Aluminium nitride (AlN) is a ceramic dielectric with high thermal conductivity. It is used in ESCs where wafer temperature uniformity and thermal stability matter more than raw material cost. It conducts heat roughly five to six times better than alumina. It expands at a rate much closer to silicon. It also behaves naturally in the Johnsen–Rahbek (JR) clamping regime. This article explains AlN’s key properties: thermal conductivity, coefficient of thermal expansion, volume resistivity and plasma resistance. It explains how each property maps to process fit. It also explains when the extra cost of AlN over alumina is justified for a given etch, PVD or CVD application.
AlN sits alongside alumina and doped ceramics in the wider dielectric trade-off covered by our pillar on ESC dielectric materials.
What aluminium nitride brings to an ESC
Aluminium nitride is a sintered technical ceramic. For ESC use it is typically supplied at 99.5–99.9% purity. Its defining property is thermal conductivity in the range of roughly 150–180 W/m·K. That is about five to six times the figure for alumina. It is the single reason a fab specifies AlN. In a plasma process the wafer absorbs significant power. The dielectric between the wafer and the cooled base plate has to move that heat through with as little temperature drop as possible. A more thermally conductive dielectric flattens the temperature profile across the wafer. It also reduces the lag between a change in ESC base temperature and the wafer responding to it. In some etch and deposition steps, critical dimensions and film stress depend on a tight, repeatable wafer temperature. For those steps, that conductivity is the deciding material property.
The second property that sets AlN apart is its coefficient of thermal expansion (CTE). AlN expands at about 4.5 ppm/K, against roughly 7–8 ppm/K for alumina. A silicon wafer expands at about 2.6–3.3 ppm/K, so AlN is the closer thermal match. Across a process temperature swing, the wafer and the AlN surface grow and shrink more nearly in step. This reduces differential expansion at the clamped interface. It also reduces the lateral slip and stress that come with that expansion. On a Coulombic alumina ESC the wider CTE gap is usually tolerable. AlN is the better choice for a thermally demanding process that cycles the ESC hard. It is also better where overlay and particle performance must be very tight.
AlN vs alumina: the property numbers that matter
The two ceramics share a great deal. Both are robust and plasma-resistant. Both are made to the same single-digit-micrometre flatness and the same purity grades. The differences that drive the choice are thermal, and they are summarised below. The clamping-regime difference follows from resistivity. Alumina is usually an insulator and runs in the Coulombic regime. AlN’s resistivity places it naturally in the JR band.
| Property | Alumina (Al₂O₃) | Aluminium nitride (AlN) |
|---|---|---|
| Thermal conductivity | ≈ 30 W/m·K | ≈ 150–180 W/m·K |
| CTE | ≈ 7–8 ppm/K | ≈ 4.5 ppm/K |
| CTE match to silicon (~2.6–3.3 ppm/K) | Wider gap | Closer match |
| Typical clamping regime | Coulombic (insulator) | Johnsen–Rahbek |
| Dielectric strength | 10–20 kV/mm (both) | |
| Purity (ESC grade) | 99.5–99.9% (both) | |
| Plasma resistance | Good | Good |
| Relative cost | Lower (standard choice) | Higher (premium) |
Dielectric strength is comparable for both ceramics at 10–20 kV/mm. So the AlN choice is not about withstanding the clamp voltage. A 300–3000 V DC supply is well inside the breakdown margin for either material at normal dielectric thickness. The choice is about heat and expansion, and about which clamping regime the ESC is built around.
Why AlN suits the Johnsen–Rahbek regime
Volume resistivity sets the clamping regime. A Coulombic dielectric is a near-perfect insulator, with resistivity above 10¹⁵ Ω·cm. A Johnsen–Rahbek dielectric is slightly conductive, at roughly 10¹⁰–10¹³ Ω·cm. Across the range of ESC dielectrics, bulk resistivity spans roughly 10⁹–10¹⁶ Ω·cm. AlN grades intended for ESC use are formulated to land in or near the JR window. In a JR ESC the charge migrates through the bulk to the surface. The effective gap collapses to the microscopic contact points. The ESC then generates far higher clamping force per volt than a Coulombic design. That lets an AlN JR ESC hold the wafer firmly at a lower voltage.
The JR regime carries its own trade-offs, and AlN’s thermal properties partly offset them. JR clamping is more temperature-sensitive, because the resistivity of the ceramic falls as it heats. It also releases more slowly, because surface charge takes time to decay. This makes dechucking more demanding. AlN’s high thermal conductivity helps here. It keeps the ESC temperature more uniform and predictable. So the resistivity, and with it the clamp force, stays more stable across the wafer. The full physics of the two regimes, including the temperature dependence of JR force, is covered in our guide to Coulombic vs Johnsen–Rahbek ESCs.
Where AlN justifies its higher cost: thermally demanding etch and PVD
AlN is not the default dielectric. Alumina is the standard choice because it is robust and cost-effective, and it meets the needs of a large share of processes. AlN justifies its higher cost where the process is thermally demanding. It is also justified where wafer temperature uniformity directly limits yield or device performance. Several situations point toward AlN.
- High-power plasma etch. The wafer absorbs large, sustained heat loads. The temperature profile must stay flat across the whole wafer to keep etch rate and critical dimensions uniform.
- Demanding PVD and CVD. Film stress, step coverage or crystallinity depend on holding the wafer at a tightly controlled temperature. Any radial gradient shows up as a within-wafer non-uniformity.
- Processes that cycle the ESC temperature hard. The closer CTE match to silicon reduces differential stress and slip at the clamped interface across each thermal swing.
- Tight overlay or particle budgets. Reduced interface stress and a more stable thermal state help keep the wafer flat. They also reduce the particle-generating micro-movement.
Where none of those pressures apply, alumina usually remains the right answer. An example is a moderate-power process with comfortable temperature margins on a mature tool. In that case, paying more for AlN gives uniformity headroom the process does not need. The decision is a process-fit question, not a blanket ranking of one ceramic above the other.
Thermal conductivity and helium backside cooling work together
AlN’s thermal conductivity does not act alone. A clamped wafer only touches the ESC at microscopic high points, so conduction through solid contact is poor. Helium is fed into the gap behind the wafer, typically at 5–40 Torr. The gas carries heat from the wafer to the temperature-controlled ESC body. The clamping force must always exceed the helium pressure. Otherwise the gas pushes the wafer off the ESC and cooling collapses. AlN improves the other half of that heat path. Once the helium has moved heat into the ceramic surface, the high-conductivity bulk takes over. It carries the heat on to the cooled base plate with a smaller temperature drop. A high-conductivity dielectric and effective helium backside cooling are complementary. AlN gives the best return where both are pushed hard. The clamp-force-versus-helium balance is the same on AlN as on any ESC. It is detailed in our guide to ESC clamping force.
Plasma resistance and surface finishing
AlN is plasma-resistant, as is alumina. For many chemistries the bare ceramic surface is adequate. Aggressive fluorine- or chlorine-based etch chemistries need more protection. A plasma-resistant coating or finish is applied to the exposed surfaces, for example yttria (Y₂O₃). It reduces erosion and the particle generation that comes with erosion. That protects both the dielectric and the wafers it processes. The clamping surface itself is finished with mesas or a textured pattern. This pattern defines the real wafer contact area and the volume available for helium. The surface is held flat to single-digit micrometres so the wafer sits flat. Surface condition is also what degrades in service. As the dielectric erodes under plasma, clamping force drifts and particle performance falls. At that point an ESC is evaluated for refurbishment.
AlN ESCs across wafer sizes and the support question
AlN ESCs are built across the production wafer sizes, from 150 mm and 200 mm through 300 mm. A great deal of installed capacity in power devices, analogue, MEMS and compound semiconductors runs on 150 mm and 200 mm tools. The original equipment makers no longer prioritise those tools. Yet they often run exactly the thermally demanding processes where an AlN dielectric was specified in the first place. That combination of a premium ceramic on a legacy tool is where second-source supply and refurbishment matter most. A new AlN ESC for a discontinued platform may have a long lead time, or it may be unavailable outright. Where the ESC body is sound, ESC repair and refurbishment can return the dielectric surface to as-new clamping and particle performance, with full functional test. That preserves the thermal advantage that justified AlN, without the cost and wait of a new OEM part. The layered construction that makes this possible is the dielectric, buried electrode, bond layer and cooled base. It is covered in our guide to ESC anatomy.
Frequently asked questions
When is AlN worth the premium over alumina?
AlN is worth the higher cost where the process is thermally demanding and wafer temperature uniformity limits yield. Examples are high-power etch, demanding PVD or CVD, and processes that cycle the ESC hard. Its conductivity of roughly 150–180 W/m·K and its closer CTE match to silicon give flatter, more stable wafer temperatures. For moderate-power processes with comfortable thermal margins, alumina usually remains the better-value choice.
What volume resistivity is needed for a Johnsen–Rahbek ESC?
A JR dielectric is slightly conductive, with a bulk resistivity of roughly 10¹⁰–10¹³ Ω·cm. A Coulombic insulator is above 10¹⁵ Ω·cm. AlN grades for ESC use are formulated to land in or near that JR window. This is why AlN ESCs generate higher clamping force per volt and run at lower voltage than Coulombic designs.
How does AlN compare with alumina for ESCs?
AlN conducts heat about five to six times better than alumina (~150–180 vs ~30 W/m·K). It expands closer to silicon (~4.5 vs ~7–8 ppm/K). This gives better temperature uniformity and less interface stress. Both share comparable dielectric strength of 10–20 kV/mm and the same purity grades. Alumina is the cost-effective standard choice; AlN is the thermally superior premium choice.