Plasma Resistant Ceramic Components for Etch Chambers
Updated: Sep 30
Plasma-Resistant Ceramic Components by Chamber Zone and Plasma Chemistry
Plasma resistant ceramic components are selected by chamber zone and plasma chemistry, not by a single ranking. Published erosion data differ in units and test conditions, so compare values only within one study.
T1. Chamber zone to candidate material
Chamber zone | Chemistry in the cited erosion data (see T2) | Electrical / RF requirement | MAC-confirmed candidate | Primary failure mode | Comparison-only material (not MAC supply) |
Focus / edge ring | CF4/CHF3 (US 10,622,194; US 8,067,067) | High-resistivity SiC, about 1 Ω·m (100 Ω·cm) or higher (WO2009079285A1) | Silicon carbide; boron carbide (US 12,548,741 lists SiC or B4C) | Erosion: SiC at about 1.6 times the alumina rate in CF4/CHF3, ungraded (US 10,622,194) | Quartz, about 2.2 times the alumina rate, ungraded (US 10,622,194) |
Showerhead / gas distribution plate | CHF3/CF4 (US 8,067,067) | Not addressed in the sources cited here | Silicon carbide faceplate (US 10,544,500 describes anodized aluminum bonded to a SiC faceplate) | Erosion: anodized Al₂O₃ 86.6 nm/min against CVD SiC 52.6 nm/min (US 8,067,067) | Anodized aluminum; plasma-sprayed Y₂O₃ at 24.3 nm/min (US 8,067,067) |
Dielectric window / lid | Fluorine plasma (Ito et al., Jpn. J. Appl. Phys. 47, 3630) | Loss tangent below 1×10⁻⁴ (US 11,998,986) | High-purity alumina | Thermal-expansion mismatch cracking (US 6,797,110); AlF₃ particle formation (Ito et al.) | Yttria and YAG coatings (US 9,869,013) |
Which chamber zone your part sits in decides the material before any supplier conversation starts
The zone decides the material because each zone imposes a different requirement. Focus rings need controlled electrical resistivity, showerheads need erosion resistance at gas-exposed faces, and dielectric windows need low dielectric loss. Selecting a supplier before fixing the zone and its requirement leaves the material choice open.
T1 (Table 1) gives sourced material and requirement data for three zones. Focus rings are described as silicon carbide or boron carbide (US 12,548,741). Showerheads are described as anodized aluminum bonded to a silicon carbide faceplate (US 10,544,500). Windows are specified by loss tangent (US 11,998,986).
SiC shows the highest plasma resistance under CF4-rich conditions and B4C is superior in Ar-rich plasma (Shim et al., Appl. Surf. Sci. 696, 162960). For the wider range, see semiconductor ceramic components.
Focus rings need conductivity control, not just erosion resistance
Silicon carbide's erosion rate in CF4/CHF3 is in T1; erosion resistance alone does not explain its use in rings. The cited focus-ring patent also specifies resistivity: a high-resistivity silicon carbide ring is described at about 1 Ω·m (100 Ω·cm) or higher (WO2009079285A1). State the required resistivity on the drawing beside the material.
You can replace an OEM part without the original drawing
Gather the OEM part number, installed material, measured critical dimensions, plasma-facing surfaces and process chemistry. The zone and chemistry narrow the material candidates above; the dimensions and plasma-facing surfaces define the drawing.
From an engineering perspective, fix the zone and its electrical requirement first; the material candidates follow from the table.
Published erosion rates rank materials differently in fluorine and chlorine chemistries
Erosion rankings depend on plasma chemistry. In one source, 99.8% alumina eroded at about 0.88 times the 92% alumina rate in CF4 but about 3.1 times in CH4/Cl₂ (US 9,869,013) — the same purity comparison reverses direction depending on chemistry.
Erosion data as published, ratios within each study only
Source | Plasma type and gas mix | Reported metric and native unit | Value as published | Ratio to alumina, within this study only |
Trench etch, CF4/CHF3 | Relative erosion rate (Al₂O₃ = 1) | Quartz 2.2; SiC 1.6; ZrO₂ 0.8; Y₂O₃ 0.19 — grade not stated in source | As published, Al₂O₃ = 1 | |
CHF3/CF4 | Etch rate, nm/min (Å/min as published in brackets) | CVD SiC 52.6 (526); anodized Al₂O₃ 86.6 (866); spray-coated Al₂O₃ 74.1 (741); bulk Y₂O₃ 12.7 (127); plasma-sprayed Y₂O₃ 24.3 (243) | Anodized Al₂O₃ = 1: CVD SiC 0.61; spray-coated Al₂O₃ 0.86; bulk Y₂O₃ 0.15; plasma-sprayed Y₂O₃ 0.28 | |
CF4 dielectric etch; ion-assisted-deposition YAG coating | Etch rate, µm/RFhr | YAG coating about 0.28. Alumina purity values are in the purity table below | Not computed | |
Cl₂ 300 sccm + N₂ 100 sccm, 13 Pa (0.1 torr), 800 W ICP, 735 °C (1,355 °F) | Etch rate, µm/hr | AlN alone 4.8; AlN + 1.0 wt% MgO 2.3 | No alumina value in this study. AlN alone = 1: AlN + 1.0 wt% MgO 0.48 |
Grades are given where the source states them; US 10,622,194 does not state grades in the excerpt reviewed.
Compare ratios within one study, not absolute rates across studies. In US 8,067,067, CVD silicon carbide etched at about 0.61 times the anodized alumina rate in CHF3/CF4. Material-level properties are on the alumina and silicon carbide pages.
Why these numbers cannot be compared across sources
The sources report relative erosion, Å/min, µm per RF-hour and µm per hour, each under its own conditions. A ratio to alumina removes the unit, not the test condition, so it cannot transfer between studies.
Where yttria and YAG sit, and why we do not supply them
Yttria and YAG appear in the erosion table as comparison values only: about 0.19 times an ungraded alumina rate for yttria (US 10,622,194) and 0.28 µm/RFhr for a YAG coating (US 9,869,013). Our confirmed materials do not include yttria, YAG or yttria coatings.
Higher alumina purity does not improve erosion resistance in every chemistry
Alumina purity by plasma chemistry
Chemistry | 92% alumina | 99.8% alumina | Direction of effect |
CF4 dielectric etch, high bias | about 1.38 µm/RFhr | about 1.21 µm/RFhr | Lower erosion at 99.8% (0.88 times) |
H₂/NF3 | about 190 nm/RFhr | about 165 nm/RFhr | Lower erosion at 99.8% (0.87 times) |
CH4/Cl₂ | about 18 nm/RFhr | about 56 nm/RFhr | Higher erosion at 99.8% (3.1 times) |
Source: US 9,869,013. Units are as published and differ by row.
T3 shows the same purity comparison reversing by chemistry, from US 9,869,013. A fluorine-plasma study of polycrystalline alumina reports etch depths that "did not appear so different" across purities, and concludes that "etch depths might not properly represent the plasma resistance of the materials." The low-purity specimen also developed "loose structures around the pores," possibly producing contamination particles (Kim et al., J. Ceram. Soc. Jpn. 117(1368):863–867).
For procurement purposes, ask any supplier quoting erosion resistance to name the source study, gas mix and unit before two figures are compared.
Plasma-exposed chamber ceramics fail through particles, contamination and cracking, each with a different cause
Three failure routes are documented for plasma-exposed ceramics: aluminum fluoride particles formed on alumina in fluorine plasma, metal contamination leached from chamber components, and cracking from thermal-expansion mismatch. Each route has a separate cause, so a material change that reduces one does not necessarily reduce the others.
Plasma resistance is associated with porosity, roughness, defect density and mechanical properties (Lee et al., Sci. Rep. 11, 10288), so composition alone does not define behavior in service.
Fluorine plasma forms AlF₃ particles on alumina
Particles found in an etching chamber were identified as AlF₃, presumably formed when the Al₂O₃ coating on the chamber wall reacted with fluorine-atom plasma in the presence of H₂O; dehydrating the chamber parts suppressed particle generation (Ito et al., Jpn. J. Appl. Phys. 47, 3630). US 6,815,072 separately names Na, K, Fe, Cr, Ni and Mg as chamber contaminants.
Window and lid cracking comes from CTE mismatch and RF loss
Cracking in alumina windows is attributed to thermal-expansion (CTE) mismatch in US 6,797,110. RF loss is separate: US 11,998,986 says members increasingly need a loss tangent below 1×10⁻⁴, and names sodium oxides as a cause of loss.
Aluminum nitride's plasma resistance depends on grade and gas chemistry
In Cl₂/N₂ plasma at 800 W ICP, 13 Pa (0.1 torr) and 735 °C (1,355 °F), aluminum nitride alone eroded at 4.8 µm/hr and aluminum nitride with 1.0 wt% MgO at 2.3 µm/hr (US 6,239,402). Applied Materials' own patent states aluminum nitride "may be plasma resistant for oxygen or nitrogen based plasmas, but may not be plasma resistant for fluorine based plasmas" (US 10,177,023). Specify the AlN grade, additive and the chamber's gas chemistry on the drawing. See the electrostatic chuck guide and the aluminum nitride page for more.
From an engineering perspective, identify the active failure route (particles, contamination or cracking) before changing the material.
What your drawing must specify before a plasma-facing ceramic part can be quoted
A plasma-facing ceramic part is quotable when the drawing identifies which surfaces see the plasma, the required purity and density, the surface finish and edge break, the dielectric loss for windows, and the cleanliness and inspection evidence needed. The values belong to the buyer's drawing, not to a supplier default.
Drawing callout checklist
Parameter to call out | Why it matters in plasma | Callable test / standard | Who sets the value |
Plasma-facing surfaces | Porosity, roughness and defect density are associated with plasma resistance (Lee et al.) | Marked on the drawing | Your drawing |
Flatness / parallelism | Fit parameter; not addressed in the cited erosion sources | Method stated on the drawing | Your drawing |
Surface finish | Roughness is associated with plasma resistance (Lee et al.) | Method stated on the drawing | Your drawing |
Edge break / chamfer | Fit and handling parameter; not addressed in the cited erosion sources | Method stated on the drawing | Your drawing |
Purity ceiling | Alumina purity changes erosion by chemistry (T3); sodium oxides raise dielectric loss (US 11,998,986) | Trace-metal method stated on the drawing | Your drawing |
Density / porosity | Pores in low-purity alumina loosened under fluorine plasma and can produce particles (Kim et al.) | Method stated on the drawing | Your drawing |
Dielectric loss (windows only) | Low loss tangent is the specified window property; below 1×10⁻⁴ in US 11,998,986 | ASTM D150 | Your drawing |
Cleanliness and packaging | Particles from chamber components are a documented contamination route (Ito et al.) | SEMI E194 | Your drawing |
Inspection evidence required | Ties reported values to the measured part | ASTM D149 (dielectric strength); ASTM C1327 (hardness), as applicable | Your drawing |
The table lists each callout with a test method the drawing can name: ASTM D149 for dielectric strength, ASTM D150 for permittivity and dissipation factor, ASTM C1327 for Vickers hardness, and SEMI E194 for assessing particulate contamination on chamber components with a liquid particle counter. Purity and trace-metal specification is covered in the high-purity ceramics guide.
Forming and finishing routes that change a plasma-facing surface
These parts are formed by pressing, injection molding or sintering, which set density and porosity, and finished by grinding of fired ceramic, which sets surface finish and edge condition. Because porosity, roughness and defect density are associated with plasma resistance (Lee et al.), name the route that produces each plasma-facing surface next to its finish requirement on the drawing.
Lifetime claims are only comparable with matching test conditions
Lifetime claims arrive as RF-hours, wafer counts or maintenance intervals. Ask for the unit, chemistry, power, bias, chamber type and failure criterion behind each. US 9,869,013 reports erosion per RF-hour, which removes exposure time but not chemistry or bias.
For procurement purposes, a drawing that carries these callouts allows a quote against stated requirements instead of a supplier's assumptions. State your target quantity — prototype or production — on the same drawing submission. Quote terms, cost, lead time, prototype handling and the documentation shipped with the parts are all assessed against what you submit; we do not publish a standing prototype, minimum-order or lead-time policy separate from a specific request.
We provide US-based engineering, application support and quality assurance from Hayward, California, with drawing review by a named materials engineer. Values on this page come from the patents and papers listed under Sources; they are not our test results.
Mark Ma, Lead Materials Engineer
Comments