CMP Polishing Pads: Types, Structure & Role in Wafer Planarization

01The Role of the Polishing Pad in CMP

The polishing pad performs three simultaneous functions in the CMP process: it distributes the abrasive slurry across the wafer–pad interface; it applies the controlled contact pressure that drives material removal via Preston’s equation (MRR = K_p × P × V); and it provides the mechanical compliance that allows the pad to conform to wafer-scale bow while maintaining local contact with surface features at the micro-scale. No other consumable in the CMP process has as complex or as multi-functional a role.

The pad properties interact with the slurry chemistry in ways that make the two consumables inseparable from a process performance perspective. Changing the pad hardness, groove pattern, or conditioning protocol changes the effective slurry delivery rate to the interface, the local contact pressure distribution, and the removal rate uniformity — even if the slurry composition is unchanged. For this reason, pad and slurry qualification are always performed together in new process development.

02Pad Materials and Manufacturing

The overwhelming majority of production CMP pads are manufactured from polyurethane — a polymer family that offers an exceptionally wide range of tunable mechanical properties (from Shore A 20 to Shore D 80), excellent chemical resistance to CMP slurry chemistries (acids, bases, peroxides, chelating agents), and the ability to be engineered with controlled porosity through foaming processes.

Polyurethane Foam CMP Pads

CMP polishing pads are manufactured by casting polyurethane foam in large-format blocks, then precision-slicing the blocks into discs of controlled thickness. The foaming process introduces micro-pores (10–50 µm diameter) distributed throughout the polyurethane matrix. These micropores serve two critical functions:

• Slurry storage and transport: The open-cell micropores act as local reservoirs that absorb and hold slurry at the pad surface, continuously supplying fresh abrasive and chemical agent to the polishing interface as the pad rotates across the wafer.
• Mechanical compliance modulation: By controlling the micropore volume fraction (typically 30–60% by volume) and cell size distribution, the pad manufacturer can tune the effective elastic modulus of the pad surface layer independently of the bulk polyurethane chemistry.

After slicing, pads are precision-machined to the target thickness (typically 1.5–3.5 mm for the top layer) and groove patterns are machined into the surface by CNC cutting. The back surface receives a pressure-sensitive adhesive layer for attachment to the platen.

Pad Hardness and Shore D Scale

CMP pad hardness is measured on the Shore D durometer scale. Hard pads used for global planarization (ILD CMP, STI CMP) typically have Shore D values of 55–65. Soft pads used for buffing steps or ultra-low-k applications have Shore D values of 25–45. The Shore D measurement is taken on the bulk pad material; the effective hardness at the polishing interface is modified by the micropore structure and the groove geometry, making the macroscopic Shore D only a starting approximation of the pad’s actual mechanical behavior at the wafer surface.

03Hard vs. Soft Pads: The Fundamental Trade-off

The fundamental physics dictates that only a pad hard enough to bridge across surface recesses — maintaining preferential contact with elevated features — can deliver global planarization. A soft pad, which deforms and fills in around individual features, polishes them at approximately the same rate as the surrounding field, preserving rather than correcting the local topography.

04Two-Layer Stacked Pad Design

The two-layer stacked pad is the production standard for all critical global planarization CMP applications — ILD, STI, W plug, and copper bulk removal. It resolves the apparent contradiction between needing a hard pad for global planarization and a compliant pad for wafer-scale bow accommodation, by separating these two functions into two distinct layers:

Hard Top Layer (Polishing Layer)

The top layer — directly contacting the wafer surface — is a hard microporous polyurethane foam (Shore D 55–65, typical thickness 1.5–2.5 mm). Its rigidity ensures that only the highest surface features on the wafer make significant contact with the pad, producing the differential contact pressure that drives global planarization. Groove patterns are machined into its upper surface. The micropores open at the surface provide slurry storage and transport.

Soft Sublayer (Compliance Layer)

The sublayer — bonded to the bottom of the hard top layer and to the platen surface — is a soft, compressible foam or felt (Shore A 20–40, thickness 0.8–1.5 mm). It provides mechanical compliance that allows the two-layer pad assembly to conform to the global bow and warp of a 300 mm silicon wafer (typically ±30–80 µm bow) without creating edge-roll-off effects or center-heavy contact pressure profiles. The sublayer also damps mechanical vibrations from the platen drive that could create periodic polishing non-uniformity (“chatter marks”) on the wafer surface.

05Groove Patterns and Slurry Transport

The groove pattern machined into the top surface of a CMP pad is one of the most important determinants of slurry distribution uniformity across the wafer. Without grooves, the pad surface would create a hydrodynamic boundary layer that prevents fresh slurry from reaching the center of the wafer at production rotation speeds, producing severe center-to-edge removal rate non-uniformity (center starvation). Grooves provide channels through which slurry can flow across the pad surface driven by centrifugal force and pressure differentials, continuously refreshing the slurry film at every point of the wafer–pad contact area.

Common Groove Pattern Designs

Groove dimensions — width (typically 0.5–1.5 mm), depth (0.4–1.2 mm), and pitch (spacing 1.5–5 mm) — are optimized for specific slurry viscosities, rotation speeds, and flow rates. Deeper grooves hold more slurry but reduce the available pad contact area; shallower grooves maximize contact area but may starve the interface at high rotation speeds.

06Pad Conditioning: Theory and Practice

Pad conditioning is one of the most critical and often underappreciated aspects of CMP process control. Without continuous conditioning, CMP performance would drift irreversibly over each pad’s lifetime as the polishing surface degrades — and pad-to-pad process reproducibility would be impossible to achieve.

The Pad Glazing Mechanism

During CMP, the mechanical interaction between the wafer surface and pad asperities, combined with slurry particle and film debris accumulation, progressively degrades the pad surface. Polishing by-products (abraded film material, reacted chemical compounds) fill the pad micropores. The surface asperities — the micro-scale protrusions that make the critical contact with the wafer — are flattened and rounded by repeated mechanical contact. The net result is “pad glazing”: the polishing surface becomes smooth, dense, and poorly wetting — completely unable to transport slurry effectively or maintain contact pressure at feature apexes. A glazed pad shows 50–90% reduction in MRR compared to a freshly conditioned pad, and severely degraded WIWNU.

Diamond Conditioning Disk Operation

The conditioning disk — a stainless steel or titanium disc (4–6 inch diameter) with CVD-deposited synthetic diamond crystals (300–500 µm average protrusion height) distributed across its lower surface — is pressed against the rotating pad surface with controlled force (typically 1–8 lbs) while the disk sweeps back and forth across the pad radius. The diamond crystals abrade the top surface of the pad, removing the glazed layer and re-exposing the fresh open-pore microstructure below. The reconstituted surface has a defined population of sharp asperities that efficiently contact the wafer and transport slurry — restoring the pad to its specification MRR and WIWNU performance.

原位置コンディショニングと原位置コンディショニング

• In-situ conditioning: The conditioning disk operates simultaneously with wafer polishing — while the wafer is on the pad. Maintains nearly constant pad state throughout the polishing run. Requires careful optimization of conditioning force and sweep rate to avoid pad wear that is faster than the desired pad lifetime. Standard for processes where run-to-run MRR stability is the top priority.
• 現場でのコンディショニング： The conditioning disk operates between wafers (no wafer on pad). Allows independent optimization of conditioning parameters without the constraint of not damaging the wafer. Results in some MRR drift between successive wafer polishing runs. Often used in combination with in-situ conditioning.

Conditioning Parameters and Their Effects

07Pad Wear, Lifetime, and End-of-Life Criteria

Polishing pads wear continuously during CMP — both from the mechanical interaction with the wafer surface and from the abrasion of the conditioning disk. The thickness of the top layer decreases over the pad’s useful life, and when the grooves become too shallow (groove depth depleted) or the pad surface properties change sufficiently to cause systematic WIWNU degradation, the pad reaches end-of-life and must be replaced.

Pad Lifetime Drivers

• Conditioning aggressiveness: The conditioning disk removes material from the pad surface with each sweep. Higher conditioning force and duty cycle accelerate pad wear. The pad wear rate must be balanced against the need to maintain fresh surface properties.
• Polishing process: Hard film polishing (tungsten, SiC) causes faster mechanical pad wear than soft film polishing (copper buff). High-pressure recipes accelerate wear relative to low-pressure gentle-buff applications.
• Pad material: Higher-durometer polyurethane pads wear more slowly but may require more aggressive conditioning to maintain surface texture.

End-of-Life Indicators

Pad end-of-life is typically determined by monitoring: (1) pad thickness (laser profilometry or contact gauge), with replacement when the top layer reaches a minimum groove depth threshold (typically when groove depth falls below 60% of initial specification); (2) systematic WIWNU drift that cannot be corrected by conditioning parameter adjustment; or (3) fixed-usage lifetime (total polished wafer count or total polishing time) established during pad qualification. Some advanced fabs implement real-time pad surface health monitoring using in-situ optical profilometry or acoustic emission sensors to detect pad degradation before it impacts wafer yield.

08Fixed Abrasive Pads

Fixed abrasive pads represent an alternative pad architecture where the abrasive particles — typically CeO₂ or Al₂O₃ — are permanently embedded in the pad matrix in a controlled, uniform distribution, rather than being delivered as a separate slurry. These pads are used in conjunction with a slurry-free or minimal-chemistry liquid (an “enabling fluid”) that provides the chemical component of CMP without the abrasive particles.

The primary advantage of fixed abrasive pads is dramatically reduced particle defect density: because the abrasive particles are fixed in the matrix, they cannot form large agglomerates that would scratch the wafer surface. This makes fixed abrasive pads the preferred choice for: post-STI silicon surface finishing (where scratch-free silicon is required before gate oxide growth); advanced copper CMP buff steps; and hybrid bonding surface preparation (where Ra < 0.3 nm and zero scratches are mandatory for void-free bond formation).

The limitation of fixed abrasive pads is their lower MRR compared to free-abrasive slurry (abrasive replenishment is limited by the fixed particle density in the pad matrix), faster pad wear rate (abrasive particles are consumed with each use), and higher unit cost. They are therefore used selectively for applications where defect performance, not throughput, is the primary optimization target.

09Pad Selection Guide by CMP Application

10JEEZ CMP Polishing Pads

JEEZ (Jizhi Electronic Technology Co., Ltd.) manufactures CMP polishing pads engineered for the full spectrum of semiconductor planarization applications. As a direct manufacturer, JEEZ produces pads in both single-layer and two-layer stacked configurations, covering oxide ILD, STI, tungsten plug, and copper CMP applications. All JEEZ pads are manufactured from high-purity polyurethane foam with controlled micropore structure, precision-machined groove patterns (X-Y grid, concentric, and spiral variants), and pressure-sensitive adhesive for platen mounting.

JEEZ pad products are validated for chemical compatibility with the full range of JEEZ CMP slurry products, and process qualification data — including MRR, WIWNU, conditioning protocol recommendations, and pad lifetime benchmarks — is available from JEEZ application engineering teams for major CMP tool platforms. JEEZ also supplies absorption and backing films for carrier head assemblies, completing the full consumable package for CMP tool operation.

よくあるご質問よくある質問