CMP Daily Abbreviations

材料移除速率下降 / 上升 / 平穩。MRR 變化通常意味 pad 或 slurry 狀態變了。

• MRR dropped from 6000 to 4500 Å/min — pad is glazing.
• MRR climbed back to baseline after fresh conditioning.

墊片表面結膠 / 整理 / 更換 / 清潔。Pad glazing 是 CMP 部門最常開的 deviation 主題。

• Pad is glazed after 250 wafers — schedule a conditioning cycle.
• Pad was swapped at 09:00 — full break-in took 30 wafers before MRR stabilized.

碟陷超標 / 在規格內 / 被遏止。Dishing 是 PQA 看到馬上會問的數字。

• Dishing exceeds spec on the wide Cu pads — 45 nm vs 30 nm target.
• Dishing is contained at < 20 nm after the downforce tune.

漿液被換 / 用完 / 認證 / 消耗。新 slurry vendor 要過完整 Qual。

• Slurry was swapped to vendor B for tungsten CMP — re-qualified with 5 monitor wafers.
• Slurry tank is depleted; pause and refill before next lot.

終點未偵測 / 提早 / 過晚觸發。EPD miss 是 CMP 最危險的事故 — over-polish。

• Endpoint was missed on wafer 13 of lot Q3-218 — over-polished by 20 s.
• EPD triggered 8 s early — likely the new pad has lower torque baseline.

刮痕被偵測到 / 清除 / 減輕。Scratch 通常是 slurry 中大顆粒造成,要查 filter。

• Scratches detected on 12 wafers post-CMP — slurry filter likely clogged.
• Scratches mitigated after replacing the inline slurry filter.

下壓力被拉高 / 降低 / 調整。Downforce 是 CMP 部門最常調的旋鈕。

• Downforce was dropped 5% to mitigate dishing on the M2 layer.
• Downforce bumped up 8% to recover MRR after the pad swap.

CMP 後殘留被清除 / 持續存在 / 被偵測到。Residue 直接影響下一層 yield。

• Post-CMP residue cleared after tuning the cleaner megasonic power.
• Residue is persistent on the M3 wafer — escalate to clean engineering.

8D · dishing excursion

Dishing on feature exceeded N nm; attributed to root_cause, confirmed by evidence.

{圖形} 上的碟陷超過 {N} nm;歸因於 {根因},以 {證據} 確認。CMP 8D D4 標準寫法。

• Dishing on wide Cu pads exceeded 45 nm; attributed to downforce overshoot on chamber A, confirmed by force log and dishing metrology.▸ 聽念法
• Dishing on the seal ring exceeded 35 nm; attributed to slurry pH drift, confirmed by inline pH monitor.▸ 聽念法

EPD failure SOP

Endpoint failure on tool: signal signal_state, over-polished by N seconds.

{機台} 終點失誤:訊號 {狀態},超磨 {N} 秒。EPD miss 報告必出現的句子。

• Endpoint failure on CMP-04: signal flat-lined at 90s, over-polished by 25 seconds.▸ 聽念法
• Endpoint failure on CMP-07: signal noisy throughout, over-polished by 18 seconds.▸ 聽念法

preventive maintenance · pad

Pad on tool is at N% wear; replacement scheduled for date; expected MRR recovery target.

{機台} 上的 pad 損耗 {N}%;預定 {日期} 更換;預期 MRR 回到 {目標}。CMP 設備保養報告句。

• Pad on CMP-04 is at 85% wear; replacement scheduled for 04/15; expected MRR recovery to 6000 Å/min.▸ 聽念法
• Pad on CMP-09 is at 70% wear; replacement scheduled for next PM window; expected MRR recovery to baseline.▸ 聽念法

change control · slurry / pad qual

After change on date, re-qualifying scope with N monitor wafers per the FMEA.

在 {變更} 於 {日期} 之後,依 FMEA 用 {N} 片監控 wafer 重新認證 {範圍}。Slurry 換 vendor 時的標準句。

• After slurry vendor swap on 03/12, re-qualifying tungsten CMP with 10 monitor wafers per the FMEA.▸ 聽念法
• After new conditioner disk on 04/01, re-qualifying all CMP chambers with 5 monitor wafers per the FMEA.▸ 聽念法

8D D3 · CMP

Containment: lots from chamber held; N wafers under clean_action pending check.

Containment:{腔體} 的批次暫停;{N} 片正在 {清潔動作},等待 {確認}。

• Containment: lots from CMP-04 held; 48 wafers under post-CMP re-clean pending residue inspection.▸ 聽念法
• Containment: lots from CMP-09 held; 12 wafers under defect map review pending FA cross-section.▸ 聽念法

Email — Pad Replacement Notice

CMP 工程師發給 OPS 跟 PIE,announce 預計的 pad 更換 + 影響。

Subject: [FYI] CMP-04 pad replacement Friday 02:00 — 6-hour downtime

Hi all,

Heads up — CMP-04 pad is at 85% wear (190 wafers since last swap). MRR has dropped from 6000 to 4500 Å/min over the past 30 wafers.

Schedule:
- Pad replacement: Friday 02:00
- Break-in: 30 dummy wafers
- Re-qualification: 5 monitor wafers
- Total downtime: ~6 hours
- Production resumes: Friday 08:00

Impact:
- 8 lots queued for CMP-04 will reroute to CMP-06 and -09 (capacity OK).
- M3 Cu CMP throughput drops by ~25% during the window.
- No customer-facing impact.

Will send qual results before going back to production.

Thanks,
CMP ENG

相關縮寫:MRR · pad · qualification

Conf Call — Dishing Excursion on M2 Cu

Yield review 上 PIE 詢問 dishing 異常,CMP 部門答 90 秒。

Thanks. Quick recap on the M2 Cu dishing excursion.

We saw dishing on the wide Cu pads (≥ 5 μm) hit 48 nm on the last three lots — spec is 30 nm. Effect is localized to the wide patterns; narrow lines are fine.

Investigation:
- All three lots ran on CMP-04 chamber A. CMP-06 and -09 wafers are within spec.
- Downforce log shows an overshoot to 4.2 psi vs 3.5 psi target on the affected lots — controller drift after the 03/12 PM.
- Slurry pH and conditioner are within spec, so they're cleared.

Root cause: downforce controller drift on chamber A.

Containment: chamber A held; downforce controller re-cal'd; 5 monitor wafers ran with dishing back to 22 nm. Released to production after the qual.

Long-term: adding the downforce as an SPC parameter on every CMP tool — currently it's only end-of-run logged. Plus tightening the post-PM qualification window.

3 lots affected, all currently held at the next inspection station. PQA confirms no customer impact yet.

Open to questions.

相關縮寫:dishing · PM · SPC · PQA

Deviation Report — Scratch Cluster (Section 4)

CMP 出現 scratch 事件的 deviation report,Section 4 根因說明。

4.0 Root Cause Analysis — Scratch Cluster on M3 CMP

4.1 Symptom
Scratch defects detected on 18 wafers across 3 consecutive lots (Q3-241 through Q3-243), all post-M3 CMP. Defect inspection shows long linear scratches (200–500 μm) on the wafer surface, concentrated in a single radial sector.

4.2 Investigation Path
- All 18 wafers ran on CMP-04 chamber B during a 4-hour window (08:00–12:00 on 03/14).
- Slurry filter on chamber B inspected; filter housing showed a 2 mm crack — large slurry agglomerates bypassed the filter.
- FA on a fail wafer confirms slurry-particle scratches; no foreign material.

4.3 Root Cause Statement
The scratch cluster is attributed to a cracked slurry filter housing on CMP-04 chamber B, allowing large slurry agglomerates to reach the wafer, confirmed by physical inspection of the filter and FA on the defect.

4.4 Containment
- All 18 affected wafers scrapped after FA confirmation.
- Chamber B held; new filter housing installed; SOP updated to inspect filter housing weekly.
- 100% scratch metrology on all WIP that ran through chamber B in the past 48 hours.

4.5 Long-Term Action
- Filter housing replaced with a stronger material (PEEK).
- FMEA updated to include "filter housing crack" as a failure mode with weekly visual inspection.

相關縮寫:scratch · FA · FMEA · SOP

讀完了?來測驗看看記住了沒。