Open Access
Add to BibSonomyAbstract
A CMOS compatible wavelength monitor comprised of two thermally tuned racetrack-ring resonators with defect mediated photodiode structures is experimentally demonstrated in monolithic silicon. Each resonator is independently tuned so as to determine an unknown input wavelength by tuning the resonance peak locations until there is overlap between the two comb spectra. The presence of two of these resonator/heater components, each with a different free spectral range, increases the unambiguous measurement range when compared to one component used on its own.
© 2013 Optical Society of America
1. Introduction
When designing silicon photonic devices the integration of electrical and optical functionality drives a great deal of research with the aim of creating devices that are compatible with CMOS process flow. Current research is aimed at improving and developing novel structures for functional high-speed modulators [1, 2], variable optical attenuators [3], efficient fiber-chip couplers [4, 5] and photodetectors [6].
The determination of unknown wavelengths is of vital importance when designing devices to be used in communication systems. While there are many ways to do this, the current push in industrial research is to accomplish this while reducing device size, lowering the power dissipated by the device, as well as reducing variations due to thermal dissipation, both of which are common problems when working on a silicon platform.
In this work we present a wavelength monitor design based on thermally tunable ring resonators. The device contains a 50/50 splitter, a pair of racetrack-ring resonators each with a local micro-heater, as well as defect mediated photodiodes that are placed onto the racetrack ring resonator waveguide. The device is depicted in Fig. 1 and is based on a tunable resonant photodiode in silicon. Previous work from our group has shown the feasibility of defect mediated ring resonant sub-bandgap detection. Doylend et al. used optical lithography (via a 2009 epixfab LETI shuttle run) to fabricate ring resonant detectors in which boron implantation was used to introduce defects [7]. Logan et al. utilized electron beam lithography and inert silicon ion implantation to create deep-levels to facilitate the detection process [8].
Fig. 1 Schematic of silicon photonic wavelength monitor showing grating-coupled input and output and 50/50 splitter with ring resonant defect-mediated photodiodes. All distance are in mm.
We experimentally demonstrate how the resonance wavelength of two independent thermally tuned racetrack-ring resonators can be used to accurately measure the wavelength of an optical input. We demonstrate the performance of this device as a wavelength monitor suitable for operation over the C-band (1530nm to 1570nm).
2. Device design
The device layout includes a dual channel coupler with a racetrack ring resonator on each of the two output arms. The length of the splitter is 0.1 mm and the gap is 0.45 μm. The resonators in the device, lower and upper path devices (LPD and UPD), have straight lengths (SL) of 0.070 mm and 0.078 mm respectively. The photodiodes are situated inside the racetrack-ring structures as shown in Fig. 1. Both racetrack ring resonators are coupled to a bus waveguide via 12 μm long directional couplers and the gap between the waveguide and the racetrack-ring resonator along the directional coupler is 350 nm as shown in Fig. 2. The bend radius within the ring is 8 μm giving circumferences of 190 μm and 205 μm for the LPD and UPD respectively. The free spectral ranges (FSR) were 3.15 nm and 2.95 nm for the lower and upper path rings respectively. The lengths of the photodiodes are 0.06 mm and defects were created by ion implantation only in the waveguide region encompassed by the photodiodes following the work of [9]. An integrated heater in the form of a 1 μm wide multilayer metal trace, consisting of Ti (30 nm), TiN (60 nm), Al/Cu (440 nm), Ti (10 nm), TiN (40 nm), was included to allow thermal tuning of the resonance wavelength. The trace was not placed directly over the waveguide due to concerns about metal-induced losses. The mask layout is shown in Fig. 2.
Fig. 2 Layout of a racetrack-ring resonator with a defect mediated photodiode within the ring. The photodiode is 60 μm in length and length of the directional coupler is 12 μm.
3. Device Fabrication
The device was fabricated with the 193 nm deep UV lithography process, carried out at CEA-LETI. The silicon on insulator (SOI) wafers consisted of 220nm top silicon on a 2 μm buried oxide. The waveguides were defined with a 170 nm etch, leaving a 50 nm silicon slab. To avoid optical absorption in the 580 nm thick metal contact layer, a 1μm oxide was deposited across the wafer. The p-i-n diodes were created by low energy phosphorous and boron doping at a dose of 2 × 1014 cm−2. After contact metallization, boron ions at 350 keV and dosage of 1 × 1013 cm−2 were implanted in the intrinsic region to introduce defects in the lattice, and subsequently annealed at 200°C for 5 min. These defects established deep levels within the bandgap [10], increasing optical absorption in the infrared [11]. The contact metallization was used to incorporate a metal trace running across the ring for resonance tuning using the thermo-optic effect. Grating couplers were incorporated onto the waveguide to enable efficient introduction and extraction of light using polarization maintaining fiber (PMF) at the input and single mode fiber (SMF) at the output. The rings had a nominal waveguide width of 500 nm and coupling gap of 350 nm.
The heater’s effect on the device was characterized using 1 μm tip radius tungsten probes to connect a dc voltage source with an ammeter in series to the metallic contact pads of the device. Light from a tunable diode laser was coupled into the chip with a PM fiber that was positioned above the input grating at an angle of 10 degrees. The transmitted light was grating-coupled into the output fiber in a similar way to the input. The output fiber was connected to a Newport fiber-optic power meter. By measuring the total loss through a straight passive test waveguide, the combined loss of the input and output grating couplers and propagation through 8.5 mm of waveguide was determined to be 22 ± 0.25 dB. This value is used in the determination of responsivity of the two photodetectors.
4. Results and Discussion
The quality factors (Q) were measured to be 10000 ± 1000 for both the LPD and UPD. Figure 3 shows wavelength scans of photocurrent and transmitted power for the 190 μm and 205 μm circumference racetrack-ring resonators. The photocurrent is greatly enhanced on resonance, and from an estimation of on-chip power, we were able to calculate the responsivity of the photodetector.
Fig. 3 Transmitted power, on the left, and diode photocurrent, on the right, as a function of wavelength for (a) 190nm and (b) 205 nm circumference racetrack-ring resonators, both structures have the photodiode biased at −2 V
The fiber-to-fiber loss of straight waveguides with no photodiodes was measured to be 22 ± 0.5 dB for unpolarized 1560 nm SMF input. The distance between the grating-couplers is 8.5 mm with a propagation loss estimated to be 0.24 dB/mm [12]. The coupling loss at the gratings was therefore taken to be 10.0 ± 0.3 dB. Though the insertion loss of the grating couplers was high, enough power was coupled into the device to allow for the observation of resonance peaks and their shift from heating for the purpose of our experiment.
The responsivity for a particular bias voltage, defined as photocurrent divided by the incident optical power on the photodiode, can be calculated for the two components of our device. The LPD (lower path device), with photocurrent and incident power of 0.11 μA and 82.5 ± 4.7 μW respectively, has a responsivity of 1.3 ± 0.1 mA/W with a reverse bias voltage of 2V. The UPD (upper path device), with photocurrent and incident power of 0.11μA and 80.0 ± 2.5 μW respectively, has a responsivity of 1.35 ± 0.05 mA/W with a same reverse bias voltage. The incident optical power was estimated from an input of 0 dBm minus 10.0 dB grating coupling loss as well as accounting for propagation loss. When accounting for the propagation loss, the light path from the input coupler to the diodes was measured to be 3.6 mm (lower path) and 4.6 mm (upper path). Figure 4a and 4b shows the current - voltage characteristic curve for LP and UP photodiodes present on the device.
Fig. 4 Absolute values of photo-current and dark-current versus voltage for the photodiodes of (a) lower path and (b) upper path with and without coupled light
Figures 5(a) and 6(a) show the transmission spectrum at different voltages applied to the heaters of the LPD and UPD. The resonances were shifted as the heater power increases and the temperature of the devices rises. In Fig. 5(b) and Fig. 6(b), a plot the relationship between the resonance wavelength shift and the heater power of the two resonators is shown, from which we calculate the linear power dependence of the thermo-optic tuning to be 0.041 nm/mW and 0.034 nm/mW respectively. The data points used in Fig. 5(b) and Fig. 6(b) are determined by applying a Lorentzian fit, in the region of the transmission peak, to determine the centre frequency for a specific applied heater power. All further analysis is done assuming that the ambient temperature of the chip is a constant. In a practical implementation of this device, either the ambient temperature will have to be help constant or the thermal tuning of the resonance wavelength will have to be calibrated and a correction applied. The tuning rate will be on the order of 80pm/C [13].
Fig. 5 (a) Transmission spectra with different power applied to the integrated micro heater of lower path resonator (b) The shift of resonance wavelength at different heater power.
Fig. 6 (a) Transmission spectra with different power applied to the integrated micro heater of upper path resonator (b) The shift of resonance wavelength at different heater power.
Knowing the dependence of the resonant wavelength on heater power, a measurement of the heater powers required to tune both rings into resonance can be used to determine the wavelength of an unknown source. As seen in Fig. 6, the UPD resonance that occurs at λ = 1553.18 nm in the absence of applied heater power and is shifted at a rate of 0.034 nm/mW as power is applied to the heater. Within the optical communications C-band there will be several additional resonances, separated from each other by the measured free spectral range of 2.95 nm. Figure 5 shows the corresponding measurement for the LPD resonance that occurs at λ = 1555.10 nm in the absence of applied heater power. For this device a tuning rate of 0.041 nm/mW was measured and the free spectral range is 3.15 nm. The plots in Fig. 7 show, for both the UPD and LPD, the relation between resonant wavelength and applied heater power for multiple adjacent resonances in a wavelength window from 1550 to 1572 nm. The slopes determined in Figs. 5(b) and 6(b) were used to model the behavior of each resonance peak in the studied wavelength range. The red and blue plots of Fig. 7 have a slope corresponding the the UPD or LPD and x-intercepts spaced by the FSR of the corresponding ring.
Fig. 7 Plot (a)-(d) illustrate the thermal tuning of resonance peaks to match a particular input wavelength, which is shown by the vertical black line. LPDHP and UPDHP are the measured lower and upper path heater powers that tune their respective rings into resonance with the input wavelengths, which are 1554nm, 1557nm, 1560nm and 1566nm in (a)-(d)
To test this structure’s effectiveness as a wavelength monitor, input light from a tunable diode laser was coupled to the input waveguide and the power emerging from the two output arms was monitored. The ring resonators on the two output arms were both tuned into resonance with the input wavelength by adjusting the electrical power to the heaters. A resonance is identified as a minimum in the optput optical power, or the a maximum photodetector photo-current. This heater power is recorded and used in the determination of the unknown wavelength. In this work, only output power was used for the determination of resonant peaks. This was done for experimental convenience however, in a practical application, the on-ring sensor would be used.
Figures 7(a)-7(d) show the dependence of the resonant wavelengths for the LPD (red) and UPD (blue) resonators on the power applied to their local heaters. The solid red and blue lines are based on the measured zero-power wavelengths and the thermal tuning rates determined from Figs. 5 and 6. For each of the four figures, a different wavelength was coupled into the device and then the minimum powers required the tune the two ring resonators into resonance were measured. These heater powers are indicated by the red and blue horizontal broken lines and the input wavelength is indicated by the vertical line.
The wavelength of the input light will correspond to the intersection point of the UPD’s horizontal power line and one of the UPD tuning curves. However, since there are a series of tuning curves in the wavelength window of interest, there will be several such intersection points and the wavelength determination is ambiguous when only considering the UPD. Similarly the intersection points of the LPD curves result in an ambiguous set of several possible wavelength values. The ambiguity is removed when these two sets of candidate wavelengths are compared there will be one value common to both of these sets, this is demonstrated in Table 1. This extends the useful range of the device from one FRS (~2nm) by at least a factor of 10.
Table 1. Results of determination of an input wavelength
Tables 1(a)-a(d) list the wavelengths at which the horizontal lines in Figs. 7(a)-(d) intersect with their corresponding tuning curves. The precision with which these wavelengths can be determined is limited by the measurements of the zero-power resonance wavelengths and the tuning rates shown in Figs. 5(b) and 6(b). The wavelengths associated with the UPD are paired with the nearest LPD wavelength and the differences are listed in the fourth column of the tables. The differences scale linearly at a rate determined by the difference in the free spectral ranges of the two resonators. In principle the wavelength of the input light source will be the wavelength at which the difference is zero. In practice, the experimental uncertainty is sufficiently low that the wavelength pair with the minimum difference can be clearly identified and then the actual wavelength can be taken to be the average value for that pair. The four test wavelengths used were 1554, 1557, 1560 and 1566nm. The largest difference between the input light wavelength and that predicted by our device was 0.075nm, giving an estimate of the precision that is possible.
For purposes of illustration in Fig. 7 and Table 1, we considered a wavelength window from 1550 to 1572. Within this window, only one pair of intersection points has a near zero wavelength difference, and the wavelength differences tune systematically by about 0.2 nm from one wavelength pair to the next. However, if the window under consideration was wide enough to contain 15 or more resonator free spectral ranges (i.e. > 40 nm) the wavelength determination would become ambiguous because a difference of one free spectral range (~3 nm) looks the same as a difference of zero. This sets the useful operating range for this wavelength monitor at about 40 nm, which is sufficient to cover the entire C band.
To determine whether or not thermal crosstalk between the two ring resonators is a complicating factor, we measured the thermal tuning rate of each resonator (UPD or LPD) as power was applied to the heater on the other resonator (LPD or UPD). The cross tuning rate was determined to be negligible. An additional complicating factor that should be considered is the possibility of a power dependent resonance wavelength. Nonlinear effects begin to appear in silicon ring resonators at optical power levels that depend on the circumference and the Q of the ring [14]. For a 5-µm-radius ring with cross sectional dimensions similar to those used in this paper, as little as 500 µW of input power can cause nonlinearities [15]. With our input power on the order of 100 µW and ring circumferences near 200 µm, we assumed that nonlinear effects were negligible.
6. Conclusion
We demonstrate a device containing two different racetrack-ring resonators to use a vernier effect for wavelength measurement. In a thermally tuned ring resonator, the heater power required to tune into resonance with an input signal can be used as a measure of the wavelength, but only over a very narrow range that is limited by the resonator’s free spectral range. By simultaneously bringing into resonance two resonators of slightly different free spectral ranges, the operating range of this type of wavelength monitor can be broadened to enable wavelength measurement to a precision of 0.075nm over an optical communication band.
Acknowledgments
The authors would like to acknowledge the support of CMC Microsystems, the Natural Sciences and Engineering Research Council of Canada and the Canadian Institute for Photonic Innovations.
References and Links
1. A. Liu, R. Jones, L. Liao, D. Samara-Rubio, D. Rubin, O. Cohen, R. Nicolaescu, and M. Paniccia, “A high-speed silicon optical modulator based on a metal-oxide-semiconductor capacitor,” Nature 427(6975), 615–618 (2004). [CrossRef] [PubMed]
2. Q. Xu, B. Schmidt, S. Pradhan, and M. Lipson, “Micrometre-scale silicon electro-optic modulator,” Nature 435(7040), 325–327 (2005). [CrossRef] [PubMed]
3. D. W. Zheng, B. T. Smith, and M. Asghari, “Improved efficiency Si-photonic attenuator,” Opt. Express 16(21), 16754–16765 (2008). [CrossRef] [PubMed]
4. D. Taillaert, P. Bienstman, and R. Baets, “Compact efficient broadband grating coupler for silicon-on-insulator waveguides,” Opt. Lett. 29(23), 2749–2751 (2004). [CrossRef] [PubMed]
5. D. I. Evans, I. Knights, A. Hopper, F. Roberts, S. Johnston, J. Day, S. Luff, J. H. Tsang, and M. Asghari, “Tapered silicon waveguides for low insertion loss highly-efficient high-speed electronic variable optical attenuators,” in Proceedings of the OFC Conference 2003. 1(1) 249–251 (2003).
6. J. J. Ackert, M. Fiorentino, D. F. Logan, R. G. Beausoleil, P. E. Jessop, and A. P. Knights, “Silicon-on-insulator microring resonator defect-based photodetector with 3.5-GHz bandwidth,” J. Nanophotonoics 5(1), 059507 (2011). [CrossRef]
7. 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]
8. 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), 1530–1532 (2010). [CrossRef]
9. A. P. Knights, J. D. B. Bradley, S. H. Gou, and P. E. Jessop, “Silicon-on-Insulator Waveguide Photo-detector with Self-ion Implantation Engineered Enhanced Infrared Response,” J. Vac. Sci. Technol. A 24(3), 783 (2006). [CrossRef]
10. J. D. B. Bradley, P. E. Jessop, and A. P. Knights, “Silicon waveguide-integrated optical power monitor with enhanced sensitivity at 1550nm,” Appl. Phys. Lett. 86(24), 241103 (2005). [CrossRef]
11. 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]
12. W. Bogaerts, R. Baets, P. Dumon, V. Wiaux, S. Beckx, D. Taillaert, B. Luyssaert, J. Van Campenhout, P. Bienstman, and D. Van Thourhout, “Nanophotonic waveguides in Silicon-on-Insulator Fabricated With CMOS Technology,” J. Lightwave Technol. 23(1), 401–412 (2005). [CrossRef]
13. G.-D. Kim, H.-S. Lee, C.-H. Park, S.-S. Lee, B. T. Lim, H. K. Bae, and W.-G. Lee, “Silicon photonic temperature sensor employing a ring resonator manufactured using a standard CMOS process,” Opt. Express 18(21), 22215–22221 (2010). [CrossRef] [PubMed]
14. W. Bogarts, P. DeHeyn, T. VanVaerenbergh, K. DeVos, S. K. Selvaraja, P. Dumon, P. Bienstman, D. VanThorhout, and R. Baets, “Silicon microring resonators,” Laser Photonics Rev. 6(1), 47–73 (2012). [CrossRef]
15. V. R. Almeida and M. Lipson, “Optical bistability on a silicon chip,” Opt. Lett. 29(20), 2387–2389 (2004). [CrossRef] [PubMed]
