Issue Browser
Volume 28, Issue 2, Feb 2007
Column
LETTERS
DCXRD Investigation of a Ge/Si(001) Island Multilayer Structure
Shi Wenhua, Xue Chunlai, Luo Liping, Wang Qiming
Chin. J. Semicond.  2007, 28(2): 145-148
Abstract PDF

A Ge/Si(001) island multilayer structure is investigated by double crystal X-ray diffraction,transmission electron microscopy,and atomic force microscopy.We fit the satellite peaks in the rocking curve by two Lorentz lineshapes,which originate from the wetting layer region and the island region.Then from the ratio of the thicknesses of the Si and Ge (GeSi) layers as determined by TEM,the Ge compositions of the wetting layer and islands are estimated to be about 0.51 and 0.67,respectively,according to the positions of the fitted peaks.This proves to be a relatively simple way to investigate the Ge/Si (001) island multilayer structure.

A Ge/Si(001) island multilayer structure is investigated by double crystal X-ray diffraction,transmission electron microscopy,and atomic force microscopy.We fit the satellite peaks in the rocking curve by two Lorentz lineshapes,which originate from the wetting layer region and the island region.Then from the ratio of the thicknesses of the Si and Ge (GeSi) layers as determined by TEM,the Ge compositions of the wetting layer and islands are estimated to be about 0.51 and 0.67,respectively,according to the positions of the fitted peaks.This proves to be a relatively simple way to investigate the Ge/Si (001) island multilayer structure.
Ohmic Contact Properties of Multi-Metal Films on n-Type 4H-SiC
Han Ru, Yang Yintang, Wang Ping, Cui Zhandong, Li Liang
Chin. J. Semicond.  2007, 28(2): 149-153
Abstract PDF

An investigation of Au/Ti/Ni and Au/Ti/Pt ohmic contacts to n-type 4H-SiC and the behavior of metal films on SiC with thermal anneals is reported.Specific contact resistance as low as 2.765e-6Ω· cm2 was achieved after rapid thermal annealing in N2 for 2min at 950℃.SIMS analysis shows that the formation of Ni silicide after annealing supported a number of carbon atoms’ outdiffusion from the SiC to form interstitial compound TiC.This process can create abundant C vacancies near the interface.It is the carbon defect layer that enhances the defect-assisted tunneling.The interface band structure within the defect level could make it clear why the metal-SiC contacts become ohmic during annealing.

An investigation of Au/Ti/Ni and Au/Ti/Pt ohmic contacts to n-type 4H-SiC and the behavior of metal films on SiC with thermal anneals is reported.Specific contact resistance as low as 2.765e-6Ω· cm2 was achieved after rapid thermal annealing in N2 for 2min at 950℃.SIMS analysis shows that the formation of Ni silicide after annealing supported a number of carbon atoms’ outdiffusion from the SiC to form interstitial compound TiC.This process can create abundant C vacancies near the interface.It is the carbon defect layer that enhances the defect-assisted tunneling.The interface band structure within the defect level could make it clear why the metal-SiC contacts become ohmic during annealing.
InP/InGaAs Heterojunction Bipolar Transistor with Base μ-Bridge and Emitter Air-Bridge
Yu Jinyong, Liu Xinyu, Su Shubing, Wang Runmei, Xu Anhuai, Qi Ming
Chin. J. Semicond.  2007, 28(2): 154-158
Abstract PDF

An InP-based single-heterojunction bipolar transistor (SHBT) with base μ-bridge and emitter air-bridge is reported.Because those bridges reduce parasitic capacitance greatly,the cutoff frequency fT of the 2μm×12.5μm InP SHBT without de-embedding reaches 178GHz.It is critical in high-speed low power applications,such as OEIC receivers and analog-to-digital converters.

An InP-based single-heterojunction bipolar transistor (SHBT) with base μ-bridge and emitter air-bridge is reported.Because those bridges reduce parasitic capacitance greatly,the cutoff frequency fT of the 2μm×12.5μm InP SHBT without de-embedding reaches 178GHz.It is critical in high-speed low power applications,such as OEIC receivers and analog-to-digital converters.
SOICMOS Integrated Circuit of Laser Range Finding Working at High Temperatures
Gao Yong, Zhang Xin, Liu Mengxin, An Tao, Liu Shanxi, Ma Liguo
Chin. J. Semicond.  2007, 28(2): 159-165
Abstract PDF

Modeling analysis of thin fully depleted SOICMOS technology has been done.Using ISETCAD software,the high temperature characteristics of an SOICMOS transistor were simulated in the temperature range of from 300 to 600K,and the whole circuit of a laser range finder was simulated with Verilog software.By wafer processing,a circuit of a laser range finder with complete function and parameters working at high temperatures has been developed.The simulated results agree with the test results.The test of the circuit function and parameters at normal and high temperature shows the realization of an SOICMOS integrated circuit with low power dissipation and high speed,which can be applied in laser range finding.By manufacturing this device,further study on high temperature characteristics of shorter channel SOICMOS integrated circuits can be conducted.

Modeling analysis of thin fully depleted SOICMOS technology has been done.Using ISETCAD software,the high temperature characteristics of an SOICMOS transistor were simulated in the temperature range of from 300 to 600K,and the whole circuit of a laser range finder was simulated with Verilog software.By wafer processing,a circuit of a laser range finder with complete function and parameters working at high temperatures has been developed.The simulated results agree with the test results.The test of the circuit function and parameters at normal and high temperature shows the realization of an SOICMOS integrated circuit with low power dissipation and high speed,which can be applied in laser range finding.By manufacturing this device,further study on high temperature characteristics of shorter channel SOICMOS integrated circuits can be conducted.
PAPERS
New Lateral Super Junction MOSFETs with n+-Floating Layeron High-Resistance Substrate
Duan Baoxing, Zhang Bo, Li Zhaoji
Chin. J. Semicond.  2007, 28(2): 166-170
Abstract PDF

A new super junction LDMOST structure that suppresses the substrate-assisted depletion effect is designed with an n+-floating layer embedded in the high-resistance p-type substrate by implanting phosphor or arsenic.This effect results from a charge imbalance between the n-type and p-type pillars when the n-type pillars are depleted by p-type substrate.The high electric field around the drain is reduced by the n+-floating layer due to the REBULF effect,which causes the redistribution of the bulk electric field in the drift region,and thus the substrate supports more biases.The new structure features high breakdown voltage,low on-resistance,and charge balance in the drift region.

A new super junction LDMOST structure that suppresses the substrate-assisted depletion effect is designed with an n+-floating layer embedded in the high-resistance p-type substrate by implanting phosphor or arsenic.This effect results from a charge imbalance between the n-type and p-type pillars when the n-type pillars are depleted by p-type substrate.The high electric field around the drain is reduced by the n+-floating layer due to the REBULF effect,which causes the redistribution of the bulk electric field in the drift region,and thus the substrate supports more biases.The new structure features high breakdown voltage,low on-resistance,and charge balance in the drift region.
Total Ionizing Dose Radiation Effects on MOS Transistors with Different Layouts
Li Dongmei, Huangfu Liying, Gou Qiujing, Wang Zhihua
Chin. J. Semicond.  2007, 28(2): 171-175
Abstract PDF

Both nMOS and pMOS transistors with two-edged and multi-finger layouts are fabricated in a standard commercial 0.6μm CMOS/bulk process to study their total ionizing dose (TID) radiation effects.The leakage current,threshold voltage shift,and transconductance of the devices are monitored before and after γ-ray irradiation.Different device bias conditions are used during irradiation.The experiment results show that TID radiation effects on nMOS devices are very sensitive to their layout structures.The impact of the layout on TID effects on pMOS devices is slight and can be neglected.

Both nMOS and pMOS transistors with two-edged and multi-finger layouts are fabricated in a standard commercial 0.6μm CMOS/bulk process to study their total ionizing dose (TID) radiation effects.The leakage current,threshold voltage shift,and transconductance of the devices are monitored before and after γ-ray irradiation.Different device bias conditions are used during irradiation.The experiment results show that TID radiation effects on nMOS devices are very sensitive to their layout structures.The impact of the layout on TID effects on pMOS devices is slight and can be neglected.
A 155Mbps 0.5μm CMOS Limiting Amplifier
Jiao Yang, Wang Zhigong, Wang Rong, Guan Zhiqiang
Chin. J. Semicond.  2007, 28(2): 176-181
Abstract PDF

This paper presents a 155Mbps limiting amplifier for STM-1 systems of SDH optical communication.It is implemented in CSMC 0.5μm CMOS technology.Under a supply voltage of 3.3V,it has a power consumption of 198mW.The core of the circuit is composed of 6 cascaded amplifiers that are in a conventional structure of differential pairs,an output buffer,and a DC offset cancellation feedback loop.The small signal gain can be adjusted from 74 to 44dB by an off-chip resistor.The chip was packaged before being tested.The experimental results indicate that the circuit has an input dynamic range of 54dB and provides a single-ended output swing of 950mV.Its output eye diagram remains satisfactory when the pseudo-random bit sequence (PRBS) input speed reaches 400Mbps.

This paper presents a 155Mbps limiting amplifier for STM-1 systems of SDH optical communication.It is implemented in CSMC 0.5μm CMOS technology.Under a supply voltage of 3.3V,it has a power consumption of 198mW.The core of the circuit is composed of 6 cascaded amplifiers that are in a conventional structure of differential pairs,an output buffer,and a DC offset cancellation feedback loop.The small signal gain can be adjusted from 74 to 44dB by an off-chip resistor.The chip was packaged before being tested.The experimental results indicate that the circuit has an input dynamic range of 54dB and provides a single-ended output swing of 950mV.Its output eye diagram remains satisfactory when the pseudo-random bit sequence (PRBS) input speed reaches 400Mbps.
A Fully Integrated CMOS Readout Circuit for Particle Detectors
Zhang Yacong, Chen Zhongjian, Lu Wengao, Zhao Baoying, Ji Lijiu
Chin. J. Semicond.  2007, 28(2): 182-188
Abstract PDF

Novel schemes for a charge sensitive amplifier (CSA) and a CR-(RC)n semi-Gaussian shaper in a fully integrated CMOS readout circuit for particle detectors are presented.The CSA is designed with poly-resistors as feedback components to reduce noise.Compared with conventional CSA,the input referred equivalent noise charge(ENC) is simulated to be reduced from 5036e to 2381e with a large detector capacitance of 150pF at the cost of 0.5V output swing loss.The CR-(RC)n semi-Gaussian shaper uses MOS transistors in the triode region in series with poly-resistors to compensate process variation without much linearity reduction.

Novel schemes for a charge sensitive amplifier (CSA) and a CR-(RC)n semi-Gaussian shaper in a fully integrated CMOS readout circuit for particle detectors are presented.The CSA is designed with poly-resistors as feedback components to reduce noise.Compared with conventional CSA,the input referred equivalent noise charge(ENC) is simulated to be reduced from 5036e to 2381e with a large detector capacitance of 150pF at the cost of 0.5V output swing loss.The CR-(RC)n semi-Gaussian shaper uses MOS transistors in the triode region in series with poly-resistors to compensate process variation without much linearity reduction.
Detailed Routing Algorithm with Optical Proximity Effects Constraint
Zhou Qiang, 蔡懿慈, Cai Yici, Zhang Wei
Chin. J. Semicond.  2007, 28(2): 189-195
Abstract PDF

We present a detailed routing algorithm considering the optical proximity effect.The light intensity is calculated beforehand and stored in look-up tables.These costs are used as a constraint to guide the sequential routing.The routing algorithm is based on constructed force directed Steiner tree routing to enhance routing efficiency.Experimental results on industrial benchmark circuits show that the presented routing algorithm can obtain much improvement considering optical effects short runtime.

We present a detailed routing algorithm considering the optical proximity effect.The light intensity is calculated beforehand and stored in look-up tables.These costs are used as a constraint to guide the sequential routing.The routing algorithm is based on constructed force directed Steiner tree routing to enhance routing efficiency.Experimental results on industrial benchmark circuits show that the presented routing algorithm can obtain much improvement considering optical effects short runtime.
Luminescence and Recombination Centers in ZnO/Si Films
Liu Cihui, Yao Ran, Su Jianfeng, Ma Zeyu, Fu Zhuxi
Chin. J. Semicond.  2007, 28(2): 196-199
Abstract PDF

D0h luminescence of ZnO films deposited on p-type Si substrates grown by MOCVD is reported.After annealing in air at 700℃ for 1h,the photoluminescence (PL) spectra,the I-V characteristics,and the deep level transient spectroscopy (DLTS) of the samples are measured.All the samples we measured have the rectification characteristic.The DLTS signals show two deep levels of E1 and E2.The Gauss fit curves of the PL spectra at room temperature show three luminescence lines,one of which is attributed to the excitation emission.The donor level E1 measured by DLTS and the other two emission lines,which are very close to each other,have a close relation with the location state donor ionization energy Ed,and are thought to be from neutral donors bound to hole emission (D0h).Moreover,the intensity of the PL spectra decreases while the relative density of E2 increases,showing that E2 has the properties of a nonradiative center.

D0h luminescence of ZnO films deposited on p-type Si substrates grown by MOCVD is reported.After annealing in air at 700℃ for 1h,the photoluminescence (PL) spectra,the I-V characteristics,and the deep level transient spectroscopy (DLTS) of the samples are measured.All the samples we measured have the rectification characteristic.The DLTS signals show two deep levels of E1 and E2.The Gauss fit curves of the PL spectra at room temperature show three luminescence lines,one of which is attributed to the excitation emission.The donor level E1 measured by DLTS and the other two emission lines,which are very close to each other,have a close relation with the location state donor ionization energy Ed,and are thought to be from neutral donors bound to hole emission (D0h).Moreover,the intensity of the PL spectra decreases while the relative density of E2 increases,showing that E2 has the properties of a nonradiative center.
Effect of Pretreatment on Thermal Donors in Neutron-Irradiated Czochralski Silicon
Deng Xiaoran, Yang Shuai
Chin. J. Semicond.  2007, 28(2): 200-203
Abstract PDF

The influence of pretreatment on the thermal donors (TD) in neutron-irradiated Czochralski silicon is investigated with four-point probe measurement and Fourier transform infrared spectrometry.The results show that the amount and the generation rate of the thermal donors are depressed after preheating treatment at 650℃.After rapid thermal processing at 1200℃ in N2 atmosphere,the number of the donors is depressed in irradiated silicon; but in Ar atmosphere,the number of the donors is depressed in non-irradiated silicon.

The influence of pretreatment on the thermal donors (TD) in neutron-irradiated Czochralski silicon is investigated with four-point probe measurement and Fourier transform infrared spectrometry.The results show that the amount and the generation rate of the thermal donors are depressed after preheating treatment at 650℃.After rapid thermal processing at 1200℃ in N2 atmosphere,the number of the donors is depressed in irradiated silicon; but in Ar atmosphere,the number of the donors is depressed in non-irradiated silicon.
Physical Vapor Transport Growth and Characterization of Large Bulk AlN Crystal
Dong Zhiyuan, Zhao Youwen, Wei Xuecheng, Li Jinmin
Chin. J. Semicond.  2007, 28(2): 204-208
Abstract PDF

AlN polycrystalline boules with a diameter of 40~50mm, a thickness of 8~10mm,and a grain size of up to 5mm have been grown with physical vapor transport technology.The crystalline quality,defects,and structure of these crystals have been investigated by using Raman scattering and cathodoluminescence spectroscopy.The electrical conductivity of the AlN crystal has been measured from room temperature to 800℃,and a deep level defect with activation energy of 0.98eV is identified.The influence of growth conditions on AlN crystalline quality is also discussed.

AlN polycrystalline boules with a diameter of 40~50mm, a thickness of 8~10mm,and a grain size of up to 5mm have been grown with physical vapor transport technology.The crystalline quality,defects,and structure of these crystals have been investigated by using Raman scattering and cathodoluminescence spectroscopy.The electrical conductivity of the AlN crystal has been measured from room temperature to 800℃,and a deep level defect with activation energy of 0.98eV is identified.The influence of growth conditions on AlN crystalline quality is also discussed.
Fabrication of 1nm/s High Deposition Microcrystalline Silicon and Its Application in Solar Cells
Zhang Xiaodan, Zhang Farong, Zhao Ying, Chen Fei, Sun Jian, Wei Changchun, Geng Xinhua, Xiong Shaozhen
Chin. J. Semicond.  2007, 28(2): 209-212
Abstract PDF

A series of microcrystalline silicon thin films were fabricated by very high frequency plasma-enhanced chemical vapor deposition at different silane concentrations using relatively high pressure and power.The deposition rate of the materials increases with the increase of the silane concentration.Through analysis of the structural and electrical properties of the materials,it can be concluded that with a high deposition rate (above 1nm/s),device-quality microcrystalline silicon is obtained.As a result,a microcrystalline silicon solar cell with 6.3% conversion efficiency is obtained.

A series of microcrystalline silicon thin films were fabricated by very high frequency plasma-enhanced chemical vapor deposition at different silane concentrations using relatively high pressure and power.The deposition rate of the materials increases with the increase of the silane concentration.Through analysis of the structural and electrical properties of the materials,it can be concluded that with a high deposition rate (above 1nm/s),device-quality microcrystalline silicon is obtained.As a result,a microcrystalline silicon solar cell with 6.3% conversion efficiency is obtained.
Low-Temperature Wafer-to-Wafer Bonding Using Intermediate Metals
Zhang Xiaoying, Chen Songyan, Lai Hongkai, Li Cheng, Yu Jinzhong
Chin. J. Semicond.  2007, 28(2): 213-216
Abstract PDF

Si/Si bonding has been achieved at low temperatures by introducing Ti/Au layers.The bonding temperature can be reduced to 414℃.The Si/Si bonding strength has been investigated by tensile strength.The results show that the bonding strength exceeds 1.27MPa.I-V tests show that the interface of Si/Ti/Au/Ti/Si is an ohmic contact,and XPS tests indicate that the interface mainly comprises Si-Au eutectic alloy.Different experiments by annealing at gradually changing temperature show that the higher the bonding temperature is,the stronger the bonding energy is.In addition,annealing at gradually changing temperature is propitious to increase bonding strength.

Si/Si bonding has been achieved at low temperatures by introducing Ti/Au layers.The bonding temperature can be reduced to 414℃.The Si/Si bonding strength has been investigated by tensile strength.The results show that the bonding strength exceeds 1.27MPa.I-V tests show that the interface of Si/Ti/Au/Ti/Si is an ohmic contact,and XPS tests indicate that the interface mainly comprises Si-Au eutectic alloy.Different experiments by annealing at gradually changing temperature show that the higher the bonding temperature is,the stronger the bonding energy is.In addition,annealing at gradually changing temperature is propitious to increase bonding strength.
Pressure-Free Localized Laser Bonding for Silicon and Glass
Ma Ziwen, Tang Zirong, Liao Guanglan, Shi Tielin, Nie Lei
Chin. J. Semicond.  2007, 28(2): 217-221
Abstract PDF

A novel pressure-free localized laser bonding process for silicon and glass has been developed.This process has been successfully realized by applying surface activation to wafers and selecting suitable laser parameters and bonding conditions.The effects of factors in this bonding method,such as laser power,scanning velocity,and base material,were also studied.The experimental results demonstrate that the bondline becomes wider at higher laser powers and smaller scanning velocities.The experiment also shows that this bond process can efficiently reduce residual stress in bonded pairs and control the bondline width.Tensile experiments indicate that the bond strength is comparable with that obtained by other bonding methods.This process can provide a simple and robust bonding method with rapid processing time and high selectivity of bonding area for the packaging and fabrication of MEMS devices.

A novel pressure-free localized laser bonding process for silicon and glass has been developed.This process has been successfully realized by applying surface activation to wafers and selecting suitable laser parameters and bonding conditions.The effects of factors in this bonding method,such as laser power,scanning velocity,and base material,were also studied.The experimental results demonstrate that the bondline becomes wider at higher laser powers and smaller scanning velocities.The experiment also shows that this bond process can efficiently reduce residual stress in bonded pairs and control the bondline width.Tensile experiments indicate that the bond strength is comparable with that obtained by other bonding methods.This process can provide a simple and robust bonding method with rapid processing time and high selectivity of bonding area for the packaging and fabrication of MEMS devices.
Duration of Startup of GaAs Wet Etching Measured by Infrared-Image of Linear Liquid Film
Liu Lin, Ye Yutang, 吴云峰, Wu Yunfeng, 陆佳佳, Fang Liang
Chin. J. Semicond.  2007, 28(2): 222-226
Abstract PDF

A new infrared measurement method for determining the duration of the startup of GaAs wet etching is proposed.The basis of this method is that the temperature variation of an etchant film is induced by the absorption or release of chemical heat during the startup of the reaction.The startup time can be determined by the real-time collection of infrared thermal images over the course of the temperature variation.Theoretical analysis and experimental results show that 2mm-wide linear liquid film is suitable for monitoring because it contains the information of temperature and space change.Thus the center of the liquid film can be set as a characteristic observation point.More accurate experimental data can be obtained because of the high sensitivity of temperature change of the extremely thin linear film made by sliding droplets as well as the avoidance of the influence of film weight on the startup time by using upright GaAs substrate.The startup time of the reaction between GaAs substrate and H2SO4∶H2O2∶H2O (=5∶1∶50) is about 0.2s,as obtained from the grey-scale variation of a certain cross section of the linear liquid film.The proposed method will stimulate the research of quick etching technology and absorption capability between solids and liquids.

A new infrared measurement method for determining the duration of the startup of GaAs wet etching is proposed.The basis of this method is that the temperature variation of an etchant film is induced by the absorption or release of chemical heat during the startup of the reaction.The startup time can be determined by the real-time collection of infrared thermal images over the course of the temperature variation.Theoretical analysis and experimental results show that 2mm-wide linear liquid film is suitable for monitoring because it contains the information of temperature and space change.Thus the center of the liquid film can be set as a characteristic observation point.More accurate experimental data can be obtained because of the high sensitivity of temperature change of the extremely thin linear film made by sliding droplets as well as the avoidance of the influence of film weight on the startup time by using upright GaAs substrate.The startup time of the reaction between GaAs substrate and H2SO4∶H2O2∶H2O (=5∶1∶50) is about 0.2s,as obtained from the grey-scale variation of a certain cross section of the linear liquid film.The proposed method will stimulate the research of quick etching technology and absorption capability between solids and liquids.
Model of Non-Normal Process Capability Indices to Semiconductor Quality Control
Wang Shaoxi, Jia Xinzhang
Chin. J. Semicond.  2007, 28(2): 227-231
Abstract PDF

It is necessary to compute the process capability index (PCI) of non-normal data when considering the level of semiconductor processes.Several main PCIs that have already been presented are analyzed,and their advantages and disadvantages are presented.Based on Chebyshev-Hermite polynomials,a model for computing non-normal PCIs is then given when regarding the fact that these four moments,i.e.mean,standard deviation,skewness,and kurtosis,are suitable to approximately characterize the data distribution properties,which is effective when data deviation is large.Finally,an example is given for proving the relational model.

It is necessary to compute the process capability index (PCI) of non-normal data when considering the level of semiconductor processes.Several main PCIs that have already been presented are analyzed,and their advantages and disadvantages are presented.Based on Chebyshev-Hermite polynomials,a model for computing non-normal PCIs is then given when regarding the fact that these four moments,i.e.mean,standard deviation,skewness,and kurtosis,are suitable to approximately characterize the data distribution properties,which is effective when data deviation is large.Finally,an example is given for proving the relational model.
A Novel Double RESURF TG-LDMOS Device Structure
Xu Shengrui, Hao Yue, Feng Hui, Li Dechang, Zhang Jincheng
Chin. J. Semicond.  2007, 28(2): 232-236
Abstract PDF

We describe a new lateral double diffused trench gate MOSFET with double RESURF technology for the first time.We simulate the breakdown voltage and capacitance,especially the influences of double RESURF technology on the breakdown voltage.Compared with conventional TG-LDMOS,the breakdown voltage of the new structure is improved by 30V with the same length of the drift region and on-state resistance,and the structure shows excellent RF characteristics.

We describe a new lateral double diffused trench gate MOSFET with double RESURF technology for the first time.We simulate the breakdown voltage and capacitance,especially the influences of double RESURF technology on the breakdown voltage.Compared with conventional TG-LDMOS,the breakdown voltage of the new structure is improved by 30V with the same length of the drift region and on-state resistance,and the structure shows excellent RF characteristics.
Regular Perturbation Method for Studying the Subthreshold Characteristics of Nano-Scaled MOSFETs
Dai Yuehua, Chen Junning, Ke Daoming
Chin. J. Semicond.  2007, 28(2): 237-240
Abstract PDF

We present a new method for studying the subthreshold characteristics of nano-scaled MOSFETs,which we call the "regular perturbation method" . The Poisson equation is solved using this method for the first time.In particular,the depletion approximation and charge-sheet model in the Poisson equation are avoided due to their invalidity in nano-scaled MOSFETs.This yields a conventional exponential form of the subthreshold current,and the subthreshold swing can be obtained analytically from this current equation.The results of the model are compared and verified with the numerical simulation.

We present a new method for studying the subthreshold characteristics of nano-scaled MOSFETs,which we call the "regular perturbation method" . The Poisson equation is solved using this method for the first time.In particular,the depletion approximation and charge-sheet model in the Poisson equation are avoided due to their invalidity in nano-scaled MOSFETs.This yields a conventional exponential form of the subthreshold current,and the subthreshold swing can be obtained analytically from this current equation.The results of the model are compared and verified with the numerical simulation.
Total Ionizing Dose Effects of Deep Submicron nMOSFET Devices
Meng Zhiqin, Hao Yue, Tang Yu, Ma Xiaohua, Zhu Zhiwei, Li Yongkun
Chin. J. Semicond.  2007, 28(2): 241-245
Abstract PDF

We study the total ionizing dose sensitivity of 0.25μm nMOSFETs.The off-state leakage current,transconductance,and the gate leakage current of the devices are monitored.Experiment shows that the effect on irradiation is mostly caused by the space charges in the field oxide,which is verified by simulation.

We study the total ionizing dose sensitivity of 0.25μm nMOSFETs.The off-state leakage current,transconductance,and the gate leakage current of the devices are monitored.Experiment shows that the effect on irradiation is mostly caused by the space charges in the field oxide,which is verified by simulation.
RF-CMOS Modeling:Parasitic Analysis for MOST On-Wafer Test Structure
Liu Jun, Sun Lingling, Xu Xiaojun
Chin. J. Semicond.  2007, 28(2): 246-253
Abstract PDF

A new model that considers the parasitic effects of a test structure when performing on-wafer S-parameter measurements on RF/microwave MOST fabricated in RF-CMOS technology is presented.Discontinuities between pads and the DUT (device under test) and between stub-interconnect metal and the DUT are considered.The parasitic between the stub-interconnect metal and the lossy substrate is modeled separately.An additional element is introduced to predict the inductive losses of the substrate.All model parameters are directly determined using a simple and analytical measurement-based method,allowing the electrical representation of the complete test structure using an equivalent circuit.The validity of the model is demonstrated by the on-wafer measurements of interconnects up to 40GHz,employing a 0.25μm RF-CMOS process supplied by CSM (Chartered Semiconductor Manufacture Ltd) 0.25μm RF-CMOS technology.

A new model that considers the parasitic effects of a test structure when performing on-wafer S-parameter measurements on RF/microwave MOST fabricated in RF-CMOS technology is presented.Discontinuities between pads and the DUT (device under test) and between stub-interconnect metal and the DUT are considered.The parasitic between the stub-interconnect metal and the lossy substrate is modeled separately.An additional element is introduced to predict the inductive losses of the substrate.All model parameters are directly determined using a simple and analytical measurement-based method,allowing the electrical representation of the complete test structure using an equivalent circuit.The validity of the model is demonstrated by the on-wafer measurements of interconnects up to 40GHz,employing a 0.25μm RF-CMOS process supplied by CSM (Chartered Semiconductor Manufacture Ltd) 0.25μm RF-CMOS technology.
Self-Seeding of DFB Laser to Improve the Frequency Response
Wang Xin, Wen Jimin, Huang Hengpei, Yuan Haiqing, 谢亮, Xie Liang
Chin. J. Semicond.  2007, 28(2): 254-257
Abstract PDF

A simple and effective self-seeding method of DFB lasers is presented.The frequency responses of a directly modulated DFB laser become very flat with this self-seeding method.Experiment was done using a three-port optical circulator and a 1×2 optical coupler.The results show that the peak of relaxation oscillation of the frequency response curves can be restrained significantly by changing the polarization of the self-seeding light.

A simple and effective self-seeding method of DFB lasers is presented.The frequency responses of a directly modulated DFB laser become very flat with this self-seeding method.Experiment was done using a three-port optical circulator and a 1×2 optical coupler.The results show that the peak of relaxation oscillation of the frequency response curves can be restrained significantly by changing the polarization of the self-seeding light.
Injection Locked Erbium-Doped Fiber Ring Laser and Its Wavelength Tuning
Zhang Xin, Zhang Tao, Chen Wei, Liu Yu, Wang Xin, Xie Liang, Zhu Ninghua, Feng Boxue
Chin. J. Semicond.  2007, 28(2): 258-264
Abstract PDF

The injection locking of an erbium-doped fiber ring laser (EDFRL) is studied,and a novel method to stabilize the wavelength and output power of tunable single-longitudinal-mode EDFRL with compound cavity structure is proposed and experimentally demonstrated.One of the longitudinal modes is stimulated by an injected continuous wave so that the mode is able to stabilize the system.An output power of >0.6dBm and an SNR of >43dB within the tuning range from 1527.4 to 1561.9nm can be achieved for this laser.A wavelength variation of less than 0.01nm,power fluctuation of less than 0.02dB,and linewidth of about 1.4kHz are also obtained.

The injection locking of an erbium-doped fiber ring laser (EDFRL) is studied,and a novel method to stabilize the wavelength and output power of tunable single-longitudinal-mode EDFRL with compound cavity structure is proposed and experimentally demonstrated.One of the longitudinal modes is stimulated by an injected continuous wave so that the mode is able to stabilize the system.An output power of >0.6dBm and an SNR of >43dB within the tuning range from 1527.4 to 1561.9nm can be achieved for this laser.A wavelength variation of less than 0.01nm,power fluctuation of less than 0.02dB,and linewidth of about 1.4kHz are also obtained.
Fabrication and Characteristics of Nano-Aperture Vertical-Cavity Surface-Emitting Laser
Gao Jianxia, Song Guofeng, Gan Qiaoqiang, Xu Yun, Guo Baoshan, Chen Lianghui
Chin. J. Semicond.  2007, 28(2): 265-268
Abstract PDF

Nano-aperture vertical-cavity surface-emitting lasers were fabricated on common 850nm VCSELs.When the size of the aperture was 400nm×400nm,the maximum far-field output power reached 0.3mW at a driving current of 25mA,and the power density was about 2mW/μm2.The fabrication process is introduced here.Spectra and lifetime characteristics of the devices are also analyzed.

Nano-aperture vertical-cavity surface-emitting lasers were fabricated on common 850nm VCSELs.When the size of the aperture was 400nm×400nm,the maximum far-field output power reached 0.3mW at a driving current of 25mA,and the power density was about 2mW/μm2.The fabrication process is introduced here.Spectra and lifetime characteristics of the devices are also analyzed.
Modeling,Fabrication,and Characterization of a Novel Optically Driven Deformable Mirror
Xiang Dong, Wang Qingling, Zhang Guangyong, Zhou Lina
Chin. J. Semicond.  2007, 28(2): 269-274
Abstract PDF

This paper presents a novel deformable mirror driven by a laser.The mirror is composed a 2mm-thick polymeric membrane mirror supported by a 6mm-thick grid of patterned photoresist and a photoconductive GaAs substrate.A very high-frequency AC bias is applied between the membrane and the photoconductive substrate.When the back of the device is illuminated by a laser,the carrier concentration in the GaAs increases.This causes an impedance redistribution between two cascaded impedances,and then the membrane is deformed because of the static electric effect.The working model of the mirror is analyzed,and the design and fabrication of the mirror are also presented.The effects of some operating parameters such as bias voltage and current frequency on the membrane deformation are tested by experiment.

This paper presents a novel deformable mirror driven by a laser.The mirror is composed a 2mm-thick polymeric membrane mirror supported by a 6mm-thick grid of patterned photoresist and a photoconductive GaAs substrate.A very high-frequency AC bias is applied between the membrane and the photoconductive substrate.When the back of the device is illuminated by a laser,the carrier concentration in the GaAs increases.This causes an impedance redistribution between two cascaded impedances,and then the membrane is deformed because of the static electric effect.The working model of the mirror is analyzed,and the design and fabrication of the mirror are also presented.The effects of some operating parameters such as bias voltage and current frequency on the membrane deformation are tested by experiment.
Coupling Theory and Simulation of a High Speed OpticalReceiver Module
Chen Shiyu, Wang Gang
Chin. J. Semicond.  2007, 28(2): 275-279
Abstract PDF

We propose an optical model of a cylindrical lens for fiber coupling based on the method of coupling an angle-ended fiber to the front facet of a chip that is used in high-speed optical receivers.We also study the effect of the packing process on coupling efficiency with the optical model and software simulation.The simulation results show that the optical coupling efficiency is determined by fiber position,the lens design of the photodiode chips,and the shape of the light source.

We propose an optical model of a cylindrical lens for fiber coupling based on the method of coupling an angle-ended fiber to the front facet of a chip that is used in high-speed optical receivers.We also study the effect of the packing process on coupling efficiency with the optical model and software simulation.The simulation results show that the optical coupling efficiency is determined by fiber position,the lens design of the photodiode chips,and the shape of the light source.
Deformation Simulation on the Beam of InP-BasedMOEMS Tunable Devices
Wu Xuming, Wang Xiaodong, 何国荣, He Guorong, Wang Qing, Cao Yulian
Chin. J. Semicond.  2007, 28(2): 280-283
Abstract PDF

Deformation of the beam in InP-based MOEMS tunable devices is frequently observed in experiments.This is because As atoms enter the beam during the growth.A method is developed to simulate the deformation,with no need to measure the gradient stress.The deformations of a one-armed beam and a two-armed beam are simulated,and the calculated values agree with the measured ones.The influences of As concentration and beam depth on the deformation are investigated,and the results show that it is useful to reduce the deformation by reducing the As concentration and increasing the beam depth.

Deformation of the beam in InP-based MOEMS tunable devices is frequently observed in experiments.This is because As atoms enter the beam during the growth.A method is developed to simulate the deformation,with no need to measure the gradient stress.The deformations of a one-armed beam and a two-armed beam are simulated,and the calculated values agree with the measured ones.The influences of As concentration and beam depth on the deformation are investigated,and the results show that it is useful to reduce the deformation by reducing the As concentration and increasing the beam depth.
Separate Absorption and Multiplication 4H-SiC UltravioletAvalanche Photodetector
Zhu Huili, Chen Xiaping, Wu Zhengyun
Chin. J. Semicond.  2007, 28(2): 284-288
Abstract PDF

A 4H-SiC avalanche photodetector (APD) with separate absorption and multiplication layers was designed and fabricated.The thicknesses of the absorption and multiplication epilayers were designed as 025 and 1μm,respectively.A multiple junction termination extension (MJTE) was used to eliminate the electric field crowding effect at the edge and to reduce the surface electric field.Dark current,photocurrent,and spectral responsivity were measured.High gain (>1e4) was achieved at a low breakdown voltage,and the dark current was on the order of about 10pA before device was punched through.The ratio of responsivity at 260nm to that at 380nm was higher than 1e3.A new phenomenon was observed in its spectral responsivity.The response wavelength peak shifted to shorter wavelength as the reverse bias was increased.Near breakdown voltage the peak of the response wavelength was located at 210nm,which was shorter than that at 0V.These results show that the samples have good performance for ultraviolet detection.

A 4H-SiC avalanche photodetector (APD) with separate absorption and multiplication layers was designed and fabricated.The thicknesses of the absorption and multiplication epilayers were designed as 025 and 1μm,respectively.A multiple junction termination extension (MJTE) was used to eliminate the electric field crowding effect at the edge and to reduce the surface electric field.Dark current,photocurrent,and spectral responsivity were measured.High gain (>1e4) was achieved at a low breakdown voltage,and the dark current was on the order of about 10pA before device was punched through.The ratio of responsivity at 260nm to that at 380nm was higher than 1e3.A new phenomenon was observed in its spectral responsivity.The response wavelength peak shifted to shorter wavelength as the reverse bias was increased.Near breakdown voltage the peak of the response wavelength was located at 210nm,which was shorter than that at 0V.These results show that the samples have good performance for ultraviolet detection.
A Novel Oscillator Based on RTD/MOSFET
Wang Wei, Niu Pingjuan, Guo Weilian, Yu Xin, Zhang Shilin
Chin. J. Semicond.  2007, 28(2): 289-293
Abstract PDF

Based on a novel oscillating mechanism,a 25MHz high-frequency oscillating circuit is achieved by combining resonant tunneling diode (RTD) characterized by intrinsic negative differential resistance (NDR) and bi-stability with a MOSFET.The mechanism of this oscillating circuit is analyzed theoretically,and the correctness of the circuit analysis is verified by advanced design system (ADS) simulation and experiment.The realization of this circuit can solve the problem of power limitation in conventional RTD-based oscillating circuits.If using a high-frequency,high-speed device such as a high electron mobility transistor (HEMT) in stead of a MOSFET,this circuit can improve its frequency greatly and is more suitable for monolithic integration.Thus it can find wide applications in radio-frequency and microwave fields.

Based on a novel oscillating mechanism,a 25MHz high-frequency oscillating circuit is achieved by combining resonant tunneling diode (RTD) characterized by intrinsic negative differential resistance (NDR) and bi-stability with a MOSFET.The mechanism of this oscillating circuit is analyzed theoretically,and the correctness of the circuit analysis is verified by advanced design system (ADS) simulation and experiment.The realization of this circuit can solve the problem of power limitation in conventional RTD-based oscillating circuits.If using a high-frequency,high-speed device such as a high electron mobility transistor (HEMT) in stead of a MOSFET,this circuit can improve its frequency greatly and is more suitable for monolithic integration.Thus it can find wide applications in radio-frequency and microwave fields.
An 80dB Dynamic Range Σ-Δ Modulator for a GSM System
Chen Jianqiu, Ren Junyan, Xu Jun, Wang Zhaogang, Li Yiran
Chin. J. Semicond.  2007, 28(2): 294-301
Abstract PDF

We propose a Sigma-Delta modulator for a GSM system,whose channel bandwidth is wider than 200kHz and dynamic range is more than 80dB.A multi-stage noise shaping structure,rather than a single-loop structure,is chosen to achieve lower OSR and higher stabilization.The design parameters are:fCLK=16MHz,OSR=32,and baseband width=250kHz.Simulation at the circuit level achieves a peak SNDR of 82dB and DR of 87dB.The chip is implemented in SMIC 0.18μm CMOS technology,and the area is 1.2mm×1.8mm.Measurement at 16MHz clock frequency and OSR of 32 achieves a peak SNDR of 75dB and DR over 80dB,which are close to the values obtained by the circuit level simulation.The chip operates under a 1.8V power supply with a power dissipation of 16.7mW.

We propose a Sigma-Delta modulator for a GSM system,whose channel bandwidth is wider than 200kHz and dynamic range is more than 80dB.A multi-stage noise shaping structure,rather than a single-loop structure,is chosen to achieve lower OSR and higher stabilization.The design parameters are:fCLK=16MHz,OSR=32,and baseband width=250kHz.Simulation at the circuit level achieves a peak SNDR of 82dB and DR of 87dB.The chip is implemented in SMIC 0.18μm CMOS technology,and the area is 1.2mm×1.8mm.Measurement at 16MHz clock frequency and OSR of 32 achieves a peak SNDR of 75dB and DR over 80dB,which are close to the values obtained by the circuit level simulation.The chip operates under a 1.8V power supply with a power dissipation of 16.7mW.
Fabrication of Monolithic Silicon Multi-Sensor on SOI Wafer
Xu Jingbo, Zhao Yulong, Jiang Zhuangde, Zhang Dacheng, Yang Fang, Sun Jian
Chin. J. Semicond.  2007, 28(2): 302-307
Abstract PDF

A monolithic silicon multi-sensor on SOI wafer that consists of a three-axis piezoresistive accelerometer,a piezoresistive absolute pressure sensor,and a silicon thermistor temperature sensor is presented.The fabrication process of the sensor is described.An effective micromachining process is developed to improve the reliability of the metal wire in the multi-sensor and to avoid adhesion between the PYREX glass and silicon mass in the process of anodic bonding.Finally,the measurement results of the sensor are shown.

A monolithic silicon multi-sensor on SOI wafer that consists of a three-axis piezoresistive accelerometer,a piezoresistive absolute pressure sensor,and a silicon thermistor temperature sensor is presented.The fabrication process of the sensor is described.An effective micromachining process is developed to improve the reliability of the metal wire in the multi-sensor and to avoid adhesion between the PYREX glass and silicon mass in the process of anodic bonding.Finally,the measurement results of the sensor are shown.
Study on the Material Removal Mechanism in Chemical Mechanical Polishing at the Molecular Scale
Wang Yongguang, Zhao Yongwu
Chin. J. Semicond.  2007, 28(2): 308-312
Abstract PDF

The mechanism of CMP material removal is investigated.Katsuki has proposed wear behavior between a single slurry particle and the wafer surface based on atomic force microscopy (AFM) measurements.The scratch depth under real CMP conduction is on the order of 1e-11m,as determined from the linear regression mechanism.A spectroscopic ellipsometry (SE) tool was used to study the relationship between the thickness of the oxidized layer and the chemical action time,which was presented by Nishizawa.Since the real CMP chemical action time is on the order of 1.0e-8s,the thickness of the oxidized layer is on the order of 1e-13m,as evaluated by the theoretical model based on the experiment.A closed-form equation supported by previously published experimental data is also derived from the material removal rate in terms of the molecular mechanism.The experimental results,in combination with the calculations,indicate that the CMP material is removed at the molecular scale.These analyses are useful to substantiate the molecular-scale mechanism of CMP material removal in addition to its underlying theoretical foundation.

The mechanism of CMP material removal is investigated.Katsuki has proposed wear behavior between a single slurry particle and the wafer surface based on atomic force microscopy (AFM) measurements.The scratch depth under real CMP conduction is on the order of 1e-11m,as determined from the linear regression mechanism.A spectroscopic ellipsometry (SE) tool was used to study the relationship between the thickness of the oxidized layer and the chemical action time,which was presented by Nishizawa.Since the real CMP chemical action time is on the order of 1.0e-8s,the thickness of the oxidized layer is on the order of 1e-13m,as evaluated by the theoretical model based on the experiment.A closed-form equation supported by previously published experimental data is also derived from the material removal rate in terms of the molecular mechanism.The experimental results,in combination with the calculations,indicate that the CMP material is removed at the molecular scale.These analyses are useful to substantiate the molecular-scale mechanism of CMP material removal in addition to its underlying theoretical foundation.