Why Might a Thinner PFA Sheath Actually Perform Worse in a Low-Pressure Plasma Environment Due to Ion Bombardment?

Low-pressure plasma settings used for semiconductor etching, plasma-enhanced chemical vapour deposition (PECVD), and surface activation immerse components in energetic ion bombardment. Ions (Ar + , O + , CF 3 + , etc) are driven across the plasma sheath potential, generally 100-1,000 V, and contact the PFA surface with energies of 50-500 eV per ion. Thinner PFA sheaths (0.5–1.0 mm) are less suitable for plasma service than thicker sheaths (1.5–2.5 mm) because the ion bombardment erodes the polymer surface layer by layer by physical sputtering and chemical etching. The thinner wall has less material before it gets to the metal core. More importantly, ion bombardment forms a changed surface layer (10-100 nm deep) with altered chemical and physical characteristics. This changed layer is often more permeable to reactive species, speeding up deterioration. Thicker PFA (2–3 mm) survives longer in low-pressure plasma, as the erosion rate is modest relative to thickness (usually 0.1–1.0 µm/hr, depending on plasma settings) and the changed surface layer does not reach through the wall. Sheaths of 1.0 mm may fail after 500–1,000 plasma hours, whereas 2.5 mm sheaths may sustain 5,000–10,000 hours.

Erosion Processes by Ion Bombardment
In low-pressure plasma (0.1-10 Torr) the PFA heater develops a negative self-bias voltage with respect to the plasma potential, because electrons are more mobile than ions. The self-bias (usually 50-500 V for RF plasmas, 100-1000 V for DC plasmas) pushes positive ions towards the PFA surface . The ions transfer momentum to the polymer upon collision, ejecting atoms and molecules, in a process termed physical sputtering. The PFA sputter yield (atoms removed per incident ion) in argon plasma is 0.5-2.0 at ion energies of 200-500 eV. The erosion rate e = Y × J_ion × (M / (ρ × N_A)) where Y is sputter yield, J_ion is ion current density (mA/cm²), M is molecular weight, ρ is density. For a typical RF plasma (13.56 MHz, 100 W, 100 mTorr Ar), J_ion ~ 1-5 mA/cm^2, Y ~ 1.0, resulting in e ~ 0.2-1.0 µm/hr. Erosion in 1000 hr. 0.2-1.0 mm The 1.0 mm sheath loses 20–100% of its thickness, but the 2.5 mm sheath loses only 8–40%.

In the presence of reactive gases (O₂, CF₄, SF₆, Cl₂) chemical etching is added to the physical sputtering. For example, the PFA surface reacts with oxygen plasma to generate volatile COF₂, CO₂ and HF. Chemical etch rate can be 2-10x higher than physical sputtering. A 1.0 mm PFA sheath can be eroded entirely in 200-500 hr in 50% O2/50% Ar plasma. The erosion is not uniform. Sharp corners erode quickly (see Article 62), and parts that face the plasma glow erode more quickly than shadowed locations. PFA wall becomes 0.2-0.5 mm thick, pinholes are formed and plasma species approach the metal core. The core is exposed to reactive plasma and corrodes rapidly (Incoloy creates volatile metal fluorides in fluorine based plasmas) causing contamination and heater failure.

Performance vs. Thickness in Plasma
Plasma Type Gas Composition Ion Energy (eV) Erosion Rate (micron/hour)Max Hours to 0.5 mm Remaining (beginning 1.0mm) Max Hours to 0.5 mm Remaining (starting 2.5 mm)Recommended Min Thickness Inert (sputtering only) Ar, He 200–500 0.2–0.5 1,000–2,500 4,000–10,000 1.5 mm
Oxidising (chemical + sputter) O2, O2/Ar 100-300 0.5-2.0 250-1,000 1,000-4,000 2.0 mm
Fluorocarbon (etching) CF4, CHF3, C4F8 200–500 1.0–5.0 100–500 400–2000 2.5 mm
Chlorine Cl₂, BCl₃ 100–300 0.3–1.0 500–1,500 2,000–6,000 2.0 mm
High-density plasma (ICP) Any 500–1,000 2.0–10.0 50–250 200–1,000 3.0 mm (if PFA used)
Plasma w/ ion beam assist Any 1,000–5,000 5.0–20.0 25–100 100–500 Not appropriate for PFA
Atmospheric plasma (low energy) Air, N₂ <50 <0.05 >10,000 >50,000 1.0 mm acceptable
Permeation Enhancement and Surface Modification
Ion bombardment does more than just sputter material away, it alters the surviving near-surface region. Ions break C-F bonds and generate free radicals that react with ambient gases. The changed layer (usually 10-100 nm thick) has a decreased fluorine concentration (fluorine depletion), a greater oxygen content (from reactivity with residual O2 or H2O) and a higher density of cross-links. The changed layer is usually more permeable to reactive species (e.g., atomic oxygen, fluorine radicals) than virgin PFA. Improved permeability speeds up degradation underneath the surface. For a thin sheath (1.0 mm) the modified layer is 10–20% of the wall thickness after 500 h, while for a thick sheath (2.5 mm) it is only 4–8%. The unaffected part of the thin sheath that remains may be too thin to constitute an effective barrier. The thin layer may become porous or brittle and allow plasma species to penetrate to the core, even without major erosion.

Field experience with semiconductor plasma etch tools indicates that PFA heaters with 1.0 to 1.5 mm walls used as in-situ heaters (immersed in the plasma) fail after 6 to 12 months. The same heaters with 2.5–3.0 mm walls will last 3–5 years. Failure is not catastrophic but incremental . Surface roughening , then pinhole formation , then metal core fluorination ( as demonstrated by purple or black discolouration of the Incoloy core . The metal fluorides are non-conductive and may lead to open circuit failure or spall off and contaminate wafers.

Recommendations for Plasma Service Design
Select the thickest PFA sheath that can be used for low-pressure plasma conditions, generally 2.5 to 3.0 mm. Thicker walls mean a longer erosion lifespan and a longer diffusion pathway for reactive substances. Trade is less heat transmission ( ΔT across sheath increases from 1.5 mm to 3.0 mm by 30 - 50 % , needing higher core temperature or larger surface area ) . For high power plasmas (>500 W, >500 eV ion energy) PFA may not be acceptable irrespective of the thickness and other materials (quartz, alumina, or silicon carbide) should be investigated. 1.5-2.0 mm PFA is appropriate for the plasma with ion energies < 100 eV (e.g. downstream plasma, remote plasma).

Additional precautionary measures: (1) Place a grounded metallic barrier (with holes) around the heater to capture energetic ions before they reach the PFA. The shield must not substantially interfere with heat transfer. (2) Rotate the heater to make the surface parallel to the electric field lines, thus reducing the effective ion flow. (3) Use a greater molecular weight (lower MFI, see Article #37) grade of PFA with improved erosion resistance. (4) For fluorocarbon plasmas, consider covering the PFA with a thin (0.1-0.2 mm) layer of PTFE or FEP, which has a lower sputter yield than PFA (because of a higher fluorine concentration). No fluoropolymer is resistant to oxygen plasmas. Use quartz or metal.

Conclusion: Thicker PFA Works Better in Plasma
In low pressure plasma conditions a thinner PFA coating (0.5-1.5 mm) actually performs worse than a thicker sheath (2.5-3.0 mm) because of ion bombardment erosion (0.1-10 µm/hour) and surface alteration that increases penetration. The thinner wall exposes the metal core earlier and the changed layer represents a bigger fraction of the remaining wall. The expected life of a standard plasma etch sheath (200-500 eV ion energy, 0.5-2.0 um/hr erosion rate) is 500-2,000 hours for a 1.0 mm sheath and 2,000-10,000 hours for a 2.5 mm sheath, 4-5 times longer. Engineers designing PFA heaters for plasma chambers must utilise the thickest practicable wall (2.5 to 3.0 mm) and accept the lower heat transfer or greater heater size. Thinner barriers are a false economy. The intuitive idea that 'less material = less deterioration' is false in plasma. It is driven by erosion, not by permeation or heat stress. Erosion takes material away from the outside in. More material means that it takes longer until the metal core is exposed. For plasma service in semiconductor fabs, 2.5 mm minimum PFA is standard. The 1.5 mm heater is deemed consumable and appropriate only for short term or low power plasma applications. Specify thickness with plasma-specific derating parameters. Verify wall thickness on receiving inspection. In plasma, thicker is not better, it is required for reliable service.