本帖最後由 hlperng 於 2018-3-12 19:36 編輯
各位老師、大德好, 小弟在 Design house 負責 HTOL 之驗證,有些疑問想在此請教各位大師, 先剪貼一段 JESD 之規範: JESD JP001 FOUNDRY PROCESS QUALIFICATION GUIDELINES (Wafer Fabrication Manufacturing Sites) Page 28 12.1.1 Operating life test requirements
Circuit bias configuration VCC_STRESS = λ VCC_OPERATING, where VCC_STRESS must not exceed the device functional limits The minimum λ allowed is 1, but may range up to that tolerated for a stress period without artificially damaging the devices under test. NOTE 1 Unless otherwise specified, the operating voltage is the maximum operating voltage specified for the device. NOTE 2 The voltage acceleration model should be obtained from specific failure data for the product under test. NOTE 3 The nomenclature VCC, or alternatively VDD, refers to the voltage(s) applied to the power supply pins.
\\(中略)
Method to be used ... b) Voltage Acceleration, AF(V) - Correct voltage acceleration factor is to be derived for the product being stressed. If the voltage acceleration is defect driven, the following model may be used: AFV= exp(α (VCC_STRESS –VCC_OPERATING) ...
以字面上來看,其規範了兩件事, 一是 Stress 電壓必須 ≧ Vddmax,亦即產品給客戶的 spec 寫到哪裡我們就必須至少 Burn 到那裏; 二是電壓加速倍率的計算,是要高過 Vddmax 才算是加速,亦即如客戶正常應用 1.2V,Spec 定義 Vddmax 1.32V, Stress 1.44V,其 (Vcc_stress-Vcc_operating)之 ΔV 只有 1.44-1.32=0.12V,不是 1.44V-1.2=0.24V
HTOL 計算電壓加速時,有關客戶端應用電壓的計算基礎,JESD 部分似乎只有看到在規範 Foundry 的 JP001 有詳細定義,但針對產品端的JESD47 (Stress-Test-Driven Qualification of Integrated Circuits) 就沒有特別定義和說明, 所以我們是 follow JP001 的規範。
但最近小弟一再被質疑,FAB 針對 device 的可靠度計算可以套用在產品端嗎? Spec 寫到哪裡就要 Burn in 到哪裡這沒有問題,但計算上為何不是以客戶端應用中心值為準? 甚至被搬出JESD74A (Early Life Failure Rate Calculation Procedurefor Semiconductor Components) 附錄上的幾個計算範例,看來其 Use voltage 就是中心值電壓,不是 Vddmax。
這讓小弟有點語塞!
|