The dielectric in an ESC is the insulating layer between the buried electrode and the wafer. Its base material is either ceramic or polymer. That choice sets almost everything that matters: plasma resistance, thermal performance, temperature range and service life. Ceramic dielectrics such as alumina, aluminium nitride and doped ceramics dominate aggressive wafer-fab plasma. Polymer and elastomer dielectrics appear mainly in lower-temperature, large-area or non-semiconductor clamps. This article compares the two material families property by property and explains where each one fits.
It is a supporting piece in our wider treatment of ESC dielectric materials, which begins with the comparison of alumina and aluminium nitride.
Why the dielectric base material matters
An ESC clamps a wafer by applying high-voltage DC to one or more electrodes buried under a thin dielectric. The voltage is 600–2,000 V DC. The resulting electrostatic attraction holds the wafer flat. The dielectric has to do three jobs at once. It must stand off that field without breaking down. It must conduct process heat from the wafer to the cooled base. It must also survive direct exposure to the plasma at its edges and through any worn spots. The base material decides how well it does all three. A ceramic and a polymer can both insulate, and both can clamp. But they operate in very different ranges of temperature, plasma exposure and durability. That difference is what separates a wafer-fab ESC from a clamp used in gentler applications.
Volume resistivity sets the clamping regime for either family. A Coulombic dielectric sits above 10¹⁵ Ω·cm and behaves as a near-perfect insulator. A Johnsen–Rahbek (JR) dielectric is slightly conductive, at roughly 10¹⁰–10¹³ Ω·cm. It gives higher force per volt, at the cost of slower release and more temperature sensitivity. Across the ESC dielectrics in service, bulk resistivity spans roughly 10⁹–10¹⁶ Ω·cm. That regime physics does not depend on the material family. It is controlled by resistivity, not by whether the body is ceramic or polymer. But each material can only hold its resistivity stable within a certain temperature and plasma envelope. That envelope is entirely a material property, and it is where the two families differ.
Plasma resistance
Plasma resistance is the property that most clearly separates ceramic from polymer dielectrics for wafer processing. Both alumina (Al₂O₃) and aluminium nitride (AlN) are plasma-resistant ceramics. They erode slowly under the energetic ion bombardment and reactive chemistry of an etch or deposition chamber. They also shed few particles as they erode. The most aggressive fluorine and chlorine etch chemistries need extra protection. A plasma-resistant coating or finish is applied to exposed surfaces; yttria (Y₂O₃) is the common choice. It slows erosion further and suppresses particle generation. The ceramic body and the protective finish work together. The result is an ESC that can sit in an etch plasma for thousands of wafers between interventions.
Polymers and elastomers do not have that margin. The reactive species and ion energy of an aggressive plasma attack organic dielectrics far faster than a sintered ceramic. The exposed surface erodes, roughens and releases particles. The dielectric thins where the plasma reaches it. In a fluorine or chlorine etch, this quickly leads to clamp drift, defectivity and failure. This single property is the reason ceramics, not polymers, are used in the chambers where the plasma does real damage.
Thermal tolerance and thermal conductivity
Plasma processes put significant heat into the wafer, and the ESC has to remove it. Two separate but related material questions follow. How well does the dielectric conduct that heat? And how hot can it run before it degrades? On both questions, ceramics are far ahead. Alumina has a thermal conductivity of about 30 W/m·K. Aluminium nitride is far higher, at roughly 150–180 W/m·K. It conducts heat around five to six times better. So it gives a flatter wafer temperature for high-power or thermally demanding processes. Both ceramics are stable at the high temperatures a plasma chamber reaches, and well beyond.
Polymer dielectrics conduct heat poorly by comparison. They also tolerate far less temperature before they soften, outgas or break down. That limits them to applications where the heat load is modest and the operating temperature stays low. A wafer-fab etch or PVD chamber needs the dielectric to move large heat fluxes and hold its properties at high temperature. A polymer cannot do this, and the wafer temperature drifts out of specification. Ceramic conductivity is also what makes effective helium backside cooling worthwhile. Helium is fed into the gap behind the wafer at 5–40 Torr. It carries heat from the wafer to the ESC surface. From there, the ceramic has to carry the heat on to the cooled base. A poorly conducting dielectric wastes that path.
Thermal expansion and wafer matching
A dielectric that runs hot also has to expand at a rate close to the silicon wafer it holds. If it does not, the two surfaces slip against each other across the temperature cycle. That slip generates particles and stress. Silicon has a coefficient of thermal expansion (CTE) of about 2.6–3.3 ppm/K. Alumina sits at roughly 7–8 ppm/K and aluminium nitride at about 4.5 ppm/K. So AlN is the closer thermal match to silicon, and it reduces differential expansion across the operating range. Both ceramics are within a few ppm/K of the wafer.
Polymers typically expand much more than either ceramic or silicon as they heat. So over any meaningful temperature excursion, the differential expansion between a polymer dielectric and the wafer is large. That mismatch works against the flat, stable, low-stress contact a wafer-fab ESC depends on. It is a further reason polymer dielectrics are limited to low-temperature use, where the excursion and the differential movement stay small. CTE matching is treated in more depth in our note on the temperature sensitivity of the two clamping regimes. In a JR ESC, that temperature sensitivity adds to any thermal mismatch.
Durability and service life
Durability follows directly from plasma and thermal tolerance. A ceramic ESC is a hard, sintered body, typically 99.5–99.9% pure. It resists abrasion from wafer handling. It holds its clamping surface flat to single-digit micrometres. It erodes only slowly under plasma. Its dielectric strength is high, in the 10–20 kV/mm range. So it stands off the clamp voltage with margin, even as the surface ages. A worn ceramic ESC can usually be reconditioned rather than scrapped, because the body survives even when the surface no longer performs.
A polymer or elastomer dielectric is softer. It erodes faster under plasma, and it is more vulnerable to mechanical wear and thermal cycling. In a gentle application that can be acceptable. A polymer clamp may simply be treated as a part that is replaced more often. In an aggressive wafer-fab chamber the same material would not last. This is why ceramic durability, not just the plasma and thermal numbers, underpins the long service intervals fabs expect. A ceramic ESC does eventually reach the end of its surface life. At that point, ESC repair and refurbishment can return it to as-new clamping and particle performance with full functional test. Most polymer clamps are disposable and do not offer that option.
Temperature range and particle behaviour
The two families occupy different temperature bands. Ceramic dielectrics operate from low temperatures up through the high temperatures of plasma processing, and beyond. Across that range they hold their resistivity, dielectric strength and dimensional stability. That breadth is what lets one ceramic ESC design serve hot etch, PVD and CVD. Polymer dielectrics are restricted to a lower, narrower band. Above it they soften, outgas or degrade. So they are inherently low-temperature components.
Particle behaviour reinforces the same split. A clean, hard ceramic surface with an appropriate plasma-resistant finish generates very few particles over a long campaign. That is essential where defectivity directly costs yield. A polymer surface erodes faster under plasma and is softer under handling. So it tends to shed more particles as it ages. Some processes treat every added particle as a potential killer defect. For them, the low and stable particle generation of a well-maintained ceramic ESC is decisive.
Ceramic vs polymer dielectrics compared
The table summarises the trade-offs. The ceramic figures are the established values for ESC-grade alumina and aluminium nitride. The polymer column is described in words only. Polymer ESC dielectrics cover a wide range of formulations, and no single set of fixed figures represents them.
| Property | Ceramic (Al₂O₃ / AlN / doped) | Polymer / elastomer |
|---|---|---|
| Plasma resistance | High; both Al₂O₃ and AlN plasma-resistant, yttria finish for aggressive etch | Low; erodes and sheds particles under aggressive plasma |
| Thermal conductivity | Al₂O₃ ≈ 30 W/m·K; AlN ≈ 150–180 W/m·K | Low (poor heat conductor) |
| CTE vs silicon (~2.6–3.3 ppm/K) | Al₂O₃ ≈ 7–8 ppm/K; AlN ≈ 4.5 ppm/K (close match) | Much higher; large differential expansion |
| Temperature range | Low through high-temperature plasma processing | Low-temperature only; softens/outgasses when hot |
| Dielectric strength | 10–20 kV/mm | Material-dependent; not characterised here |
| Durability / particles | Hard, low-particle, long service life; refurbishable | Softer, higher particle shedding, shorter life |
| Clamping regime | Coulombic (> 10¹⁵ Ω·cm) or JR (10¹⁰–10¹³ Ω·cm) by resistivity | Set by resistivity as for any dielectric |
| Typical use | Aggressive wafer-fab plasma: etch, PVD, CVD, implant | Low-temperature, large-area or non-semiconductor clamps |
Typical applications for each
For aggressive wafer-fab plasma the answer is settled: ceramics dominate, and for good reason. Plasma etch, PVD, CVD and ion implantation all combine reactive chemistry, ion bombardment and high heat load. Only a ceramic dielectric survives that environment while holding wafer temperature, clamp force and particle performance to specification over a long campaign. Alumina is the robust, cost-effective standard choice, and it is usually Coulombic. Aluminium nitride is chosen where its much higher thermal conductivity and closer CTE match justify its higher cost. Doped or composite ceramics place the resistivity in the JR window for higher force per volt. The choice within the ceramic family is a real engineering decision. But for these chambers the choice stays within the ceramic family.
Polymer and elastomer dielectrics have their place outside that envelope. Some applications run cool, cover a large area, or sit outside semiconductor wafer processing altogether. Where plasma exposure is also mild or absent, a polymer dielectric can clamp adequately at lower cost and complexity. Some low-temperature or large-area clamps use exactly this approach. A polymer dielectric is not a substitute for ceramic in an aggressive etch or deposition chamber. The plasma, the heat and the particle budget rule it out. So a fab engineer evaluating dielectric materials for a 150 mm, 200 mm or 300 mm plasma tool is choosing among ceramics. The polymer option is relevant only at the gentler edge of the wider clamping field.
Frequently asked questions
Ceramic or polymer dielectric for an ESC?
For semiconductor wafer processing the dielectric is ceramic: alumina, aluminium nitride or a doped ceramic. Ceramics resist aggressive plasma, conduct heat well, tolerate high temperature and shed few particles. Polymer and elastomer dielectrics are used only in lower-temperature, large-area or non-semiconductor clamps where plasma exposure is mild.
Why do ceramics dominate aggressive plasma wafer processing?
Ceramics combine plasma resistance, high temperature tolerance, good thermal conductivity and low particle generation. Alumina conducts heat at ≈ 30 W/m·K and AlN at ≈ 150–180 W/m·K. A polymer erodes, softens and sheds particles far faster under the reactive chemistry and ion bombardment of an etch or deposition plasma. So it cannot hold clamp force, wafer temperature or defectivity to specification.
Can a polymer ESC handle high temperatures?
No. Polymer and elastomer dielectrics are inherently low-temperature components. Above a modest range they soften, outgas or degrade, and their high thermal expansion does not match the silicon wafer. High-temperature plasma processing requires a ceramic dielectric. A ceramic holds its resistivity, dielectric strength and dimensional stability across the full operating range.