SEMICONDUCTOR FAILURE ANALYSIS
半导体失效分析:定位、截面与证据链准备Semiconductor failure analysis: localization, cross-sections and evidence planning
把“为什么失效”拆成可复核的现象、对照、定位、取样、分析方法和结论边界,而不是让单张显微图承担全部判断。Turn “why did it fail?” into reviewable observations, controls, localization, sampling, methods and conclusion boundaries instead of asking one image to prove everything.
这类分析解决什么问题What failure analysis addresses
失效分析用于把异常现象与可验证证据建立关联:例如开路、短路、参数漂移、局部烧毁、分层、裂纹或污染。合理的路径通常从保留原始状态的外观与电学记录开始,再根据失效位置、材料堆栈和可承受的破坏性,安排显微、成分、截面或其他专项分析。分析结果应说明观察到什么、在什么条件下观察到,以及哪些因果关系仍需进一步验证。Failure analysis links an observed anomaly to checkable evidence: opens, shorts, drift, local damage, delamination, cracks or contamination. A defensible route preserves as-received visual and electrical records first, then selects microscopy, composition, cross-section or other methods based on location, stack and allowable destructiveness.
适合什么需求Good For
适合研发样片、封装器件、薄膜结构或批次对比中出现异常后,整理初步失效定位与分析路线;也适合在送检前把“找什么、和什么比、允许破坏到哪一步”写成可报价、可追溯的需求包。For anomalous research samples, packaged devices, film structures or lot comparisons; also for turning what to find, what to compare and allowed destructive steps into a traceable request.
需要准备什么Prepare First
提供样品编号和数量、失效与对照样本、器件/封装结构和材料栈、已知工艺或使用历史、异常发生条件、原始电学/光学/热学记录、失效坐标或疑似区域、已有图片、是否允许开封/切割/研磨/FIB,以及需要保留的样品与数据交付格式。Provide sample IDs and count, failed and control samples, device/package structure and stack, known process or use history, anomaly conditions, raw test records, failure coordinates, existing images, allowed decapsulation/cutting/polishing/FIB, and required retained samples and data format.
常见风险Common Risks
只有失效样品没有对照、样品在送检前被反复探测或清洗、定位坐标与方向未保存、先做破坏性制样、将相关性直接写成根因,都会削弱证据链。失效分析不是对材料、工艺、可靠性或责任的自动判定;结论强度必须与样本数、方法能力和对照证据相匹配。No control sample, repeated probing or cleaning before intake, missing orientation, destructive preparation too early and treating correlation as root cause all weaken the evidence chain. Failure analysis does not automatically establish material, process, reliability or liability conclusions.
怎样写清送样与验收Make handoff and acceptance explicit
先约定问题陈述、对照组、分析顺序和每一步的停止条件。验收可分为:样品接收与链路可追溯、原始状态和定位记录完整、已执行方法及其限制写明、原始图像/谱图/测试数据可交付,以及结论与观察、推断和待验证事项清楚分栏。若分析会消耗样品,先确认优先级和备份数量。Agree on the problem statement, controls, analysis order and stop criteria. Separate traceable receipt, complete as-received and localization records, stated method limits, deliverable raw data, and clearly labeled observations, inferences and open checks. Confirm priority and backups before consuming samples.
与已有表征页的边界Boundary with related characterization pages
本页聚焦失效问题的证据链和方法排序,并不替代具体仪器操作。需要局部截面和层间观察时,先看FIB 截面制样;需要元素或颗粒分布时看EDS 能谱元素分析;需要表面化学态或深度信息时看XPS 表面分析和ToF-SIMS 深度剖析。各方法是否适用、检测限和结论仍由实际平台按样品确认。This page covers evidence planning and method ordering, not operation of a specific instrument. Use the FIB, EDS, XPS and ToF-SIMS pages for their distinct preparation intents. Actual suitability, limits and conclusions require confirmation by the performing facility.
公开资料Public references
核查日期:2026-11-15。以下资料用于说明失效分析的通用证据与方法概念;本文为素微项目沟通清单,不构成任何平台能力、检测限、根因、可靠性或责任认定承诺。Checked November 15, 2026. These references support general failure-analysis evidence and methods; this checklist is project guidance, not a capability, detection-limit, root-cause, reliability or liability claim.