Correlation between predicted cause of SRAM failures and in-line defect data

A correlation has been made between the bitmap data from an SRAM and the in-line defect data as measured on a KLA2122 and Tencor7700. The SRAM was a dedicated design for yield enhancement in a 0.35 μm technology. Extra design features were added to encourage the change of having defect on particular places and discourage it on safe designed places. From the failure signature of a memory cell (0 or 1) and its failure extent (single cell, double cell, bitline, wordline (WL), …) one can predict the process-related cause of the failure. A special test program has been written which translates the electrical data from the failing cells into its process defect.The failing bits from the SRAM have been transferred into a KLA results file and added as an extra inspection to the defect database. With a defect source analysis it was possible to find out if the electrical failing bits were seen as a defect in the line and at which steps. With this analysis it is possible to find out if the predicted cause of the process defects from the test program is confirmed by the performed in-line inspections. With an intensive inspection plan about half of the electrical defects were seen in the line. For a large amount of these defects their predicted cause are indeed matching with the inspected layer. Moreover, quite some unknown failures can be explained by the in-line inspections. This correlation work makes it possible to prioritize in tackling the most killing defect sources.     

Maximum likelihood estimation for failure analysis [IC yield]

This paper presents an iterative maximum likelihood (ML) estimation method for statistical analysis of yield loss. By means of inductive fault analysis (IFA) and circuit simulation, the mapping between defect types to the corresponding fault signature is constructed. Using the count of each fault signature occurrence, which is provided by a tester on defective ICs, the most likely causes of low yield are identified automatically without the need for physically deprocessing the defective IC's. We present an experiment on an SRAM cell array to illustrate the effectiveness of the iterative ML algorithm     

低电压SRAM测试电路设计与实现

本文基于SMIC 40nm LL CMOS工艺对一款256Kb的低电压8T SRAM芯片进行测试电路设计与实现,重点研究低电压SRAM的故障模型和测试算法,并完成仿真验证与分析。电路主要包括DFT电路和内建自测试电路两部分,前者针对稳定性故障有着良好的覆盖率,后者在传统March C+算法基础上,提出了一种新的测试算法,March-Like算法,该算法能够实现更高的故障覆盖率。仿真结果表明,本文设计的DFT电路能够减小稳定性故障的最小可检测电阻,提高了稳定性故障的测试灵敏度;March-Like算法可以检测到低电压SRAM阵列中的写破坏耦合故障、读破坏耦合故障和写干扰故障。     

Numerical Estimation of Yield in Sub-100-nm SRAM Design Using Monte Carlo Simulation

This paper describes a method to numerically calculate the design margin and to estimate the yield associated with the read access failure for sub-100-nm SRAM. Process variations at sub-100 nm not only affect SRAM cells but also periphery circuits, such as the sense amplifier (SA) and the tracking scheme. Simulation that incorporates both SRAM cells and surrounding circuits is either accurate but computationally expensive (comprehensive Monte Carlo simulation), or overly simple (fixed corner design) and unable to capture crucial statistical variation concern, dominant in sub-100-nm designs. By mathematically combining the separate Monte Carlo simulation results of SRAM cells and each peripheral block, we show that the distribution of the SA input voltage can be estimated accurately in a case where fixed corner simulation underestimates by 19%. We also present the yield equation by combining the SA input voltage and the SA offset distribution, which can be used to choose the design point. In addition, yield sensitivities are derived from the yield data to make sure that the yield has good dependence to design variables.      

基于制造成品率模型的集成电路早期可靠性估计

缺陷是影响集成电路成品率与可靠性的主要因素.本文在区分缺陷与故障两个概念的基础上,将缺陷区分为成品率缺陷(硬故障)、可靠性缺陷(软故障)和良性缺陷.利用关键区域的面积,给出了一个缺陷成为"硬故障"或"软故障"的概率,给出了精度较高的IC成品率预测模型.利用成品率缺陷与可靠性缺陷之间的关系,给出了工艺线生产的产品的失效率与该工艺线制造成品率之间的定量关系.在工艺线稳定的条件下,通过该工艺线的制造成品率可以利用该关系式可以有效的估计出产品的失效率,可以有效地缩短了新产品的研发周期.     

Back-end soft and hard defect monitoring using a single test chip

The objective of this paper is to present a mixed test structure designed to characterize yield losses due to hard defect and back-end process variation (PV) at die and wafer level. A brief overview of the structure, designed using a ST-Microelectronics’ 130 nm technology, is given. This structure is based on a SRAM memory array for detecting hard defects. Moreover each memory cell can be configured in the Ring Oscillator (RO) mode for back-end PV characterization. The structure is tested in both modes (SRAM, RO) using a single test flow. The test data analysis method is presented and applied to experimental results to confirm the ability of the structure to monitor PV and defect density.     

Dynamic Power Supply Current Testing of CMOS SRAMs

We describe the design and implementation of a dynamic power supply current sensor which is used to detect SRAM faults such as disturb faults as well as logic cell faults. A formal study is presented to assess the parameters that influence the sensor design. The sensor detects faults by detecting abnormal levels of the power supply current. The sensor is embedded in the SRAM and offers on-chip detectability of faults. The sensor detects abnormal dynamic current levels that result from circuit defects. If two or more memory cells erroneously switch as a result of a write or read operation, the level of the dynamic power supply current is elevated. The sensor can detect this elevated value of the dynamic current. The dynamic power supply current sensor can supplement the observability associated with any test algorithm by using the sensor as a substitute for the read operations. This significantly reduces the test length and the additional observability enhances defect coverages.     

基于OMP的SRAM成品率分析的分组建模方法

静态随机存取存储器(SRAM)电路的失效是极小概率事件,并且不同电路条件下的失效区域边界可能相距很远。因此,为了更高效地获得更精准的SRAM成品率预测结果,提出一种基于正交匹配追踪(OMP)算法的SRAM性能分组建模方法,并应用于典型SRAM电路成品率的预测。此方法主要根据不同SRAM电路条件下失效区域边界距离的差异将仿真数据划分为多组,之后利用OMP算法对不同组的数据分别建立多项式模型,该模型可用于对SRAM电路的成品率进行快速分析预测。与标准蒙特卡洛统计算法及基于OMP的单一建模方法相比,基于OMP的分组建模方法不仅可以缩短建模时间,提高建模准确度,还能够获得更加精确的SRAM成品率预测结果。     

Localization and physical analysis of a complex SRAM failure in 90nm technology

In this paper the analysis process of a complex SRAM failure in 90nm technology is introduced in detail. Using a correlation method, it could be traced back to a failure with an increased supply current. With the help of MCT emission microscopy and thermal laser stimulation (TLS) the defects were localized at both edges of every failing SRAM block. Further inspection by passive voltage contrast (PVC) and atomic force prober (AFP) current imaging led to a localization down to contact level. In the TEM analysis high angle annular dark field scanning TEM (HAADF STEM) was used to obtain better material contrast. CoSi residues were found at the wall of spacers of the failing FETs. Further surface parallel TEM inspection confirmed the hypothesis of a new type of bridging defect, i.e. CoSi stringers along word lines in SRAM cells, which has not been observed before to our knowledge. The process adjustment in the fab to avoid this failure led to a significant yield improvement. The abstract should be 75-200 words long, summarizing the work and placing it in an appropriate context.     

SRAM单元单粒子翻转效应的电路模拟

在三维器件数值模拟的基础上,以经典的双指数模型为原型通过数值拟合得到了单粒子效应瞬态电流脉冲的表达式,在理论分析的基础上,引入了描述晶体管偏压和瞬态电流关系的方程,并将其带入电路模拟软件HSPICE中进行SRAM存储单元单粒子翻转效应的电路模拟,最后分别使用电路模拟和混合模拟两种方法得到了存储单元的LET阈值,通过在精度和时间开销上的对比,验证了这种模拟方法的实用性.     

A SPICE-Like 2T-FLOTOX Core-Cell Model for Defect Injection and Faulty Behavior Prediction in eFlash

The embedded Flash technology can be subject to complex defects creating functional faults. In this paper, we describe the different steps in the electrical modeling of 2T-FLOTOX core-cells for a good understanding of failure mechanisms. At first, we present a first order electrical model of 2T-FLOTOX core-cells which is characterized and compared with silicon data measurements based on the ATMEL 0.15 µm eFlash technology. Next, we propose a study of resistive defect injections in eFlash memories to show the interest of the proposed simulation model. At the end of the paper, a table summarizes the functional fault models for different resistive defect configurations and experimental set-ups. According to these first results and with additional analysis on actual defects presented in [3] we are then able to enhance existing test solutions for eFlash testing.     

驱动能力对SOI SRAM单元单粒子效应的影响仿真

采用silvaco软件对抗辐射不同沟道宽度的PD SOI NMOS器件单元进行了三维SEU仿真,将瞬态电流代入电路模拟软件HSPICE中进行SRAM存储单元单粒子翻转效应的电路模拟。通过这种电路模拟的方法,可以得到SRAM存储单元的LET阈值。通过对比LET阈值的实际测量值,验证了这种方法的实用性,并对不同驱动能力的SRAM单元进行了翻转效应的对比。在NMOS和PMOS驱动比相同的情况下,沟道宽度越大,SRAM的翻转LET阈值反而越高。     

Yield-oriented computer-aided defect diagnosis

Any good yield-oriented defect strategy must have two main components-(a) the ability to perform rapid defect diagnosis for yield learning, and (b) the ability to efficiently extract defect parameters from the manufacturing line. In this work, an inductive fault analysis (IFA)-based defect methodology is investigated to see if it meets the above requirements. Using an SRAM test vehicle as an example, the research analyzes whether computer-generated mappings between defect types and tester fail data can provide sufficient resolution for both, defect diagnosis and defect parameter characterization     

SRAM design on 65-nm CMOS technology with dynamic sleep transistor for leakage reduction

A 70-Mb SRAM is designed and fabricated on a 65-nm CMOS technology. It features a 0.57-/spl mu/m/sup 2/ 6T SRAM cell with large noise margin down to 0.7 V for low-voltage operation. The fully synchronized subarray contains an integrated leakage reduction scheme with dynamically controlled sleep transistor. SRAM virtual ground in standby is controlled by programmable bias transistors to achieve good voltage control with fine granularity under process skew. It also has a built-in programmable defect "screen" circuit for high volume manufacturing. The measurements showed that the SRAM leakage can be reduced by 3-5/spl times/ while maintaining the integrity of stored data.     

Weak Cell Detection in Deep-Submicron SRAMs: A Programmable Detection Technique

Embedded SRAM bit count is constantly growing limiting yield in systems-on-chip (SoCs). As technology scales into deep sub-100-nm feature sizes, the increased defect density and process spreads make stability of embedded SRAMs a major concern. This paper introduces a digitally programmable detection technique, which enables detection of SRAM cells with compromised stability [with data retention faults (DRFs) being a subset]. The technique utilizes a set of cells to modify the bitline voltage, which is applied to a cell under test (CUT). The bitline voltage is digitally programmable and can be varied in wide range, modifying the pass/fail threshold of the technique. Programmability of the detection threshold allows tracking process variations and maintaining the optimal tradeoff between test quality and test yield. The measurement results of a test chip presented in the paper demonstrate the effectiveness of the proposed technique.     

Ultra large scale static rams

Using a patented defect avoidance technique, high yield production of high density SRAM devices (ULSI SRAMs) can be achieved one process generation ahead of the rest of the industry. Production wafer yields as high as 100% and long-term average yields above 80% are reported on Inova's monolithic, 1.2μ, 320 square mm, one megabit SRAM demonstrating a practical method of achieving wafer scale integration. A yield model is presented and used to determine the optimized architecture and redundancy scheme for Inova's four megabit SRAM and to predict yield as a function of defect density. Achievement of a working 8M-bit experimental device using a 1.2μ process is also reported.     

28 nm低功耗工艺SRAM失效分析

在28 nm低功耗工艺平台开发过程中,对1.26 V测试条件下出现的SRAM双比特失效问题进行了电性能失效模式分析及物性平面和物性断面分析.指出失效比特右侧位线接触孔底部空洞为SRAM制程上的缺陷所导致.并通过元素成分分析确定接触孔底部钨(W)的缺失,接触孔底部外围粘结阻挡层的氮化钛(TiN)填充完整.结合SRAM写操作的原理从电阻分压的机理上解释了较高压下双比特失效,1.05 V常压下单比特不稳定失效,0.84 V低电压下失效比特却通过测试的原因.1.26 V电压下容易发生的双比特失效是一种很特殊的SRAM失效,其分析过程及结论在集成电路制造行业尤其是对先进工艺制程研发过程具有较好的参考价值.     

Design Considerations for PD/SOI SRAM: Impact of Gate Leakage and Threshold Voltage Variation

We present a critical study of the impact of gate tunneling currents on the yield of 65-nm partially depleted/silicon-on-insulator (PD/SOI) SRAM designs. A new gate leakage monitor structure is developed to obtain device-specific gate leakage characteristics of the SRAM cells. This allows us to explore the design space accurately with reliable process information at an early stage. By relying on supply voltage-dependent analysis, it is shown that the gate-leakage impact on the cell yield can be nonmonotonic and substantial even for nondefective devices. It is also shown that design optimizations such as increased operating voltages or shorter hierarchical bitline architecture can help alleviate the gate-leakage impact on yield. Mixture importance sampling is used to estimate yield in terms of cell writability and stability. Threshold voltage variations to model random fluctuation effects are extrapolated from hardware results.     

The effect of the number of defect mechanisms on fault clusteringand its detection using yield model parameters

The superposition principle is used to analyze faults of different mechanisms. The sum of the individual cluster coefficients of each mechanism is approximately equal to the cluster coefficient for all mechanisms combined. This technique is used on defect density test structures as well as bit failures from SRAM chips. A regression analysis of empirical data is used to demonstrate this concept for the defect density test chips. The actual and, the model fault densities are compared and show excellent agreement. As a comparative analysis, a quadrant technique was used to compile a frequency distribution of electrical faults and a nonlinear least-squares technique is applied to the distribution to estimate the parameters in the gamma and Poisson distributions. These results are compared to the cluster parameters from the summation technique and the technique using moment estimates. All three estimates are in very good agreement. The application of this model to actual chip yields is shown not only to be more accurate but also to contain information about the relative number of fault generating mechanisms for the mask level of interest in the process     

Analyzing data for CMOS leakage faults

Stuck-on and leakage faults are among the major failure modes for CMOS ICS. The only definite way to test for these faults is to monitor the supply current under different test vectors. To evaluate the effectiveness of this test technique, distribution of normal and abnormal conductance values has to be obtained. This paper presents two procedures which can be used to extract the required information from measurable data. Experimental data for a number of CMOS chips has been analyzed and the results are presented.