Robust Optimization Of A Silicon Lateral Pin Photodiode

Transcription

1 Australian Journal of Basic and Applied Sciences, 6(8): , 2012 ISSN Robust Optimization Of A Silicon Lateral Pin Photodiode 1 S. Kalthom Tasirin, 2 P. Susthitha Menon, 1 Ibrahim Ahmad and 1 S.Fazlili Abdullah 1 Department of Electronics and Communication Engineering College of Engineering Universiti Tenaga Nasional Jalan IKRAM-UNITEN Kajang, Selangor, Malaysia 2 Institute of Microengineering and Nanoelectronics (IMEN) Universiti Kebangsaan Malaysia Bangi, Selangor, Malaysia Abstract: The objective of this paper is to optimize the process parameters for a Si lateral p-in photodiode to obtain high responsivity, frequency response, quantum efficiency and low transient time Four process parameters were chosen, namely the intrinsic region length, photo-absorption layer thickness, the incident optical power and the bias voltage. The Taguchi method technique was used to optimize the experiment. Two noise factors were used that consist of four measurements for each row of experiment in the L 9 array. ATHENA and ATLAS module from Silvaco Int. were used for fabrication simulation and electrical characterization. The results obtained for responsivity, quantum efficiency, frequency response and transient time after the optimization approach were 0.62A/W, 96.37%, 13.1 GHz and x respectively which correspond to the optimization value for intrinsic region length of 6 µm, photo-absorption layer thickness of 50 µm, incident optical power of 1 mw/cm 2 and bias voltage of 3 V. As a conclusion, the optimum solution in achieving the desired high speed photodiode was successfully predicted by using Taguchi optimization method. The percent of improvement for quantum efficientcy is 25%. Key words: Taguchi method, photodetector device, ATHENA, ATLAS. INTRODUCTION Lateral pin photodiodes based on Si substrates have been used for high data rate applications where the main focus of device structure and design has been to achieve wide bandwidth operation while increasing the quantum efficiency, responsivity and frequency response. In order to keep the cost of the short-haul data communication system low, the Si lateral pin photodiode is best to be monolithically integrated. In literature, the Si photodiode in standard 0.18 µm CMOS technology exhibited frequency response of 8.7 GHz and responsivity of A/W at 11.4-V bias (Fang-ping Chou et al., 2010). In previous research (S.K. Tasirin, et al., 2012; Menon, P. S. 2005; Menon, P. S. & Shaari, S. 2005), the results of quantum efficiency and transient time were 76.6% and 1.54 x 10-7 s. In another research (Fujikata, J. et al., 2008) was reported the response time of bulk CMOS was 0.19 µs and the response time of SOI CMOS was 0.08 µs for a photodiode effective dimension 20 µm x 20 µm. This paper investigates the optimization of the process parameters on the frequency response and responsivity of a Si lateral pin-photodiode based on the device which was developed previously (S.K. Tasirin, et al., 2012; Menon, P. S. 2005; Menon, P. S. & Shaari, S. 2005). A two dimensional model of a Si lateral pinphotodiode operating at the optical wavelength, λ from 700 nm to 800 nm was developed using an industrial based numerical software. Statistical optimization of the Si lateral pin-photodiode model was executed using L 9 orthogonal array by using Taguchi optimization method. Taguchi method provide the most efficient and viable solution in such cases with minimal experimental trials. The optimization of design parameters is one of the vital industrial functions to improve the product performance as well as to save manufacturing cost. In another research (F.Salehuddin, et al., 2011), Taguchi method was used in the optimization of gate oxide and silicide thickness for 45nm NMOS device. The optimization method of designing experiments based on Taguchi Methods for optimal solution in producing 32nm CMOS technology transistor with desired leakage current was reported (H.A. Elgomati, et al., 2011). In this work, both numerical modeling and Taguchi optimization was used. Improvement was obtained for the quantum efficiency and transient response were obtained from the best setting of the process parameters after the optimization approach. This methodology aids in cost-effective device optimization prior to the actual fabrication of the device. MATERIAL AND METHODS Corresponding Author: S. Kalthom Tasirin, Department of Electronics and Communication Engineering College of Engineering Universiti Tenaga Nasional Jalan IKRAM-UNITEN Kajang, Selangor, Malaysia sitikalthom@gmail.com 275

2 The Si lateral pin-photodiode was simulated on silicon substrate (n + -type) in two dimensional using ATHENA software from Silvaco Int. Then, the n-well was developed using phosphorus diffusion with dopant concentration of 2.02 x cm -3 on the left side of the photodiode with diffusion temperature of C for50 seconds. Whilst the p-well was diffused with boron concentrationn of 8.09 x10 19 cm -3 on the other side of the photodiode with temperature of C for 120seconds. SiO 2 layer with thickness of 280 nm was deposited on the silicon substrate to act as a passivation layer. The electrode contacts with thickness and length of 500 nm and 6000 nm respectively, were processed by depositing aluminumm on the n-well and p-well areas of the silicon photodiode. The developed device structure is shown in Figure 1. Fig. 1: The device structure of Si lateral pin-photodiode The device s electrical and optical characteristics were executed using the ATLAS module from Silvaco Int. The responsivity of a pin-photodiode is given (Menon, P.S.K. et al., 2011): ( 1) Wheree I s is the source photocurrent, I T is the cathode current and λ is the optical wavelength. Calculation for the total quantum efficiency (%) is given by equation (2) (Menon, P. S. 2005): ( 2) The frequency response is defined as (Menon, P.S.K. et al., 2011): ( 3) Wheree I R is the real cathode current and I Ro is the real component current. Whilst, the transient response, t r is given by equation (4) (Menon, P. S. 2005): ( 4) L 9 Orthogonal Array by Taguchi Optimization Method: In this research, four factors were identified namely the intrinsic region length, photo-absorption layer thickness, incident optical power and bias voltage which were tested at three levels using the L9 orthogonal array by Taguchi optimization method. These factors are portrayed in Figure 1. Whilst, the noise factors were the time and temperaturee of the n-well diffusion were tested at two levels. By using the S/N ratio and ANOVA Pareto, it allows us to make accurate conclusion for the experiment either the factor is giving dominant effect or minimum effect. To find the optimum factors and levels, signal to noise ratio (SNR) of larger the better (LTB) was applied to examine the performance factors of the devicee namely the frequency response and the responsivity. The value for the variation was chosen according to previous research. Using L9orthogonal array method, nine sets of experiments were used to vary the parameter for the intrinsic region length between 6 µm to 16 µm, the photo-absorption layer thickness between 15 µm to 80 µm, the incident optical power between 1 mw/cm2 to 20 mw/cm2 and the bias voltage between 1 V to 3V as shown in Table 1. Then, the parameters of the noise factors for the diffusion time and temperature were tested at two levels will create four measurements for each row of L9orthogonal array. Table 1: Control factors and their levels Set Parameter Level Reference

3 1 2 3 A Intrinsic region length (µm) (Michelly de Souza, et al., 2011) B Photoabsorption layer thickness (µm) (Yibin Bai, et al., 2008) C Incident optical power (mw/cm2) Michelly de Souza, et al., D Bias voltage (V) (Fang-ping Chou2010) The frequency response and responsivity were studied as the output characteristic of the pin-photodiode. Table 2 shows the L9orthogonal array to be inserted into Variant 4 factor 3 level. Table 2: Variant 4 factor 3 level of Taguchi optimization method Exp.No. Control factors A B C D Results: The simulation results for the frequency response and responsivity of the Si lateral pin-photodiode is shown in Table 3 and Table 4 respectively. The frequency response and responsivity of the Si lateral pin-photodiode is attributed to SNR of larger-the-better in Taguchi optimization method. The SNR η can be expressed as: 1 1 SN 10log (5) n y 2 where µ is the mean and σ is the variance. By applying Eq. (5), η for each device was calculated and given in Table 5. The effect of each process parameter on the SNR at different levels can be separated because the experimental design is orthogonal. The SNR values for each level of the process parameters are summarized in Table 6. In addition, the total mean of SNR for these 9 experiments has been also calculated and listed in Table 6. Table 3: Frequency response values for Si lateral pin-photodiode Exp.No. Frequency response (GHz) Table 4: values for Si lateral pin-photodiode Exp.No. (A/W) Table 5: SNR for frequency response and responsivity Exp. No. SNR (db) 277

4 Frequency response Table 6: SNR of the frequency response and responsivity for each level. Response Process SNR (larger-the-best) Total Maxmin parameter Level 1 Level 2 Level 3 mean SNR A Frequency B response C D A B C D Discussion: Analysis of Variance (anova): A better feel for the relative effect of the different process parameter on the responsivity and frequency response were obtained by decomposition of variance, which is called analysis of variance (ANOVA) (Abdullah, H. et al. 2009). The priority of the process parameters with respect to the responsivity and frequency response were investigated to determine more accurately the optimum combinations of the process parameters. The result of ANOVA for the Si lateral pin-photodiode device is presented in Table 7. Table 7: Result of ANOVA for responsivity and frequency response in Si lateral pin-photodiode Response Process parameter Degree of Sum of freedom square Mean square Factor effect on SNR (%) A Frequency response B C D A B C D Statistically, F-test provides a decision at some confidence level as to whether these estimates are significantly different. The percent factor effect on SNR indicates the priority of a factor (process parameter) to reduce variation. For a factor with a high percent contribution, it will have a great influence on the performance (Ugur Esme. 2009). For the responsivity characteristic, the incident power optical was found to be the major factor affecting the responsivity (56%), whereas bias voltage and photo-absorption layer thickness were the second ranking factor (22%). The percentage effect on SNR for the intrinsic region length is low at 0%. For the frequency response characteristic, intrinsic region length was found to be the most significant factor affecting 100% of the device performance. The optimized factors for responsivity and frequency response in Si lateral pin-photodiode device which had been suggested by Taguchi optimization method are shown in Table 8. Table 8: Best setting of the process parameters Response Process parameter Unit Best value Intrinsic region length (µm) 6 Frequency response & Photoabsorption layer thickness (µm) 50 responsivity Incident optical power (mw/cm2) 1 Bias voltage (V) 3 From the above parameters as shown in Table 8, the final simulation was performed to verify the accuracy of the Taguchi optimization method prediction. In this research, intrinsic region length has the strongest effect on the frequency response characteristics and incident power optical has the strongest effect on the responsivity of the Si lateral pin-photodiode device. The best result for responsivity after the optimization approaches is 0.62 A/W at optical wavelength of 0.85nm is 38% better than the previously developed device where the responsivity was 0.45 A/W (Menon, P.S. 2005). Figure 2 shows the responsivity of the device after optimization. 278

5 Fig. 2: of Si lateral pin photodiode after optimization. The best result for frequency response after the optimization approach is13.1ghz. Previously, frequency response of 2.27 MHz was achieved at an optical wavelength of 850 nm for a 30 µm intrinsic width of a Si lateral pin-photodiode before prior to optimization (Menon, P.S. 2005).. The result is shown in Figure 3. Fig. 3: The -3dB frequency response. The best result for total quantum efficiency using the best setting after optimization approaches is 96.37% compare to 76.6% at 850nm optical wavelength for the device prior to optimization (Menon, P.S. 2005).. The best result of total quantum efficiency of photodiode is as in Figure 4. Fig. 4: The graph of total quantum efficiency %

6 The transient response for Si lateral pin photodiode using the best setting after optimization approach is x s and this is shown in Figure 5. This result is much better that the result of transient response for the device prior to optimization which is x 10-7 s (Menon, P.S. 2005).. Fig. 5: The graph of transient time response. The summary table for the comparison result of a Si Lateral PIN Photodiode characteristics between the device after and prior to optimization is shown in Table 9. Table 9: Summary table. Characteristics Frequency Response Quantum Efficiency Transient Time After Optimization A/W 13.1 GHz 96.4% x 10-11s Prior Optimization 0.45 A/W 2.27 MHz 76.6% 1.54 x 10-7s Conclusion: As a conclusion, the optimum solution in achieving the desired high speed photodiode was successfully predicted using Taguchi optimization method. Theree are many physical limitations involved as the size gets smaller approaching the molecular or atomic limitations of the substrate and dopant. In this work, Taguchi optimization method was used to optimize the responsivity and frequency response of a Si lateral pin- prior device. The incident optical power and the intrinsic region length were identified as control factors that have the strongest effect on the frequency response and responsivity of the device. ACKNOWLEDGEMENT The authors would like to thank the Public Service Department of Malaysia (JPA) for their financial support, Photonic Lab, Institute of Microengineerin ng and Nanoelectronics (IMEN), Universiti Kebangsaan Malaysia (UKM) and Pusat Latihan Teknologi Tinggi, Taiping (ADTEC) for the technical and moral support throughoutt the project. photodiode. Therefore, the optical characteristic results of the Si pin-photodiode become better than that of the REFERENCES Abdullah, H. et al Simulation of fabrication process VDMOSFET Transistorr Using Silvaco Software. European Journal of Scientific Research, 29(4): X. Elgomati, H.A., I. Ahmad, F. Salehuddin, F.A. Hamid, A. Zaharin, B.Y. Majlis, P.R. Apte, Optimal Solution in Producing 32nm CMOS Technology Transistor With Desired Leakage Current Semiconductor Physics, Quantum Electronics & Optoelectronics, 14( (2): Fang-ping Chou; Ching-Wen Wang; Guan-Yu Chen; Yue-ming Hsin, An 8.7 GHzSi photodiode in standard 0.18-μm CMOS technology, Opto Electronics and Communications Conference (OECC), th, : 5-9. Fujikata, J., K. Nose, J. Ushida, K. Nishi, M. Kinoshita, T. Shimizu, T. Ueno, D. Okamoto, A. Gomyo, M. Mizuno, T. Tsuchizawa, T. Watanabe, K. Yamada, S. Itabashi, K. Ohashi, Waveguide-integrated Si nano- IV Photonics 5th IEEE International Conference, pp: : photodiode with surface-plasmon antenna and its application to on-chip optical clock signal distribution, Group

7 Menon, P.S. & S. Shaari, Surface versus lateral illumination effects on an interdigitated Si planar PIN photodiode (Poster). Proceedings of the SPIE Symposium on Optics and Photonics: Infrared and Photoelectronic Imagers and Detector Devices, 2005, San Diego, USA, 5881: 58810S pp: 1-8. Menon, P.S., Pembangunan diodfoto planar p-i-n silikon (Development of silicon-based p-i-n photodiode), MSc Thesis. Universiti Kebangsaan Malaysia. Menon, P.S.K., A.A. Kandiah. Ehsan & S. Shaari, Concentration-dependent minority carrier lifetime in an In(0.53)Ga(0.47)As interdigitated lateral PIN photodiode based on spin-on chemical fabrication methodology. International Journal of Numerical Modelling: Electronic Networks, Devices and Fields., 24(5): Michelly de Souza, Olivier Bulteel, Denis Flandre and Marcelo A. Pavanello., Temperature and silicon film thickness influence on the operation of lateral SOI PIN photodiodes for detection of short wavelength, Journal of Integrated Circuits and Systems,v.6/n.1: Salehuddin, F., I. Ahmad, F.A. Hamid, A. Zaharim, Influence of HALO andsource/drain Implantation on Threshold Voltage in 45nm PMOS Device Australian Journal of Basic and Applied Sciences, 5(1): 55-61, 2011 ISSN Tasirin, S.K., P.S. Menon, I. Ahamd, S.F. Abdullah, Optimization of Process Parameters For Si Lateral PIN Photodiode. International Conference of Mathematical Aplications in Engineering. Ugur Esme, Application of Taguchi Method for the Optimization of ResistanceSpot Welding Process. The Arabian Journal for Science and Engineering, 34: 2B. Yibin Bai, Jagmohan Bajaj, James W. Beletic, Mark, C. Farris, Atul Joshi, StefanLauxtermann, Anders Petersen, George Williams, Teledyne Imaging Sensors:Silicon CMOS imaging technologies for x-ray, UV, visible and near infrared Proceedings of the SPIE Conference on Astronomical Instrumentation (2008, Marseille, France). 281