互连铝通孔电迁移特征及其可靠性研究

研究了超大规模集成电路铝互连系统中铝通孔的电迁移失效机理及其可靠性寿命评价技术.试验采用CMOS和BiCMOS两种工艺各3组的铝通孔样品,分别在三个温度、恒定电流的加速条件下试验,以通孔开路为电迁移失效判据,最后得到了在加速条件下互连铝通孔的电迁移寿命,其结果符合标准的威布尔分布,试验准确可行.通过电迁移模型对试验数据进行了拟合,得到了激活能、电流加速因子和温度加速因子,计算出了正常工作条件下通孔电迁移的寿命,完成了对铝通孔电迁移的研究和寿命评价.     

铝硅铜合金和铝铜合金多层结构的电迁移行为

本文给出了由同一工艺线上加工的铝硅和铝硅铜金属互连线的电迁移加速寿命试验结果，后者寿命比前者要高一个数量级，我们对此作了简要的说明，并介绍了铝铜多层结构互连的的民迁移性能。     

MCM-C基板上的多层互连及厚膜电阻的可靠性研究

重点研究了MCM-C基板中多层互连和厚膜电阻的可靠性.试验采用温度应力和电应力的双应力加速寿命试验.试验发现,温度小于180℃时互连失效在MCM-C基板失效中占主要地位,膜电阻失效相对互连失效可忽略不计.在温度高于180℃时膜电阻失效将起较大作用,即膜电阻比互连温度加速系数要大.重点计算了膜电阻和互连寿命分布及加速系数.     

集成电路互连引线电迁移的研究进展

随着大规模集成电路的不断发展,电迁移引起的集成电路可靠性问题日益凸现.本文介绍了电迁移的基本理论,综述了集成电路互连引线电迁移的研究进展.研究表明,互连引线的尺寸、形状和微观组织结构对电迁移有重要影响;温度、电流密度、应力梯度、合金元素及工作电流模式等也对电迁移寿命有重要影响.同时指出了电迁移研究亟待解决的问题.     

VLSI金属互连线1/f γ噪声指数与电迁移失效

对VLSI的金属互连线实施高应力下的加速寿命试验和常规应力下的噪声频谱测量,得到了金属薄膜1/f ^γ噪声的频率指数γ在电迁移演化过程中的变化规律,发现γ指数在寿命试验的某个时间点发生突变,从1.0上升到1.6以上.这种突变可以归因于电迁移诱发空洞形成过程的起点,因而是金属薄膜结构开始发生不可逆结构变化的标志.1/ f ^γ噪声指数因子可望成为金属薄膜电迁移损伤程度或寿命的一个表征参量.     

多层金属化系统中的蓄水池效应

对于W通孔多层金属化系统来说,金属离子蓄水池效应对其电迁移寿命的影响很大.设计了12种不同的蓄水池结构,并进行电迁移实验;考察了蓄水池面积、通孔位置、数目及大小等对互连线的电迁移寿命的影响,得出蓄水池的面积是影响电迁移寿命的主要因素.     

VLSI金属互连线电迁移噪声检测敏感性的逾渗模拟

在电迁移物理机制的基础上结合逾渗理论,建立了一种金属互连线电迁移的逾渗模型。基于该模型,采用蒙特卡罗方法模拟了超大规模集成电路（VLSI）金属互连线电迁移过程中电阻和低频噪声参数的变化规律。结果表明,与传统的电阻测量方法相比,低频噪声表征方法对电迁移损伤更敏感,检测的效率更高。该研究结果为低频噪声表征VLSI金属互连线电迁移损伤的检测方法提供了理论依据。     

多层金属化系统中蓄水池效应对电迁移寿命的影响

在W通孔的多层金属化系统中,金属离子的蓄水池效应对其电迁移寿命的影响很大,文中设计制作了12种不同的蓄水池结构,并进行了电迁移实验.着重考察蓄水池面积、通孔位置、通孔数目对互连线电迁移寿命的影响,得出蓄水池的面积是影响电迁移寿命的主要因素.     

多层金属化系统中蓄水池效应对电迁移寿命的影响

在W通孔的多层金属化系统中,金属离子的蓄水池效应对其电迁移寿命的影响很大,文中设计制作了12种不同的蓄水池结构,并进行了电迁移实验.着重考察蓄水池面积、通孔位置、通孔数目对互连线电迁移寿命的影响,得出蓄水池的面积是影响电迁移寿命的主要因素.     

电迁移致无铅钎料微互连焊点的 脆性蠕变断裂行为

研究了电迁移条件下不同电流密度(0.8~1.27×10⁴A/cm²)和通电时间(0~96 h)对无铅钎料模拟微互连焊点的蠕变断裂行为的影响.研究结果发现,电迁移作用加速了焊点的蠕变断裂过程,随着电迁移通电时间的延长及电流密度的增加,其蠕变应变速率显著增大,而蠕变寿命逐渐缩短;电迁移还导致焊点蠕变断裂机制发生明显变化,在高电流密度或长时间通电的电迁移后,微互连焊点在服役条件下会发生由延性断裂向脆性断裂的转变.     

Lifetime and drift velocity analysis for electromigration in sputtered Al films,multilayers, and alloys

Besides high lifetimes and a low standard deviation of failure times, the electromigration drift velocity is an important parameter for metallization reliability. Here, lifetimes should be high and drift velocity low. For pure Al films, only properly chosen and optimized sputter conditions yield good film quality and electromigration data comparable to evaporated stripes. Annealing and oxide covering improve lifetime and reduce drift velocity of pure Al structures. For redundant AlTiAl multilayers, annealing yielded an enormous lifetime enhancement of 3 orders of magnitude. However, these stripes are not suited for practical applications because of whisker formation and of an increased drift velocity! These disadvantages do not appear for a lifetime-improving TiN intermediate layer instead of Ti. A further method of lifetime enhancement is Cu-addition which reduces the drift of stripes. Finally, for (Al + Cu)TiN(Al + Cu)-structures, lifetime is improved even further in combination with low drift velocity.     

Electromigration and mechanical stress in aluminium conductor tracks passivated by anodisation

The electromigration lifetime of integrated circuit metallisation is commonly found to increase if the metallisation is covered with a passivation layer such as glass. We have investigated the electromigration behaviour of aluminium conductor stripes under different thicknesses of oxide passivation formed by anodisation. The ionic drift velocity of anodised stripes was measured using the Blech-Kinsbron edge displacement technique (Thin Solid Films 25 327 (1975)) and it was found that the drift velocity decreased with increasing anodisation thickness. Stripes tested with a reversed current drifted backwards with an initial velocity which exceeded the original forward velocity. These results are related to a change in the self-diffusivity of aluminium as a result of high compressive stresses imposed by the anodised layer, and the distribution of stress in a drifting stripe is discussed. A general conclusion is that passivation layers can support compressive stresses sufficient to retard mechanical failure processes in metallisation in two distinct ways, either by raising the threshold stress for the nucleation of damage such as hillocks or whiskers, or by reducing the diffusivity in the metallisation.     

Electromigration mechanisms in aluminum lines

Special test structures were used for investigating electromigration mechanisms. Large-grained Al lines of different lengths and widths were interconnected by varying TiN auxiliary layers. Test current densities lay between 4 × 10⁵ A/cm² and 6 × 10⁵ A/cm² at 200°C. Considering electromigration threshold, grain boundary electromigration was eliminated and interface electromigration appeared, affecting the conductive Al/TiN interface. Interface electromigration clearly contributes to the mass flow of Al lines, and thus can be detrimental for the reliability of metallization. The interface diffusion activation energy is comparable to the grain boundary activation energy. Contrary to a conductive interface, the technical Al surface does not contribute to mass flow. The elimination of interface effects finally brings out homogeneous bulk electromigration. The drift velocity was directly measured after a stress period of 8300 hours at 200°C. For a current density of 4 × 10⁵ A/cm² bulk drift velocity was 7 × 10^?12 cm/s, while grain boundary electromigration surpassed this value by a factor of 300. Electromigration threshold was ascertained for grain boundary as well as for interface and bulk diffusion.     

A direct measurement of electromigration induced drift velocity in Cu dual damascene interconnects

A dual damascene structure with an additional 25 nm Ta diffusion barrier embedded into the upper Cu layer was fabricated to measure the drift velocity of electromigration. The embedded diffusion barrier layer successfully confined void growth into a long and regular shape between the SiN layer and embedded Ta layer. Edge depletion was observed to initiate from the cathode end and elongate into a long and regular shape due to the confinement of the intermediate Ta diffusion barrier layer. With this test structure, electromigration induced drift displacement can be accurately measured with a linear dependence on time. Measurement was conducted at a series of temperatures to obtain the Cu/capping interface diffusion controlled activation energy.     

1-D model of current distribution and resistance changes forelectromigration in layered lines

We present a one-dimensional (1-D) analytical model incorporating contact resistance to calculate the current distribution between an Al line and refractory metal underlayer in a drift velocity measurement structure for electromigration. The current profile so calculated is more accurate than the usual assumption that the current all flows in the Al portion, and it should give a more accurate and reliable relation between drift velocity and current density. It should also be useful in correlating local current density with voiding. Various experimentally observed voiding processes are included in the model; they reproduce and predict different types of changes in total line resistance, most notably the step structure observed in electromigration     

Electromigration in Double-Layer Metallization

The electromigration of the top stripe in aluminum double-layer metallization systems was investigated. The current density dependence and the activation energy characterization are important in double-layer metallization. The step-coverage and coating effects of SiN is better than that of Si02. New phenomena associated with electromigration have been observed as follows: 1. The mean lifetime is affected by the material of the dielectric layer. This material effect might be related to the hardness of layer. 2. The mean lifetime due to electromigration depends on the magnitude and polarity of the electric field applied between adjacent stripes. Ordinary stress tests for electromigratfon are done where current is conducted only in the top stripe and not in the bottom stripe. Our results show that this situation is realistic under conditions existing in microelectronic circuits. The proposed method for stress testing should be used to simulate actual condition in microelectronic circuits. We emphasize that the stress test method used to disclose this electric field effect is important for accelerated stress testing, especially for metallization in VLSI circuits and multi-layer systems. The failure mechanism due to the electric field effect can be explained in terms of the applied electric field deflecting current-carrying electrons in the metal stripe, and is independent of leakage current between stripes.     

Relationship between texture and electromigration lifetime in sputtered AI-1% Si thin films

The relationship among the grain structure, texture, and electromigration lifetime of four Al-1% silicon metallizations produced under similar sputtering conditions was explored. The grain sizes and distributions were similar and the grain structure was near-bamboo for all metallizations. All metallizations exhibited a near-(111) fiber texture, as determined by the pole figure technique. Differences in electromigration behavior were noted. Three of the metallizations exhibited a bimodal failure distribution while the fourth was monomodal and had the longest electromigration lifetime. The electromigration lifetime was directly related to the strength of the (111) fiber texture in the metallization as anticipated. However, whereas the grain size distribution has an effect on the electromigration lifetime when metallization lines are several grains wide, the electromigration lifetime of these near-bamboo metallizations appeared independent of the grain structure. It was also observed that a number of failures occurred in the 8 μm interconnect supplying the 5 μm wide test lines. This apparently reflects an increased susceptibility of the wider interconnect lines to electromigration damage.     

The isocurrent test: A promising tool for wafer-level evaluation of the interconnect reliability

A highly accelerated wafer-level electromigration test, the isocurrent test, is presented. A constant high current is used to give both the current and temperature stress to a 4-point resistor with a bonding pad layout which minimizes temperature gradients. The test is used to evaluate unpassivated Al---Cu and passivated Al---Si---Cu lines of different line width. Log normal failure distributions are obtained and the line width dependence of the MTTF and DTTF is similar to that observed in classical electromigration tests. A storage test at 250 °C clearly deteriorates the lifetime of 0.5 and 0.7 μm passivated lines. This is probably due to stress induced void formation.