{"title":"设计参数对芯片级封装可靠性影响的概率分析","authors":"J. Wilde, E. Zukowski","doi":"10.1109/ESIME.2006.1644001","DOIUrl":null,"url":null,"abstract":"This paper describes two important issues associated with CSP package reliability. Failure due to thermomechanical stress is one of the dominant failure causes of CSPs. FE-simulations are frequently used to analyse local stresses and strains in soldered joints under thermal loads. Consequently stress and strain data are a basis for the lifetime estimation of the assemblies. In order to make a design optimisation it is necessary to identify those design and manufacturing parameters that will affect reliability. Among the factors which have influence on life-time the principal ones are materials, geometry, and assembly process parameters. In assembly technology, geometrical features can be reproduced only within certain tolerances. Also the material properties of all parts are subjected to a certain scatter. The same holds for boundary conditions, such as thermal loads. Only some of the parameters that can affect reliability can be modelled directly in a finite elements analysis. The principal variables were analysed by physical investigation of a large number of assemblies. This analysis yielded variation ranges to be used in the FEA. From FE simulation results sensitivities of lifetimes on these variables were computed and ranked. A second aspect is the statistical nature of reliability data. Thermal cycling experiments typically lead to a wide distribution of lifetimes. Therefore a designer should also be capable of the computation of generic distribution functions. This is a technique which has not yet been established. Therefore we worked out a method which combines the scatter ranges of the input variables with probabilistic methods like Monte Carlo simulation and also with FEA in order to allow for a statistical lifetime prediction. In this way, it was possible to compute also the uncertainty ranges of output parameters. In our case the influence of input parameters on thermal-cycling lifetime was estimated. Parametric studies were conducted to study qualitative and quantitative effects of several parameters on the lifetime distribution. The Weibull function is well suited to describe the simulated life distributions. Our results revealed that the slope of the Weibull curves from simulations is similar to experimental values. Therefore it was concluded that the dominant sensitivities have been modelled. Summarising, in this work a probabilistic simulation method was developed in order to calculate realistic failure distributions and to compare these directly with experimental failure data. In this way we contribute to the concept of \"simulation rather than testing\"","PeriodicalId":60796,"journal":{"name":"微纳电子与智能制造","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2006-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":"{\"title\":\"Probabilistic Analysis of the Influences of Design Parameter on the Reliability of Chip Scale Packages\",\"authors\":\"J. Wilde, E. Zukowski\",\"doi\":\"10.1109/ESIME.2006.1644001\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"This paper describes two important issues associated with CSP package reliability. Failure due to thermomechanical stress is one of the dominant failure causes of CSPs. FE-simulations are frequently used to analyse local stresses and strains in soldered joints under thermal loads. Consequently stress and strain data are a basis for the lifetime estimation of the assemblies. In order to make a design optimisation it is necessary to identify those design and manufacturing parameters that will affect reliability. Among the factors which have influence on life-time the principal ones are materials, geometry, and assembly process parameters. In assembly technology, geometrical features can be reproduced only within certain tolerances. Also the material properties of all parts are subjected to a certain scatter. The same holds for boundary conditions, such as thermal loads. Only some of the parameters that can affect reliability can be modelled directly in a finite elements analysis. The principal variables were analysed by physical investigation of a large number of assemblies. This analysis yielded variation ranges to be used in the FEA. From FE simulation results sensitivities of lifetimes on these variables were computed and ranked. A second aspect is the statistical nature of reliability data. Thermal cycling experiments typically lead to a wide distribution of lifetimes. Therefore a designer should also be capable of the computation of generic distribution functions. This is a technique which has not yet been established. Therefore we worked out a method which combines the scatter ranges of the input variables with probabilistic methods like Monte Carlo simulation and also with FEA in order to allow for a statistical lifetime prediction. In this way, it was possible to compute also the uncertainty ranges of output parameters. In our case the influence of input parameters on thermal-cycling lifetime was estimated. Parametric studies were conducted to study qualitative and quantitative effects of several parameters on the lifetime distribution. The Weibull function is well suited to describe the simulated life distributions. Our results revealed that the slope of the Weibull curves from simulations is similar to experimental values. Therefore it was concluded that the dominant sensitivities have been modelled. Summarising, in this work a probabilistic simulation method was developed in order to calculate realistic failure distributions and to compare these directly with experimental failure data. In this way we contribute to the concept of \\\"simulation rather than testing\\\"\",\"PeriodicalId\":60796,\"journal\":{\"name\":\"微纳电子与智能制造\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2006-04-24\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"3\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"微纳电子与智能制造\",\"FirstCategoryId\":\"1087\",\"ListUrlMain\":\"https://doi.org/10.1109/ESIME.2006.1644001\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"微纳电子与智能制造","FirstCategoryId":"1087","ListUrlMain":"https://doi.org/10.1109/ESIME.2006.1644001","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Probabilistic Analysis of the Influences of Design Parameter on the Reliability of Chip Scale Packages
This paper describes two important issues associated with CSP package reliability. Failure due to thermomechanical stress is one of the dominant failure causes of CSPs. FE-simulations are frequently used to analyse local stresses and strains in soldered joints under thermal loads. Consequently stress and strain data are a basis for the lifetime estimation of the assemblies. In order to make a design optimisation it is necessary to identify those design and manufacturing parameters that will affect reliability. Among the factors which have influence on life-time the principal ones are materials, geometry, and assembly process parameters. In assembly technology, geometrical features can be reproduced only within certain tolerances. Also the material properties of all parts are subjected to a certain scatter. The same holds for boundary conditions, such as thermal loads. Only some of the parameters that can affect reliability can be modelled directly in a finite elements analysis. The principal variables were analysed by physical investigation of a large number of assemblies. This analysis yielded variation ranges to be used in the FEA. From FE simulation results sensitivities of lifetimes on these variables were computed and ranked. A second aspect is the statistical nature of reliability data. Thermal cycling experiments typically lead to a wide distribution of lifetimes. Therefore a designer should also be capable of the computation of generic distribution functions. This is a technique which has not yet been established. Therefore we worked out a method which combines the scatter ranges of the input variables with probabilistic methods like Monte Carlo simulation and also with FEA in order to allow for a statistical lifetime prediction. In this way, it was possible to compute also the uncertainty ranges of output parameters. In our case the influence of input parameters on thermal-cycling lifetime was estimated. Parametric studies were conducted to study qualitative and quantitative effects of several parameters on the lifetime distribution. The Weibull function is well suited to describe the simulated life distributions. Our results revealed that the slope of the Weibull curves from simulations is similar to experimental values. Therefore it was concluded that the dominant sensitivities have been modelled. Summarising, in this work a probabilistic simulation method was developed in order to calculate realistic failure distributions and to compare these directly with experimental failure data. In this way we contribute to the concept of "simulation rather than testing"