Detection of Field Failure Chips by Ensemble Learned from Different Chip Areas

Author(s):  
Min-Soo Kim ◽  
Jong-Seok Lee ◽  
Jung-Hoon Chun
Keyword(s):  
2018 ◽  
Vol 18 (3) ◽  
pp. 250-259 ◽  
Author(s):  
Yangwoo Seo ◽  
Kyeshin Lee ◽  
Younho Lee ◽  
Jeyong Kim

2018 ◽  
Author(s):  
Wentao Qin ◽  
Scott Donaldson ◽  
Dan Rogers ◽  
Lahcen Boukhanfra ◽  
Julien Thiefain ◽  
...  

Abstract Many semiconductor products are manufactured with mature technologies involving the uses of aluminum (Al) lines and tungsten (W) vias. High resistances of the vias were sometimes observed only after electrical or thermal stress. A layer of Ti oxide was found on such a via. In the wafer processing, the post W chemical mechanical planarization (WCMP) cleaning left residual W oxide on the W plugs. Ti from the overlaying metal line spontaneously reduced the W oxide, through which Ti oxide formed. Compared with W oxide, the Ti oxide has a larger formation enthalpy, and the valence electrons of Ti are more tightly bound to the O ion cores. As a result, the Ti oxide is more resistive than the W oxide. Consequently, the die functioned well in the first test in the fab, but the via resistance increased significantly after a thermal stress, which led to device failure in the second test. The NH4OH concentration was therefore increased to more effectively remove residual W oxide, which solved the problem. The thermal stress had prevented the latent issue from becoming a more costly field failure.


2014 ◽  
Vol 8 (3) ◽  
pp. 1713-1727 ◽  
Author(s):  
Zhisheng Ye ◽  
Hon Keung Tony Ng
Keyword(s):  

2017 ◽  
Vol 2017 (1) ◽  
pp. 000517-000525
Author(s):  
Josh Liew ◽  
Otto Fanini

Abstract The oil and gas industry subsurface formation evaluation instruments experience significant challenging vibrations and shock levels. Equipment development requirements for these instruments include reliability and durability under these severe operating conditions. The engineering design for this equipment utilizes tools that enable the estimation of service lifetime, maintenance cycles, and related costs. These engineering tools model failure modes and their acceleration factors and how failures interact under certain circumstances. Laboratory test data and operations failure history are used to validate the model results. Incorporating equipment operational failure history into the reliability design after failure analysis, enables determination of failure modes and the length of stress level exposure. Before the equipment is commissioned to field operational service, it is subjected to a batch of environmental qualification tests under objective pass-fail criteria. The environmental qualification test conditions and adopted stress levels are acquired through measurements made with sensors (temperature, acceleration, and shock) in the equipment assembly during field operational conditions in targeted environments. After the equipment passes the qualification tests and final inspection, it is commissioned to field commercial service. This paper studies the development of specific equipment failures during operational field deployment after they were subjected to standard environmental qualifications tests. Various investigative actions focused on determining the cause and circumstances that led to the unexpected field failure. Results helped to introduce corrective updates to the equipment design and manufacturing, durability and reliability design models and procedures, environmental stress levels, and corresponding qualification test conditions. The equipment failures were examined, and comprehensive 3D custom vibration and stress modeling were conducted for the entire equipment assembly and each assembly module. The modeling results pinpointed and confirmed the high stress levels in the failure areas. These high stress levels exceeded the assembly construction strength thresholds, causing failures. The equipment assembly was modified and reinforced to properly support the detected stresses and provide the required lifetime reliability and durability for the operational service. A full 3D model of the equipment assembly was used for the vibration and bending load analysis including all mechanical assembly parts, electronics modules, couplings, and attachments. The 3D model was meshed with Tet and Hex elements in ANSYS application software, failure-prone and critical regions were meshed with finer divisions. In this analysis the electronics modules assembly were considered with all parts, attachments, structural frames, linkages, carriers, and printed circuit board (PCB) modules properly attached and connected to the main chassis structural carrier. Geometries, mass, module and assembly attachments, and material properties were assigned to components in this model. External loads and boundary conditions environmentally imposed to the assembly were applied in the model. Environmental conditions, shock, and vibration (x, y, and z) recorded from similar equipment deployed in subsurface operations in equivalent wells and geological formations were used in the modeling parameters. Displacement modeling data and analysis was performed for all mechanical structural components, PCB electronics module assemblies and assembly components, and module electronics component attachments. A model harmonic analysis under static conditions was performed to detect the oscillatory modes and vibratory resonances and the extent of oscillatory displacements. A structural and main carrier chassis modal analysis was conducted for the entire model, identifying the dominant oscillatory modes and natural structural oscillatory frequencies. The displacement can be used for detection of maximum allowable plastic deformation threshold and cyclic fatigue analysis of attachments, structural support members, and linkages for equipment service lifetime durability and reliability assessment. Past field instrumented operational conditions with documented failures and lab characterization of failure modes along with failure behavior and failure triggering thresholds have provided limits for the mechanical and electronics assembly technology with maximum acceleration level of random vibration and maximum equivalent stress level tolerated by the equipment's structural assembly, standard design techniques, and materials. With these structural stress and displacement limits the 3D modeling results were inspected for the entire assembly, identifying the points in the mesh model where these limits were exceeded. The inspection determined that these recommended limits had been exceeded according to the model results, placing a reduced importance to the adjustment of tolerable maximum stresses and displacements. The mesh points with excessive stress and displacement-induced fatigue coincided with the areas where field failure had been detected in examined field failed units. Because of this modeled assembly performance result and details from the externally imposed operational shocks and vibration, the equipment mechanical and electronics assembly structural design were re-engineered to produce an updated model simulation results that did not exceed the demonstrated cumulative failure threshold stresses in lab tests and field operations. The modified equipment assembly was built and environmentally re-tested in the lab environment with more instrumentation points and scrutiny around the failure critical areas. The test results were successful. After deployment of the new and updated equipment assembly version, its field deployment has not observed similar field failures compared with the previous design version. These modeling and engineering tools, qualification test procedures, and methods can be used to validate a new design or understand the most effective and economical approach to iterate the design before it is launched to field operations or after a field failure.


Author(s):  
Roozbeh Bakhshi ◽  
Peter Sandborn

With renewable energy and wind energy in particular becoming mainstream means of energy production, the reliability aspect of wind turbines and their sub-assemblies has become a topic of interest for owners and manufacturers of wind turbines. Operation and Maintenance (O&M) costs account for more than 25% of total costs of onshore wind projects and these costs are even higher for offshore installations. Effective management of O&M costs depends on accurate failure prediction for turbine sub-assemblies. There are numerous models that predict failure times and O&M costs of wind farms. All these models have inputs in the form of reliability parameters. These parameters are usually generated by researchers using field failure data. There are several databases that report the failure data of operating wind turbines and researches use these failure data to generate the reliability parameters through various methods of statistical analysis. However, in order to perform the statistical analysis or use the results of the analysis, one must understand the underlying assumptions of the database along with information about the wind turbine population in the database such as their power rating, age, etc. In this work, we analyze the relevant assumptions and discuss what information is required from a database in order to improve the statistical analysis on wind turbines’ failure data.


2019 ◽  
Vol 26 (1) ◽  
pp. 87-103
Author(s):  
Rajkumar Bhimgonda Patil

Purpose Reliability, maintainability and availability of modern complex engineered systems are significantly affected by four basic systems or elements: hardware, software, organizational and human. Computerized Numerical Control Turning Center (CNCTC) is one of the complex machine tools used in manufacturing industries. Several research studies have shown that the reliability and maintainability is greatly influenced by human and organizational factors (HOFs). The purpose of this paper is to identify critical HOFs and their effects on the reliability and maintainability of the CNCTC. Design/methodology/approach In this paper, 12 human performance influencing factors (PIFs) and 10 organizational factors (OFs) which affect the reliability and maintainability of the CNCTC are identified and prioritized according to their criticality. The opinions of experts in the fields are used for prioritizing, whereas the field failure and repair data are used for reliability and maintainability modeling. Findings Experience, training, and behavior are the three most critical human PIFs, and safety culture, problem solving resources, corrective action program and training program are the four most critical OFs which significantly affect the reliability and maintainability of the CNCTC. The reliability and maintainability analysis reveals that the Weibull is the best-fit distribution for time-between-failure data, whereas log-normal is the best-fit distribution for Time-To-Repair data. The failure rate of the CNCTC is nearly constant. Nearly 66 percent of the total failures and repairs are typically due to the hardware system. The percentage of failures and repairs influenced by HOFs is nearly only 16 percent; however, the failure and repair impact of HOFs is significant. The HOFs can increase the mean-time-to-repair and mean-time-between-failure of the CNCTC by nearly 65 and 33 percent, respectively. Originality/value The paper uses the field failure data and expert opinions for the analysis. The critical sub-systems of the CNCTC are identified using the judgment of the experts, and the trend of the results is verified with published results.


2018 ◽  
Vol 139 ◽  
pp. 512-520
Author(s):  
Gyungsik Yun ◽  
Hee-Won Jung ◽  
Sungbum Park

2010 ◽  
Vol 458 ◽  
pp. 173-178
Author(s):  
Zhen Zhou ◽  
L.N. Zhang ◽  
Y. Qin ◽  
D.Z. Ma ◽  
B. Niu

Characteristics of field failure data are analyzed in this paper. The failure data and sales record of LZL-type mass flowmeter are used to infer life distribution of this conduct. The lines can be fitted in coordinates of six distribution using least square and the residual sum of squares are compared, the minimum correspond is the best distribution type. The results show that the life distribution style of this conduct is the two parameter exponential distribution, which is the base to analyze and predict failure development, research failure mechanism and draw up maintenance policy.


Sign in / Sign up

Export Citation Format

Share Document