Open Access
Add to BibSonomyAbstract
We describe, model and demonstrate a tunable micro-ring resonator integrated monolithically with a photodiode in a silicon waveguide device. The photodiode is made sensitive to wavelengths at and around 1550nm via the introduction of lattice damage through selective ion implantation. The ring resonator enhances detector responsivity in a 60 μm long waveguide photodiode such that it is 0.14 A/W at −10Vbias with less than 0.2 nA leakage current. The device is tunable such that resonance (and thus detection) can be achieved at any wavelength from 1510 – 1600 nm. We also demonstrate use of the device as a digital switch with integrated power monitoring, 20 dB extinction, and no optical power tapped from the output path to the photodiode. A theoretical description suggests that for a critically coupled resonator where the round trip loss is dominated by the excess defects used to mediate detection, the maximum responsivity is independent of device length. This leads to the possibility of extremely small detector geometries in silicon photonics with no requirement for the use of III-V materials or germanium.
©2010 Optical Society of America
1. Introduction
As the size and cost of optical on-chip functionality continues to decrease, silicon photonics has emerged as the method of choice for monolithically integrating optical and electrical devices. Among the many optical functions that have been demonstrated either in silicon or in silicon-based hybrids are fiber-chip couplers [1,2], lasers [3], detectors [4], variable optical attenuators [5], and high-speed modulators [6,7].
In the case of terminal detection, silicon-based devices in the C-band have employed either germanium-on-silicon [8], hybrid integration of III-V materials with silicon [9], or defect-mediated sub-band detection [10]. The latter offers significant advantages in processing simplicity since it can be incorporated into a waveguide-based diode by the addition of only two back-end process steps (ion implantation and low temperature annealing) [11].
One of the disadvantages with this approach is the required tradeoff between device length and minimum detectable power (MDP), which determines signal-to-noise ratio (SNR) for a given optical power and responsivity. For the defect-mediated photodetectors described in this work MDP is determined by the leakage current and the responsivity, where the latter (for a given internal quantum efficiency) depends on the fraction of input optical power that is absorbed. The optical absorption per unit length increases with implantation dose [12], likely due to the corresponding increase in overall defect density in the waveguide. Therefore for low implantation doses, responsivity can be increased either by increasing the implantation dose or by increasing the length of the implanted region in the waveguide such that a higher fraction of the input power is absorbed and converted to photocurrent.
Leakage current also increases with implantation dose [11] however, and so the improvement in responsivity associated with higher implantation doses may be accompanied by a corresponding increase in leakage current and thus a degradation in MDP. Maintaining a low implantation dose and lengthening the device also has an adverse impact on device density on the chip and overall chip size.
For example, the device originally reported in the ground-breaking paper by Geis et al [10] with a length of 0.25 mm has responsivity of 0.12 A/W at 1550 nm and −10V bias and 0.2 μA dark current (MDP ~1.7 μW). At −20V bias the responsivity increased to approximately 1 A/W, but dark current also increased to more than 10 μA (MDP ~10 μW). Geis et al also reported longer devices (3 mm) with 0.5 A/W responsivity at 0V bias (10 nA leakage current) for MDP ~2 nW [13]; an improved 3 mm design with responsivity of 0.8 A/W and 0.5 nA leakage current at −5V bias (MDP ~0.6 nW) [14]; and 300 μm devices exhibiting avalanche gain and very high responsivity (~50 A/W) but also somewhat higher leakage current (0.1 mA) and hence MDP higher than 1 μW [14].
One approach to alleviate this tradeoff between device length, attenuation, and leakage current is the use of a resonator to effectively multiply the optical power incident on the photodiode, thus improving responsivity without having to increase either device length or defect density [15]. This approach has been demonstrated for ring resonators in silicon which use surface defect states to render the device sensitive to sub-bandgap equivalent wavelengths [16]. In that case, the small number of defect states and the small overlap of the optical mode with their location in the waveguide resulted in responsitivies of 0.25 mA/W at −15V bias with 2.5 nA dark current, or MDP of 10 μW. Also of relevance to the current work is a recently submitted report by Logan et al. [17], in which a sub-bandgap ring resonant detector structure was defined using electron-beam lithography. In the current work we report the integration of a defect-mediated photodetector in silicon within a ring resonator for cavity enhancement of the photocurrent. The device has a responsivity which compares well with those reported previously for defect-mediated silicon photodetectors at standard telecommunications wavelengths while occupying significantly less chip surface area. It is operable from 1510 nm to 1600 nm and was fabricated entirely in a standard industrial CMOS fabrication facility.
2. Theoretical considerations
2.1 Defect-mediated detection as a function of device length
Although silicon has a bandgap of 1.1 eV and is therefore generally transparent to wavelengths around 1550 nm (equivalent photon energy around 0.8 eV), energy states within the bandgap (due to lattice damage) allow such photons to be absorbed via Shockley-Read-Hall recombination. For a relatively short length of material which contains a uniform distribution of defects, mid-gap electronic mediated carrier generation will be proportional to the incident optical power according to Eq. (1):
where iph is photocurrent, dz is the propagation length, P(z) is optical power, and γ represents carrier generation efficiency of the traps and is a function of trap density, optical cross-section, photon energy, and the proportion of photogenerated carriers successfully extracted before recombination. The optical power then decreases exponentially with length (for uniform propagation loss) according to Eq. (2):and integrating over length L yields total photocurrent according to Eq. (3):It may be deduced from Eq. (3) that for an absorption of 45 dB/cm (α = −10.4 cm−1) photocurrent will be dependent on device length up to 1 mm, after which two thirds of the optical power has already been absorbed. Further device length thus provides steadily diminishing additional photocurrent.
2.2 Photodiode integrated within a ring resonator
Relative power in a ring resonator is given by Eq. (4):
where Pres is power in the resonator (prior to round trip loss), P0 is input power, k is the coupling between bus waveguide and the resonator, ε is the round trip fractional power loss, and θ is the round trip phase. At critical coupling k = ε and at resonance cos θ = 1, so Eq. (4) can be simplified to Eq. (5):Now if we introduce a defect-mediated photodetector into the resonator, we can substitute Pres from Eq. (5) into Eq. (3) as Pinc. Furthermore if the round-trip loss (RTL) in the resonator is dominated by loss due to defect-mediated absorption in the photodiode, then fractional round-trip loss ε = (1- e-αL) and we derive the expression in Eq. (6) for photocurrent:
As shown in Eq. (6), for a critically coupled ring resonator at resonance with RTL dominated by optical absorption of the photodetector, the photocurrent is not only independent of diode length but can, in principle - by comparing to Eq. (3) - equal the maximum responsivity achievable by extending a straight photodiode until all optical power is absorbed. This somewhat remarkable result can be seen intuitively by considering that as photodiode length is decreased, the reduction in generated photocurrent is compensated by a corresponding increase in cavity-enhanced power. It follows that by incorporating a defect-mediated photodiode with a ring resonator tuned to resonance and critically coupled, the device can be extremely compact and still achieve high responsivity.
3. Device fabrication and description of results
3.1 Device fabrication
Fabrication was carried out at LETI, using the ePIXfab platform [18]. Silicon-on-insulator (SOI) <100> wafers with a top silicon thickness of 220 nm and 2 μm buried oxide were etched 170 nm to form optical rib waveguides with a thin slab for electrical contact. Low energy phosphorus and boron implantation on either side of the waveguide, followed by an activation anneal, formed the diodes. After contact metallization the devices were masked with a thick resist and implanted with boron at 350 keV to a dose of 1 x 1013 cm−2 to introduce volume defects. At this energy the boron is situated below the buried oxide, and so takes no part in the device operation beyond the introduction of defects during its implantation. The contact metallization was also used to incorporate a metal trace running across the ring for resonance tuning using the thermo-optic effect. A schematic of the device is shown in Fig. 1 .
Fig. 1 Schematic of the silicon photonic resonator-enhanced defect-mediated photodetector. GCI = grating-coupled input; GCO = grating-coupled output.
3.2 Detector responsivity and excess loss in a straight waveguide device
Fiber-to-fiber loss of −16 ± 1 dB on unimplanted straight waveguides was measured for unpolarized 1550 nm SMF input. We assume 3 dB input loss due to the polarization-selective coupling of the input grating, and then 6.5 dB coupling loss at each grating. This loss value is greater than the value of 6 dB reported by Laere et al [19] for similar grating couplers. Total input loss is therefore taken to be 9.5 ± 1 dB.
Responsivity and excess loss of several 1 mm straight photodiodes (i.e. diodes which were not incorporated in a resonant structure) were measured. For the lowest loss devices, responsivity was consistently measured as 0.13 ± 0.02 A/W at −10 V bias and input power ranging from −18 dBm to + 2 dBm (responsivity is here defined as photocurrent divided by input optical power after subtracting 9.5 dB input coupling loss) with 4.5 dB (45 dB/cm) excess loss (resulting from the introduction of the implantation defects). Quantum efficiency is therefore 10%. This result is very similar to that reported by Geis et al [10] at −10 V bias for a similar device with one quarter the length but a higher implantation dose and using silicon instead of boron for the defect implantation species. Dark current at −10 V bias for the diodes reported in the current work is less than 0.2 nA.
3.3 Photocurrent and optical transmission of the ring-resonator photodiode
Photocurrent and optical transmission were measured for a ring resonator device with near-critical coupling at 1550 nm. The test setup is shown in Fig. 2 . The photocurrent and optical transmission spectrum for wavelengths near 1550 nm is shown in Fig. 3 .
Fig. 2 Schematic of the test setup used to characterize responsivity, optical transmission, and tuning efficiency. The resistor on the DUT represents the thermo-optic tuner. ECL = external cavity laser, PS = polarization scrambler, Amm. = ammeter.
The optical throughput resonance FWHM is 75 pm at 1548.79 nm consistent with a Q-factor of 21000. The directional coupler k was measured independently using passive devices and was found to be between 2% and 7% at 1550 nm. A value of k = 6.9% was used to provide the best fit to the optical transmission shown in Fig. 4 ; this value together with Eq. (4) suggests a round-trip loss of 0.35 dB for the ring.
Fig. 4 Output excess loss and photodiode responsivity for the resonance closest to 1550 nm. For this plot, grating coupler loss of 7.5 dB was used to enable a match of the modeled excess loss to measured values.
Using the previously measured 45 dB/cm value to calculate loss for the 60 μm in-ring photodiode yields 0.27 dB, which suggests that there is approximately 0.1 dB additional bend and scattering loss in the resonator apart from the implant optical absorption.
It should be noted that while the near-critical coupling achieved here enhances the photodiode responsivity, it necessarily limits bandwidth to a FWHM of 9 GHz. This places an inherent limitation on the data rates for which the device can be used. Designing the resonator to be over-coupled rather than critically coupled [i.e. k > ε in Eq. (4)] can increase bandwidth by sacrificing responsivity. Alternatively, adding defects by employing a higher implantation dose in the ring will increase ring loss, thus lowering the Q of the ring. This increases bandwidth at critical coupling with minimal impact on overall responsivity as indicated by Eq. (6) assuming that the ratio γ/α (carrier generation efficiency vs. absorption) remains relatively unchanged.
3.4 Responsivity, excess loss, and MDP in the resonator-enhanced photodiode
Using Eq. (3) and the measured responsivity described in section 3.2, we may derive a value for γ of 1.40 A/W·cm. Substituting this value together with Pres from Eq. (4) for Pinc in Eq. (3), and using the k and RTL values calculated in section 3.3, we can model the expected responsivity of the resonator-enhanced photodiode. This modeled result is shown in Fig. 4 plotted together with the experimentally measured responsivity of the device.
As expected, the resonator-enhanced photodiode shows superior responsivity at resonance (0.14 A/W) with a 10x reduction in length compared to the straight waveguide device, or a 3x reduction in length compared to the device originally reported by Geis et al [10]. Since the current result was achieved without increasing the damage inducing implantation dose, dark current is kept to a value of 0.2 nA at −10 V and hence we report MDP of 1.4 x 10−9 W at resonance near 1550 nm. This is comparable to the lowest MDP reported to date for a defect-mediated silicon photodetector at 1550 nm [14], but with a 30x reduction in overall device length (50x reduction in actual diode size) thus offering an advantage in terms of device density and/or chip size without sacrificing signal-to-noise ratio.
3.5 Resonance wavelength tuning
We use a resistive metal track which overlays the ring resonator to dissipate thermal power into the SOI and thus shift the resonance wavelength. Figure 5 shows the responsivity spectra obtained while tuning the location of the resonances. A plot of relative phase shift versus i2R dissipated power (inset, Fig. 5) yields a value of 44 mW/π for the tuning efficiency with the device thermally isolated.
Fig. 5 Thermo-optic tuning of the resonance peak wavelengths through a full free spectral range. Inset shows tuning versus thermal power with a tuning efficiency of 44 mW/π.
Tuning efficiency is degraded to 97 mW/π for a device thermally grounded to a heat sink. These values can be improved by designing the heater to run alongside or directly over the waveguide rather than merely crossing it as shown in Fig. 1.
The tuning capability allows the resonance to be adjusted across the full free spectral range of the device such that any wavelength between resonances can be aligned with a resonance peak for optimal responsivity. Measurements of responsivity from 1510 nm to 1600 nm show variation of less than ± 20% after normalization to the coupler grating transmission.
Tuning also allows the device to be operated as a “black-box” digital switch. The diode acts as a monitor, and the optical output can be tuned on and off resonance for digital operation with greater than 20 dB extinction. This architecture has the additional advantage that the photodiode does not tap any optical power from the output path since the photodiode solely taps optical power stored in the ring. This allows for the minimization of excess loss. Switching speed could be increased by using carrier injection or depletion for a ring-resonator high-speed modulator as demonstrated by Lipson et al. [7] with the added advantage of an integrated tap monitor for bias control.
4. Conclusion
A tunable ring resonator-enhanced defect-mediated photodetector employing defects introduced into the volume of the silicon waveguide by ion implantation, has been modeled and demonstrated. Modeling shows that cavity enhancement at critical coupling eliminates the dependence of responsivity on photodiode length without the need for increased defect densities, thus allowing improved minimum detectable power. The device is entirely compatible with fabrication processes used for standard silicon devices with no recourse to the integration of III-V materials or germanium. The device demonstrated here has a total size of 0.2 mm x 0.1 mm, leakage current less than 0.2 nA for −10V bias, responsivity of at least 0.1 A/W and MDP less than 2 nW for wavelengths from 1510 nm to 1600 nm.
Acknowledgments
The authors would like to thank Doug Bruce, Jason Ackert, and Richard Jones for useful discussions, Dan Deptuck for help with mask and process design, and Brad Robinson for help with test setup. The authors would like to acknowledge useful discussions with Dylan Logan with regard to ring resonant detectors fabricated using electron-beam lithography. The authors would also 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. 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]
2. I. Day, I. Evans, A. Knights, F. Hopper, S. Roberts, J. Johnston, S. Day, J. Luff, H. Tsang, and M. Asghari, “Tapered silicon waveguides for low insertion-loss highly efficient high-speed electronic variable attenuators,” Proc. IEEE Opt. Fiber Commun. Conf. 1, Mar. 23–28, pp. 249–251 (2003).
3. A. W. Fang, H. Park, O. Cohen, R. Jones, M. J. Paniccia, and J. E. Bowers, “Electrically pumped hybrid AlGaInAs-silicon evanescent laser,” Opt. Express 14(20), 9203–9210 (2006). [CrossRef] [PubMed]
4. J. B. D. Bradley, P. E. Jessop, and A. P. Knights, “Silicon waveguide integrated optical power monitor with enhanced sensitivity at 1550 nm,” Appl. Phys. Lett. 86(24), 241103 (2005). [CrossRef]
5. D. W. Zheng, B. T. Smith, and M. Asghari, “Improved efficiency Si-photonic attenuator,” Opt. Express 16(21), 16754–16765 (2008). [CrossRef] [PubMed]
6. 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]
7. Q. Xu, B. Schmidt, S. Pradhan, and M. Lipson, “Micrometre-scale silicon electro-optic modulator,” Nature 435(7040), 325–327 (2005). [CrossRef] [PubMed]
8. M. Morse, O. Dosunmu, G. Sarid, and Y. Chetrit, “Performance of Ge-on-Si p-i-n photodetectors for standard receiver modules,” IEEE Photon. Technol. Lett. 18(23), 2442–2444 (2006). [CrossRef]
9. H. Park, A. W. Fang, R. Jones, O. Cohen, O. Raday, M. N. Sysak, M. J. Paniccia, and J. E. Bowers, “A hybrid AlGaInAs-silicon evanescent waveguide photodetector,” Opt. Express 15(10), 6044–6052 (2007). [CrossRef] [PubMed]
10. M. W. Geis, S. J. Spector, M. E. Grein, R. T. Schulein, J. U. Yoon, D. M. Lennon, S. Deneault, F. Gan, F. X. Kaertner, and T. M. Lyszczarz, “CMOS-compatible all-Si high-speed waveguide photodiodes with high responsivity in near-infrared communication band,” IEEE Photon. Technol. Lett. 19(3), 152–154 (2007). [CrossRef]
11. J. K. Doylend, A. P. Knights, B. J. Luff, R. Shafiiha, M. Asghari, and R. M. Gwilliam, “Modifying functionality of variable optical attenuator to signal monitoring through defect engineering,” Electron. Lett. 46(3), 234 (2010). [CrossRef]
12. P. J. Foster, J. K. Doylend, P. Mascher, A. P. Knights, and P. Coleman, “Optical attenuation in defect-engineered silicon rib waveguides,” J. Appl. Phys. 99(7), 073101 (2006). [CrossRef]
13. M. W. Geis, S. J. Spector, M. E. Grein, R. J. Schulein, J. U. Yoon, D. M. Lennon, C. M. Wynn, S. T. Palmacci, F. Gan, F. X. Käertner, and T. M. Lyszczarz, “All silicon infrared photodiodes: photo response and effects of processing temperature,” Opt. Express 15(25), 16886–16895 (2007). [CrossRef] [PubMed]
14. 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]
15. M. Casalino, L. Sirleto, L. Moretti, F. D. Corte, and I. Rendina, “Design of a silicon resonant cavity enhanced photodetector based on the internal photoemission effect at 1.55 um,” J. Opt. A, Pure Appl. Opt. 8(10), 909–913 (2006). [CrossRef]
16. H. Chen, X. Luo, and A. W. Poon, “Cavity-enhanced photocurrent generation by 1.55 um wavelengths linear absorption in a p-i-n diode embedded silicon microring resonator,” Appl. Phys. Lett. 95(17), 171111 (2009). [CrossRef]
17. D. F. Logan, Department of Engineering Physics, McMaster University, 1280 Main Street West, Hamilton, Ontario, L8S4L7, Canada, and P. Velha, M. Sorel, R. M. De La Rue, A. P. Knights and P. E. Jessop, are preparing a manuscript to be called “Defect-enhanced Silicon-on-insulator Waveguide Resonant Photodetector with High Sensitivity at 1.55μm”.
19. F. Van Laere, G. Roelkens, J. Schrauwen, D. Taillaert, P. Dumon, W. Bogaerts, D. Van Thourhout, and R. Baets, “Compact Grating Couplers Between Optical Fibers and Silicon-on-Insulator Photonic Wire Waveguides with 69% Coupling Efficiency”, Optical Fiber Communication Conference (OFC), PDP15 (2006).
