG-like modes [28]. The field-amplitude distributions inside the waveguide reveal multiple high-order transverse modes, suggesting a multimode waveguide, as shown in the zoom-in views of Figs. 2(b) and 2(c). Similar to the WGMs in circular microdisks, the WG-like modes in the microspiral resonator have a weak field-amplitude distribution in the center region of the disk, as shown in Figs. 2(b) and 2(c). In order to quantify the field-intensity distribution in the radial direction, we define the normalized azimuthal-angle-integrated field-intensity distribution in the radial direction as follows: 2π

F (r) =

|H

z

( r , θ ) |2 rdθ

0

r0

 dr  | H 0

(2)

2π

( r , θ ) | rdθ 2

z

0

where r = r(θ) follows Eq. (1). Figure 3(a) shows the calculated F(r) for the two resonances at 1549.2 nm and 1549.9 nm. In order to spatially overlap the current injection with the high-Q modes of the microspiral cavity, we design ring-shaped p-contacts on top of the microspiral rim region. We use three different designs of ring-shaped p-contacts with a width of Wp = 4 μm and various inner-ring radius Rp (defined at θ = 0), as schematically shown in Fig. 1(c). The inner boundary of the pcontact metal also follows Eq. (1), with r0 substituted by Rp. For the microspiral with r0 = 30 μm, we choose Rp equals to 24, 20 and 14 μm, corresponding to an outer-, a middleand an inner-ring-shaped p-contact. For the microspiral with r0 = 40 μm, we choose Rp equals to 34, 28 and 20 μm, corresponding to an outer-, a middle- and an inner-ring-shaped pcontact. We leave a margin of at least 2 μm from the microdisk edge in order to ease the

#200669 - $15.00 USD (C) 2014 OSA


contact lithography alignment and to avoid short-circuit to the substrate. We fix the n-contact to be a c-shaped ring with the arc opens in the notch-waveguide region.

Fig. 3. (a) 2D-FDTD simulated normalized azimuthal-angle-integrated internal-field intensity radial distributions for the modes at 1549.2 nm and 1549.9 nm. (b) Finite-element-method (FEM) simulated normalized azimuthal-angle-integrated injection current radial distributions for different injection designs.

We simulate the injection current density distribution using a commercially available semiconductor device simulation tool (Silvaco). We assume a vertical p-i-n diode in a circular microdisk with a two-dimensional cross-section comprising a p+-InGaAs, a p-InP, an i-AlGaInAs and an n-InP layer sitting on an n-InP substrate. The layer thicknesses and the doping concentrations follow from our fabricated devices. The p- and n-contact metal dimensions and spacing also follow the device design. We define the normalized azimuthalangle-integrated injection current density distribution in the radial direction as follows: 2π

I (r) =

 J ( r )rdθ 0

r0

2π

0

0

(3)

 dr  J ( r )rdθ where r = r(θ) follows Eq. (1), J(r) is the injection current density distribution in the radial direction in the middle of the i-AlGaInAs region. Figure 3(b) shows the simulated I(r) for a 30μm-radius disk, assuming the three different ring-injection designs and the disk injection design (with a 2μm margin away from the microdisk edge). In order to approximately evaluate the current injection efficiency to the microdisk modes, we define a spatial overlap factor Γ between F(r) and I(r) as follows: r0

Γ=

 F ( r )I ( r )dr 0

r0

F 0

r0

2

(4)

( r )dr  I 2 ( r )dr 0

Table 1 summarizes the calculated Γ for the highest-Q and second-highest-Q WG-like modes in microspiral resonators with r0 = 30 and 40 μm and w = 1.5 and 2 μm. The inner-ring injections in all cases exhibit the lowest Γ values (< 0.53). For the disk injection, the calculated Γ values suggest smaller current injection efficiencies to the microdisk modes than those via the outer- and middle-ring-shaped injections. The outer-ring injections in all cases exhibit the highest Γ values (≥ 0.81). Our simulation results thus suggest the outer-ring pcontact design is desirable for efficient current injection to the high-Q microdisk modes.

#200669 - $15.00 USD (C) 2014 OSA


Table 1. Calculated spatial overlap factor Γ for various spatially selective injection designs in the microspiral disks

r0 = 30 μm, w = 1.5 μm r0 = 30 μm, w = 2 μm

r0 = 40 μm, w = 1.5 μm r0 = 40 μm, w = 2 μm

Resonance mode

Outer-ring (Rp = 24 μm)

Middle-ring (Rp = 20 μm)

Inner-ring (Rp = 14 μm)

Disk injection

Highest-Q 2nd-highest-Q Highest-Q 2nd-highest-Q

0.81 0.91 0.83 0.93 Outer-ring (Rp = 34 μm) 0.82 0.91 0.85 0.92

0.79 0.73 0.80 0.86 Middle-ring (Rp = 28 μm) 0.73 0.75 0.74 0.75

0.52 0.44 0.52 0.53 Inner-ring (Rp = 20 μm) 0.29 0.31 0.30 0.32

0.72 0.67 0.74 0.77 Disk injection 0.63 0.67 0.65 0.68

Highest-Q 2nd-highest-Q Highest-Q 2nd-highest-Q

3. Fabrication of AlGaInAs/InP microspiral disk lasers We fabricate the microspiral disk lasers on a commercially available AlGaInAs/InP wafer with eight pairs of compressively strained multiple-quantum-wells (MQWs). The quantum wells (QWs) and barrier layers with thicknesses of about 6 nm and 9 nm, respectively, are sandwiched between two 100nm AlGaInAs separate confinement heterostructure (SCH) layers. The upper layers are p-InP cladding and p+-InGaAs contact layers with a total thickness of 1.5 μm. A 700nm SiO2 is deposited by plasma-enhanced chemical vapor deposition (PECVD) on the laser wafer. The laser cavity patterns are transferred onto the SiO2 layer using i-line (365 nm) contact photolithography and reactive ion etching (RIE) techniques. The patterned SiO2 is used as a hard mask for the following inductive-coupled plasma (ICP) dry etching process. AlGaInAs/InP laser wafer has a total etched depth of about 3.5 μm.

Fig. 4. (a)-(d) SEM images of the waveguide-coupled 30μm-radius microspiral disk after ICP etching. (a) Top view, (b) side view, (c) cross-sectional view of the 2μm-wide notch-coupled waveguide and (d) cross-sectional view of the 1.5μm-wide notch-coupled waveguide. (e), (f) Optical microscope images of the microspiral disk laser after p- and n-contact metal deposition for (e) outer-ring contact design and (f) disk contact design. (g) Optical microscope image of the microspiral disk laser with patterned p- and n-electrodes. G: ground, S: signal.

Figures 4(a) and 4(b) show the scanning-electron microscope (SEM) images of the topview and the cavity sidewall of the 30μm-radius microspiral with a 2μm-wide waveguide after the ICP etching. The cavity sidewall suggests surface roughness and a slight vertical curvature. We attribute the surface roughness to the photolithography mask resolution and the dry etching process. Figures 4(c) and 4(d) show the cross-section for the fabricated

#200669 - $15.00 USD (C) 2014 OSA


waveguides with a designed width of 2 and 1.5 μm, respectively. We notice a ~200nm undercut below the p+-InGaAs layer. We deposit Ti/Pt/Au and Ni/Ge/Au by electron-beam evaporation and lift-off processes as p- and n-contact metals, respectively. Figures 4(e) and 4(f) show the optical microscope images of the microspiral disk laser with a ring-shaped (Rp = 24 μm) and disk injection after p- and n-contact metal deposition. A 500nm SiO2 insulating layer is deposited on the laser wafer, and a contact window is opened on top of the p- and n-contact metals using RIE etching process. Ti/Al electrode metal is deposited and patterned to form a ground-signalground (GSG) pad. The laser sample is cleaved, with a typical passive waveguide length of about 100 μm for measurements. The output-waveguide facet is not coated with antireflection (AR) coating. Therefore, the undesirable reflection at the waveguide facet reduces the output-coupled power. However, as the output-waveguide is not injected and thus highly absorbing, the reflection does not effectively contribute as an additional optical feedback. Figure 4(g) shows the optical microscope image of the waveguide-coupled microspiral disk laser with patterned electrodes. 4. Experimental results 4.1 DC characteristics We characterize the lasing properties of the fabricated waveguide-coupled microspiral disk lasers with DC electrical injection. We use a 50μm-core-diameter multimode fiber to buttcouple to the cleaved end of the output-waveguide in order to collect the laser light. We mount the sample-under-test on a thermoelectric cooler (TEC) with a typical fixed stage temperature of 15 – 20°C. We realize room-temperature cw electrically injected lasing for the microspiral disk lasers with a radius of 30 and 40 μm. We observe lasing up to a maximum stage temperature of about 55°C.

Fig. 5. Fiber-coupled laser output power versus injection current of the waveguide-coupled microspiral disk lasers with the outer-ring (solid line), middle-ring (dashed line) and disk injection (dotted line) designs. (a) r0 = 30 μm, ɛ = 0.05, w = 1.5 μm, (b) r0 = 30 μm, ɛ = 0.067, w = 2μm, (c) r0 = 40 μm, ɛ = 0.037, w = 1.5 μm and (d) r0 = 40 μm, ɛ = 0.05, w = 2 μm.

Figure 5 shows the measured total output-coupled power as a function of the injection current at a stage temperature of 20°C for 30μm- and 40μm-radius microspiral disk lasers.

#200669 - $15.00 USD (C) 2014 OSA


Both microdisks are butt-coupled with a waveguide with widths of 1.5 and 2 μm. The microspiral disks with an outer- or middle-ring or disk injection show lasing, while the disks with an inner-ring injection (with Rp = 14 μm for the 30μm-radius microspiral and Rp = 20 μm for the 40μm-radius microspiral) show no lasing. We attribute the latter to the low spatial overlap factor between the current injection and the high-Q WG-like modes, as shown in Table 1. Tables 2 and 3 summarize the measured lasing thresholds and slope-efficiencies for the different waveguide-coupled microspiral disk lasers. The devices with an outer-ring injection always exhibit the lowest lasing threshold and the highest slope-efficiency because of the highest spatial overlap factor. The disk injections always display the highest lasing threshold and the lowest slope-efficiency because of the wasted injection current in the center area of the microdisk. The measured results are consistent with the calculated spatial overlap factors summarized in Table 1. The lasing thresholds upon ring-shaped injection are approximately proportional to the microdisk radius. We also note that the microdisk lasers of the same radius but different waveguide widths exhibit similar lasing thresholds, but the microdisks with w = 2 μm generally display a larger slope-efficiency. The measured total output-coupled powers of our devices are, however, relatively low. We attribute the low output power partly to the undesirable absorption along the 100μm-long passive waveguide, and partly to an unoptimized waveguide-to-fiber coupling efficiency and the end-face reflection at the multimode output-waveguide facet. We do not plot the estimated laser power-injection current (P-I) curves for the microlasers because the estimated laser power has a large uncertainty due to the estimated output-waveguide loss. Table 2. Measured lasing thresholds for different waveguide-coupled microspiral disk lasers (unit: mA)

r0 = 30 μm, w = 1.5 μm r0 = 30 μm, w = 2 μm r0 = 40 μm, w = 1.5 μm r0 = 40 μm, w = 2 μm

Outer-ring (Rp = 24 μm) 26 28 Outer-ring (Rp = 34 μm) 34 36

Middle-ring (Rp = 20 μm) 34 36 Middle-ring (RP = 28 ΜM) 47 47

Disk injection 37 42 Disk injection 57 68

Table 3. Measured slope-efficiencies for different waveguide-coupled microspiral disk lasers (unit: mW/A)


Outer-ring (Rp = 24 μm) 0.455 0.724 Outer-ring (Rp = 34 μm) 0.345 0.850

Middle-ring (Rp = 20 μm) 0.419 0.649 Middle-ring (Rp = 28 μm) 0.282 0.456

Disk injection 0.136 0.269 Disk injection 0.219 0.156

We measure the lasing spectra using an optical spectrum analyzer (OSA). Figures 6(a) and 6(b) show the measured lasing spectra from 1540 to 1580 nm upon injection currents of 40 mA and 60 mA at a stage temperature of 15°C. Upon injection current of 40 mA, the laser shows a main lasing peak at 1547 nm (denoted with a *), with a 6dB side-mode suppression ratio (SMSR) and a peak output-coupled power of ~-38 dBm. The FSR is 3.7 nm. Upon injection current of 60 mA, the laser shows a red-shifted main lasing peak at 1565 nm (denoted with a *), with an improved SMSR of 17 dB and a larger peak output-coupled power exceeding −30 dBm. The FSR for the red-shifted lasing mode is ~3.8 nm. The measured FSR value is smaller than the simulated FSR value of ~4.3 nm [shown in Fig. 2(a)] because the waveguide and material dispersions are neglected in the simulation. There are two primary lasing modes that are very close to each other, as shown in the lasing emission spectra in Fig. 6. We note that the main lasing mode (on the short-wavelength side) always dominates #200669 - $15.00 USD (C) 2014 OSA


the other one by at least 10 dB. Therefore, only the main lasing mode should significantly contribute to the observed lasing threshold current values and the coupled-power-current measurements.

Fig. 6. Measured lasing spectra of the 30 μm-radius microdisk laser with w = 2 μm and the outer-ring injection upon injection currents of (a) 40 mA and (b) 60 mA. Inset in (a): Measured lasing peak wavelength versus injection current. Line: linear fit. Inset in (b): Top-view infrared image of the microspiral disk laser upon a 60mA injection.

We attribute the lasing spectral red-shift to an injection-current-induced thermal effect. We measure the peak wavelengths of the lasing modes around 1565 nm versus the injection current (at a fixed stage temperature of 15°C) and the stage temperature (at a fixed injection current of 60 mA). The lasing wavelength red-shift is proportional to the injection current at a slope of ~0.041 nm/mA, as shown in the inset of Fig. 6(a). The lasing mode wavelength redshift is proportional to the stage temperature at a slope of ~0.12 nm/K (not shown). Thus, we observe a current-induced thermal heating effect of 0.34 K/mA. The current-induced thermal loading also results in shifting the gain spectrum, leading to the observed mode-hoping from ~1547 nm - ~1565 nm. The inset of Fig. 6(b) shows the measured top-view infrared image of the microdisk lasing upon a 60mA injection. We observe a largely uniform scattering from the microdisk sidewall along the rim, suggesting a WG-like mode-field distribution in the microspiral disk. We then obtain a group refractive index of ~3.6 from the FSR, assuming the lasing mode is WG-like. The extracted group refractive index agrees well with the group refractive index obtained from a 170 μm-long Fabry-Pérot (FP) laser (not shown) fabricated on the same laser wafer. However, based on the measured FSR value comparing with the simulation, together with the measured top-view light-scattering image of the microdisk laser emission, it remains difficult to conclude which mode is lasing.

Fig. 7. Measured photocurrent from an on-chip photodiode waveguide-coupled to a 30μmradius microspiral disk laser versus injection current. Line: visual guide. Inset: Top-view optical microscope image of the microspiral disk laser “1” interconnected with an on-chip photodiode “2”. G: ground; S: signal.

#200669 - $15.00 USD (C) 2014 OSA


In order to estimate the waveguide-coupled lasing power and the fiber-coupling loss for the microspiral disk lasers, we fabricate a waveguide-coupled microspiral disk laser directly coupled with an integrated double-notch-shaped on-chip AlGaInAs/InP photodiode. The inset of Fig. 7 shows the top-view optical microscope image of such an interconnected microdisk laser-photodiode. The 100μm-long waveguide connecting the laser and the photodiode is not injected. We measure the photocurrent from the photodiode upon a 0V bias, while forwardbiasing the microspiral disk laser. The photodiode dark current upon 0V bias is ~nA. Figure 7 shows the measured photocurrent as a function of the injection current for a 30μm-radius microspiral disk laser, with w = 2 μm and the outer-ring injection design. The measured photocurrent suggests a lasing threshold consistent with that shown in the outputcoupled power measurement [Fig. 5(b)]. Assuming an ideal photoresponsivity of 1.25 A/W around 1550 nm, we estimate the received laser power at the on-chip photodiode is at least 200 μW upon a 70mA injection. Comparing with the measured fiber-coupled laser power [Fig. 5(b)], we then estimate the fiber-coupling loss is about 10 dB. This suggests that the microspiral disk laser butt-coupled with an un-injected 100μm-long waveguide has a slopeefficiency of about 7 mW/A (approximately an order improvement from the measured fibercoupled slope-efficiency, as shown in Table 3). From the literature reporting an III-V-onsilicon singlemode non-injected waveguide absorber with MQWs [35], we estimate our uninjected waveguide absorption loss to be ~3 – 7 dB for a waveguide length of ~100 μm. However, as our output-waveguide supports multiple modes [see Figs. 2(b) and 2(c)], the above estimation only works well for the fundamental mode but not necessarily for the higher-order modes. 4.2 Small-signal response We study the small-signal response for the waveguide-coupled microspiral disk lasers, without using the on-chip integrated photodiode. We find a leakage from the on-chip photodiode to the microspiral disk laser upon connecting the AC ground to the photodiode, and thus cannot observe a modulated photocurrent. We attribute the leakage to the heavily ndoped substrate. The DC-bias current is combined with the modulation signal using an 18GHz bias-T and fed to the laser through a 40GHz radio-frequency probe. The microlaser output is coupled into a lensed single-mode polarization-maintaining fiber and amplified by about 30 dB using an erbium-doped fiber amplifier (EDFA). The amplified output is spectrally filtered using a tunable band-pass filter in order to suppress the EDFA amplified spontaneous emission (ASE) noise.

Fig. 8. Measured small-signal frequency responses of the 30μm-radius microspiral lasers with w = 2 μm upon different injection designs. (a) Outer-ring injection with Rp = 24 μm, (b) middle-ring injection with Rp = 20 μm, and (c) disk injection. Inset in (a): Fitted curve for the measured S21 response at a 70mA injection.

We measure the small-signal modulation responses using a 20GHz network analyzer. Figure 8 shows the measured small-signal modulation responses for the 30μm-radius microspiral disk laser with w = 2 μm and ɛ = 0.067, using different injection designs at a stage temperature of 20°C. Figures 8(a) to 8(c) show the measured responses with the outer-ring, middle-ring and disk injection, respectively. The measured S21 is normalized to the low-

#200669 - $15.00 USD (C) 2014 OSA


frequency response at 100 MHz. The 3dB bandwidth increases with the injection current, and saturates at about 10.7 GHz at the injection current of 110 mA for the laser with the outer-ring injection. The maximum bandwidths are only 7.7 GHz and 8.9 GHz for the lasers with the middle-ring and disk injection, respectively. By further increasing the injection current, we note that the main lasing peak shifts out of the EDFA gain spectral band due to the semiconductor gain shift upon current-induced thermal loading [see Fig. 6]. Table 4 summarizes the measured maximum 3dB bandwidths of the microspiral disk lasers with different injection designs for the 30μm- and 40μm-radius microdisks with w = 1.5 and 2 μm. The lasers with the outer-ring injection always exhibit the highest bandwidths compared with the middle-ring and disk injection. The measured maximum bandwidths for the 30μm- and 40μm-radius microspiral disk lasers are 10.7 and 9.0 GHz, respectively. Table 4. Measured maximum 3dB bandwidths for different waveguide-coupled microspiral disk lasers (unit: GHz)

r0 = 30 μm, w = 1.5 μm* r0 = 30 μm, w = 2 μm r0 = 40 μm, w = 1.5 μm r0 = 40 μm, w = 2 μm

Outer-ring (Rp = 24 μm) 10.4 10.7 Outer-ring (Rp = 34 μm) 9.0 -



*This is a different sample obtained from the same wafer from that used in the DC measurement.

We study the dynamic response as a function of the modulation frequency H(f) of the lasers. We express H(f) as follows [36]: H(f )=

f R4 1 + (2π fRC ) 2 ( f R2 − f 2 ) 2 + f 2γ 2 / (2π ) 2 1

(5)

where R is the resistance of the laser, C is the capacitance of the laser, fR is the relaxation oscillation frequency, and γ is the damping factor. We extract the fR and γ values upon various injection currents by fitting with Eq. (5) the measured S21 responses, ranging from 1 GHz to the frequency where the response drops by 20 dB below the relaxation resonance peak according to the measured RC time constant, as shown in the inset of Fig. 8(a). The measured capacitance and resistance (at 1 kHz) of the 30μm- and 40μm-radius microdisk lasers are 1.8 pF and 9 Ω, and 2.6 pF and 8 Ω, respectively, corresponding to RC-cutoff frequencies of about 10 GHz and 8 GHz, respectively. The relatively large capacitances are due to the relatively large microdisk area and the metal pads. We can approximately express fR as follows [36]: fR =

1 v g aN p ( ) = D ( I − I th )1/2 2π τp

(6)

where τp is the photon lifetime, a is the differential gain versus carrier density, Np is the photon density, vg is the group velocity, Ith is the threshold current, and D is the D-factor, which indicates the efficiency of the direct modulation. Figure 9 shows the measured injection current dependences of fR at a fixed stage temperature of 20°C for the 30μm- and 40μm-radius microdisk lasers with w = 1.5 and 2 μm and different injection designs. We do not include measurement results for the 40μm-radius microspiral disk lasers with w = 2 μm and the outer-ring injection as the lasing wavelength has shifted out of the EDFA gain range. Table 5 summarizes the extracted D-factor values from Fig. 9 using Eq. (6) for the different waveguide-coupled microspiral disk lasers. The D-factors of the 30μm- and 40μmradius microspiral disk lasers range from about 1.04 to 0.68 GHz/mA1/2 and 0.78 to 0.5 GHz/mA1/2, respectively, for the outer-ring injection to the disk injection. The lasers with the

#200669 - $15.00 USD (C) 2014 OSA


outer-ring injection always have the largest D-factors and thus modulation efficiencies due to the highest spatial overlap factor Γ. The measured highest fR values are ~8 GHz, obtained at the bias current of around 100 mA for the 30μm-radius microspiral lasers with the outer-ring injection, as shown in Figs. 8(a) and 8(b). We attribute the saturation of fR observed at higher bias currents to the rise of the laser temperature.

Fig. 9. Extracted relaxation oscillation frequencies versus square root of the bias current above the lasing threshold for different injection designs. Black squares: out-ring injection; Red circles: middle-ring injection; Blue triangles: disk injection. Dashed lines: linear fits from the origin. (a) r0 = 30 μm, ɛ = 0.05, w = 1.5 μm, (b) r0 = 30 μm, ɛ = 0.067, w = 2 μm, (c) r0 = 40 μm, ɛ = 0.037, w = 1.5 μm and (d) r0 = 40 μm, ɛ = 0.05, w = 2 μm. Inset in (b): Extracted damping factor as a function of the squared resonant oscillation frequency for the outer-ring injection. Solid line: linear fit Table 5. Extracted D-factors for different waveguide-coupled microspiral disk lasers (unit: GHz/mA-1/2)





The damping factor is proportional to f R2 and is given by [36]

γ = γ 0 + Kf R2

(7)

where γ0 is the damping factor offset and K is a constant. We extract the K-factor with a linear fit of γ versus f R2 The K-factor gives the maximum 3dB bandwidth f3dBmax as follows [36]: f 3dB max =

#200669 - $15.00 USD (C) 2014 OSA

2 2π K

(8)


The inset of Fig. 9(b) shows the extracted damping factor as a function of f R2 for the 30μm microspiral disk laser with a waveguide width of 2 μm. We obtain a K-factor of 0.32 ns, which suggests a maximum intrinsic 3dB bandwidth of 28 GHz using Eq. (8). However, our measurements reveal that the measured maximum bandwidths of our microspiral disk lasers (with lasing peak wavelengths within the EDFA gain band) are primarily limited by the RC bandwidth. By reducing the microdisk area along with a proper RF design [37], we believe it is feasible to further improve the direct-modulation bandwidth towards 25 GHz.

4.3 Data modulation We measure the data modulation of the 30μm-radius microspiral disk laser with a waveguide width of 2 μm. We use a 30Gbit/s pseudorandom bit sequence (PRBS) generator to apply to the laser non-return-zero (NRZ) signals with a pattern length of 231− 1. The microlaser is biased at 90 mA and modulated by 10 Gbit/s and 15 Gbit/s signals with a peak-to-peak voltage (Vpp) of 2 V. We measure the modulated laser output waveforms using a 20GHz digital sampling oscilloscope. Figures 10(a) and 10(b) show the measured optical waveforms at 10 Gbit/s and 15 Gbit/s, respectively. Figures 10(c) and 10(d) show the measured eye diagrams at 10 Gbit/s and 15 Gbit/s, respectively. Our measured eye diagrams show an extinction ratio (ER) of 5 dB at 10 Gbit/s and 15 Gbit/s. We attribute the noise in the measured eye diagrams to the EDFA ASE noise within the tunable filter pass-band. The fiber-coupled lasing peak power of ~-35 dBm is only slightly higher than the EDFA ASE noise level of ~-40 dBm.

Fig. 10. Measured direct-modulated laser output waveforms and eye diagrams for the 30μmradius microspiral lasers with w = 2 μm and the outer-ring injection at (a), (c) 10 Gbit/s and (b), (d) 15 Gbit/s. Table 6. Calculated energy consumption per bit @ 10 Gbit/s for different waveguidecoupled microspiral disk lasers (unit: pJ/bit)



Middle-ring (Rp = 20 μm) 9.0 13.3 Middle-ring (Rp = 28 μm) -

Disk injection 16.8 19.1 Disk injection 27.4 -

We estimate the energy consumption per bit for data modulation of the lasers as follows:

#200669 - $15.00 USD (C) 2014 OSA


Energy consumption per bit =

VI Data rate

(9)

where V and I are the applied DC-bias voltage and injection current, respectively. We consider the data rate at 10 Gbit/s. This corresponds to a bandwidth of about 7.6 GHz assuming the available data rate is about 1.3 × of the 3dB bandwidth [38]. Table 6 summarizes the calculated energy consumption per bit at 10 Gbit/s for different waveguidecoupled microspiral disk lasers. The outer-ring injection lasers always show the lowest calculated energy consumption per bit. We only include the calculated energy consumption per bit for those lasers with the measured maximum bandwidths above 7 GHz (see Table 4). Table 7 compares this work with the state-of-the-art on direct-modulated microresonator lasers in the literature. We demonstrate the highest data transmission rate of 15 Gbit/s for the direct-modulated semiconductor microresonator lasers to our knowledge. We obtain a 3dB bandwidth exceeding 10 GHz, which is larger than most of the demonstrated devices. However, the energy consumption per bit is relatively large because of the large size of our lasers, which are mainly limited by the resolution of our contact photolithography. Table 7. Comparison of direct-modulated semiconductor microresonator lasers

Cavity shape

This work

Injection design

Microspiral disk

A.Kapsalis, et al. [30] L.Liu, et al. [31] X.M.Lv, et al. [33]

Microring Circular disk Circular disk

Ringshaped Diskshaped Ringshaped Diskshaped Diskshaped

Bandwidth (GHz)

Data rate (Gbit/s)

Calculated energy consumption per bit at 10 Gbit/s (pJ/bit)

10.7

15

8.7

8.9

-

19.1

70

-

7

-

3.75

3.5

3

-

10-15

13

12.5

2-4

Platform

Radius (μm)

AlGaInAs QWs

30

InGaAs QWs InAsP QWs on SOI AlGaInAs QWs

5. Conclusion We have demonstrated room-temperature continuous-wave electrically injected AlGaInAs/InP waveguide-coupled microspiral disk lasers with spatially selective injection. Compared to the conventional disk injection design for microdisk lasers, our outer-ring injection design for the 30μm-radius microspiral disk lasers with a 2μm-wide directly gapless coupled waveguide has shown a reduced lasing threshold current from 42 to 28 mA, an improved slope-efficiency from 0.269 to 0.724 mW/A, an improved 3dB bandwidth from 8.9 to 10.7 GHz with an open eye diagram at 15 Gbit/s, and suggested a reduced energy consumption per bit at 10 Gbit/s from 19.1 to 8.7 pJ/bit. In order to further scale down the device and connect it to silicon-based on-chip optical links, we are now working on III-V-onSi microspiral disk lasers using Benzocyclobutene (BCB) bonding [31]. Acknowledgments This work was supported by the National Science Foundation of China (NSFC) and the Research Grants Council (RGC) of the Hong Kong Special Administrative Region under Grant N_HKUST606/10. Yuede Yang acknowledges the support from The Hong Kong Scholars Program 2011-2013.

#200669 - $15.00 USD (C) 2014 OSA


off chip optical interconnects.

All-optical interconnects using Airy beams.

Sol-Gel Derived Active Material for Yb Thin-Disk Lasers.

Advanced modulation formats for high-performance short-reach optical interconnects.

Precise Spatially Selective Photothermolysis Using Modulated Femtosecond Lasers and Real-time Multimodal Microscopy Monitoring.

Lasers for coherent optical satellite links with large dynamics.

Plasma optical modulators for intense lasers.

A multi-pass pumping scheme for thin disk lasers with good anti-disturbance ability.

mode-division multiplexed on-chip optical interconnects.

Thermally insensitive determination of the linewidth broadening factor in nanostructured semiconductor lasers using optical injection locking.

Fast phase response and chaos bandwidth enhancement in semiconductor lasers subject to optical feedback and injection.

Photonic microwave stabilization for period-one nonlinear dynamics of semiconductor lasers using optical modulation sideband injection locking.

Multimode regimes in quantum cascade lasers with optical feedback.

Monolithic III-V on Silicon Plasmonic Nanolaser Structure for Optical Interconnects.

Suspended ultra-small disk resonator on silicon for optical sensing.

Q-switched Nd:YAG optical vortex lasers.

Visible-Frequency Metasurface for Structuring and Spatially Multiplexing Optical Vortices.

intra chip interconnects.

High Photoresponsivity Ge-dot PhotoMOSFETs for Low-power Monolithically-Integrated Si Optical Interconnects.

Thermally-drawn fibers with spatially-selective porous domains.

Rapid measurement of T1 with spatially selective pre-inversion pulses.

Reconfigurable spatially-diverse optical vector network analyzer.

Birefringence compensated silicon nanowire arrayed waveguide grating for CWDM optical interconnects.

All-optical spectral linewidth reduction of lasers for coherent optical communication.