论文部分内容阅读
可靠性管理部门所面临的一项任务,就是开展对于走在现代工艺水平某些方面前面的新IC元件系列及其制作工艺的令人信服的可靠性方案研究,这是因为新材料的质量水平以及这些材料的结合使用,对于尺寸公差提出的新要求,工艺改变带来的新影响等等,将导致出现一些新的难于预料的失效机理。然而,在新产品的开发和计划的初期,管理部门须拥有一可靠性方案这一点已被担当的风险和投资费用之大所证实。HP公司台式计算机分公司的技术部,根据经验的可靠性模型和可靠性成本模型的组合,开发了一种可避开问题中未确定的技术方面而提出可普遍采纳的方案的一种途径。模型的参数随时间的变化以及它们的值反应了不同的可靠性设计和制造筛选情况,它们都可估算到合理的精度,而这反过来又可导致最佳方案的研究。自然,能够把可靠性成本降至最低限度并能在研究与发展、制造和保用三方面进行最佳资金分配的可靠性方案通常是会被有关各方面采纳的。本文通过所用的实际参数和进行的最佳化研究,说明了这些模型的逻辑和数学结构以及他们在开发HP公司的NMOC工艺和芯片电路系列中的应用。然后将研究的结果(失效率、预计的重新设计周期的数量和筛选方案)与制造的元件的现场使用所观察到的实际可靠性进行了比较。这一比较有利地证实了所提出
One of the tasks facing reliability management is to develop a compelling and reliable solution to the new family of IC components and their fabrication processes that precede some aspects of modern technology because of the quality level of new materials As well as the combination of these materials, new requirements for dimensional tolerances, new effects of process changes, etc., will lead to new and unpredictable failure mechanisms. However, early in the development and planning of new products, management has to demonstrate the reliability and cost of ownership of a reliability program. Based on the combination of an empirical reliability model and a reliability cost model, the technology department at HP Computer’s Desktop Computer Branch developed a way to propose a universally accepted solution that avoids the undefined technical aspects of the problem. Changes in model parameters over time and their values reflect different reliability designs and manufacturing screening scenarios, all of which can be estimated to reasonable accuracies, which in turn can lead to best-case studies. Naturally, reliability programs that minimize the cost of reliability and provide the best possible allocation of funds for research, development, manufacturing and preservation are often adopted by all parties concerned. This article illustrates the logical and mathematical structure of these models and their application to the development of HP’s NMOC process and chip circuit families using the actual parameters used and the optimization studies conducted. The results of the study (failure rate, number of expected redesign cycles, and screening protocol) were then compared with the actual reliability observed for the field use of manufactured components. This compares favorably with the proposed