Issue Browser
Volume 23, Issue 9, Sep 2002
CONTENTS
Double Covalency Factors and Double ξ Model in Study on Optical and Magnetic Properties of Diluted Magnetic Semiconductors ZnX∶Co2+
SHI Si-qi, LEI Min-sheng, OUYANG Chu-ying
Chin. J. Semicond.  2002, 23(9): 897-901
Abstract PDF

The optical absorption spectra of the covalent crystals ZnX (X=S, Se) doped with Co2+ are studied using the double covalency factors, which considers the anisotropic distortion of eg and t2g orbits for d electron. When the paramagnetic g factor is calculated, the contributions of the spin orbit coupling from the ligand ions are taken into account besides that from the central ion, which is the double ξ model. The calculated results indicate that the theoretical values coincide with the experimental values very well. This suggests that the method presented in this paper could be more valid to some strongly covalent crystals.

The optical absorption spectra of the covalent crystals ZnX (X=S, Se) doped with Co2+ are studied using the double covalency factors, which considers the anisotropic distortion of eg and t2g orbits for d electron. When the paramagnetic g factor is calculated, the contributions of the spin orbit coupling from the ligand ions are taken into account besides that from the central ion, which is the double ξ model. The calculated results indicate that the theoretical values coincide with the experimental values very well. This suggests that the method presented in this paper could be more valid to some strongly covalent crystals.

Fabrication of Hydrogenated Microcrystalline Silicon Thin Films at Low Temperature by VHF-PECVD
YANG Hui-dong, WU Chun-ya, MAI Yao-hua, LI Hong-bo, XUE Jun-ming, LI Yan, REN Hui-zhi, ZHANG Li-zhu, GENG Xin-hua, XIONG Shao-zhen
Chin. J. Semicond.  2002, 23(9): 902-908
Abstract PDF

Using H2 diluted silane, series of μc-Si∶H films are fabricated at low temperature with VHF PECVD. The thickness measurements reveal that the deposition rates are obviously enhanced with higher plasma excitation frequency or working pressure, but increase firstly and then decrease with the increase of plasma power density. Raman spectra show that the crystallinity and the average grain sizes of the films strongly depend on the temperature of substrate and the concentration of silane. However, the plasma excitation frequency only has effect on the crystallinity,and a maximum occurs during the further increase of plasma excitation frequency. From XRD and TEM experiments,three preferential crystalline orientations (111), (220) and (311) are observed, and the average grain sizes are different for every crystalline orientation.

Using H2 diluted silane, series of μc-Si∶H films are fabricated at low temperature with VHF PECVD. The thickness measurements reveal that the deposition rates are obviously enhanced with higher plasma excitation frequency or working pressure, but increase firstly and then decrease with the increase of plasma power density. Raman spectra show that the crystallinity and the average grain sizes of the films strongly depend on the temperature of substrate and the concentration of silane. However, the plasma excitation frequency only has effect on the crystallinity,and a maximum occurs during the further increase of plasma excitation frequency. From XRD and TEM experiments,three preferential crystalline orientations (111), (220) and (311) are observed, and the average grain sizes are different for every crystalline orientation.

3-D Simulation of FINFET
LIU En-feng, LIU Xiao-yan, HAN Ru-qi
Chin. J. Semicond.  2002, 23(9): 909-913
Abstract PDF

An SOI MOSFET with FINFET structure is simulated using a 3-D simulator. I-V characteristics and subthreshold characteristics, as well as the short channel effect (SCE) are carefully investigated. SCE can be well controlled by reducing fin height. Good performance can be achieved with thin height, so fin height is considered as a key parameter in device design. Simulation results show that FINFETs present performance superior to conventional single gate devices.

An SOI MOSFET with FINFET structure is simulated using a 3-D simulator. I-V characteristics and subthreshold characteristics, as well as the short channel effect (SCE) are carefully investigated. SCE can be well controlled by reducing fin height. Good performance can be achieved with thin height, so fin height is considered as a key parameter in device design. Simulation results show that FINFETs present performance superior to conventional single gate devices.

Poly-Silicon Micromachined Switch
ZHANG Zheng-yuan, WEN Zhi-yu, XU Shi-liu, ZHANG Zheng-fan, LI Kai-cheng, HUANG Shang-lian
Chin. J. Semicond.  2002, 23(9): 914-920
Abstract PDF

By using LPCVD SiO2 and poly silicon as sacrificial layer and cantilever respectively, a poly silicon micromachined RF MEMS (radio frequency microelectronic mechanical system) switch is fabricated. During the fabrication process, the stress of poly-silicon is released to prevent poly-silicon membrane from bending, and the issue of compatibility between RF switch and IC process technology is also resolved. The low residual tensile stress poly-silicon cantilever is obtained by the optimization. The switch is tested, and the preliminary test results show: the pull down voltage is 89 V, and the switch speed is about 5 μs. The switch provides the potential to build a new fully monolithic integrated RF MEMS for radar and communications applications.

By using LPCVD SiO2 and poly silicon as sacrificial layer and cantilever respectively, a poly silicon micromachined RF MEMS (radio frequency microelectronic mechanical system) switch is fabricated. During the fabrication process, the stress of poly-silicon is released to prevent poly-silicon membrane from bending, and the issue of compatibility between RF switch and IC process technology is also resolved. The low residual tensile stress poly-silicon cantilever is obtained by the optimization. The switch is tested, and the preliminary test results show: the pull down voltage is 89 V, and the switch speed is about 5 μs. The switch provides the potential to build a new fully monolithic integrated RF MEMS for radar and communications applications.

SiGe HBT Class AB Power Amplifier for Wireless Communications
JIA Hong-yong, LIU Zhi-nong, LI Gao-qing, QIAN Pei-xin
Chin. J. Semicond.  2002, 23(9): 921-924
Abstract PDF

Good performance SiGe power amplifiers applicable to wireless communications are demonstrated. The output power can reach more than 30 dBm in class B mode. And in class AB mode, the output power at 1dB compression point (P1dB) is 24 dBm, the output third order intercept (TOI) power is 39 dBm under Vcc of 4 V. The highest power added efficiency (PAE) and PAE at 1 dB compression point are 34% and 25%, respectively. The adjacent channel power rejection for CDMA signal is more than 42 dBc, which complies with IS95 specification.

Good performance SiGe power amplifiers applicable to wireless communications are demonstrated. The output power can reach more than 30 dBm in class B mode. And in class AB mode, the output power at 1dB compression point (P1dB) is 24 dBm, the output third order intercept (TOI) power is 39 dBm under Vcc of 4 V. The highest power added efficiency (PAE) and PAE at 1 dB compression point are 34% and 25%, respectively. The adjacent channel power rejection for CDMA signal is more than 42 dBc, which complies with IS95 specification.

Scaling of Hydrogenic Impurity Binding Energy and Virial Theorem in Semiconductor Quantum Wells and Wires
LIU Jian-jun, SU Hui, YANG Guo-chen
Chin. J. Semicond.  2002, 23(9): 925-929
Abstract PDF

The scaling rule for hydrogenic impurity binding energy and the Virial theorem value (potential- to kinetic-energy) in quantum wells and wires of rectangular cross section with three different section ratios are studied through the variational method. The results reveal that (1) A parameter (impurity effective Bohr radius aim) that the impurity binding energy strongly depend on does exist; (2) The Virial theorem value is nonconstant but approaches 2 from above when the wire width is very small or large and the Virial theorem value versus aim does not increase monotonically due to the consideration of the correlation effect between the confined and non confined directions.

The scaling rule for hydrogenic impurity binding energy and the Virial theorem value (potential- to kinetic-energy) in quantum wells and wires of rectangular cross section with three different section ratios are studied through the variational method. The results reveal that (1) A parameter (impurity effective Bohr radius aim) that the impurity binding energy strongly depend on does exist; (2) The Virial theorem value is nonconstant but approaches 2 from above when the wire width is very small or large and the Virial theorem value versus aim does not increase monotonically due to the consideration of the correlation effect between the confined and non confined directions.

Correlation Between Er3+ Emission and Microstructure of Erbium-Implanted a-SiOx∶H Film
CHEN Zhang-yong, CHEN Wei-de, WANG Yong-qian, SONG Shu-fang, XU Zhen-jia, GUO Shao-ling
Chin. J. Semicond.  2002, 23(9): 930-934
Abstract PDF

An investigation on the correlation between Er3+ emission and the microstructure of erbium implanted hydrogenated amorphous silicon oxide (a SiOx ∶H) films is presented. FTIR spectra experimental results indicate that the a SiOx ∶H films are a mixture of two phases, an hydrogenated amorphous SiOx matrix and silicon rich domains embedded therein. The changes in the nonradiative centers in the silicon rich domains have a strong influence on Er3+ emission.

An investigation on the correlation between Er3+ emission and the microstructure of erbium implanted hydrogenated amorphous silicon oxide (a SiOx ∶H) films is presented. FTIR spectra experimental results indicate that the a SiOx ∶H films are a mixture of two phases, an hydrogenated amorphous SiOx matrix and silicon rich domains embedded therein. The changes in the nonradiative centers in the silicon rich domains have a strong influence on Er3+ emission.

Investigation of Mode and Modulation Responses in Semiconductor PS-DFB Lasers: Use of a Vector-Newton Method
ZHANG Ye-jin, ZHU Lin, GAO Zhi-guo, CHEN Ming-hua, DONG Yi, CHEN Wei-you, LIU Shi-yong, XIE Shi-zhong
Chin. J. Semicond.  2002, 23(9): 935-940
Abstract PDF

A numerical model for steady state analysis and analytical expressions for the AM and FM modulation responses of DFB lasers are presented. The small signal modulation responses of 3 phase shift (PS) DFB is investigated for the first time. A new method (Vector Newton method) to obtain multiple longitudinal mode of DFB lasers is used. It is demonstrated that this method is suitable for obtaining multiple solution of high nonlinear equations. Longitudinal photon density distribution and multiple longitudinal mode of 3PS DFB and simple DFB lasers are analyzed. The results show that modulation response characteristics of 3PS DFB laser is as good as that of DFB, and PS can weaken the longitudinal spatial hole burning (LHSB) effect and is in favor of single longitudinal mode operating of lasers.

A numerical model for steady state analysis and analytical expressions for the AM and FM modulation responses of DFB lasers are presented. The small signal modulation responses of 3 phase shift (PS) DFB is investigated for the first time. A new method (Vector Newton method) to obtain multiple longitudinal mode of DFB lasers is used. It is demonstrated that this method is suitable for obtaining multiple solution of high nonlinear equations. Longitudinal photon density distribution and multiple longitudinal mode of 3PS DFB and simple DFB lasers are analyzed. The results show that modulation response characteristics of 3PS DFB laser is as good as that of DFB, and PS can weaken the longitudinal spatial hole burning (LHSB) effect and is in favor of single longitudinal mode operating of lasers.

Theory Analysis for Semiconductor Optical Amplifier with Large 3dB Bandwidth
ZHANG Rui-ying, DONG Jie, ZHANG Jing, FENG Zhi-wei, WANG Wei
Chin. J. Semicond.  2002, 23(9): 941-946
Abstract PDF

Graded strained bulk like structure as semiconductor optical amplifier active region is proposed for the first time.Large band width and polarization insensitivity characteristics for such active structure are analyzed from the point of its gain spectra and energy band diagram.Graded tensile strain induced in the active structure is to enhance the TM mode material gain,obtain polarization insensitivity,expand the bandwidth for TE mode gain,and relax the limitation for the stripe width.Unstrained layer induced is primarily used to enhance the polarization insensitive mode gain and improve the active region crystal quality.

Graded strained bulk like structure as semiconductor optical amplifier active region is proposed for the first time.Large band width and polarization insensitivity characteristics for such active structure are analyzed from the point of its gain spectra and energy band diagram.Graded tensile strain induced in the active structure is to enhance the TM mode material gain,obtain polarization insensitivity,expand the bandwidth for TE mode gain,and relax the limitation for the stripe width.Unstrained layer induced is primarily used to enhance the polarization insensitive mode gain and improve the active region crystal quality.
Computer-Aided Design of RF-MEMS Switch
WEI Hua-zheng, GUO Fang-min, LAI Zong-sheng, ZHU Shou-zheng, ZHU Zi-qiang, LI Xiao-wei, CHENG Zhi-qun, SUN Xiao-wei
Chin. J. Semicond.  2002, 23(9): 947-951
Abstract PDF

A RF MEMS switch is designed by computer aided design (CAD) method.The transmission line loss of the coplanar waveguide (CPW) and the equivalent circuit model of the microelectromechanical systems (MEMS) switch are simulated using Agilent ADS software.The actuation voltage of the switch is also analysed by ANSYS software.Significant simulation results of the shunt capacitive switch,operating at 35GHz,are obtained.

A RF MEMS switch is designed by computer aided design (CAD) method.The transmission line loss of the coplanar waveguide (CPW) and the equivalent circuit model of the microelectromechanical systems (MEMS) switch are simulated using Agilent ADS software.The actuation voltage of the switch is also analysed by ANSYS software.Significant simulation results of the shunt capacitive switch,operating at 35GHz,are obtained.
Solving Process of Trap Density and Extracting of Correlative Parameters in Thin Gate Dielectric
LIU Hong-xia, HAO Yue
Chin. J. Semicond.  2002, 23(9): 952-956
Abstract PDF

TDDB evaluation experiments are implemented on the thin gate oxides MOS capacitor, and a method of precise measurement and characterization the trap density and accumulative failure are presented. Based on dynamic equilibrium equation in the process of trapped charges, the method can obtain the trap density by measuring the change of gate voltage of MOS capacitance under constant current stress and the change of high frequency C-V curve before and after the stress. The dynamitic parameters of characterization the trap density can be extracted from the experiment. On the base of experiment, the accumulation failure of devices can be evaluated precisely.

TDDB evaluation experiments are implemented on the thin gate oxides MOS capacitor, and a method of precise measurement and characterization the trap density and accumulative failure are presented. Based on dynamic equilibrium equation in the process of trapped charges, the method can obtain the trap density by measuring the change of gate voltage of MOS capacitance under constant current stress and the change of high frequency C-V curve before and after the stress. The dynamitic parameters of characterization the trap density can be extracted from the experiment. On the base of experiment, the accumulation failure of devices can be evaluated precisely.

Ga0.49In0.51P/GaAs HBT Small-Signal Model Extraction Using a Improved Genetic Algorithm

SHI Rui-ying, LIU Xun-chun, QIAN Yong-xue, SHI Hua-fen
Chin. J. Semicond.  2002, 23(9): 957-961
Abstract PDF

An improved global optimum genetic algorithm are described, reproduction, crossover and mutation operators are automatically optimized in the algorithm. The time searching the best of all reproduction, crossover and mutation operators is saved and the speed of parameter extraction is quickened. All 16 parameters of the Ga0.49In0.51P/GaAs hetero junction bipolar transistor (HBT) small signal equivalent circuit model are extracted from S parameters in the range of 1~26.5 GHz. Excellent agreement between the modeled and measured device performance is achieved in a wide range of operating conditions.

An improved global optimum genetic algorithm are described, reproduction, crossover and mutation operators are automatically optimized in the algorithm. The time searching the best of all reproduction, crossover and mutation operators is saved and the speed of parameter extraction is quickened. All 16 parameters of the Ga0.49In0.51P/GaAs hetero junction bipolar transistor (HBT) small signal equivalent circuit model are extracted from S parameters in the range of 1~26.5 GHz. Excellent agreement between the modeled and measured device performance is achieved in a wide range of operating conditions.

GaAsSb/InP HBT Growth on InP Substrates
XU Xian-gang, LIU Zhe, CUI De-liang
Chin. J. Semicond.  2002, 23(9): 962-965
Abstract PDF

InP/GaAs0.5Sb0.5/InP DHBT (double heterojunction bipolar transistor) is grown by MOCVD (metalorganic chemical vapor deposition). The influence of material quality on device performance is studied. DHBTs with high quality epilayers show good direct current and microwave performance, which are in agreement with energy band engineering. The collector and base current ideality factors are 1.00 and 1.06, respectively. The DHBTs breakdown voltage can be as high as 15 V, and microwave measurements indicate a current gain cutoff frequency fT of over 100 GHz .

InP/GaAs0.5Sb0.5/InP DHBT (double heterojunction bipolar transistor) is grown by MOCVD (metalorganic chemical vapor deposition). The influence of material quality on device performance is studied. DHBTs with high quality epilayers show good direct current and microwave performance, which are in agreement with energy band engineering. The collector and base current ideality factors are 1.00 and 1.06, respectively. The DHBTs breakdown voltage can be as high as 15 V, and microwave measurements indicate a current gain cutoff frequency fT of over 100 GHz .

A Modified DSOI Device
JIANG Bo, HE Ping, TIAN Li-lin, LIN Xi
Chin. J. Semicond.  2002, 23(9): 966-971
Abstract PDF

A new device structure called DSOI (drain/source on insulator) is proposed to alleviate the thermal transfer problem and floating body effects in SOI (silicon on insulator) device.The purpose of present work is to modify DSOI structure to get the best device electrical capability.Simulation results approve that this modified structure has better electrical performance than prototype.

A new device structure called DSOI (drain/source on insulator) is proposed to alleviate the thermal transfer problem and floating body effects in SOI (silicon on insulator) device.The purpose of present work is to modify DSOI structure to get the best device electrical capability.Simulation results approve that this modified structure has better electrical performance than prototype.
MEMS Phase Shifter with AlSi Alloy Membrane on High-Resistivity Silicon
SHI Yan-ling, QING Jian, XIN Pei-sheng, ZHU Zi-qiang, LAI Zong-sheng
Chin. J. Semicond.  2002, 23(9): 972-976
Abstract PDF

The design and fabrication of low loss distributed MEMS phase shifter are presented. The principle for this phase shifter is that the phase velocity can be varied by mean of snapping down the membrane of MEMS loading capacitors of the transmission line. Its MEMS capacitors and control voltage are discussed in detail.Several different membranes are used as the membranes and Al0.96Si0.04 alloy shows the best performance with pulling down voltage of the MEMS phase shifter no more than 40 V. There is a phase shift when the control voltage is only 10 V. With the phase shifter fabricated on the high resistivity silicon (ρ > 4000 Ω·cm), the insert loss is very low, no more than 3 dB from 1 GHz to 40 GHz. The phase shift will be more than 25° at 25 V.

The design and fabrication of low loss distributed MEMS phase shifter are presented. The principle for this phase shifter is that the phase velocity can be varied by mean of snapping down the membrane of MEMS loading capacitors of the transmission line. Its MEMS capacitors and control voltage are discussed in detail.Several different membranes are used as the membranes and Al0.96Si0.04 alloy shows the best performance with pulling down voltage of the MEMS phase shifter no more than 40 V. There is a phase shift when the control voltage is only 10 V. With the phase shifter fabricated on the high resistivity silicon (ρ > 4000 Ω·cm), the insert loss is very low, no more than 3 dB from 1 GHz to 40 GHz. The phase shift will be more than 25° at 25 V.

Inner Die Crack Simulation of vfBGA
PANG En-wen, LIN Jing, YU Fang, ZONG Xiang-fu
Chin. J. Semicond.  2002, 23(9): 977-982
Abstract PDF

The ANSYS software is employed to analyze the inner die crack of vfBGA.A suitable 3D model and the minimum stress causing die crack are achieved by modeling and estimating.The crucial cause of inner die crack is predicted and found after the simulation of the test process.It is the great stress induced by the abnormal contact between the testing equipment and vfBGA unit that causes the inner die crack.

The ANSYS software is employed to analyze the inner die crack of vfBGA.A suitable 3D model and the minimum stress causing die crack are achieved by modeling and estimating.The crucial cause of inner die crack is predicted and found after the simulation of the test process.It is the great stress induced by the abnormal contact between the testing equipment and vfBGA unit that causes the inner die crack.
Model Library of Φ76mm GaAs Devices
CHEN Xin-yu, CHEN Xiao-jian, HAO Xi-ping, LI Fu-xiao, JIANG You-quan
Chin. J. Semicond.  2002, 23(9): 983-987
Abstract PDF

Based on simulating measured S parameter and DC I-V characteristics, the engineering model of GaAs devices (MESFET, PHEMT) is extracted using microwave on wafer testing technology, and a model library based on Φ76mm GaAs process line is established. These models are confirmed by the design and fabricated of several GaAs MMICs. The results of calculation and those of simulation are in good agreement with each other.

Based on simulating measured S parameter and DC I-V characteristics, the engineering model of GaAs devices (MESFET, PHEMT) is extracted using microwave on wafer testing technology, and a model library based on Φ76mm GaAs process line is established. These models are confirmed by the design and fabricated of several GaAs MMICs. The results of calculation and those of simulation are in good agreement with each other.

150Ms/s、6bit Digital CMOS Folding A/D Converter with Current-Mode Interpolating
LIU Fei, JI Li-jiu
Chin. J. Semicond.  2002, 23(9): 988-995
Abstract PDF

A 150 Msamples/s, 6 bit CMOS folding and current mode interpolating analog to digital is designed in a 1.2 μm digital CMOS technology. A low power, high speed regenerated current comparator is proposed. By adopting Domino logic circuit, a very simple and flexible decoder is realitied with high speed and low power. The latency between input signal and output code is less than 2 clock cycles. The ADC only uses a single clock and its complement which simplifies the whole circuit. The converter power dissipation is simulated as 185 mW from a 5 V supply.

A 150 Msamples/s, 6 bit CMOS folding and current mode interpolating analog to digital is designed in a 1.2 μm digital CMOS technology. A low power, high speed regenerated current comparator is proposed. By adopting Domino logic circuit, a very simple and flexible decoder is realitied with high speed and low power. The latency between input signal and output code is less than 2 clock cycles. The ADC only uses a single clock and its complement which simplifies the whole circuit. The converter power dissipation is simulated as 185 mW from a 5 V supply.

Power Optimization Methods of Energy Recovery Circuits
DAI Hong-yu, ZHANG Sheng, ZHOU Run-de
Chin. J. Semicond.  2002, 23(9): 996-1000
Abstract PDF

The non adiabatic loss in energy recovery circuit is proportional to CLΔV2. Two methods are presented to lower the two factors of CL and ΔV. High efficient energy recovery logic (HEERL) circuit utilizes bootstrap effect to decrease node residential voltage ΔV. Improved energy recovery logic (IERL) adds extra recovery path to improve the recovery efficiency. At the same time the control node has CAΔV2 non adiabatic loss, but the total circuit power is saved. Compared with other energy recovery circuits, the two circuits presented show more than 50% power saving with only small area loss.

The non adiabatic loss in energy recovery circuit is proportional to CLΔV2. Two methods are presented to lower the two factors of CL and ΔV. High efficient energy recovery logic (HEERL) circuit utilizes bootstrap effect to decrease node residential voltage ΔV. Improved energy recovery logic (IERL) adds extra recovery path to improve the recovery efficiency. At the same time the control node has CAΔV2 non adiabatic loss, but the total circuit power is saved. Compared with other energy recovery circuits, the two circuits presented show more than 50% power saving with only small area loss.

Design of High Brightness Cubic-GaN LED Grown on GaAs Substrate
SUN Yuan-ping, ZHANG Ze-hong, ZHAO De-gang, FENG Zhi-hong, FU Yi, ZHANG Shu-ming, YANG Hui
Chin. J. Semicond.  2002, 23(9): 1001-1005
Abstract PDF

The feasibility of improving the light extraction efficiency of GaN/GaAs optical devices by using wafer bonding technique is calculated by the principles of optical thin films. The light extraction efficiency can be improved by 2.65 times when a thin Ni layer is used as an adhesive layer and Ag layer as a reflective layer. Experimental results show that the reflectivity at 459.2 nm of the bonded samples is improved by 2.4 times than the as grown samples.

The feasibility of improving the light extraction efficiency of GaN/GaAs optical devices by using wafer bonding technique is calculated by the principles of optical thin films. The light extraction efficiency can be improved by 2.65 times when a thin Ni layer is used as an adhesive layer and Ag layer as a reflective layer. Experimental results show that the reflectivity at 459.2 nm of the bonded samples is improved by 2.4 times than the as grown samples.

CMP Slurry of Copper Interconnection for ULSI
WANG Xin, LIU Yu-ling
Chin. J. Semicond.  2002, 23(9): 1006-1008
Abstract PDF

In order to avoid many scratches on copper surface caused by hard mechanical alumina abrasives in Cu CMP process, a new Cu CMP slurry is introduced in which organic amine is used for complex agent and colloidal silica (SiO2) is used for abrasive particles.According to mechanism of strong complex, the issuses that copper layer is removed slowly by colloidal silica and silicon glue generates in solution are solved. The experimental result shows that the slurry is suitable for the first polishing of Cu CMP process with faster removal rate of Cu and high selectivity between Cu/Ta/SiO2.

In order to avoid many scratches on copper surface caused by hard mechanical alumina abrasives in Cu CMP process, a new Cu CMP slurry is introduced in which organic amine is used for complex agent and colloidal silica (SiO2) is used for abrasive particles.According to mechanism of strong complex, the issuses that copper layer is removed slowly by colloidal silica and silicon glue generates in solution are solved. The experimental result shows that the slurry is suitable for the first polishing of Cu CMP process with faster removal rate of Cu and high selectivity between Cu/Ta/SiO2.