How to Select a Copper CMP Slurry: Specifications Checklist & Supplier Evaluation Guide
Selecting the right copper CMP slurry is a high-stakes decision. A slurry that underperforms on dishing, defectivity, or WIWNU can cost millions of dollars in yield loss before the root cause is identified and the slurry is replaced. A slurry that is over-specified for the application wastes procurement budget without improving device performance. This guide provides a systematic, engineering-driven selection framework — covering key performance specifications, the three-stage qualification protocol, supplier evaluation criteria, total cost of ownership modelling, and application-specific considerations — for process and procurement engineers at IDMs, foundries, and advanced packaging facilities in 2026.
Table des matières
- 01.Building a Selection Framework
- 02.Step 1 vs. Step 2: Different Slurries, Different Criteria
- 03.Key Performance Specifications Checklist
- 04.Three-Stage Qualification Protocol
- 05.Supplier Evaluation: Beyond the Data Sheet
- 06.Total Cost of Ownership Model
- 07.Application-Specific Considerations
- 08.How JEEZ Addresses Each Selection Criterion
- 09.Questions fréquemment posées
1. Building a Selection Framework
A rigorous copper CMP slurry selection process begins not with a data sheet comparison but with a clear definition of the application requirements. Before evaluating any candidate slurry, the process and integration engineers responsible for the CMP module should document:
- Technology node and ILD stack: Node (28 nm, 7 nm, 3 nm class), ILD type (TEOS, OSG, ELK), ILD elastic modulus, and maximum allowable polishing pressure.
- Barrier/liner material: Ta/TaN, Co, Ru, or barrierless — determines Step 2 slurry chemistry compatibility requirements.
- Pattern geometry: Minimum metal pitch, maximum feature width, copper density range across the die — determines dishing and erosion tolerance requirements.
- Throughput requirements: Target wafers-per-hour (WPH) for the CMP module, which sets minimum removal rate requirements.
- Defect budget: Maximum allowable post-CMP defect density at the relevant inspection sensitivity, derived from the yield model for the target product.
- Supply chain requirements: Geographic supply region, qualification timeline, minimum order quantities, and any environmental/chemical compliance restrictions (BTA-free, PFAS-adjacent restrictions).
This requirements document becomes the primary evaluation filter: candidate slurries that cannot meet any hard requirement on this list are eliminated before consuming evaluation resources.
2. Step 1 vs. Step 2: Different Slurries, Different Criteria
Step 1 and Step 2 slurry selection require almost entirely different evaluation criteria because they perform fundamentally different tasks. Conflating the two leads to misaligned procurement decisions.
Step 1 Selection Priorities
Copper removal rate (nm/min) at target pressure. WIWNU (1σ). Abrasive type and ELK compatibility (pressure limit). POU oxidizer stability. Step 1 defect density (some tolerance — Step 2 partially cleans). Slurry flow rate efficiency (consumption cost). Pad lifetime impact.
Step 2 Selection Priorities
Cu:barrier selectivity (≥5:1). Cu:ILD selectivity (≥50:1). Static etch rate (dishing control). Barrier material compatibility (Ta/TaN vs. Co/Ru). Step 2 defect density (very tight — no further clean after Step 2). Erosion in dense arrays. pH and stability.
A common procurement mistake is to evaluate Step 1 and Step 2 slurries from the same supplier on the same set of metrics — then be surprised when the Step 1 winner (on removal rate) is unacceptable for Step 2 (on selectivity). Always evaluate Step 1 and Step 2 candidates separately against their respective requirement sets, and also evaluate them together as a matched pair against integrated dishing/erosion/endpoint performance on patterned wafers.
3. Key Performance Specifications Checklist
| Specification | Step 1 Target | Step 2 Target | Measurement Method |
|---|---|---|---|
| Cu Removal Rate | ≥300 nm/min (≥2 psi) or ≥180 nm/min (≤1 psi ELK) | 50–150 nm/min | 4-point probe / blanket wafer |
| WIWNU (1σ, 300 mm) | <5% | <3% | 49-point or 121-point mapping |
| Cu:Ta Selectivity | N/A (no barrier exposed) | >5:1 | Blanket wafers, individual RR ratio |
| Cu:TEOS Selectivity | N/A | >50:1 | Blanket wafers, individual RR ratio |
| Static Etch Rate | <2 nm/min | <0.5 nm/min | Immersion test, 4-point probe |
| Defect Density (≥0.1 µm) | <50/cm² | <10/cm² | KLA-Tencor SP5 or equivalent |
| Particle Size D50 | 60–150 nm | 60–120 nm | DLS (dynamic light scattering) |
| Particle Size D99 | <500 nm | <400 nm | DLS or laser diffraction |
| Zeta Potential | |ζ| >25 mV (colloidal stability) | |ζ| >30 mV | Electrophoretic light scattering |
| pH Stability | ±0.3 units over shelf life | ±0.2 units over shelf life | pH meter, multiple time points |
| Metal Ion Purity | Fe, Al, Na, K <50 ppb each | Fe, Al, Na, K <20 ppb each | ICP-MS |
| H₂O₂ Stability (POU) | ±10% at 60 min after mixing | ±10% at 60 min after mixing | Titrimetric or colorimetric assay |
| Dishing (100 µm pad, patterned) | N/A | <50 nm | Profilometry on SEMATECH wafer |
| Erosion (70% density, patterned) | N/A | <15 nm | Profilometry on SEMATECH wafer |
Specification tip: Request the slurry supplier’s full lot release specification — not just the typical performance data sheet. The specification defines the guaranteed range across all production lots; the typical data shows only one measurement point. A slurry with excellent typical performance but wide lot-to-lot specification limits will produce high process variability in production.
4. Three-Stage Qualification Protocol
A rigorous copper CMP slurry qualification proceeds through three evaluation stages, each requiring increasing investment but providing increasing confidence in production performance. Skipping stages — particularly by jumping from blanket wafer data to production adoption — is a common source of unexpected yield loss after slurry introduction.
Stage 1 — Blanket Wafer Screening (2–4 weeks, 20–50 wafers)
Polish blanket copper, Ta, TaN (and Co/Ru if applicable), TEOS oxide, and ELK wafers at the target process conditions. Measure removal rate, WIWNU, static etch rate, and selectivity ratios. Purpose: rapidly screen out candidates that fail fundamental kinetic or selectivity requirements before committing patterned wafer budget. Cost: low. Go/no-go decision at the end of Stage 1.
Stage 2 — Patterned Wafer Evaluation (4–8 weeks, 50–150 wafers)
Polish SEMATECH 854 or MIT mask patterned test wafers (or fab-proprietary test reticles) through the complete two-step CMP sequence. Measure dishing, erosion, within-die planarity across all feature width and density combinations in the test mask. Characterize post-CMP defect density and defect type distribution by wafer map. Purpose: assess real-world performance on realistic circuit-like topographies. Cost: moderate. Eliminate candidates with unacceptable patterned performance.
Stage 3 — Product Wafer Qualification (8–20 weeks, 200–500 wafers)
Polish actual production-layer wafers under full two-step process conditions including endpoint detection, post-CMP cleaning, and downstream processing through the next metal level. Measure electrical yield (via resistance, line resistance, opens/shorts), TDDB dielectric reliability, and electromigration lifetime on process control monitor (PCM) structures. Cost: high. Successful completion defines the production process control limits and the approved supplier qualification.
5. Supplier Evaluation: Beyond the Data Sheet
Performance specifications are necessary but not sufficient for supplier evaluation. A slurry that meets all specifications in Stage 1 but is produced by a supplier with inconsistent manufacturing quality will deliver high lot-to-lot variability in production — the most damaging form of process instability in a high-volume fab. Supplier evaluation should address:
- Manufacturing quality system: ISO 9001 certification is baseline. Look for semiconductor-specific quality management systems (SEMI E10, IATF 16949 for automotive-grade supply chains) and SPC (statistical process control) data on critical slurry parameters (particle size, pH, oxidizer concentration) across a minimum of 20 production lots.
- Raw material traceability: Can the supplier trace each slurry lot to the specific raw material batches (abrasive source lot, BTA lot, oxidizer grade) used in its production? Traceability is essential for root cause investigation when a production excursion is detected.
- Lot-to-lot consistency data: Request Cpk (process capability index) data for key slurry parameters across ≥20 production lots. Cpk >1.33 for particle size D50, pH, and oxidizer concentration at POU is a reasonable minimum requirement for a high-volume logic supplier.
- Application engineering support: Does the supplier provide process engineers who can support slurry integration, optimize process parameters on the customer’s specific CMP tool type, and assist in troubleshooting production excursions? Slurry supply without application support is a significant risk for advanced-node integration.
- Supply chain resilience: Single-plant suppliers are a supply chain risk. Assess manufacturing capacity, backup raw material sourcing, and the supplier’s track record for meeting delivery commitments during industry upturns.
6. Total Cost of Ownership Model
The purchase price per liter of copper CMP slurry rarely determines the optimal procurement decision. A comprehensive TCO model must include all process cost contributors influenced by slurry performance:
| TCO Component | How Slurry Affects It | Relative Weight |
|---|---|---|
| Slurry purchase cost | Direct — price per liter × consumption per wafer | Moyen |
| Slurry consumption per wafer | Flow rate required for target RR and WIWNU; dilution ratio at POU | Moyen |
| Polishing pad lifetime | Slurry chemistry affects pad wear rate; gentler slurry = longer pad life | Medium-High |
| CMP tool throughput (WPH) | Higher RR = shorter cycle time = higher WPH = lower capital cost per wafer | Haut |
| Defect-driven yield loss | Every 1/cm² reduction in defect density at advanced nodes = significant die yield improvement | Très élevé |
| Post-CMP cleaning cost | Slurries requiring aggressive cleaning chemistry add cleaning chemical cost; slurry compatibility determines cleaning sequence complexity | Low-Medium |
| Rework and scrap cost | Slurry-driven dishing/erosion excursions may require rework or scrap of affected lots | Haut |
| Qualification cycle time | Supplier support quality affects time-to-qualification; faster qual = earlier volume ramp revenue | Medium-High |
The most common TCO miscalculation is underweighting defect-driven yield loss. At a production cost of $10,000 per 300 mm wafer and a die size of 100 mm² (typical for advanced logic), a 1/cm² reduction in post-CMP defect density reduces the probability of a defect falling on any given die by approximately 1 × 10⁻² — depending on critical area, this might translate to 0.5–2% die yield improvement. Across 10,000 wafer starts per month, this is potentially $0.5M–$2M per month in additional value, dwarfing any slurry price difference between candidates.
7. Application-Specific Considerations
Leading-Edge Logic (7 nm and Below)
Priority: lowest possible defect density, ELK compatibility (<1 psi), Co/Ru liner compatibility, tightest dishing tolerance (<3 nm on 20 nm features). Recommended evaluation criteria weight: defect density 40%, dishing 30%, erosion 20%, throughput 10%. BTA-free formulations increasingly requested for environmental compliance.
DRAM Copper Interconnect
DRAM copper CMP typically operates at older nodes (25–40 nm effective pitch) but with extremely tight electrical uniformity requirements driven by the 1T1C cell’s sensitivity to access transistor resistance variation. Priority: WIWNU <2%, lot-to-lot consistency (Cpk >1.5 for key parameters), and very high throughput (DRAM fabs operate with thin margins per die). Recommended evaluation criteria weight: WIWNU 35%, lot-to-lot consistency 30%, throughput 25%, defect 10%.
Advanced Packaging (Hybrid Bonding, RDL CMP)
Hybrid bonding CMP requires precision recess control (see advanced node article). RDL (redistribution layer) copper CMP typically operates at feature widths of 2–10 µm at low pressure, where the primary concern is surface planarity for subsequent photolithography rather than dishing at sub-100 nm dimensions. RDL CMP tolerates higher defect counts than front-end logic but requires very high planarity uniformity across the full package substrate area (often >300 mm × 300 mm panel formats at advanced packaging facilities).
8. How JEEZ Addresses Each Selection Criterion
Jizhi Electronic Technology Co., Ltd. (JEEZ) designs its copper CMP slurry portfolio with each of the above selection criteria explicitly addressed:
- Performance coverage: CuB (Step 1 alumina), CuB-S (Step 1 colloidal silica, ELK-safe), CuS (Step 2 standard Ta/TaN), CuS-Co (Step 2 Co/Ru-compatible), CuELK (low-pressure advanced node), CuHB (hybrid bonding precision recess), CuG (BTA-free).
- Specification transparency: Full lot release specifications with Cpk data available for all production parameters across the last 24 production months; Certificate of Analysis with every shipment.
- Qualification support: Dedicated application engineers support Stage 1 through Stage 3 qualification — providing blanket wafer test protocols, SEMATECH patterned wafer evaluation support, and on-site process window optimization assistance.
- Supply chain: Multiple manufacturing sites; ISO 9001 certified; full raw material traceability to source lot; standard lead times of 4–6 weeks with emergency buffer stock available for key customers.
- Environmental compliance: BTA-free CuG series available; full REACH compliance documentation; safety data sheets (SDS) in all required regional languages.
9. Frequently Asked Questions
How long does a full copper CMP slurry qualification typically take?
A complete three-stage qualification — blanket wafer screening, patterned wafer evaluation, and product wafer qualification with electrical yield and reliability data — typically takes 16–28 weeks from sample receipt to production approval at a leading-edge logic fab. DRAM and mature-node fabs may compress this to 8–16 weeks with streamlined product wafer qualification (PCM-only, no reliability) for low-risk node/slurry combinations. Parallel-path qualification (running Stage 2 and 3 simultaneously with different wafer sets) can reduce elapsed time by 4–6 weeks but requires more wafer investment upfront.
Should I evaluate Step 1 and Step 2 slurries from the same supplier?
Not necessarily — the optimal Step 1 slurry and the optimal Step 2 slurry for your application may come from different suppliers. The practical consideration is that using matched slurry pairs from one supplier simplifies troubleshooting (one point of contact for integrated process questions) and may provide commercial pricing advantages. The technical priority should always be to qualify the best-performing slurry for each step independently, then evaluate the pair integration — not to constrain selection to one supplier without evidence that their paired offering is the best available combination for your specific application.
What data should I request from a copper CMP slurry supplier before beginning qualification?
Request the following before committing to qualification: (1) full lot release specification document for all critical parameters; (2) Cpk data across ≥12 production lots for particle size D50, pH, and H₂O₂ concentration at POU; (3) blanket wafer performance data on copper, oxide/ELK, and barrier (at your process pressure and speed conditions if possible, or at closest available conditions); (4) patterned wafer dishing and erosion data on a comparable test mask; (5) material safety data sheet (SDS) and REACH compliance declaration; and (6) sample quality and certificate of analysis from the specific lot being sent for evaluation.
How do I calculate the slurry cost per wafer for TCO comparison?
Slurry cost per wafer = (slurry flow rate, mL/min) × (polishing time, min) × (price per mL). For POU mixing: add the oxidizer cost separately at its consumption rate. Divide total slurry + oxidizer cost by the number of wafers polished per pad change cycle (pad lifetime in wafers) and add the pro-rated pad cost per wafer for a complete consumables cost per wafer figure. The slurry flow rate is itself a function of the specific slurry formulation — some formulations achieve target WIWNU at lower flow rates than others — so always use the experimentally determined minimum viable flow rate from your process conditions, not the supplier’s nominal recommendation.
Ready to Evaluate JEEZ Copper CMP Slurry?
Jizhi Electronic Technology Co., Ltd. (JEEZ) provides sample quantities with full certificates of analysis, Cpk data, and application engineering support for all qualification stages. Tell us your node, barrier stack, target performance requirements, and timeline — and we will match you with the right product from our CuB, CuS, CuELK, CuHB, or CuG series.
Start Your Copper CMP Slurry Evaluation →
