We have fabricated a waveguide integrated monolithic silicon infrared detector. The photodiode consists of a p-i-n junction across a silicon-on-insulator (SOI) rib waveguide. Absorption is due to surface-states at the silicon/air interface of the waveguide. A 2 mm long detector shows a response of 0.045 A/W (calculated as a function of coupled light) and is capable of operation at 10 Gb/s at a reverse bias voltage of 2 V.
© 2014 Optical Society of America
Silicon is transforming the integrated photonics landscape. Steady improvement in the performance of building-block devices, coupled with the availability of complimentary-metal-oxide-semiconductor (CMOS) compatible fabrication facilities provides a roadmap to dense integration of optical devices on the silicon-on-insulator (SOI) platform . Silicon photonics offers a solution to the capacity limits of electrical interconnects  and among other applications, a wide variety of uses for sensing in mechanical, chemical and medical settings [3–5].
The SOI platform offers the capability to create densely routed low loss waveguides, however the integration of active devices, such as lasers, requires additional materials. For photodetection, hybrid material integration offers the best performance with the epitaxial growth of germanium on silicon as a favoured approach . Detection can also be achieved with a monolithic silicon approach. Introducing defects into waveguides with ion implantation can result in high speed response [7–9] and sensitivity to wavelengths at which germanium is transparent. For example, recent work has shown acceptable responsivity of these photodetectors at a wavelength of 1.9 µm . Such detectors are still inferior to germanium in several aspects, such as the need for a high operating bias voltage [6,8]. However the fabrication simplicity of monolithic detectors ensures they remain of interest.
Monolithic detection of infrared wavelengths in silicon has been shown to be possible without the introduction of defects through ion implantation. Two-photon absorption can occur for instance. This is a power dependent phenomenon and therefore can be enhanced with a resonant device . Also, at the surface of a (unpassivated) crystalline material there are necessarily some unsatisfied (dangling) bonds. Such states resemble a plane of defects, present due to discontinuation of the lattice structure. These surface state defects have been shown to provide absorption at 1550 nm . In that case, photon absorption and the resulting change in conductivity was measured by electrically contacting a silicon waveguide through narrow silicon ‘wings’. The narrow wings isolated the optical mode from the metal contacts while still providing electrical contact. Power monitoring has also been demonstrated by measuring the change in capacitance as a result of change in free carrier concentration due to surface state defect absorption .
We have extended the work reported in  to a novel geometry with significant improvement in performance resulting from two key features. A p-i-n junction has been implemented across the waveguide, and a selective oxide etch centred on the waveguide has left the surface unpassivated. This provides a considerable enhancement in responsivity when compared to the results of  and operation is demonstrated at 10 Gb/s with a relatively low bias of 2 V.
2. Fabrication and measurement
The current devices were fabricated at the Institute for Microelectronics in Singapore (IME-A*STAR) using 248 nm ultraviolet lithography. Waveguides were formed in SOI which had a 220 nm thick top silicon layer over 2 µm of oxide. An etch of 130 nm formed the waveguides, which had a nominal width of 500 nm. A second etch of 70 nm was used to form gratings for coupling light from a remote fiber. The grating lines had a uniform period of 610 nm and were laid out in a curved focusing structure to reduce the footprint . The p-i-n junction was formed with boron and phosphorous implantation with a target concentration of 8 x 1019 cm−3. The implanted regions began 500 nm from the waveguide sidewall and extended outwards to the metal contacts. An oxide layer was deposited, and aluminium was used to contact the silicon. After the metal processes the oxide immediately above the p-i-n junction was selectively etched to the waveguide resulting in an unpassivated surface. Scanning electron microscope images of the device are shown in Fig. 1. In Fig. 1(a) a top view of the exposed waveguide can be seen, while in Fig. 1(b) a cross section cut with a focused ion beam is shown.
All measurements were carried out with a tuneable laser locked at a wavelength of 1530 nm, corresponding to the minimum loss of the input gratings. In order to vary input power an erbium-doped fiber amplifier (EDFA) was used along with a variable fiber-optic attenuator. Waveguide and device loss was determined by measuring devices with varying length. Eye diagram measurements employed the tuneable laser, EDFA and a bit pattern generator (BPG) operating at 10 Gb/s. A 20 GHz compatible bias tee was used to apply a reverse bias voltage to a 40 GHz rated RF probe. An RF amplifier with a bandwidth of 45 GHz boosted the detector signal prior to measurement on an oscilloscope.
3. Experimental results
3.1 DC response
The devices were characterized first for DC operation, Fig. 2 shows the current-voltage characteristic for a 2 mm long photodiode with an estimated 220 µW of power coupled into the waveguide (labeled ‘without oxide’). The detector produces a responsivity of 0.09 A/W at a reverse bias of 25 V and 0.045 A/W at 2V. Control measurements showed that waveguides with an oxide cladding layer (i.e. a passivated surface) have a measured loss of 3 ± 1 dB/cm, while waveguides without the oxide cladding show a loss of 6 ± 1 dB/cm. Using this absorption of 6 dB/cm, the internal responsivity of the 2 mm long device can be calculated. With 220 µW of coupled power, over a length of 2 mm 53 µW is absorbed and 20 µA of photocurrent is produced corresponding to 0.37 A/W at 25 V, or equivalent internal quantum efficiency of 30% at 1530 nm.
To demonstrate the effect of the oxide removal, Fig. 2 also shows the current-voltage characteristic for a p-i-n diode that did not have the waveguide silicon surface exposed with an oxide etch. This device shows a reduction of approximately two orders of magnitude in photocurrent compared with the device with the exposed surface. Dark current is also lower for the passivated device (as one might expect), by up to two orders of magnitude for high reverse bias. The difference can be attributed to the unsatisfied bonds at the silicon surface, which in the presence of air forms a low quality shallow oxide layer.
3.2 High speed performance
The dynamic photodetector response was observed at 10 Gb/s stimulated by a 231-1 PRBS on-off keying signal. Eye diagrams captured from an oscilloscope are shown in Fig. 3 while operating with a reverse bias voltage of 2 V (Fig. 3(a)) and 10 V (Fig. 3 (b)). Increasing the bias voltage had little effect on the rise and fall time of approximately 60 ps, although an increase in signal amplitude was consistent with the DC measurement.
3.3 Input power variation
The photodiode linearity (i.e. response versus input optical power) was measured with the photodiode current versus waveguide coupled power shown in Fig. 4. A linear response was observed up to a waveguide coupled power of 4 dBm (the maximum power available in our set-up). This characteristic enables a broad range of operation and confirms the expectation that two photon absorption does not play a significant role .
The introduction of the p-i-n junction and oxide opening has resulted in an increased responsivity over previous work on integrated surface-state mediated detectors reported by Baehr-Jones . There, 0.036 A/W at 11 V was reported for a 1.5 mm long device for coupled power below 1 µW. We observe 0.09 A/W for a 2 mm device with a bias of 25 V and 0.045 A/W at 2 V, while operating with a much higher coupled optical power of 2.5 mW. We attribute the increased efficiency to the reduction of recombination. The devices in  require electronic carriers to travel several microns in narrow undoped silicon wires. In contrast, the use of a rib waveguide and p-i-n structure reduces the maximum carrier travel distance to 1 µm, the waveguide width plus the doping-waveguide separation. The p-i-n junction also allows for operation at 10 Gb/s, as the increased electric field within the absorbing region provides a high carrier drift velocity. The 10 Gb/s operation is achievable at a bias of just 2 V, an important figure for compatibility with CMOS electronics.
There are several factors that can limit the operation frequency of the device, including carrier transit time, defect lifetime and the resistance-capacitance (RC) time constant. The device is not limited by carrier transit time, with undoped silicon we assume saturation velocity is reached at 1 x 107 cm/s, giving transit times less than 10 ps. An unpassivated surface provides a semi-continuous density of states, enabling rapid decay. An estimate of the surface free carrier lifetime provides a proxy figure for the maximum defect lifetime. Using a surface recombination model in , we have estimated the lifetime of a surface free carrier to be less than 15 ps. The RC limited bandwidth is calculated by assuming a 50 Ω load and a measured capacitance of approximately 1 pF for a 2 mm long device. This results in a RC limited bandwidth of 3.1 GHz and is not incompatible with the bit pattern measurements.
While these surface state mediated detectors do not achieve the higher responsivity at high bias of silicon waveguide detectors employing bulk defect states , their responsivity is greater than such devices in the low bias regime (i.e. <10 V reverse bias). This is a result of the high density of defects in the bulk silicon (relative to the unperturbed bulk of the surface state mediated detectors) which reduces carrier mobility and increases dark current.
4.2 Applications and suggested improvements
The current devices also seem attractive for further development because of their extreme ease of fabrication and potential for integration with complex circuits. The low absorption of the surface states and low operating bias voltage makes them particularly suitable for power monitoring applications . We also note that these detectors have potential application if integrated into an evanescent sensor geometry. For such a low speed application, increasing the length of the device to several mm’s would not cause undue concern. Such a sensor would have the potential for increased functionality over those previously reported , because the detection mechanism is likely a sensitive function of the surface condition.
The overall efficiency of these detectors could be improved in several ways. Since the optical absorption is occurring at the surface, the power density overlap with the surface determines the responsivity, meaning sensitivity will change depending on polarization and waveguide geometry. Employing different waveguide geometries, such as a slot waveguide , could increase surface overlap. The surface itself could also be altered. Increased roughness would increase the total surface area and provide more absorption, as the devices are not limited by optical scattering loss. Finally, performance will vary depending on the level of surface passivation. The results presented are for a native oxide, alternative surface treatments would result in different levels of optical absorption, recombination and quantum efficiency.
A monolithic-silicon CMOS compatible photodetector utilizing surface state defects and operating in the C-band at 10 Gb/s has been demonstrated. The sensitivity has been increased over previous efforts to 0.045 A/W at 2 V by employing a p-i-n junction in combination with selective exposure of the silicon waveguide. The junction also enables operation at 10 Gb/s at a reverse bias voltage of 2 V. The ease of fabrication makes this detector especially attractive for power monitoring applications.
The authors thank Dan Deptuck and Jessica Zhang of CMC Microsystems for facilitating device design and fabrication, Julia Huang and the Canadian Centre for Electron Microscopy for imaging and Martin Gerber for useful discussion. The authors acknowledge the support of the Natural Sciences and Engineering Research Council of Canada and CMC Microsystems.
References and links
1. M. Hochberg and T. Baehr-Jones, “Towards fabless silicon photonics,” Nature Phot. 4(8), 492–494 (2010). [CrossRef]
2. R. Beausoleil, “Large-scale integrated photonics for high-performance interconnects,” ACM J. Emerg. Technol . 7(2), Article 6 (2011).
3. C. Lee and J. Thillaigovindan, “Optical nanomechanical sensor using a silicon photonic crystal cantilever embedded with a nanocavity resonator,” Appl. Opt. 48(10), 1797–1803 (2009). [CrossRef] [PubMed]
4. S. Janz, D.-X. Xu, M. Vachon, N. Sabourin, P. Cheben, H. McIntosh, H. Ding, S. Wang, J. H. Schmid, A. Delâge, J. Lapointe, A. Densmore, R. Ma, W. Sinclair, S. M. Logan, R. Mackenzie, Q. Y. Liu, D. Zhang, G. Lopinski, O. Mozenson, M. Gilmour, and H. Tabor, “Photonic wire biosensor microarray chip and instrumentation with application to serotyping of Escherichia coli isolates,” Opt. Express 21(4), 4623–4637 (2013). [CrossRef] [PubMed]
6. J. Michel, J. Liu, and L. C. Kimerling, “High performance Ge-on-Si photodetectors,” Nature Phot. 4(8), 527–534 (2010). [CrossRef]
7. 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]
8. J. J. Ackert, A. S. Karar, D. J. Paez, P. E. Jessop, J. C. Cartledge, and A. P. Knights, “10 Gbps silicon waveguide-integrated infrared avalanche photodiode,” Opt. Express 21(17), 19530–19537 (2013). [CrossRef] [PubMed]
9. R. R. Grote, K. Padmaraju, B. Souhan, J. B. Driscoll, K. Bergman, and R. M. Osgood Jr., “10 Gb/s error-free operation of all-silicon ion-implanted-waveguide photodiodes at 1.55 µm,” IEEE Photon. Technol. Lett. 25(1), 67–70 (2013). [CrossRef]
10. B. Souhan, C. P. Chen, R. R. Grote, J. B. Driscoll, N. Ophir, K. Bergman, and R. M. Osgood, “Error free operation of an all-Si waveguide photodiode at 1.9 µm,” IEEE Photon. Technol. Lett. 25(21), 2031–2034 (2013). [CrossRef]
11. H. Chen and A. W. Poon, “Two-photon absorption photocurrent in p-i-n embedded silicon microdisk resonators,” Appl. Phys. Lett. 96(19), 191106 (2010). [CrossRef]
13. F. Morichetti, S. Grillanda, M. Carminati, G. Ferrari, M. Sampietro, M. J. Strain, M. Sorel, and A. Melloni, “Non-Invasive on-chip light observation by contactless waveguide conductivity monitoring,” IEEE J. Sel. Top. Quantum Electron. 20(4), 8201710 (2014). [CrossRef]
14. F. Van Laere, T. Claes, J. Schrauwen, S. Scheerlinck, W. Bogaerts, D. Taillaert, L. O’Faolain, D. Van Thourhout, and R. Baets, “Compact focusing grating couplers for silicon-on-insulator integrated circuits,” IEEE Photon. Technol. Lett. 19(23), 1919–1921 (2007). [CrossRef]
15. A. B. Sproul, “Dimensionless solution of the equation describing the effect of surface recombination on carrier decay in semiconductors,” J. Appl. Phys. 76(5), 2851 (1994). [CrossRef]
16. K. Padmaraju, D. F. Logan, X. Zhu, J. J. Ackert, A. P. Knights, and K. Bergman, “Integrated thermal stabilization of a microring modulator,” Opt. Express 21(12), 14342–14350 (2013). [CrossRef] [PubMed]
17. A. Densmore, D. X. Xu, P. Waldron, S. Janz, P. Cheben, J. Lapointe, A. Delage, B. Lamontagne, J. H. Schmid, and E. Post, “A silicon-on-insulator photonic wire based evanescent field sensor,” IEEE Photon. Technol. Lett. 18(23), 2520–2522 (2006). [CrossRef]