We propose GeOx passivation by plasma post-oxidation for dark-current suppression in a germanium (Ge) photodetector (PD). A GeOx/Ge interface exhibits a significantly lower interface trap density than SiO2/Ge and Al2O3/Ge interfaces. GeOx passivation on a Ni/Ge Schottky diode decreases the dark current under −1 V bias by more than one order of magnitude compared with Al2O3 passivation, which is attributed to the reduction in the surface leakage current. We also evaluated the Ge surface potential to study its effect on the surface leakage current. It was found that the surface leakage is suppressed when the accumulation condition of the Ge surface is enhanced as a result of fixed charges in the passivation layer. Thus, we have revealed the importance of a low interface trap density at the Ge surface and a suitable number of fixed charges in the passivation layer for achieving a low dark current in Ge metal-semiconductor-metal (MSM) PDs. Finally, we have examined the effect of GeOx passivation on a normal-incidence Ge MSM PD. We observed a significant decrease in the dark current in the GeOx-passivated samples, and a dark current of 97 nA under −1 V bias was achieved under the optimal GeOx passivation conditions.
© 2015 Optical Society of America
The tremendous growth in internet traffic has triggered a growing interest in developing higher-performance optical communication systems, in which photodetectors (PDs) with high speed, high responsivity, and low power consumption are expected to be key elements [1,2]. Germanium (Ge) has been perceived as one of the most promising materials for on-chip photodetectors owing to its large absorption coefficient at near-infrared wavelengths, low cost, and compatibility with current silicon complementary metal-oxide-semiconductor (CMOS) technologies [3–5]. Furthermore, a Ge metal-semiconductor-metal (MSM) PD has major advantages in optical fiber communication applications for low cost and fast fabrication due to process simplicity of MSM PDs [6, 7]. However, the large dark current in Ge MSM PDs is a critical technology issue since it severely degrades device performance . This large dark current is also unfavorable for large-scale optical systems for which low power consumption is strongly required in each component.
To reduce the large dark current in MSM PDs, considerable effort has been made to suppress the bulk leakage current through the Schottky junction between the metal and Ge, which is dominated by the thermal emission process, mainly by increasing the Schottky barrier height [9–11]. However, the dark-current density of Ge MSM PDs is still rather large, typically on the order of 100 mA/cm2 [5, 9–14].
The surface leakage current (JS), such as the trap-assisted tunneling current, can also significantly contribute to the dark current, especially for MSM PDs with interdigitated finger electrodes, in which the ratio of the surface area to the volume is large. We consider that the large number of defects on the Ge surface, which is evaluated as the interface trap density (Dit), can serve as leakage paths for the surface leakage current. Actually, the difficulty of passivating a Ge surface results in a relatively high Dit between the Ge surface and most passivation layers, which is usually on the order of 1012 - 1013 cm−2eV−1 [15–17]. Recently, surface passivation by thermally grown GeO2 has shown great potential for the reduction of JS for pn-junction Ge PDs  since the GeO2/Ge interface formed by thermal oxidation above 550 °C provides a good interface property with a low Dit on the order of 1011 cm−2eV−1 . Nonetheless, the GeO2 grown by high-temperature oxidation contains a large number of fixed charges, which cause an unexpectedly large shift in the flat-band voltage (VFB) of the GeO2/Ge structure , resulting in an increase in the dark current . Thus, an alternative passivation method that can be carried out at a lower temperature is required, especially for the dark-current suppression of Ge MSM PDs.
In this work, we have investigated the effect of GeOx passivation by a plasma post-oxidation process on Ge MSM PDs using bulk Ge wafer. Metal-oxide-semiconductor (MOS) capacitors were fabricated in advance to study the properties of interfaces between Ge and different passivation layers. Schottky diodes with a specifically designed junction pattern with different peripheral lengths and the same area are also fabricated to extract the surface leakage component from the dark current. The impact of interface properties such as the interface trap density (Dit) and the surface potential (ψs) on the surface leakage current is studied to determine suitable passivation conditions for low-dark-current operation. We also fabricated Ge MSM PDs to examine the effects of passivation on dark-current suppression. The dark current of a GeOx-passivated Ge MSM PD with a suitable surface potential is successfully reduced by more than one order of magnitude as compared with that of a conventional SiO2-passivated device.
2. Effects of GeOx passivation on Ge surface
Instead of the conventional SiO2 passivation of a Ge surface by plasma-enhanced chemical vapor deposition (PECVD), we propose the GeOx passivation by an electron cyclotron resonance (ECR) plasma post-oxidation process . The process flow of GeOx passivation is shown in Fig. 1(a). First, a 1-nm-thick Al2O3 layer is deposited on a Ge surface by atomic layer deposition (ALD) at 300 °C. Then, ECR plasma of a mixture gas of Ar (9 SCCM) and O2 (3 SCCM) (SCCM denotes cubic centimeter per minute at STP) with a 650 W microwave generated by JSW Afty AFTEX-2300 system is irradiated on the Al2O3/Ge samples at 300 °C for 10 sec to form a 1.2-nm-thick GeOx layer between the Al2O3 and Ge. This GeOx layer primarily contributes to the passivation of the Ge surface. Meanwhile, the Al2O3 layer deposited in advance protects the GeOx layer from air exposure, since the Ge oxide is vulnerable to moisture and degrades rapidly if directly exposed to air. It should be noted that ALD Al2O3 is chosen because it is a uniform and dense film enough for protecting the GeOx layer from water, and meanwhile it forms a stable interface between Al2O3/GeOx layers. In a previous study , a GeOx/Ge interface formed by GeOx passivation exhibited superior properties with a low Dit on the order of 1011 cm−2eV−1. Additionally, the low temperature of GeOx passivation enables a small VFB shift of the MOS structure.
To study the properties of interfaces between Ge and different passivation layers, MOS capacitors with a Ni/passivation layer/Ge structure are fabricated. In this study, we examined 20 nm Al2O3, 20 nm Al2O3/1.2 nm GeOx, 20 nm SiO2, and 20 nm SiO2/1 nm Al2O3/1.2 nm GeOx passivation layers. The properties of interfaces between Ge and different passivation layers can be evaluated as Dit, obtained by the low temperature conductance method . Figure 1(b) shows the Dit distribution evaluated from the capacitors with different passivation layers. The Dit distribution of a capacitor with GeO2 passivation performed by thermal oxidation at 550 °C is also plotted for reference.
From the Dit distributions result, it is clearly shown that the capacitors with GeOx passivation (namely, a GeOx/Ge interface) have a lower Dit value than the Al2O3- and SiO2-passivated capacitors (namely, Al2O3/Ge and SiO2/Ge interfaces) by at least one order of magnitude. Meanwhile, the GeOx-passivated capacitors have the same Dit level as the capacitor with the thermally grown GeO2 passivation, indicating that the GeOx/Ge interface formed by GeOx passivation has comparable properties with the thermal GeO2/Ge interface.
3. Schottky diode characteristics and extraction of surface leakage component
Since a metal/Ge Schottky junction is the main component of Ge MSM PDs, Schottky diodes are fabricated to study the effects of surface passivation on dark-current suppression. We fabricated Ni/Ge Schottky diodes with different passivation methods on (001) n-type bulk Ge substrates with a doping concentration of 1 × 1016 cm−3. After the precleaning of the Ge surface by HCl solution, surface passivation is performed. Then the Schottky junction pattern is defined on the passivation layer by conventional photolithography and the junction area is opened by BHF wet etching, which is followed by the thermal evaporation of Ni and the lift-off process. Finally Al is deposited as the back contact electrode by thermal evaporation.
Figure 2 shows the current-voltage (I-V) characteristics of Schottky diodes with the same junction pattern and different passivation layers of (1) 20 nm Al2O3 and (2) 20 nm Al2O3/1.2 nm GeOx. From the I-V results, it is clearly observed that the dark current of the GeOx-passivated sample is much lower than that of the Al2O3-passivated sample by one order of magnitude under −1 V bias. Note that both samples are fabricated simultaneously after the surface passivation process, meaning that the Ni/Ge junction properties are similar. Therefore, the decrease in the dark current should mainly be due to the introduction of the GeOx passivation layer, which exhibits a low Dit as shown in Fig. 1(b).
To further analyze the effects of passivation on the surface leakage current, we extracted the surface leakage component from the measured dark current. Assuming that the surface leakage component (JS) is proportional to the peripheral length of the Schottky junction pattern (LPeripheral) while the bulk leakage component (JB) is proportional to the junction area (SArea), the total dark current can be expressed as
Therefore, by applying some specifically designed Schottky junction patterns with various peripheral lengths and the same junction area, the JS value can be obtained from the slope of the total dark current plotted as a function of the peripheral length. Figure 3(a) shows the extraction of JS using a set of Schottky diodes with passivation layers of (1) 20 nm Al2O3 and (2) 20 nm Al2O3/1.2 nm GeOx from the dark current under −1 V bias, and the results are summarized in Fig. 3(b).
It is clearly shown that the JS has been reduced by one order of magnitude in the sample with GeOx passivation, which mainly contributes to the subsequent decrease in the total dark current. This result supports our suggestion that the surface leakage current, which may be attributed to trap-assisted tunneling, can be significantly suppressed by an improved interface with lower Dit.
4. Evaluation of Ge surface potential and its effect on surface leakage current
In addition to the interface traps, the fixed charges contained in the passivation layer may also affect the surface leakage current since they change the Ge surface condition. These fixed charges originate from the fabrication process including oxide growth, ionizing radiation, and bombardment with high-energy photons or particles . Meanwhile, the level of the fixed charges is modifiable by either changing the process temperature or performing sample annealing after the formation of passivation layer [24, 25]. In this study, we have confirmed that the level of the fixed charges in the passivation layer was reproducible if the same process conditions were applied.
The Ge surface condition is evaluated as the Ge surface potential (ψs). Considering the structure of Ge with a passivation layer deposited on its surface, charge neutrality is always preserved in the passivation layer/Ge interface, which requiresFig. 4. By solving Poisson’s equation, Qs can be expressed by Eqs. (2) and (3).
Capacitance-voltage (C-V) measurement of the MOS capacitor provides the value of the Qox because the existence of fixed charges will result in a parallel shift of the C-V curve , and a corresponding shift of the gate voltage (Vg) at a certain capacitance. Thus, Qox can be obtained from 
Using the metal work function of Ni of 5.1 eV , we first calculate the ideal C-V curve of the MOS capacitor in accordance with Ref . Then, we obtain the measured high-frequency C-V curve at 1 MHz . Figure 5(a) shows an example of the calculated ideal C-V curve and the measured C-V curve of the MOS capacitor with the passivation layer of 20 nm Al2O3/1.2 nm GeOx. As shown in Fig. 5(a), Vg at a certain capacitance can be obtained by comparing the ideal and measured C-V curves. Note that the same capacitance value from the ideal and measured C-V curves should correspond to the same ψs on the Ge surface .
Because of the existence of interface traps on the passivation layer/Ge interface, the fixed charges include some charged interface traps, the number of which depends on ψs. Therefore Qox is also a function of ψs similarly to Qs. Figure 5(b) shows plots of the absolute values of Qs and Qox as function of ψs calculated by Eqs. (3) and (4). The ψs value can be obtained from the point of intersection between the curves of Qox(ψs) and Qs(ψs), where the charge neutrality requirement is met as Eq. (2). Through this procedure, we can obtain ψs for each passivated sample.
To investigate the effect of ψs on the dark current, MOS capacitors fabricated with various passivation methods and having various passivation layer thicknesses are prepared. Since the SiO2 layer deposited by PECVD contains positive charges inside , when assuming that the charges are uniformly distributed in the SiO2 layer, we can intentionally change the amount of fixed charges by changing the thickness of the passivation layer, in particular, the thickness of SiO2 layer. By applying the above method, ψs is evaluated for these MOS capacitors with different passivation methods as a function of the passivation layer thickness as shown in Fig. 6(a). It is found that ψs is positive regardless of the passivation layer, suggesting that all the Ge surfaces are under a weak accumulation condition as shown in Fig. 4. However, the accumulation level is reduced with increasing passivation layer thickness owing to the fixed charges in the passivation layer.
To evaluate the surface leakage current, Ni/Ge Schottky diodes with exactly the same passivation layers are also fabricated by applying the process described in the previous section. Similarly, JS is extracted by the same method depicted in Fig. 3(a). As show in Fig. 6(b), the surface leakage current tends to increase with the thickness of the passivation layer. Combining the results for ψs evaluated from the capacitors and those for JS extracted from the Schottky diodes, the relationship between ψs and JS is obtained, as shown in Fig. 6(c). It is clearly found that the surface leakage current decreases when the accumulation level of the Ge surface is enhanced. This is because when the accumulation level is enhanced, fewer generation-recombination centers can contribute to the trap-assisted tunneling process , thus suppressing the surface leakage current. The SiO2-passivated samples exhibit larger JS than the GeOx-passivated samples, which is attributed to the difference in the value of Dit as shown in Fig. 3. Thus, these results reveal that a low interface trap density and the accumulation condition should be simultaneously achieved for low-dark-current operation in Ge MSM PDs.
5. Characteristics of Ge MSM PDs
On the basis of the findings in the previous sections, normal-incidence Ge MSM PDs are fabricated on an n-Ge substrate to examine the effect of passivation on the dark-current suppression. The fabrication process for the MSM PDs is similar to that for the Schottky diodes. The only difference is a second patterning and lift-off process to form Ni contact pads on the dielectric passivation layer after the formation of the Ni/Ge Schottky junction.
The dark current of the MSM PDs is first studied by obtaining the I-V characteristics. Figure 7(a) shows I-V curves of the MSM PDs with different passivation layers. The inset shows a top-view microscopic image of a fabricated device.
It is clearly observed that the MSM PDs with the GeOx passivation exhibit a much lower dark current than the samples without the GeOx passivation, by more than one order of magnitude under −1 V bias, which is consistent with the results obtained from the Schottky diodes in Fig. 6(c). The lowest dark current of 97 nA under −1 V bias is achieved in the sample with the Al2O3/GeOx passivation layer, which has the finger width of 1 μm, finger spacing of 2 μm, and active PD area of 1225 μm2. The GeOx passivation effectively suppress the dark current in Ge MSM PDs, making it even promising to reduce the shot noise when operating in the avalanche mode and improve its sensitivity [29, 30].
The photocurrent of the MSM PDs is also measured using a normal incidence light input from a single-mode fiber connected to a laser source. Figure 7(b) shows the I-V characteristic of the MSM PD with the 10 nm Al2O3/1.2 nm GeOx passivation layer under 0 dBm illumination at a wavelength of 1550 nm. The dark current is also plotted for reference. A constant photocurrent is obtained regardless of the bias voltage, and an on-off ratio of more than 103 is achieved. The corresponding responsivity is approximately 0.4 A/W, which is a reasonable value considering the shadowing effect of the interdigitated electrodes.
We have investigated the effect of GeOx passivation by the plasma post-oxidation method on dark-current suppression in Ge MSM PDs. Firstly we studied the effect of GeOx passivation on Ni/Ge Schottky diodes. The introduction of a GeOx layer resulted in the reduction of the total dark current by over one order of magnitude, which was mainly attributed to the decrease in the surface leakage current owing to the superior GeOx/Ge interface. We also studied the effect of fixed charges in the passivation layer on the surface leakage current by using MOS capacitors and Schottky diodes passivated under various conditions. By evaluating the relationship between the Ge surface potential and the surface leakage current, we found that the surface leakage current was suppressed when the accumulation condition on the Ge surface was enhanced, which was explained by the reduction of the trap-assisted tunneling current through the generation-recombination centers. We also applied GeOx passivation with the optimized number of fixed charge to a normal-incidence Ge MSM PD. The reduction of the dark current by the GeOx passivation was confirmed and a dark current of 97 nA with an on-off ratio of over 103 was achieved. These results suggest the importance of a low interface trap density at the Ge surface and a suitable amount of fixed charges in the passivation layer for low-dark-current operation in Ge MSM PDs. Moreover, since a similar surface leakage mechanism is always involved in any Ge PDs, we believe a well-controlled Ge surface condition is mandatory to achieve a low-dark-current operation in both Ge MSM PDs and Ge PIN PDs fabricated even on Ge epitaxially grown on Si substrate. As a result, GeOx passivation, which can be performed at a low temperature of approximately 300 °C, was found to be a promising method for dark-current suppression in Ge PDs.
This work was partly supported by the New Energy and Industrial Technology Development Organization (NEDO) “Integrated Photonics-Electronics Convergence System Technology (PECST)” project and a MEXT Grant-in-Aid for Scientific Research (S).
References and links
1. L. C. Kimerling, L. Dal Negro, S. Saini, Y. Yi, D. Ahn, S. Akiyama, D. Cannon, J. Liu, J. G. Sandland, D. Sparacin, J. Michel, K. Wada, and M. R. Watts, “Monolithic silicon microphotonics,” Chapter in Silicon Photonics, Top. Appl. Phys. 94, 89–120 (2004).
2. A. Saha and N. Manna, Optoelectronics and Optical Communication (University Science, 2011).
3. L. C. Kimerling, D. Ahn, A. B. Apsel, M. Beals, D. Carothers, Y. K. Chen, T. Conway, D. M. Gill, M. Grove, C. Y. Hong, M. Lipson, J. Liu, J. Michel, D. Pan, S. S. Patel, A. T. Pomerene, M. Rasras, D. K. Sparacin, K. Y. Tu, A. E. White, and C. W. Wong, “Electronic-photonic integrated circuits on the CMOS platform,” Proc. SPIE 6125, 6–15 (2006).
4. D. Ahn, C. Y. Hong, J. Liu, W. Giziewicz, M. Beals, L. C. Kimerling, J. Michel, J. Chen, and F. X. Kärtner, “High performance, waveguide integrated Ge photodetectors,” Opt. Express 15(7), 3916–3921 (2007). [CrossRef] [PubMed]
6. L. Vivien, M. Rouvière, J. M. Fédéli, D. Marris-Morini, J. F. Damlencourt, J. Mangeney, P. Crozat, L. El Melhaoui, E. Cassan, X. Le Roux, D. Pascal, and S. Laval, “High speed and high responsivity germanium photodetector integrated in a silicon-on-insulator microwaveguide,” Opt. Express 15(15), 9843–9848 (2007). [CrossRef] [PubMed]
8. J. Michel, J. Liu, and L. C. Kimerling, “High-performance Ge-on-Si photodetectors,” Nat. Photonics 4(8), 527–534 (2010). [CrossRef]
9. J. Oh, S. K. Banerjee, J. C. Campbell, J. Oh, S. K. Banerjee, and J. C. Campbell, “Metal-germanium-metal photodetectors on heteroepitaxial Ge-on-Si with amorphous Ge Schottky barrier enhancement layers,” IEEE Photon. Technol. Lett. 16(2), 581–583 (2004). [CrossRef]
10. K.-W. Ang, S.-Y. Zhu, J. Wang, K. T. Chua, M.-B. Yu, G.-Q. Lo, and D.-L. Kwong, “Novel silicon-carbon (Si:C) Schottky barrier enhancement layer for dark-current suppression in Ge-on-SOI MSM photodetectors,” IEEE Electron Device Lett. 29(7), 704–707 (2008). [CrossRef]
11. K.-W. Ang, M.-B. Yu, S.-Y. Zhu, K. T. Chua, G.-Q. Lo, and D.-L. Kwong, “Novel NiGe MSM photodetector featuring asymmetrical Schottky barriers using sulfur co-implantation and segregation,” IEEE Electron Device Lett. 29(7), 708–710 (2008). [CrossRef]
12. S. Assefa, F. Xia, S. Bedell, Y. Zhang, T. Topuria, P. Rice, and Y. Vlasov, “CMOS-integrated 40GHz germanium waveguide photodetector for on-chip optical interconnects,” Opt. Fiber Commun. Conf., OMR4 (2009). [CrossRef]
13. D. Buca, S. Winnerl, S. Lenk, S. Mantl, and C. Buchal, “Metal–germanium–metal ultrafast infrared detectors,” J. Appl. Phys. 92(12), 7599 (2002). [CrossRef]
14. M. Rouvière, L. Vivien, X. Le Roux, J. Mangeney, P. Crozat, C. Hoarau, E. Cassan, D. Pascal, S. Laval, J.-M. Fédéli, J.-F. Damlencourt, J. M. Hartmann, and S. Kolev, “Ultrahigh speed germanium-on-silicon-on-insulator photodetectors for 1.31 and 1.55μm operation,” Appl. Phys. Lett. 87(23), 231109 (2005). [CrossRef]
15. C.-C. Cheng, C.-H. Chien, G.-L. Luo, J.-C. Liu, C.-C. Kei, D.-R. Liu, C.-N. Hsiao, C.-H. Yang, and C.-Y. Chang, “Characteristics of atomic-layer-deposited Al2O3 high-k dielectric films grown on Ge substrates,” J. Electrochem. Soc. 155(10), G203–G208 (2008). [CrossRef]
16. V. V. Afanas’ev, Y. G. Fedorenko, and A. Stesmans, “Interface traps and dangling-bond defects in (100) Ge/HfO2,” Appl. Phys. Lett. 87(3), 032107 (2005). [CrossRef]
17. C. Claeys and E. Simoen, Germanium-Based Technologies: From Materials to Devices (Elsevier, 2007).
18. M. Takenaka, K. Morii, M. Sugiyama, Y. Nakano, and S. Takagi, “Dark current reduction of Ge photodetector by GeO₂ surface passivation and gas-phase doping,” Opt. Express 20(8), 8718–8725 (2012). [CrossRef] [PubMed]
19. H. Matsubara, T. Sasada, M. Takenaka, and S. Takagi, “Evidence of low interface trap density in GeO2/Ge metal-oxide-semiconductor structures fabricated by thermal oxidation,” Appl. Phys. Lett. 93(3), 032104 (2008). [CrossRef]
20. Y. Oniki, H. Koumo, Y. Iwazaki, and T. Ueno, “Evaluation of GeO desorption behavior in the metal/GeO2/Ge structure and its improvement of the electrical characteristics,” J. Appl. Phys. 107(12), 124113 (2010). [CrossRef] [PubMed]
21. R. Zhang, T. Iwasaki, N. Taoka, M. Takenaka, and S. Takagi, “Al2O3/GeOx/Ge gate stacks with low interface trap density fabricated by electron cyclotron resonance plasma postoxidation,” Appl. Phys. Lett. 98(11), 112902 (2011). [CrossRef]
22. E. Nicollian and A. Goetzberger, “The Si-SiO2 interface electrical properties as determined by the metal-insulator-silicon conductance technique,” J. Bell Syst. Tech. 46(6), 1055–1133 (1967). [CrossRef]
23. Y. Taur and T. Ning, Fundamentals of Modern VLSI Devices (Cambridge University, 1998).
24. V. E. Drozd, A. P. Baraban, and I. O. Nikiforova, “Electrical properties of Si-Al2O3 structures grown by ML-ALE,” Appl. Surf. Sci. 82-83, 583–586 (1994). [CrossRef]
25. T. Sameshima, K. Sakamoto, Y. Tsunoda, and T. Saitoh, “Improvement of SiO2 properties and silicon surface passivation by heat treatment with high-pressure H2O vapor,” Jpn. J. Appl. Phys. 37(Part 2, No. 12A), L1452–L1454 (1998). [CrossRef]
26. K. C. Chiang, C. H. Cheng, K. Y. Jhou, H. C. Pan, C. N. Hsiao, C. P. Chou, S. P. McAlister, A. Chin, and H. L. Hwang, “Use of a high-work-function Ni electrode to improve the stress reliability of analog SrTiO3 metal-insulator-metal capacitors,” IEEE Electron Device Lett. 28(8), 694–696 (2007). [CrossRef]
27. L. Terman, “An investigation of surface states at a silicon/silicon oxide interface employing metal-oxide-silicon diodes,” Solid-State Electron. 5(5), 285–299 (1962). [CrossRef]
28. A. G. Aberle, S. Glunz, and W. Warta, “Impact of illumination level and oxide parameters on Shockley-Read-Hall recombination at the Si/SiO2 interface,” J. Appl. Phys. 71(9), 4422 (1992). [CrossRef]
29. N. C. Harris, T. Baehr-Jones, A. E.-J. Lim, T. Y. Liow, G. Q. Lo, and M. Hochberg, “Noise characterization of a waveguide-coupled MSM photodetector exceeding unity quantum efficiency,” J. Lightwave Technol. 31(1), 23–27 (2013). [CrossRef]