每日资讯Daily Notes
从虚拟光刻到先进封装：把仿真筛选与样品验收分开确认From virtual lithography to advanced packaging: separate simulation screening from sample acceptance
依据 imec 与 Applied Materials 的公开资料整理。仿真和设备信息可以帮助提出问题，不能替代目标样品的工艺与量测验收。Based on public material from imec and Applied Materials. Simulation and equipment information can frame questions, but cannot replace process and metrology acceptance for a target sample.
虚拟筛选：先定义模型边界，再缩小试验范围Virtual screening: define model boundaries before narrowing experiments
imec 于 10 月 22 日发布的合作内容提到，物理仿真和加速计算可用于评估光刻中的成像、掩模、光刻胶、随机效应与工艺窗口，并在实体硬件出现前帮助比较方案。项目中应随结果保留仿真输入、材料模型、版图版本和输出判据；模型给出的优先级不等同于样品已通过工艺资格或器件测试。An imec partner article published October 22 describes using physics simulation and accelerated computing to evaluate lithography imaging, masks, resists, stochastic effects and process windows. Retain simulation inputs, material models, layout revisions and output criteria; a model ranking is not process qualification or device-test acceptance.
封装量测：能力信息仍要落回样品与规格Packaging metrology: capability information must return to the sample and specification
Applied Materials 的公开资料列出面向先进封装的 CMP、电子束量测与缺陷分析系统，并说明其面对厚度变化、翘曲和异质基板等挑战。公开产品信息可补全量测问题；具体样品的厚度均匀性、关键尺寸、缺陷阈值、抽样位置和判定规则，仍须在实际工艺流和测试计划中确认。Applied Materials describes CMP, e-beam metrology and defect analysis for advanced packaging. Product information can improve metrology questions; sample-specific thresholds, sites and decision rules require an actual process flow and test plan.
一个工艺观察Process Note
把“预测—试片—量测—判定”留成可追溯链路Keep “prediction–test wafer–metrology–decision” traceable
素微编辑观察：把仿真版本、输入参数、假设条件、掩模/版图版本、试片批次、工艺实际值、量测原始数据和验收结论放进同一版本表。这样能分清预测与实测，也便于在结果偏离时定位材料、图形、设备或量测环节。Place model revision, inputs, assumptions, mask/layout revision, test-wafer lot, actual process values, raw metrology and acceptance conclusion in one versioned record.
项目准备提醒Project Prep
提交试制需求时，附上仿真依据与可测的验收阈值For prototyping, attach simulation rationale and measurable acceptance thresholds
请提供材料与层结构、设计和掩模版本、模型或既有数据的适用范围、目标尺寸与允许偏差、关键量测位置、测试条件、对照样品、缺陷或失效阈值，以及期望交付的数据格式。Provide material and stack, design and mask revisions, model/prior-data scope, target dimensions and tolerances, metrology sites, test conditions, controls, defect thresholds and desired data format.
来源Sources