ID 原文 译文
43106 利用 Sentaurus TCAD 软件对不同尺寸的纳米线器件性能进行仿真，采用控制变量法，以 0. 5 nm 为步长，分别将纳米线高度及宽度从4 nm增加至 8 nm。 The performances of different sizes of nanowire devices weresimulated by Sentaurus TCAD software.Using the control variable method, the height and width of thenanowires were increased from 4 nm to 8 nm in steps of 0.5 nm, respectively.
43107 仿真结果表明，当纳米线高度及宽度分别取 4 nm 和 6. 5 nm 时，可最大程度规避微尺度空间效应对载流子迁移率的影响，并有效提升散热能力， The simulation results showthat when the height and width of the nanowires are 4 nm and 6.5 nm, respectively, the influence of themicro-scale spatial effect on the carrier mobility can be largely avoided, and the heat dissipationcapability is effectively improved.
43108 使器件开态电流增加44.4%，沟道热学电阻减小 60. 3%。 The on-state current of the device increases 44.4% and the channelthermal resistance decreases 60.3%.
43109 此外，设置纳米线高度为 4 nm，依次将顶部/中部/底部沟道的纳米线宽度从 6. 5 nm 增加至 8 nm， In addition, when the height of the nanowire is 4 nm, the width ofthe nanowires of the top /middle /bottom channel is sequentially increased from 6.5 nm to 8 nm.
43110 发现当底部沟道的纳米线宽度相等时，增加靠近体硅处的沟道宽度更有利于改善器件的电热性能。 And it isfound that when the width of the nanowires of the bottom channel is equal, the increase of the channelwidth close to the bulk silicon is more conducive to improve the electrothermal performance of the device.
43111 在硅通孔 ( TSV) 制造工艺中，TSV 不可避免会出现电阻开路和电流泄漏同时存在的复合故障， In the manufacturing process of through-silicon via (TSV) , TSVs inevitably appear composite faults with both open-circuit and current leakage.
43112 且相比 TSV 单一故障，复合故障会大大降低三维集成电路的可靠性。 Compared with TSV single faults, compositefaults greatly reduce the reliability of three-dimensional (3D) integrated circuits.
43113 以 TSV 作为环形振荡器的负载，以环形振荡器的振荡周期与占空比为测试参数，提出了一种基于粒子群优化( PSO) 的最小二乘支持向量机 ( LSSVM) 的故障诊断模型。 A fault diagnosis modelbased on the least squares support vector machine (LSSVM) optimized by particle swarm optimization(PSO) was proposed.TSV was used as a load of the ring oscillator, and the oscillation period and dutycycle of the ring oscillator were used as test parameters.
43114 利用不同故障类型的振荡周期与占空比的数据来训练 LSSVM，采用 PSO 优化 LSSVM 的结构参数，提高了模型诊断的效率与正确率。 LSSVM was trained with data of oscillationperiods and duty cycles of different fault types, and the structural parameters of the LSSVM were optimized by PSO, which improved the efficiency and accuracy of the model diagnosis.
43115 仿真结果表明，该方法不仅能够检测出故障，还可以将故障进行分类，即开路故障、泄漏故障以及不同程度的复合故障。 The simulation resultsshow that the method can detect faults and classify them into open faults, leakage faults and compositefaults with different degrees.