A 512-bit EEPROM IP was designed by using just logic process based devices.To limit the voltages of the devices within 5.5 V,EEPROM core circuits,control gate(CG) and tunnel gate(TG) driving circuits,DC-DC converters:...A 512-bit EEPROM IP was designed by using just logic process based devices.To limit the voltages of the devices within 5.5 V,EEPROM core circuits,control gate(CG) and tunnel gate(TG) driving circuits,DC-DC converters:positive pumping voltage(VPP=4.75 V) ,negative pumping voltage(VNN=4.75 V) ,and VNNL(=VNN/2) generation circuit were proposed.In addition,switching powers CG high voltage(CG_HV) ,CG low voltage(CG_LV) ,TG high voltage(TG_HV) ,TG low voltage(TG_LV) ,VNNL_CG and VNNL_TG switching circuit were supplied for the CG and TG driving circuit.Furthermore,a sequential pumping scheme and a new ring oscillator with a dual oscillation period were proposed.To reduce a power consumption of EEPROM in the write mode,the reference voltages VREF_VPP for VPP and VREE_VNN for VNN were used by dividing VDD(1.2 V) supply voltage supplied from the analog block in stead of removing the reference voltage generators.A voltage level detector using a capacitive divider as a low-power DC-DC converter design technique was proposed.The result shows that the power dissipation is 0.34μW in the read mode,13.76μW in the program mode,and 13.66μW in the erase mode.展开更多
With the introduction of EUV lithography,the photolithographic process in 5 nm logic process can be simplified to use mostly single exposure method.In a typical 5 nm logic process,the contact-poly pitch(CPP)is 44-50 n...With the introduction of EUV lithography,the photolithographic process in 5 nm logic process can be simplified to use mostly single exposure method.In a typical 5 nm logic process,the contact-poly pitch(CPP)is 44-50 nm,the minimum metal pitch(MPP)is around 30-32 nm.And the overlay budget is estimated to be 2.5 nm(on product overlay).Although the optical resolution of a 0.33NA exposure tool(such as ASML NXE3400)can reach below 32 nm pitch,stochastics in the EUV absorption in photoresists has limited its application to smaller pitches.For the CPP mentioned above,one can use 193 nm immersion lithography with Self-Aligned Double Patterning(SADP)technique to provide good image contrast(or CDU,LWR)as well as good overlay,as for the 10 and 7 nm generations.In the BEOL,however,the 30-32 pitch cannot be realized by a single EUV exposure with enough printing defect process window.If this pitch is to be done by 193 nm immersion lithography,more than 6-8 exposures are needed with very complicated overlay result.For EUV,this can be done through self-aligned LELE with both good CD and overlay control.We have done an optimization of the photolithographic process parameters for the typical metal with a self-developed aerial image simulator based on rigorous coupled wave analysis(RCWA)algorithm and the Abbe imaging routine with an EUV absorption model which describes stochastics.We have calibrated our model with wafer exposure data from several photoresists under collaboration with IMEC.As we have presented last year,to accommodate all pitches under a logic design rule,as well as to provide enough CDU for the logic device performance,in DUV lithography,a typical minimum exposure latitude(EL)for the poly and metal layers can be set at,respectively,18%and 13%.In EUV,due to the existence of stochastics,13%EL,which corresponds to an imaging contrast of 40%,seems not enough for the metal trenches,and to obtain an imaging contrast close to 100%,which yields an EL of 31.4%means that we need to relax minimum pitch to above 41 nm(conventional imaging limit for 0.33NA).This is the best imaging contrast a photolithographic process can provide to reduce LWR and stochastics.In EUV,due to the significantly smaller numerical apertures compared to DUV,the aberration impact can cause much more pronounced image registration error,in order to satisfy 2.5 nm total overlay,the aberration induced shift needs to be kept under 0.2 nm.We have also studied shadowing effect and mask 3D scattering effect and our results will be provided for discussion.展开更多
A single poly EEPROM cell circuit sharing the deep N-well of a cell array was designed using the logic process. The proposed cell is written by the FN tunneling scheme and the cell size is 41.26 μm2, about 37% smalle...A single poly EEPROM cell circuit sharing the deep N-well of a cell array was designed using the logic process. The proposed cell is written by the FN tunneling scheme and the cell size is 41.26 μm2, about 37% smaller than the conventional cell. Also, a small-area and low-power 512-bit EEPROM IP was designed using the proposed cells which was used for a 900 MHz passive UHF RFID tag chip. To secure the operation of the cell proposed with 3.3 V devices and the reliability of the used devices, an EEPROM core circuit and a DC-DC converter were proposed. Simulation results for the designed EEPROM IP based on the 0.18μm logic process show that the power consumptions in read mode, program mode and erase mode are 11.82, 25.15, and 24.08 ~tW, respectively, and the EEPROM size is 0.12 mm2.展开更多
How to design a multicast key management system with high performance is a hot issue now. This paper will apply the idea of hierarchical data processing to construct a common analytic model based on directed logical k...How to design a multicast key management system with high performance is a hot issue now. This paper will apply the idea of hierarchical data processing to construct a common analytic model based on directed logical key tree and supply two important metrics to this problem： re-keying cost and key storage cost. The paper gives the basic theory to the hierarchical data processing and the analyzing model to multieast key management based on logical key tree. It has been proved that the 4-ray tree has the best performance in using these metrics. The key management problem is also investigated based on user probability model, and gives two evaluating parameters to re-keying and key storage cost.展开更多
Modal logic characterization in a higher-order setting is usually not a trivial task because higher-order process-passing is quite different from first-order name-passing. We study the logical characterization of high...Modal logic characterization in a higher-order setting is usually not a trivial task because higher-order process-passing is quite different from first-order name-passing. We study the logical characterization of higherorder processes constrained by linearity. Linearity respects resource-sensitiveness and does not allow processes to duplicate themselves arbitrarily. We provide a modal logic that characterizes linear higher-order processes,particularly the bisimulation called local bisimulation over them. More importantly, the logic has modalities for higher-order actions downscaled to resembling first-order ones in Hennessy-Milner logic, based on a formulation exploiting the linearity of processes.展开更多
模型劫持攻击是一种新型攻击方式,通过植入特定词语,能够隐蔽地控制模型执行与原始任务截然不同的劫持任务,使模型拥有者的训练算力成本增加的同时面临潜在的法律风险。目前,已有研究针对德-英文语言翻译模型探索了这一攻击方式,但在中...模型劫持攻击是一种新型攻击方式,通过植入特定词语,能够隐蔽地控制模型执行与原始任务截然不同的劫持任务,使模型拥有者的训练算力成本增加的同时面临潜在的法律风险。目前,已有研究针对德-英文语言翻译模型探索了这一攻击方式,但在中文自然语言处理(natural language processing,NLP)领域尚属空白。中文语言的独特性使得其面临不同于其他语言环境的安全挑战,因此亟需开发针对中文模型的攻击评估方法。基于上述事实,提出了一种基于中文逻辑词的模型劫持攻击方法Cheater,用于评估中文模型的安全性。Cheater针对中-英文NLP任务,首先使用公共模型对劫持数据进行伪装生成过渡数据,再通过在过渡样本中嵌入中文逻辑词的方式对其进行改造生成毒性数据,最后利用毒性数据完成对目标模型的劫持。实验表明,对于Bart[large]模型,Cheater在0.5%的数据投毒率下攻击成功率可以达到90.2%。展开更多
基金Project supported by the Second Stage of Brain Korea 21
文摘A 512-bit EEPROM IP was designed by using just logic process based devices.To limit the voltages of the devices within 5.5 V,EEPROM core circuits,control gate(CG) and tunnel gate(TG) driving circuits,DC-DC converters:positive pumping voltage(VPP=4.75 V) ,negative pumping voltage(VNN=4.75 V) ,and VNNL(=VNN/2) generation circuit were proposed.In addition,switching powers CG high voltage(CG_HV) ,CG low voltage(CG_LV) ,TG high voltage(TG_HV) ,TG low voltage(TG_LV) ,VNNL_CG and VNNL_TG switching circuit were supplied for the CG and TG driving circuit.Furthermore,a sequential pumping scheme and a new ring oscillator with a dual oscillation period were proposed.To reduce a power consumption of EEPROM in the write mode,the reference voltages VREF_VPP for VPP and VREE_VNN for VNN were used by dividing VDD(1.2 V) supply voltage supplied from the analog block in stead of removing the reference voltage generators.A voltage level detector using a capacitive divider as a low-power DC-DC converter design technique was proposed.The result shows that the power dissipation is 0.34μW in the read mode,13.76μW in the program mode,and 13.66μW in the erase mode.
文摘With the introduction of EUV lithography,the photolithographic process in 5 nm logic process can be simplified to use mostly single exposure method.In a typical 5 nm logic process,the contact-poly pitch(CPP)is 44-50 nm,the minimum metal pitch(MPP)is around 30-32 nm.And the overlay budget is estimated to be 2.5 nm(on product overlay).Although the optical resolution of a 0.33NA exposure tool(such as ASML NXE3400)can reach below 32 nm pitch,stochastics in the EUV absorption in photoresists has limited its application to smaller pitches.For the CPP mentioned above,one can use 193 nm immersion lithography with Self-Aligned Double Patterning(SADP)technique to provide good image contrast(or CDU,LWR)as well as good overlay,as for the 10 and 7 nm generations.In the BEOL,however,the 30-32 pitch cannot be realized by a single EUV exposure with enough printing defect process window.If this pitch is to be done by 193 nm immersion lithography,more than 6-8 exposures are needed with very complicated overlay result.For EUV,this can be done through self-aligned LELE with both good CD and overlay control.We have done an optimization of the photolithographic process parameters for the typical metal with a self-developed aerial image simulator based on rigorous coupled wave analysis(RCWA)algorithm and the Abbe imaging routine with an EUV absorption model which describes stochastics.We have calibrated our model with wafer exposure data from several photoresists under collaboration with IMEC.As we have presented last year,to accommodate all pitches under a logic design rule,as well as to provide enough CDU for the logic device performance,in DUV lithography,a typical minimum exposure latitude(EL)for the poly and metal layers can be set at,respectively,18%and 13%.In EUV,due to the existence of stochastics,13%EL,which corresponds to an imaging contrast of 40%,seems not enough for the metal trenches,and to obtain an imaging contrast close to 100%,which yields an EL of 31.4%means that we need to relax minimum pitch to above 41 nm(conventional imaging limit for 0.33NA).This is the best imaging contrast a photolithographic process can provide to reduce LWR and stochastics.In EUV,due to the significantly smaller numerical apertures compared to DUV,the aberration impact can cause much more pronounced image registration error,in order to satisfy 2.5 nm total overlay,the aberration induced shift needs to be kept under 0.2 nm.We have also studied shadowing effect and mask 3D scattering effect and our results will be provided for discussion.
基金Project(10039239) supported by the Industrial Strategic Technology Development Program Funded by the Ministry of Knowledge Economy, Korea
文摘A single poly EEPROM cell circuit sharing the deep N-well of a cell array was designed using the logic process. The proposed cell is written by the FN tunneling scheme and the cell size is 41.26 μm2, about 37% smaller than the conventional cell. Also, a small-area and low-power 512-bit EEPROM IP was designed using the proposed cells which was used for a 900 MHz passive UHF RFID tag chip. To secure the operation of the cell proposed with 3.3 V devices and the reliability of the used devices, an EEPROM core circuit and a DC-DC converter were proposed. Simulation results for the designed EEPROM IP based on the 0.18μm logic process show that the power consumptions in read mode, program mode and erase mode are 11.82, 25.15, and 24.08 ~tW, respectively, and the EEPROM size is 0.12 mm2.
基金Supported by the National High-Technology Re-search and Development Programof China(2001AA115300) the Na-tional Natural Science Foundation of China (69874038) ,the Nat-ural Science Foundation of Liaoning Province(20031018)
文摘How to design a multicast key management system with high performance is a hot issue now. This paper will apply the idea of hierarchical data processing to construct a common analytic model based on directed logical key tree and supply two important metrics to this problem： re-keying cost and key storage cost. The paper gives the basic theory to the hierarchical data processing and the analyzing model to multieast key management based on logical key tree. It has been proved that the 4-ray tree has the best performance in using these metrics. The key management problem is also investigated based on user probability model, and gives two evaluating parameters to re-keying and key storage cost.
基金the National Natural Science Foundation of China(Nos.61202023,61261130589 and61173048)the PACE Project(No.12IS02001)the Specialized Research Fund for the Doctoral Program of Higher Edueation of China(No.20120073120031)
文摘Modal logic characterization in a higher-order setting is usually not a trivial task because higher-order process-passing is quite different from first-order name-passing. We study the logical characterization of higherorder processes constrained by linearity. Linearity respects resource-sensitiveness and does not allow processes to duplicate themselves arbitrarily. We provide a modal logic that characterizes linear higher-order processes,particularly the bisimulation called local bisimulation over them. More importantly, the logic has modalities for higher-order actions downscaled to resembling first-order ones in Hennessy-Milner logic, based on a formulation exploiting the linearity of processes.
文摘模型劫持攻击是一种新型攻击方式,通过植入特定词语,能够隐蔽地控制模型执行与原始任务截然不同的劫持任务,使模型拥有者的训练算力成本增加的同时面临潜在的法律风险。目前,已有研究针对德-英文语言翻译模型探索了这一攻击方式,但在中文自然语言处理(natural language processing,NLP)领域尚属空白。中文语言的独特性使得其面临不同于其他语言环境的安全挑战,因此亟需开发针对中文模型的攻击评估方法。基于上述事实,提出了一种基于中文逻辑词的模型劫持攻击方法Cheater,用于评估中文模型的安全性。Cheater针对中-英文NLP任务,首先使用公共模型对劫持数据进行伪装生成过渡数据,再通过在过渡样本中嵌入中文逻辑词的方式对其进行改造生成毒性数据,最后利用毒性数据完成对目标模型的劫持。实验表明,对于Bart[large]模型,Cheater在0.5%的数据投毒率下攻击成功率可以达到90.2%。
