Development of low-loss photonic components in the ultraviolet (UV) band will open new prospects for classical and quantum optics. Compared with other integrated platforms, aluminum nitride (AlN) is particularly attractive as it features an enormous bandgap of and intrinsic and susceptibilities. In this work, we demonstrate a record quality factor of (optical loss ) at 390 nm based on single-crystalline AlN microrings. The low-loss AlN waveguide represents a significant milestone toward UV photonic integrated circuits as it features full compatibility for future incorporation of AlGaN-based UV emitters and receivers. On-chip UV spectroscopy, nonlinear optics, and quantum information processing can also be envisioned.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Integrated photonic components have gained remarkable progress in the telecom band, thanks to the maturity of silicon photonics . Nonetheless, extending the operating wavelength to the ultraviolet (UV) range still remains non-trivial, yet is significant for on-chip UV spectroscopy, biochemical sensing, nonlinear optics, and quantum information processing [2–4]. Since the relatively narrow bandgap () of silicon limits its utility at short wavelengths, the development of a wide-bandgap photonic platform, such as aluminum nitride (AlN), is desirable. It is known that AlN exhibits a large bandgap of , and is thereby transparent to light of wavelength () above 200 nm and free of two-photon absorption above 400 nm. This wideband transparency allows AlN photonic components to interact with ions in the UV and visible regions, including ytterbium () at 369.5 nm, nitrogen vacancy centers in diamond at 637 nm, and rubidium () at 778.1 nm, as required for on-chip quantum computing  and precision optical clocks . Compared with silicon nitride and diamond, the intrinsic susceptibility of AlN also makes it a competing platform for on-chip nonlinear interactions [8–10] and quantum frequency conversion [11,12].
Although polycrystalline AlN has proved as a viable platform for chip-scale nonlinear optics and optomechanics , it suffers from a large waveguide attenuation of 650 dB/cm at 400 nm due to defect-related absorption and scattering , thereby hampering its applications at UV wavelengths. Recently, an impressive intrinsic -factor () of 24 k at 369.5 nm was measured in nanocrystalline AlN microrings via a swept UV laser , yet is likely still limited by the roughness of the waveguide sidewall and the nanocrystalline film morphology. In contrast, single-crystalline AlN epitaxially grown on sapphire exhibits superior film quality and poises to become a versatile photonic platform at short wavelengths. Compared with other integrated platforms, crystalline AlN shows unprecedented advantages to leverage UV photonic integrated circuits for implementing passive/active optoelectronic functionalities by incorporating Al(Ga)N-based UV emitters and detectors . In the telecom band, a high of up to has been recorded in crystalline AlN microrings , yielding nonlinear photonic devices such as broadband Kerr frequency combs  and high-efficiency Raman lasers . Nonetheless, the current -factors of recorded in crystalline AlN-based UV resonators at 310–411 nm [20–24] are far from the maximum attainable -factors as limited by the Rayleigh scattering loss ().
In this Letter, we investigate the waveguide losses in the UV band for a high-quality single-crystalline AlN film. Based on microring architectures, we achieve a record of 210 k at 390 nm (optical loss of ). We also show UV transmittance, which covers a 1 nm range for both transverse electric (TE) and transverse magnetic (TM) modes. By comparing the -factors of AlN microrings of different widths, the influence of sidewall scattering is confirmed. A reduced Rayleigh scattering is verified by feeding the chip with 455 nm light, where an improved of 398 k (optical loss of ) is attained. The high- AlN microrings are promising for cavity-enhanced second harmonics  and atom-photonic integration  in the UV and blue regions.
Figure 1(a) depicts the cross sections and simulated modal profiles of AlN-on-sapphire microrings. The fundamental modes at 390 nm are well confined within the resonators, but become more susceptible to sidewall scattering at a smaller waveguide width. Due to the narrow sweeping range () of our UV laser source (limited by the phase-matching bandwidth of second-harmonic conversion), we adopt a relatively large radius of 30 μm to obtain a small free spectral range (FSR) such that multiple UV resonances can be characterized. Meanwhile, we utilize two cascaded microrings with slightly varied radii to provide resonant features distinguishable from the transmittance background. For effective waveguide-to-microring coupling at the challenging UV band, we employ a weakly tapered gap coupler to wrap the microring at an angle of 20°  and optimize the center gap (0.12–0.15 μm) using a FIMMPROP software. Although the AlN resonators manifest a strong normal dispersion in the UV region, which suppresses efficient four-wave mixing, nonlinear photonic applications can be accessed by exploiting the intrinsic susceptibility of AlN (Supplement 1).
In the experiment, single-crystalline AlN of a thickness of is grown on -plane sapphire by metal-organic chemical vapor deposition. The microrings and associated feeding waveguides are then defined by a 100 kV electron-beam lithography (EBL) system (Raith EBPG 5000+) with a negative FOx-16 resist. Since the AlN-on-sapphire wafer is highly insulating, we spin 300 nm of poly(4-styrenesulfonic acid) (PSSA) on top of the FOx-16 resist and then sputter 10 nm of gold to mitigate charging effects. The PSSA is water-soluble and helps remove the gold after EBL writing. To ensure a high-contrast e-beam pattern, which is particularly important for UV resonators with small coupling gaps, we utilize a 25% tetramethylammonium hydroxide developer . Then the pattern is transferred to the AlN layer by optimized -based inductively coupled plasma etching. Finally, the wafer is embedded in by plasma-enhanced chemical vapor deposition and is then cleaved to expose the waveguide facets.
Figure 1(b) presents a scanning electron microscope (SEM) image of one of the cascaded AlN microrings. Figure 1(c) is a zoom-in view of Fig. 1(b), highlighting a weakly tapered gap coupler (center separation, 0.14 μm), which is constructed by tapering the coupling waveguide width from 0.32 to 0.35 μm when approaching the microring from both sides. To ensure effective fiber-to-chip coupling for both TE and TM modes, the coupling waveguide is finally tapered to an expanded width of 1 μm at the chip facets, as shown in Fig. 1(d).
Figure 1(e) illustrates the experimental setup for characterizing the devices. We construct a sweeping UV laser by frequency-doubling of a Ti-sapphire laser (M2 SolsTiS, 700–1000 nm) to around 390 nm via a lithium triborate (LBO) crystal. The wavelength of the Ti-sapphire laser is precisely determined with a high-resolution (0.1 pm) wavemeter. Subsequently, we use a UV fiber port to collect the UV beam into a single-mode UV fiber, followed by a UV collimator and a UV lens for coupling the light into the chip. To determine the input polarization, we adjust a fiber polarization controller while monitoring the power after a UV linear polarizer (not shown). The output light from the chip is collected with a lensed fiber and then sent to a UV-sensitive photodetector (PD). Finally, a data acquisition card is employed to simultaneously record the transmittance from the PD and the wavelength from the Ti-sapphire laser.
It is noteworthy that the UV beam generated exhibits a wavelength-dependent output orientation [indicated by purple dashed lines in Fig. 1(e)] due to the angle-dependent phase matching of the LBO to the pump beam, which induces a varied coupling efficiency of the free-space UV beam into the fiber port upon sweeping the wavelength. For instance, when optimizing the alignment at 390 nm, we observe a large transmission background variation of in the wavelength range of 389.5–390.5 nm (Supplement 1). To address this issue, we divide the 1 nm scanning range into several segments and optimize the alignment separately to suppress the power fluctuation in the fiber.
Figure 2(a) shows the concatenated transmittance of the AlN microring (width, 0.8 μm; gap, 0.14 μm) at TM polarization, where a sweeping rate of 2 GHz/s is employed (and hereafter) to ensure a high wavelength resolution. Note that the background fluctuation in the full scanning region is suppressed by 13 dB when applying the approach described above. Due to the adoption of two cascaded microrings, we observe four cascaded dips (magenta and blue) spaced by their predicted FSRs (see Supplement 1), corresponding, respectively, to on-resonance fundamental and first-order modes, while the large background resonances are attributed to the power fluctuation during the UV laser scan. Based on Figs. 2(b) and 2(c), the loaded -factors () of resonances I and II in Fig. 2(a) are, respectively, 86 k and 103 k, corresponding to of 156 k and 189 k in an under-coupled condition. The high extinction ratio (ER) of 22 dB at a gap of 0.14 μm suggests a nearly critical coupling for the mode. Figure 2(d) shows the transmittance recorded at TE polarization, indicative of cascaded fundamental resonances with respective FSRs of 657.4 and 657.9 GHz. In Figs. 2(e) and 2(f), the of resonances I and II in Fig. 2(d) are, respectively, 144 k and 136 k, corresponding to of 186 k and 210 k. The highest -factor at 390 nm attained in this work is a six-fold improvement over the result ( at 400 nm) in nanocrystalline AlN .
The propagation loss is then derived to be for the mode () based on the expression  ( and being the optical frequency and microring radius, respectively). This value is significantly smaller than the results of 75 dB/cm at 369.5 nm for nanocrystalline AlN microrings  and 650 dB/cm at 400 nm for polycrystalline AlN straight waveguides . We believe that the low UV propagation loss achieved in this work is attributed to the excellent film quality of the single-crystalline AlN, the engineered microring geometries, and the optimized fabrication process. The maximum -factors that can be achieved in our AlN chips are found to be limited by the sidewall and Rayleigh scattering losses, as described later.
We assess the sidewall-induced waveguide loss by comparing the -factors of devices with different widths (0.4, 0.6, and 0.8 μm). Figure 3(a) shows the measured resonance of the mode for the 0.6-μm-wide microring. The is measured to be 122 k, corresponding to a of 175 k, which is slightly lower than the value in Fig. 2 for the 0.8-μm-wide device. Nonetheless, a notably reduced of 83 k is observed for the mode in the 0.4-μm-wide microring [Fig. 3(b)]. We have then plotted the extracted against ring width for both and modes in Fig. 3(c), in which a degraded -factor with ring width is clearly observed. We further plot the corresponding against ring width in Fig. 3(d). Notwithstanding a small ring width of 0.4 μm, we still attain a low of and for, respectively, the and modes. The experimental results suggest that the sidewall scattering loss at 390 nm can be mitigated with a relatively large waveguide width in our AlN platform.
Theoretically, in optical waveguides is given by Supplement 1). This is consistent with the extracted of the mode in Fig. 3(d), whereas the of the mode tends to follow the same relationship except at a width of 0.8 μm. The discrepancy can be interpreted when accounting for an over-coupled condition at this ring width, which yields a of 226 k in Fig. 2(c), corresponding to of 7.7 dB/cm for the mode. It also agrees with the fact that the mode is less susceptible to sidewall roughness than the mode, and is, thus, supposed to exhibit a lower propagation loss. When employing a ring width of above 0.8 μm, a saturation of the -factor is anticipated from the simulation (Supplement 1).
We then explore the influence of Rayleigh scattering by characterizing the -factors of the 0.8-μm-wide AlN microring using 455 nm light. We employ another LBO crystal in Fig. 1(e) to attain a continuously tunable blue laser (center at 455 nm) and determine the input polarization with a linear polarizer that covers this region. Figures 4(a), 4(b), and 4(c) depict the measured resonances of the mode at gaps of 0.13, 0.14, and 0.15 μm, respectively. The increasing ER with gap indicates an over-coupled microring at 455 nm. At a gap of 0.14 μm, the extracted of 398 k in Fig. 4(b) is two-fold higher than that at 390 nm [Figs. 2(e) and 2(f)], corresponding to a of based on the recorded FSR of for the mode. The high -factors and the low propagation loss of the AlN resonators at 455 nm are also the state of the art up to date. By including the -factors attained from our visible and near-infrared AlN microrings, we derive an appropriate (Supplement 1), suggesting a Rayleigh scattering-dominated loss () in the UV and blue AlN microrings (width, 0.8 μm). In Fig. 4(d), we summarize the reported at a wavelength below 480 nm for different types of AlN microcavities. Clearly, the of the AlN microrings at 390 and 455 nm in this work are greatly improved over the values reported in the literature.
In conclusion, we demonstrate ultra-high- UV microrings based on single-crystalline AlN and investigate possible loss mechanisms. We achieve a record of 210 k at 390 nm, corresponding to a low propagation loss of . Except for sidewall roughness, we attribute the Rayleigh scattering to the loss in the UV region based on a higher of 398 k () obtained at 455 nm. Our results pave the avenue to leveraging crystalline AlN-based integrated photonic components in the UV and blue regions for a variety of nonlinear and quantum photonic applications by taking advantage of its excellent optical property and intrinsic and susceptibilities.
Defense Advanced Research Projects Agency (DARPA) (W31P4Q-15-1-0006, HR0011-16-C0118); Air Force Office of Scientific Research (AFOSR) (FA9550-15-1-0029); Army Research Office (ARO) (W911NF-14-1-0563); National Science Foundation (NSF) (EFMA-1640959); David and Lucile Packard Foundation.
The authors thank Michael Power and Dr. Michael Rooks for assistance in the device fabrication.
See Supplement 1 for supporting content.
1. A. Rickman, Nat. Photonics 8, 579 (2014). [CrossRef]
2. Y. Ozaki and S. Kawata, Far- and Deep-Ultraviolet Spectroscopy (Springer, 2015).
3. M. Nissen, B. Doherty, J. Hamperl, J. Kobelke, K. Weber, T. Henkel, and M. A. Schmidt, Sensors 18, 478 (2018). [CrossRef]
4. R. Krischek, W. Wieczorek, A. Ozawa, N. Kiesel, P. Michelberger, T. Udem, and H. Weinfurter, Nat. Photonics 4, 170 (2010). [CrossRef]
5. H. Yamashita, K. Fukui, S. Misawa, and S. Yoshida, J. Appl. Phys. 50, 896 (1979). [CrossRef]
6. C. Monroe and J. Kim, Science 339, 1164 (2013). [CrossRef]
7. S. B. Papp, K. Beha, P. Del’Haye, F. Quinlan, H. Lee, K. J. Vahala, and S. A. Diddams, Optica 1, 10 (2014). [CrossRef]
8. X. Guo, C.-L. Zou, and H. X. Tang, Optica 3, 1126 (2016). [CrossRef]
9. J. B. Surya, X. Guo, C.-L. Zou, and H. X. Tang, Opt. Lett. 43, 2696 (2018). [CrossRef]
10. X. Guo, C.-L. Zou, H. Jung, and H. X. Tang, Phys. Rev. Lett. 117, 123902 (2016). [CrossRef]
11. X. Guo, C.-L. Zou, C. Schuck, H. Jung, R. Cheng, and H. X. Tang, Light: Sci. Appl. 6, e16249 (2017). [CrossRef]
12. H. Rütz, K.-H. Luo, H. Suche, and C. Silberhorn, Phys. Rev. Appl. 7, 024021 (2017). [CrossRef]
13. C. Xiong, W. H. Pernice, X. Sun, C. Schuck, K. Y. Fong, and H. X. Tang, New J. Phys. 14, 095014 (2012). [CrossRef]
14. M. Stegmaier, J. Ebert, J. Meckbach, K. Ilin, M. Siegel, and W. Pernice, Appl. Phys. Lett. 104, 091108 (2014). [CrossRef]
15. T.-J. Lu, M. Fanto, H. Choi, P. Thomas, J. Steidle, S. Mouradian, W. Kong, D. Zhu, H. Moon, K. Berggren, J. Kim, M. Soltani, S. Preble, and E. Dirk, Opt. Express 26, 11147 (2018). [CrossRef]
16. D. Li, K. Jiang, X. Sun, and C. Guo, Adv. Opt. Photon. 10, 43 (2018). [CrossRef]
17. X. Liu, C. Sun, B. Xiong, L. Wang, J. Wang, Y. Han, Z. Hao, H. Li, Y. Luo, J. Yan, T. W. Wei, Y. Zhang, and J. Wang, Opt. Express 25, 587 (2017). [CrossRef]
18. X. Liu, C. Sun, B. Xiong, L. Wang, J. Wang, Y. Han, Z. Hao, H. Li, Y. Luo, J. Yan, T. W. Wei, Y. Zhang, and J. Wang, ACS Photon. 5, 1943 (2018). [CrossRef]
19. X. Liu, C. Sun, B. Xiong, L. Wang, J. Wang, Y. Han, Z. Hao, H. Li, Y. Luo, J. Yan, T. W. Wei, Y. Zhang, and J. Wang, Optica 4, 893 (2017). [CrossRef]
20. D. Sam-Giao, D. Néel, S. Sergent, B. Gayral, M. Rashid, F. Semond, J.-Y. Duboz, M. Mexis, T. Guillet, C. Brimont, S. David, X. Checoury, and P. Boucaud, Appl. Phys. Lett. 100, 191104 (2012). [CrossRef]
21. S. Sergent, M. Arita, S. Kako, K. Tanabe, S. Iwamoto, and Y. Arakawa, Appl. Phys. Lett. 101, 101106 (2012). [CrossRef]
22. F. Tabataba-Vakili, I. Roland, T.-M. Tran, X. Checoury, M. El Kurdi, S. Sauvage, C. Brimont, T. Guillet, S. Rennesson, J.-Y. Duboz, F. Semond, B. Gayral, and P. Boucaud, Appl. Phys. Lett. 111, 131103 (2017). [CrossRef]
23. M. Mexis, S. Sergent, T. Guillet, C. Brimont, T. Bretagnon, B. Gil, F. Semond, M. Leroux, D. Néel, S. David, X. Chécoury, and P. Boucaud, Opt. Lett. 36, 2203 (2011). [CrossRef]
24. M. Bürger, G. Callsen, T. Kure, A. Hoffmann, A. Pawlis, D. Reuter, and D. As, Appl. Phys. Lett. 103, 021107 (2013). [CrossRef]
25. J. D. Thompson, T. Tiecke, N. P. de Leon, J. Feist, A. Akimov, M. Gullans, A. S. Zibrov, V. Vuletić, and M. D. Lukin, Science 340, 1202 (2013). [CrossRef]
26. D. T. Spencer, J. F. Bauters, M. J. Heck, and J. E. Bowers, Optica 1, 153 (2014). [CrossRef]
27. W. Henschel, Y. Georgiev, and H. Kurz, J. Vac. Sci. Technol. B 21, 2018 (2003). [CrossRef]
28. R. J. Deri and E. Kapon, IEEE J. Quantum Electron. 27, 626 (1991). [CrossRef]