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．     

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.     

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.     

Al互连线和Cu互连线的显微结构

利用电子背散射衍射(EBSD)技术，测量了由反应离子刻蚀工艺(RIE)制备的Al互连线和大马士革工艺(Damascene)制备的Cu互连线的显微结构，包括晶粒尺寸、晶体学取向和晶界特征.分析了Cu互连线线宽，及Al和Cu互连线退火工艺对互连线显微结构及电徙动失效的影响.     

通孔尺寸对铜互连应力迁移失效的影响

在200℃温度下进行了700h双层铜互连(M1/M2)的应力迁移加速老化试验, 结合有限元分析和聚焦离子束(focused-ion-beam,简称FIB)技术研究了通孔直径分别为500和350nm的铜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理, 并分析了通孔尺寸对铜互连应力迁移的影响. 结果表明,M1互连应力和应力梯度在通孔底部边缘处达到极大值. 应力梯度在应力诱生空洞成核过程中起主导作用, 由张应力产生的过剩空位在应力梯度作用下沿Cu M1/SiN界面作扩散运动并在应力梯度极大值处成核生长成空洞. 由于M1互连应力沿横向方向变化较快, 因此应力诱生空洞的横向生长速率较大. 当通孔直径增大时,互连应力和应力梯度值增大, 并导致应力诱生空洞的生长速率上升. 关键词： 铜互连 应力迁移 应力诱生空洞 失效     

组分对Cu-Zn-In-S/ZnS核壳量子点发光性能的影响

利用热注射法通过调控Cu/Zn比例制备了不同组分的Cu-Zn-In-S/ZnS核壳量子点,通过紫外-可见吸收光谱以及稳态和时间分辨光谱分析Cu/Zn比例对量子点发光性能的影响.结果表明,不同组分Cu-Zn-In-S/ZnS核壳量子点呈现闪锌矿结构且晶粒尺寸接近;随着Cu/Zn比例的减小,Cu-Zn-In-S/ZnS核壳量子点的带隙变宽,导致吸收光谱发生蓝移;当Cu/Zn比例从6/1减小到1/6时,量子点的发光峰位从640nm蓝移529nm.由于Zn2+替代Cu+能够减少Cu原子缺陷的形成,从而提高了量子点的荧光效率;当Cu/Zn=1/6时,样品中观测到Cu+离子发光和较长的荧光寿命.     

有丢失物缺陷的铜互连线中位寿命的定量研究

本文研究了六层互连线上的丢失物缺陷对互连电迁移中位寿命的影响,提出了各层互连线缺陷处的温度模型和缺陷在不同互连层的中位寿命模型,能够定量地计算缺陷对互连电迁移中位寿命的影响,给出了提高互连线中位寿命的方法.研究结果表明：互连线宽度与缺陷处互连线有效宽度的比值越大,互连线寿命越短;缺陷处的温度越高,互连线寿命越短.在互连线参数变化明显的层与层之间,互连线寿命受比值和温度的双重影响,寿命急剧下降.根据该物理模型可以准确计算出互连线具体的温度和寿命数据,可以直接指导集成电路的设计和工艺制造. 关键词： 丢失物缺陷 中位寿命 可靠性 铜互连     

纳米多晶铜的超塑性变形机理的分子动力学探讨

在聚能装药爆炸压缩形成射流的过程中, 伴随着金属药型罩的晶粒细化, 从原始晶粒30-80 μm细化到亚微米甚至纳米量级, 从微观层面研究其细化机理和动态超塑性变形机理具有很重要的科学意义. 采用分子动力学方法模拟了不同晶粒尺寸下纳米多晶铜的单轴拉伸变形行为, 得到了不同晶粒尺寸下的应力-应变曲线, 同时计算了各应力-应变曲线所对应的平均流变应力. 研究发现平均流变应力最大值出现在晶粒尺寸为14.85 nm时. 通过原子构型显示, 给出了典型的位错运动过程和晶界运动过程, 并分析了在不同晶粒尺寸下纳米多晶铜的塑性变形机理. 研究表明: 当晶粒尺寸大于14.85 nm时, 纳米多晶铜的变形机理以位错运动为主; 当晶粒尺寸小于14.85 nm时, 变形机理以晶界运动为主, 变形机理的改变是纳米多晶铜出现软化现象即反常Hall-Petch关系的根本原因. 通过计算结果分析, 建立了晶粒合并和晶界转动相结合的理想变形机理模型, 为研究射流大变形现象提供微观变形机理参考.     

晶粒尺寸对纳米多晶铁变形机制影响的模拟研究

利用分子动力学模拟方法研究了拉伸荷载作用下晶粒尺寸对纳米多晶铁变形机制的影响.研究结果表明杨氏模量随着晶粒尺寸的减小而减小.当晶粒尺寸小于15.50 nm时,纳米多晶铁的峰值应力和晶粒尺寸之间遵循反常的Hall-Petch关系,此时晶粒旋转和晶界迁移是其塑性变形的主要变形机制;随着晶粒尺寸的增大,变形孪晶和位错滑移在其塑性变形过程中逐渐占据主导地位.裂纹的形成是导致大晶粒尺寸模型力学性能降低的主要因素.纳米多晶铁在塑性变形中会出现孪晶界的迁移和退孪晶现象.此外还研究了温度对纳米多晶铁变形机制的影响.     

金属互连电迁移噪声的多尺度熵复杂度分析

针对传统频域方法用于分析金属铝互连中的噪声信号的局限性,本文采用多尺度熵方法分析电迁移噪声时间序列.结果表明在电迁移早期,噪声信号较不规律,复杂度较大;随空洞成核的发生,噪声信号规律性增强,复杂度明显减小,反映出随电迁移过程的进行系统的混乱度不断减小.通过与传统表征参量的对比,证明多尺度熵能够对电迁移失效过程进行表征. 关键词： 电迁移 噪声 多尺度熵 复杂度     

Effects of Cr dopant on the microstructure and electromigration performance of Cu interconnects

Cu and Cu(Cr) alloy films were deposited on SiO₂ substrates by magnetron sputtering. The microstructure and electromigration performance of films were investigated. A small amount of Cr refines the Cu grains and improves the surface morphology of Cu films. After annealing at 450 °C, in contrast to the Cu film with large lateral grown grains, the Cu(Cr) alloy film exhibits fine columnar grains with a 1 1 1 preferred orientation. Most of Cr in the annealed Cu(Cr) film has segregated at the film surface and the film/substrate interface. The grain boundary grooving at the film/substrate interface is completely prohibited for Cu(Cr) films. As a result, the electromigration lifetimes of annealed Cu(Cr) lines are 10–100 times longer than those of annealed Cu lines. The final resistivity of the annealed Cu(Cr) film is 2.55 μΩ cm which is close to that of the annealed Cu film. With the improved surface morphology and high electromigration resistance, the dilute Cu(Cr) alloy film can be a viable interconnect material or a seed layer in the Cu-damascene technology.     

Improved electromigration failure in Al based interconnects

Electromigration (EM) failure in Al interconnects is significantly improved by inserting a WN film between Al and the interlayer dielectric: over 90% of test samples failed with the Al/TiN/Ti interconnects, whereas the failure rate of the Al film on WN is reduced to less than 13% under the stress con‐ ditions of 9 MA/cm² and 225 °C, and the EM lifetime is also much extended at the same conditions. Experimental results suggest that higher activation energy, no hillocks and compressive stress are responsible for the improved electromigration performance in the Al/WN interconnect. (© 2007 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

Microstructure Measurement Techniques for Studying Electromigration in ULSI Interconnects

Electromigration reliability remains a major threat to microelectronic circuits. Microstructure of a thin film conductor used in integrated circuit affects the electromigration lifetime significantly. A wealth of knowledge is acquired on thin film microstructure and electromigration in metallic interconnects from relevant studies in past few decades. However, it is noticed that the various techniques to measure microstructure-related attributes of thin film metallization are not presented in the context of electromigration, since these measurement techniques have their own importance. On the other hand, aggressive scaling of interconnect line-width down to nano regime, poses new challenges to microstructure characterization techniques. This article connects these two aspects of electromigration study, e.g., the characterization of microstructure and measurement techniques for the influential microstructural attributes especially for Cu-based interconnects. The microstructure-related parameters, attributes, and their impacts on electromigration lifetime are discussed. The sample preparation and various techniques to measure attributes of microstructure are presented in detail. This article describes the current state-of-the-art for the advancement of studying microstructure dependent electromigration reliability.     

Numerical Simulation of Grain-Boundary Grooving by Level Set Method

A numerical investigation of grain-boundary grooving by means of a level set method is carried out. An idealized polycrystalline interconnect which consists of grains separated by parallel grain boundaries aligned normal to the average orientation of the surface is considered. Initially, the surface diffusion is the only physical mechanism assumed. The surface diffusion is driven by surface-curvature gradients, while a fixed surface slope and zero atomic flux are assumed at the groove root. The corresponding mathematical system is an initial boundary value problem for a two-dimensional equation of Hamilton–Jacobi type. The results obtained are in good agreement with both Mullins analytical “small-slope” solution of the linearized problem (W. W. Mullins, 1957, j. Appl. Phys. 28, 333) (for the case of an isolated grain boundary) and with the solution for a periodic array of grain boundaries (S. A. Hackney, 1988, Scripta Metall. 22, 1731). Incorporation of an electric field changes the problem to one of electromigration. Preliminary results of electromigration drift velocity simulations in copper lines are presented and discussed.     

Local plasticity of Al thin films as revealed by x-ray microdiffraction

Grain-to-grain interactions dominate the plasticity of Al thin films and establish effective length scales smaller than the grain size. We have measured large strain distributions and their changes under plastic strain in 1.5-microm-thick Al 0.5% Cu films using a 0.8-microm-diameter white x-ray probe at the Advanced Light Source. Strain distributions arise not only from the distribution of grain sizes and orientation, but also from the differences in grain shape and from stress environment. Multiple active glide plane domains have been found within single grains. Large grains behave like multiple smaller grains even before a dislocation substructure can evolve.     

Microstructure of aluminium-copper thin films and its relation to electromigration

The microstructure of polycrystalline aluminium — 7·6 wt % copper thin films evaporated from the starting Al-Cu alloy was studied using electron microscopy and microanalysis. The structural properties of films were confronted with electromigration lifetime measurements of Al-Cu thin film conductors. In the films Al₂Cu precipitates were observed. Thermodynamically stable -Al₂Cu precipitates were found at grain boundaries, triple points and on the film surface. Within the grains -Al₂Cu precipitates were detected. When the temperature of deposition and/or the time of annealing of the films at 480 °C in nitrogen increased, the growth of grains and of the total volume of the precipitate phase was observed. Under these conditions precipitates tended to grow preferentially at the film surface and they ceased to influence the grain size distribution. The distribution approached then the log-normal distribution function. With regard to the structural investigations it is anticipated that the best electromigration resistance among the studied samples should possess conductors deposited at 250 °C as well as conductors deposited at 150 °C and 250 °C and annealed 10 min at 480 °C. The lifetimes of conductors having this type of microstructure and tested at the current density of 3×10⁶ A/cm² had values of 210÷3500 hrs, corresponding to the testing temperature of 150 °C. Higher lifetimes were observed with wider (20 m) and shortest (120 m) conductors encapsulated in packages filled with nitrogen. The activation energy for the failure process was 0·65 eV in this case.     

Effect of copper surface morphology on grain size uniformity of graphene grown by chemical vapor deposition

The graphene grain boundaries (GGBs) of polycrystalline graphene grown by chemical vapor deposition (CVD) typically constitute a major reason of deterioration of the electrical properties of graphene-based devices. To reduce the density of GGB by increasing the grain size, CVD growth conditions with a reduced CH₄ flow rate have been widely applied and, recently, electropolishing of copper (Cu) foil substrates to flatten the surface has been undertaken prior to graphene growth. In this study, we show that polycrystalline graphene layer grown on typical Cu foil features two heterogeneous regions with different average grain sizes: small-grain regions (SGRs) and large-grain regions (LGRs). Statistical analysis of the grains of the graphene layers grown under different process conditions showed that SGRs (which form on Cu striations) limit the average grain size, the ability to control the grain size through adjustment of growth conditions, and global grain-size uniformity. Analysis showed that the surface-flattening process significantly improves grain-size uniformity, and monolayer coverage, as well as the average grain size. These results suggest that a process for flattening the surfaces of Cu substrates is critical to controlling the quality and uniformity of CVD-grown graphene layers for practical device applications.     

Enhanced Raman scattering from nano-SnO2 grains

We present the Raman spectra of nano-SnO_{2} grains with sizes from 4nm to 80nm excited by 532nm and 1.06μm lines. The enhanced Raman scattering of the nanograins is observed for both exciting lines when the grain size is less than 8nm. The less the grain size is, the more intensely the Raman scattering is enhanced. According to our results, the enhancements of the Raman intensity are a few tenfolds and different for different exciting lines when the grain size is 4nm. It can be attributed to enhanced Raman scattering by electron－hole pair excitations in the nanograins that originate from sub-microscopic (10nm) size and other defect- and surface-related features. A critical size that divides respective predominance of bulk properties and the defect-, surface-, and size-related features can be determined to be about 8nm.     

晶粒棱长、尺寸与拓扑学特征之间关系的Monte Carlo仿真研究

采用Potts模型Monte Carlo方法研究了晶粒棱长、尺寸与拓扑学特征之间的统计关系.结果表明,晶粒棱长与晶粒面数之间呈线性统计关系,并且平均N面体晶粒模型和Poisson-Voronoi组织均支持该结论.不同时刻的晶粒长大仿真数据表明,在准稳态晶粒长大阶段晶粒棱长的分布具有自相似性.个体晶粒的平均棱长随晶粒面数（或晶粒尺寸）的增加而逐渐增大,这说明一些理论模型中采用的“不同面数的晶粒平均棱长均相等”的假设具有局限性.仿真数据和纯铁实验数据均表明,晶粒尺寸与晶粒面数之间的统计关系表现为一条单调递增的凸曲线. 关键词： 晶粒棱长 晶粒尺寸 拓扑学 Monte Carlo仿真