Cu互连应力迁移温度特性研究

提出了一种基于扩散-蠕变机制的空洞生长模型, 结合应力模拟计算和聚焦离子束分析技术研究了Cu互连应力诱生空洞失效现象, 探讨了应力诱生空洞的形成机制并分析了空洞生长速率与温度、应力梯度和扩散路径的关系. 研究结果表明, 在Cu M1互连顶端通孔拐角底部处应力和应力梯度达到极大值并观察到空洞出现. 应力梯度是决定空洞成核位置及空洞生长速率的关键因素. 应力迁移是空位在应力梯度作用下沿主导扩散路径进行的空位积聚与成核现象, 应力梯度的作用与扩散作用随温度变化方向相反, 并存在一个中值温度使得应力诱生空洞速率达到极大值. 关键词： Cu互连 应力迁移 应力诱生空洞 失效     

热-电应力下Cu/Ni/SnAg_(1.8)/Cu倒装铜柱凸点界面行为及失效机理

微互连铜柱凸点因其密度高、导电性好、噪声小被广泛应用于存储芯片、高性能计算芯片等封装领域,研究铜柱凸点界面行为对明确其失效机理和组织演变规律、提升倒装封装可靠性具有重要意义.采用热电应力实验、在线电学监测、红外热像测试和微观组织分析等方法,研究Cu/Ni/SnAg_1.8/Cu微互连倒装铜凸点在温度100—150℃、电流密度2×10~4—3×10~4 A/cm~2热电应力下的互连界面行为、寿命分布、失效机理及其影响因素.铜柱凸点在热电应力下的界面行为可分为Cu_6Sn_5生长和Sn焊料消耗、Cu_6Sn_5转化成Cu_3Sn、空洞形成及裂纹扩展3个阶段,Cu_6Sn_5转化为Cu_3Sn的速率与电流密度正相关.热电应力下,铜凸点互连存在Cu焊盘消耗、焊料完全合金化成Cu_3Sn、阴极镍镀层侵蚀和层状空洞4种失效模式.基板侧Cu焊盘和铜柱侧Ni镀层的溶解消耗具有极性效应,当Cu焊盘位于阴极时,电迁移方向与热迁移方向相同,加速Cu焊盘的溶解以及Cu_3Sn生长,当Ni层为阴极时,电迁移促进Ni层的消耗,在150℃,2.5×10~4 A/cm~2下经历2.5h后,Ni阻挡层出现溃口,导致Ni层一侧的铜柱基材迅速转化成(Cu_x,Ni_y)_6Sn_5和Cu_3Sn合金.铜柱凸点互连寿命较好地服从2参数威布尔分布,形状参数为7.78,为典型的累积耗损失效特征.研究结果表明:相比单一高温应力,热电综合应力显著加速并改变了铜柱互连界面金属间化合物的生长行为和失效机制.     

铜互连电迁移失效阻变特性研究

基于铜互连电迁移失效微观机理分析建立一种Cu／SiCN互连电迁移失效阻变模型，并提出一种由互连阻变曲线特征参数即跳变台阶高度与斜率来获取失效物理参数的提取方法．研究结果表明，铜互连电迁移失效时间由一定电应力条件下互连阴极末端晶粒耗尽时间决定．铜互连电迁移失效一般分为沟槽型和狭缝型两种失效模式．沟槽型空洞失效模式对应的阻变曲线一般包括跳变台阶区和线性区两个特征区域．晶粒尺寸分布与临界空洞长度均符合正态对数分布且分布参数基本一致．阻变曲线线性区斜率与温度呈指数函数关系．利用阻变模型提取获得的电迁移扩散激活能约为0．9eV,与Black方法基本一致．     

基于微观结构的Cu互连电迁移失效研究

提出了一种基于微观晶粒尺寸分布的Cu互连电迁移失效寿命模型. 结合透射电子显微镜和统计失效分析技术, 研究了Cu互连电迁移失效尺寸缩小和临界长度效应及其物理机制. 研究表明, 当互连线宽度减小, 其平均晶粒尺寸下降并导致互连电迁移寿命降低. 小于临界长度的互连线无法提供足够的空位使得铜晶粒耗尽而发生失效. 当互连长度大于该临界长度时, 在整个电迁移测试时间内, 部分体积较小的阴极端铜晶粒出现耗尽情况. 随着互连长度的增加该失效比例迅速增大, 电迁移失效寿命减小. 当互连长度远大于扩散长度时, 失效时间主要取决于铜晶粒的尺寸, 且失效寿命和比例随晶粒尺寸变化呈现饱和的波动状态. 关键词： Cu 互连 电迁移 微观结构     

面心立方铜晶体中空洞生长与贯通的尺寸效应

利用分子动力学方法模拟沿拉伸方向排布的两个空洞在单轴拉伸作用下的动力学行为.着重研究不同尺寸空洞对其拉伸贯通过程的影响.结果表明,不同尺度的空洞都是通过空洞表面发射位错环长大与贯通的.空洞在弹性阶段沿加载方向缓慢长大,在塑性阶段沿垂直方向生长后形成类八面体形状.随空洞尺寸的减小,临界屈服应力逐渐增大.当半径较大时,位错对称成核、迁移,空洞沿加载方向被拉长,演化过程相似;当半径较小时,位错不对称成核,空洞沿垂直方向被拉长.空洞生长分为弹性变形、独立长大、融合贯通和平稳生长四个阶段.独立生长阶段随尺寸的减小逐渐缩短甚至消失.     

倒装芯片上金属布线/凸点互连结构中原子的定向扩散

采用倒装芯片互连凸点串联回路研究了高温、高电流密度条件下倒装芯片上金属布线/凸点互连结构中原子的定向扩散现象,分析了互连结构中受电应力和化学势梯度作用的各相金属原子的扩散行为.在电迁移主导作用下,Ni(V)镀层中的Ni原子的快速扩散导致原本较为稳定的Ni(V)扩散阻挡层发生快速的界面反应,造成Al互连金属与焊料的直接接触.Al原子在电子风力作用下沿电子流方向向下迁移造成窗口附近焊料中Al原子含量逐步上升,同时,空位的反向迁移、聚集形成过饱和,导致Al互连中形成大面积空洞.焊料中的Sn,Pb原子在化学势梯度 关键词： 倒装芯片 凸点 电迁移 扩散     

金属互连电迁移噪声的分形特征

为了研究金属互连电迁移失效机理并寻找新的电迁移表征参量，应用分形理论，通过电子扩散轨迹分形维数，将电迁移噪声时间序列分形维数与晶粒间界分形维数相联系，确定了噪声时间序列分形维数在电迁移演变中的变化趋势.研究结果表明，在金属互连电迁移前期，晶粒间界形貌越来越复杂，致使噪声时间序列的分形维数逐渐增大；成核后，由于空位凝聚成空洞，晶粒间界形貌变得较成核前规则，致使噪声时间序列的分形维数减小；成核时刻是其折点.实验结果证明理论分析的正确性，噪声时间序列的分形维数可望作为金属互连电迁演变的表征参量.     

外加应力作用下UO2中空洞演化过程的相场模拟

本工作建立了外加应力作用下UO₂中空洞演化的相场模型.首先,使用摄动迭代法求解了弹性平衡方程,对外加应力下单个空洞周围的应力分布进行了计算,结果表明空洞边缘有应力集中现象,模拟得到的应力分布和解析解一致.然后,利用相场方法模拟了不同外加应力下单个空洞的演化过程,结果表明随着外加应力的增大,空洞的生长速度加快.最后,研究了外加应力对多晶体系中晶粒长大和空洞演化的影响,结果表明,不同晶粒内的应力大小不同,应力越小的晶粒越容易长大,尺寸越大的空洞的边缘应力也越大.晶间空洞与弯曲晶界存在相互作用,一方面晶界附近的空洞会生长成透镜状,另一方面空洞对晶界也有钉扎作用,能减缓晶界的迁移.此外,外加应力会加速多晶系统中空洞的生长,并且本文计算得到了外加应力与空洞半径的关系,发现外加应力越大,空洞的生长越快.     

高导无氧铜的高压与高应变率本构模型研究

基于Y/G及G/B为常数的假设,构建了高导无氧铜的七种高压与高应变率本构模型.对于高导无氧铜进行了平面冲击波试验,采用纵向与横向锰铜应力计记录了试件中的纵向与横向应力,从而得到了屈服应力历史.用所构建的七种本构模型进行了数值模拟,并与高导无氧铜的平面冲击波试验结果进行比较.结果表明,平面冲击波载荷下高导无氧铜的屈服强度对于压力、密度、温度以及塑性应变的依赖性是本构描述的关键.而由Hopkinson试验取得的高导无氧铜高应变率本构模型,并不适合描述平面冲击波载荷下的本构特性. 关键词： 本构模型 高导无氧铜 平面冲击波试验 锰铜应力计     

超大规模集成电路互连电迁移自由体积电阻模型

将晶核析出的Avrami 方程应用于描述超大规模集成电路中金属Al薄膜互连电迁移过程中电阻的演变. 根据电子散射理论，晶界电阻主要起源于晶界处空位或者空洞对电子的散射. 为了描述这些离子的特征，引入了自由体积的概念，将晶界处电子散射这个复杂的过程简化用自由体积的有效散射截面来描述，从而建立了自由体积与电阻变化的定量关系，统一描述了电迁移过程中不同阶段的电阻变化. 数值模拟结果表明，在第一个空洞成核时刻电阻会发生急剧变化，这一结果已被实验所证实. 关键词： 电迁移 Al互连 电阻变化     

CoxCu1-x复相纳米线阵列的制备及其磁性的研究

利用电化学沉积方法在同一种富Co²⁺溶液Co²⁺/Cu²⁺=10∶1中,利用不同的沉积电位成功地制备了一系列不同成分(x=0.38—0.87)和复合相结构的Co_xCu_1-x纳米线阵列.发现随着纳米线中Cu含量的变化,Co_xCu_1-x纳米线的复相结构随之发生规律的变化,最终导致纳米线的磁性也随之规律的变化.随着纳米线中Cu含量的不断增加,一部分Cu与Co形成面心立方结构(fcc)的CoCu固溶体,减弱了磁晶各向异性与形状各向异性的竞争,从而提高样品的方形度;一部分Cu以fcc结构的Cu单质的形式存在于纳米线中,并随着Cu颗粒大小的不同分别起到破坏磁晶各向异性和钉扎畴壁的作用,从而增加纳米线的方形度和矫顽力.对比不同成分的样品,发现Co_xCu_1-x纳米线的方形度和矫顽力的最大值分别出现在Co₇₅Cu₂₅和Co₆₀Cu₄₀中,并且由于其特殊的复相结构致使它们的值要好于相同直径的单相结构的结果. 关键词： 纳米线 电化学沉积 磁性     

Vertical assembly of carbon nanotubes for via interconnects

The via interconnects are key components in ultra-large scale integrated circuits(ULSI).This paper deals with a new method to create single-walled carbon nanotubes(SWNTs) via interconnects using alternating dielectrophoresis(DEP).Carbon nanotubes are vertically assembled in the microscale via-holes successfully at room temperature under ambient condition.The electrical evaluation of the SWNT vias reveals that our DEP assembly technique is highly reliable and the success rate of assembly can be as high as 90%.We also propose and test possible approaches to reducing the contact resistance between CNT vias and metal electrodes.     

High-resolution in-beam particle spectroscopy --New results on prompt proton emission from 58Cu

Prompt proton decay lines in ⁵⁸Cu have been studied by means of high-resolution in-beam particle-γ coincidence spectroscopy using the GAMMASPHERE Ge-detector array in conjunction with a dedicated set of ancillary detectors including four ΔE-E silicon-strip telescopes. High-spin states in ⁵⁸Cu have been populated via the heavy-ion fusion-evaporation reaction ²⁸Si(³⁶Ar, 1α1p1n) at 148 MeV beam energy. The full-width at half maximum for the proton peak could be reduced significantly compared to earlier experiments. The results indicate that only one prompt proton decay branch exists in the decay-out of the well-deformed band of ⁵⁸Cu. Received: 11 March 2002 / Accepted: 10 April 2002     

Micro-Raman investigation of stress distribution in laser drilled via structures

Through-wafer vertical electrical interconnects (vias) with diameters varied from 15 to 80 μm were formed on Si substrates using a UV diode-pumped solid state laser (355 nm). Micro-Raman spectroscopy was employed for the investigation of stress and structural changes induced in silicon within the heat-affected zone due to laser machining. A maximum stress of ∼300 MPa, as a result of laser drilling, was observed close to the via edge. It was found that the stress decays within a distance of 1-3 μm from the via’s side-wall and that the laser machining did not lead to the formation of amorphous silicon around the via structures.     

Modeling of supersonic plasma flow in the scrape-off layer

The substantial drop of the plasma temperature along magnetic field lines with increasing plasma density is one of the main features in tokamak divertors. As a result the temperature gradient at the divertor plates becomes very steep and the boundary condition normally applied for the parallel Mach number M at the target, M_t = 1, cannot be satisfied. In this case the value of M_t based on the general form of the Bohm criterion, M_t  1, has to be determined from the continuity of plasma parameters.In the present paper a new approach to resolve the Mach number at the target for such a situation is proposed. This method avoids the singularity problem that arises by treating the particle balance and parallel motion equations in a differential form. Instead, the integral representation of the equations is formulated for an arbitrary form of particle and momentum sources. The approach can also take into account transport perpendicular to the magnetic field lines.The proposed method is demonstrated on the example of a one-dimensional stationary model for the scrape-off layer (SOL) plasma and includes the continuity-, parallel momentum- and heat transfer equations. The recycled neutrals are described in the diffusion approximation. In the case of low density the normal condition M_t = 1 is satisfied and the results are in agreement with the two-point model. At high enough plasma density solutions with the supersonic flow at the divertor plates, M_t > 1, are found. These states correspond to a partially detached plasma with a temperature of a few eV.     

A Fast Variant of H Spectroscopic U-FLARE Imaging Using Adjusted Chemical Shift Phase Encoding

So far, fast spectroscopic imaging (SI) using the U-FLARE sequence has provided metabolic maps indirectly via Fourier transformation (FT) along the chemical shift (CS) dimension and subsequent peak integration. However, a large number of CS encoding steps N_ω is needed to cover the spectral bandwidth and to achieve sufficient spectral resolution for peak integration even if the number of resonance lines is small compared to N_ω and even if only metabolic images are of interest and not the spectra in each voxel. Other reconstruction algorithms require extensive prior knowledge, starting values, and/or model functions. An adjusted CS phase encoding scheme (APE) can be used to overcome these drawbacks. It incorporates prior knowledge only about the resonance frequencies present in the sample. Thus, N_ω can be reduced by a factor of 4 for many ¹H in vivo studies while no spectra have to be reconstructed, and no additional user interaction, prior knowledge, starting values, or model function are required. Phantom measurements and in vivo experiments on rat brain have been performed at 4.7 T to test the feasibility of the method for proton SI.     

Korrelationseffekte in den Auger-Spektren von Edelgasen

The possibility of studying correlation effects through Auger electron spectrometry has been shown recently by Krause, Carlson and Moddeman in the case of theK Auger spectrum of neon. As a further example we have measured theM _4,5 Auger spectrum of krypton with high resolution. Correlation effects have been found via the strong deviations of relative intensities of Auger diagram lines (e.g.M _4,5 N ₁ N _2,3(¹ p ₁)) from theoretical values and via the occurence of double Auger transitions, where one electron is emitted and another is excited. A critical examination of high resolution Auger spectra of noble gases, which has been measured so far, has shown that several non diagram lines can be assigned to double Auger transitions of the above kind.     

Properties of reactively sputtered W–B–N thin film as a diffusion barrier for Cu metallization on Si

Thin films of W–B–N (10 nm) have been evaluated as diffusion barriers for Cu interconnects. The amorphous W–B–N thin films were prepared at room temperature via reactive magnetron sputtering using a W₂B target at various N₂/(Ar + N₂) flow ratios. Cu diffusion tests were performed after in-situ deposition of 200 nm Cu. Thermal annealing of the barrier stacks was carried out in vacuum at elevated temperatures for one hour. X-ray diffraction patterns, sheet resistance measurement, cross-section transmission electron microscopy images, and energy-dispersive spectrometer scans on the samples annealed at 500°C revealed no Cu diffusion through the barrier. The results indicate that amorphous W–B–N is a promising low resistivity diffusion barrier material for copper interconnects.     

Chromatic Spatial Summation at Equiluminance

The effect of spatial summation on chromatic detection at suprathreshold levels for the red-green and yellow-blue chromatic opponent mechanisms has been studied using simple reaction time (RT) as a measurement of the response. We varied the target size in the experiments from 8 min of arc in diameter to 10 deg, and the hue-substitution paradigm was adopted to generate only chromatic changes in the cone input. Equiluminant stimuli were distributed along three tritan and three red-green confusion lines. Three reference stimuli were chosen as both reference and adapting stimuli, in this case to determine whether the chromatic adaptation affected the spatial summation. The results showed that the mean value of RT at a constant cone-input variation, either of the L #x2212; 2M channel or of the S #x2212; (L #x002B; M) channel, against the target size decreased until a stimulus size was reached, the RT being constant for greater sizes. The spatial summation was fulfilled for a maximum target size which was less for the L #x2212; 2M channel than for the S #x2212; (L #x002B; M) channel, this size decreasing as the cone-input variation rose. Furthermore, there was no spatial summation for high-enough variations in the cone input. In addition, the chromatic mechanisms appear to organize their spatial integration with the adapting level.