Open Access
Add to BibSonomyAbstract
We present a study on the effects of inert ion implantation of Silicon-On-Insulator (SOI) racetrack resonators. Selective ion implantation was used to create deep-level defects within a portion of the resonator. The resonant wavelength and round-trip loss were deduced for a range of sequential post-implantation annealing temperatures from 100 to 300 °C. As the devices were annealed there was a concomitant change in the resonance wavelength, consistent with an increase in refractive index following implantation and recovery toward the pre-implanted value. A total shift in resonance wavelength of ~2.9 nm was achieved, equivalent to a 0.02 increase in refractive index. The excess loss upon implantation increased to 301 dB/cm and was reduced to 35 dB/cm following thermal annealing. In addition to providing valuable data for those incorporating defects within resonant structures, we suggest that these results present a method for permanent tuning (or trimming) of ring resonator characteristics.
©2011 Optical Society of America
1. Introduction
Silicon photonics continues to attract a great deal of attention as a likely candidate technology for the implementation of optical interconnections [1]. Together with its compatibility with standard Complimentary-Metal-Oxide-Semiconductor (CMOS) processing methods, the scalability of silicon has a potential for terabit data transmission when deployed inside a Wavelength Division Multiplexing (WDM) optical communication system paradigm. To this end, the SOI ring (or race-track) resonator is a key building block offering the ability to modulate, route and detect individual channels.
The deliberate introduction of ion implantation induced defects into silicon waveguides has been demonstrated to expand the optical functionality of silicon-based optical integrated circuits, for example through the modification of carrier lifetime [2] and the fabrication of waveguide photodetectors with sensitivity to sub-band wavelengths [3].
This report presents quantitative data on the impact of low concentrations of deep-levels on the real part of the refractive index of silicon waveguides for wavelengths around 1550 nm, obtained from the study of the effects of inert ion implantation on race-track resonators. The data presented is of general interest to those wishing to incorporate defects into ring resonators, such as in the case of the resonant detectors, recently reported by Doylend et al. [4] Logan et al. [5] and Preston et al. [6].
We also note the potential for defect implantation as a means of device trimming. One of the major issues facing the implementation of ring resonator structures is fabrication tolerance. These devices rely on evanescent field coupling, which strongly depends on the waveguide width and separation from the ring. Small variations that arise in fabrication prevent a resonant structure achieving its design specification within accepted tolerance. In this case, active tuning, often through application of the thermo-optic effect, is required to modify the resonance. This can be accomplished by local heating of the ring with a resistive metal strip [7]. Another method, limited to rib waveguides, is to use a p-i-n diode structure over the waveguide [8,9]. Applying a bias changes the concentration of free carriers in the ring, thus altering the refractive index [10]. These methods yield adequate tuning but introduce added complexity to the device. For large devices containing multiple rings [11] active tuning becomes less attractive as each device requires its own independent tuner and control circuit. The tuners consume chip area and power, reducing the advantages gained by using resonant structures over other geometries such as Mach-Zender Interferometers (MZI) [8].
We demonstrate that the resonance location of silicon race-tracks can be shifted by introducing deep-levels within the silicon bandgap via ion implantation and subsequent annealing. This represents a method to passively, and permanently tune a resonant structure. The tuning is accompanied by excess loss, but in many cases this trade-off may be within device tolerance limits. These first results on defect mediated tuning suggest further investigation to explore the limits of this technique, particularly in conjunction with selective annealing as can be achieved via focussed laser activation [12].
2. Device description and experimental procedure
2.1 Formation of the waveguide structures
The structures used in this experiment were double-bus SOI racetrack resonators, consisting of 30 µm radius curves and either 15 or 40 µm coupling regions. Fabrication was carried out with the passive SOI platform of IMEC in Leuven, Belgium, and facilitated via CMC Microsystems. The SOI wafers were comprised of a 220 nm thick top layer of silicon over a 2 µm buried oxide, and waveguides were patterned with 193 nm deep UV lithography [13].
Waveguide width was varied during fabrication by adjusting the photo-exposure; those used in this study were estimated to be approximately 450 nm wide. Coupling in and out of the waveguides was achieved through shallow-etch integrated grating couplers [14].
2.2 Ion implantation
Ion implantation was done at relatively high energy so that the implant species would penetrate through to the oxide, leaving only structural, point defects in the waveguide layer. Photolithography was used to define implantation windows as to avoid the coupling regions. By avoiding the couplers we maintain the same coupling coefficients, which allows the change in loss and the Quality factor (Q) from the implant to be observed. Figure 1 (a) shows a scanning electron microscope (SEM) image of a racetrack resonator, while Fig. 1 (b) shows the photomask layout of the implant windows.
Fig. 1 (a) SEM image of a racetrack resonator (b) The ion implantation photomask overlaid on the racetrack design. The implantation windows avoid the couplers and bus waveguides. (SEM image taken at the Canadian Center for Electron Microscopy)
Multiple chips were implanted with varying implant dosage below the amorphization threshold [15]. The ions implanted were either boron at 350 keV or silicon at 700 keV, energies sufficient for the ions to reside in the oxide layer. Subsequently in this work each chip will be referred to in a manner which encompasses the implantation procedure it underwent in terms of implant dosage, implant species, and a letter to signify variations in resonator length; for example device A/1.75E14-Si refers to a chip with an implant dosage of 1.75x1014 cm−2, with silicon ions (at 700 keV). The prefix ‘A’ designates a resonator circumference of 218 µm, while ‘B’ indicates 268 µm.
2.3 Characterization
After the implantation, cleaning the masking resist involved chip immersion in a H2SO4 + H2O2 or ‘piranha etch’ solution. The heat from this reaction formed the baseline annealing temperature of 100 °C. Once implanted and cleaned, each resonator was annealed in sequential steps of 25 °C, to a maximum of 300 °C. Measurements between annealing steps were taken using a New Focus Model 6427 external cavity tunable diode laser, with a cleaved polarization maintaining single mode fiber to couple light to the chip. A single mode fiber was also used to couple light from the chip, where a Newport 818-1S1 power meter took readings. Total coupling loss in and out of the device is estimated to be 11 dB. Repeated measurements showed a maximum temperature drift of ± 0.1 nm, this error is small relative to the total resonance shift observed.
3. Results
3.1 Experimental results
Ion implantation and subsequent annealing has two pronounced effects on the resonator characteristics. The first is a resonance shift observed after implantation, which decreases in magnitude as the ring is annealed. The resonance wavelength thus progressively moves to lower wavelengths after exposure to higher annealing temperatures. The second effect is that the Q of the ring is reduced after implantation. This is unsurprising given the previously reported increase in optical loss with increasing concentration of ion implantation defects in a silicon waveguide [16]. As the annealing temperature is increased we observed the Q increase towards the pre-implanted value.
An example of optical spectra for a single chip is shown in Fig. 2 where the transmitted power of device A/3E14-Boron after various post-implantation annealing temperatures is plotted.
Fig. 2 Resonance shifting for device A/3E14-Boron. The trend of this result is representative of other devices, including those that received silicon. As the annealing temperature is increased the resonance peaks shift to lower wavelengths and the Q-factor increases.
The effect of implantation dose on the resonance shift and propagation loss is summarised in Fig. 3 . In Fig. 3(a), we show a comparison of resonance shift versus annealing temperature for three nominally identical resonators (from three different chips) which received different implant dosage. There is a clear increase in the resonance shift with increasing implantation dose, indicating a direct relationship between the real component of the refractive index (RI) and the concentration of deep levels introduced by the implantation. The shift in RI is positive after ion implantation, with a progressive recovery toward the value for unimplanted waveguides as annealing is increased. This is discussed in detail in section 3.2 Modelling the RI shift.
Fig. 3 Resonance shift (relative to the implanted state), as a function of annealing: (a) three identical devices, were implanted with different doses of silicon at 700 keV; (b) the total optical loss of the race-tracks. Note: Marker size is indicative of the uncertainty.
To extract the propagation loss from measured data such as that shown in Fig. 2, the resonance spectra were fitted with the analytical through-port transmission function for double bus resonators with identical couplers, Eq. (1):
where Pin is input power, Pout is output power, t is the transmission coefficient, γ is the insertion loss, A = exp(αL) with L as the resonator length, α is the loss coefficient, and δ = 2π/(λneffL) is the phase. After fitting the results, it was found that the transmission coefficients remained within the fitting error (as expected given that the coupling region was masked during implantation), but the loss decreased with annealing. Figure 3(b) shows the extracted values for excess propagation loss as a function of annealing temperature, for the same devices as Fig. 3 (a). It is instructive to compare the loss values for the lowest temperature anneal (which is likely to be representative of the loss in the as-implanted condition) to values extracted from the empirical model proposed by Foster et al. [16]. The usefulness of the Foster model is that it allows the comparison of expected loss for a wide range of implantation conditions (i.e. variations in ion species, ion dose, and ion energy). For the silicon implanted samples the values of expected loss would be 23, 96 and 152 dBcm−1 for increasing ion dose. These compare to measured values (shown in Fig. 3b) of 71, 239 and 301dBcm−1. The values measured here are thus approximately twice those which might be expected from the empirical model proposed by Foster et al. We feel that this discrepancy is not wholly unexpected given the different experimental conditions used by Foster et al. in their work, from which the empirical model was developed. We also note that the Foster model has previously underestimated experimental optical loss [16].As the devices are annealed the loss trends downward, in a manner consistent with the shift in resonance. The response of resonance wavelength and propagation loss to annealing is similar. This is not surprising given that these properties represent the real and complex components of the refractive index respectively, related via the Kramer-Kronig relations. The precipitous decrease in both properties at temperatures > 200°C suggest that the deep-level responsible for the initial post implantation change is the silicon divacancy, previously observed to be the primary optically active defect formed after relatively low dose inert ion implantation in a silicon waveguide [17].
The increase in propagation loss may also be represented by the Q-factor of the resonator. This was extracted by curve fitting and plotted in Fig. 4 for two representative devices. An increase in blue shift of the resonance peak (compared with the as-implanted device annealed at 100°C) corresponds to an increasing Q as the defects are removed.
Fig. 4 Quality Factor and resonance shift (relative to the implanted state) versus annealing temperature for resonator (a) A/1.5E14-Si and (b) B/1.25E12-Si
3.2 Modelling the RI shift
The change of the effective index of the SOI waveguide, versus the change in the silicon index, is plotted at a representative wavelength of 1563 nm in Fig. 5 (a) . The slope is dneff / dnSi = 1.177. This quantity is also known as the effective index susceptibility, previously used to describe the effective index changes in SOI waveguides due to the core index changes (due to temperature) or cladding changes [18]. Figure 5 (b) shows the mode profile is increasingly localized in the centre of the waveguide as index increases. This is similar to the effect of modal dispersion and gives rise to an increase in the effective index that is surprisingly 17.7% larger than the change in the silicon index itself.
Fig. 5 a) The change in the waveguide effective index as a function of the change in the silicon index of refraction for a 450 x 220 nm waveguide at 1563 nm; b) The mode profile in the in-plane direction, at the centre of the waveguide (at a height of 110 nm), for different silicon indices. n Si = 3.48 (red) and 3.47 (blue), n SiO2 = 1.444. 450 x 220 nm waveguide. This waveguide has an ng (1563 nm) = 4.55.
Here, we derive expressions for the change in the resonance condition as a function of the change in index of refraction. First, consider a resonator with a uniform non-dispersive medium, with a length L. For a mode index m, the mode condition is given by Eq. (2).
A change in index of refraction leads to a change in resonance wavelength, Eq. (3):
In a resonator with dispersion (material, waveguide), the mode condition is given by Eq. (4):
and the free-spectral range is Eq. (5):To analyze the effect of the change in silicon index of refraction, we need to consider the change in the mode profile that changes the index of refraction. Thus, we can write the expression for effective index as, Eq. (6):
Using the mode condition (Eq. (4), we equate the mode at the initial wavelength to the shifted wavelength, shown in Eq. (7).
This allows us to find the wavelength shift, which simplifies to Eq. (8):
We note that in the case where the effective index change is equal to the silicon index change, i.e., neglecting effective index susceptibility, the wavelength shift reduces to Eq. (9). This relation was presented previously in [19]:
In essence, the shift in resonance wavelength is determined by three factors: a) the shift in the material index of refraction, b) the material and waveguide dispersion – a change in wavelength causes a further change in effective index, and c) the change in material index changes the mode profile resulting in an additional effective index change. It should be noted that the same effect is found in the temperature dependence of waveguide-based resonators [18,20].
Using Eq. (8) we can now calculate the maximum index shift for the implanted resonators as a function of dose. The group index, n g = 4.545, was found from the resonator free-spectral range, length and operating wavelength λ0 = 1563nm. From Fig. 3 the doses used were 1.25 x 1012 cm−2, 7.5 x 1013 cm−2 and 1.75 x 1014 cm−2. The resonance wavelengths were shifted, respectively, Δλ = 0.7 nm, 2.2 nm and 2.9 nm. With approximately one third of the waveguide being implanted, this leads to increases in the silicon refractive index of Δn Si = 0.005, 0.016 and 0.021 respectively.
3.3 Summary
The results described above provide an important insight into the impact of defects on the characteristics of a ring resonator. The deliberate introduction of defects into a resonant structure to date has been limited to the development of resonant enhanced detectors [4,5]. However, we suggest here that defect injection and selective annealing offers an intriguing solution to post fabrication trimming of the response of resonators across a complete wafer. Clearly, blanket ion implantation and traditional thermal annealing will simply result in a systematic shift in the characteristics of all resonators on a single wafer. However, selective laser-assisted annealing may provide local removal of defects on an areal scale of tens of square microns [12]. This would then facilitate test and trimming at the wafer scale of each resonator. The most obvious disadvantage of this approach is the increase in round-trip loss with defect introduction. Future work will help elucidate the limitations of this trimming technique.
4 . Conclusion
SOI double bus racetrack resonators have undergone selective ion implantation to introduce deep level defects into the silicon waveguide. We have observed a shift in resonance of nearly 3 nm after ion implantation of silicon at an energy of 700 keV and a dose of 1.5x1014 cm−2. The magnitude of the resonance shift corresponds to the excess loss within the resonator. This is evidence for a refractive index increase of 0.021 associated with the concentration of structural point defects within the silicon waveguide. This shift in the real part of the refractive index, when combined with selective laser assisted annealing, presents a method for permanent tuning of resonators on the wafer level scale.
Acknowledgements
The authors would like to thank Jack Hendriks at the University of Western Ontario for ion implantation, Chris Brooks at McMaster University for photolithography assistance and Dan Deptuck at CMC Microsystems for facilitating the device design and Greg Wojcik for useful discussions. We also thank Lumerical Solutions Inc. for providing the mode solver MODE solutions. The authors acknowledge the support of CMC Microsystems, the Canadian Institute for Photonic Innovations and the Natural Sciences and Engineering Research Council of Canada.
References and Links
1. L. Pavesi, and G. Guillot, “Optical Interconnects: The Silicon Approach,”,Vol. 119 Springer Series in Optical Sciences, (Springer-Verlag 2006).
2. N. M. Wright, D. J. Thomson, K. L. Litvinenko, W. R. Headley, A. J. Smith, A. P. Knights, J. H. B. Deane, F. Y. Gardes, G. Z. Mashanovich, R. Gwilliam, and G. T. Reed, “Free carrier lifetime modification for silicon waveguide based devices,” Opt. Express 16(24), 19779–19784 (2008). [CrossRef] [PubMed]
3. M. W. Geis, S. J. Spector, M. E. Grein, J. U. Yoon, D. M. Lennon, and T. M. Lyszczarz, “Silicon waveguide infrared photodiodes with > 35 GHz bandwidth and phototransistors with 50 AW−1 response,” Opt. Express 17(7), 5193–5204 (2009). [CrossRef] [PubMed]
4. J. K. Doylend, P. E. Jessop, and A. P. Knights, “Silicon photonic resonator-enhanced defect-mediated photodiode for sub-bandgap detection,” Opt. Express 18(14), 14671–14678 (2010). [CrossRef] [PubMed]
5. D. F. Logan, P. Velha, M. Sorel, R. M. De La Rue, A. P. Knights, and P. E. Jessop, “Defect-enhanced Silicon-on-insulator Waveguide Resonant Photodetector With High Sensitivity at 1.55 µm,” IEEE Photon. Technol. Lett. 22(20), 1530 (2010). [CrossRef]
6. K. Preston, Y. H. D. Lee, M. Zhang, and M. Lipson, “Waveguide-integrated telecom-wavelength photodiode in deposited silicon,” Opt. Lett. 36(1), 52–54 (2011). [CrossRef] [PubMed]
7. I. Kiyat, A. Aydinli, and N. Dagli, “Low-Power Thermooptical Tuning of SOI Resonator Switch,” IEEE Photon. Technol. Lett. 18(2), 364–366 (2006). [CrossRef]
8. W. M. J. Green, M. J. Rooks, L. Sekaric, and Y. A. Vlasov, “Ultra-compact, low RF power, 10 Gb/s silicon Mach-Zender modulator,” Opt. Express 15(25),17106–17113 (2007). [CrossRef] [PubMed]
9. G. T. Reed, G. Mashanovich, F. Y. Gardes, and D. J. Thompson, “Silicon Optical Modulators,” Nat. Photonics 4(8), 518–526 (2010). [CrossRef]
10. R. A. Soref and B. R. Bennett, “Electrooptical Effects in Silicon,” IEEE J. Quantum Electron. 23(1), 123–129 (1987). [CrossRef]
11. Q. Xu, B. Schmidt, J. Shakya, and M. Lipson, “Cascaded silicon micro-ring modulators for WDM optical interconnection,” Opt. Express 14(20), 9431–9435 (2006). [CrossRef] [PubMed]
12. C. W. White, J. Narayan, and R. T. Young, “Laser Annealing of Ion-Implanted Semiconductors,” Science 204(4392), 461–468 (1979). [CrossRef] [PubMed]
13. W. Bogaerts, R. Baets, P. Dumon, V. Wiaux, S. Beckx, D. Taillaert, B. Luyssaert, J. Van Campenhout, P. Bientsman, and D. Van Thourhout, “Nanophotonic Waveguides in Silicon-on-Insulator Fabricated with CMOS Technology,” J. Lightwave Technol. 23(1), 401–412 (2005). [CrossRef]
14. D. Taillaert, F. Van Laere, M. Ayre, W. Bogaerts, D. Van Thourout, P. Bientsman, and R. Baets, “Grating Couplers for Coupling between Optical Fibers and Nanophotonic Waveguides,” Jpn. J. Appl. Phys. 45(No. 8A), 6071 (2006). [CrossRef]
15. L. Pelaz, L. A. Marqués, and J. Barbolla, “Ion-beam-induced amorphization and recrystallization in silicon,” J. Appl. Phys. 96(11), 5947 (2004). [CrossRef]
16. P. J. Foster, J. K. Doylend, P. Mascher, A. P. Knights, and P. G. Coleman, “Optical attenuation in defect engineered silicon rib waveguides,” J. Appl. Phys. 99(7), 073101 (2006). [CrossRef]
17. J. K. Doylend, “Defect-mediated photodetectors for silicon photonic circuits” PhD. Thesis, McMaster University (2010).
18. T. Baehr-Jones, M. Hochberg, C. Walker, E. Chan, D. Koshinz, W. Krug, and A. Scherer, “Analysis of the tuning sensitivity of silicon-on-insulator optical ring resonators,” J. Lightwave Technol. 23(12), 4215 (2005). [CrossRef]
19. F. Y. Gardes, A. Brimont, P. Sanchis, G. Rasigade, D. Marris-Morini, L. O’Faolain, F. Dong, J. M. Fedeli, P. Dumon, L. Vivien, T. F. Krauss, G. T. Reed, and J. Martí, “High-speed modulation of a compact silicon ring resonator based on a reverse-biased pn diode,” Opt. Express 17(24), 21986–21991 (2009). [CrossRef] [PubMed]
20. N. Rouger, L. Chrostowski, and R. Vafaei, “Temperature effects on silicon-on-insulator (SOI) racetrack resonators: A coupled analytic and 2-D finite difference approach,” J. Lightwave Technol. 28(9), 1380–1391 (2010). [CrossRef]
