UHV starts at the material certificate
I have scrapped more UHV flanges to material issues than to bad machining. A helium leak test at 1e-9 atm-cc/s that should have passed came back marginal because the 316L bar had elevated sulfur content — it machined beautifully, but the high-sulfur inclusions created micro-porosity at weld heat-affected zones and produced an inconsistent electropolished surface. For UHV work below 1e-10 mbar, specify 316L VAR (vacuum arc remelted) or 316L ESR with sulfur below 0.005%. The extra material cost is trivial compared to a rejected assembly.
Aluminum 6061-T6 works for rough and high vacuum above 1e-7 mbar but outgasses more and cannot be baked above 120 C. 304L is an option for larger HV chambers; 316L is standard for CF knife edges and bake-out above 150 C. OFHC copper is used for gaskets and specialty components but must stay annealed and scratch-free.
- 316L VAR/ESR: default for UHV below 1e-9 mbar, low sulfur for consistent electropolishing
- 304L: acceptable for HV chambers above 1e-7 mbar, watch carbon content for weld corrosion
- 6061-T6 aluminum: rough/high vacuum only, limit bake-out temperature
- OFHC copper: gasket material and heat-sink components, keep annealed and scratch-free
Surface finish is a vacuum requirement, not cosmetics
Surface roughness directly determines leak rate and outgassing. On a CF knife-edge seal, the copper gasket plastically deforms into the knife edge. A single circumferential feed mark 0.02 mm deep across the knife edge creates a leak channel that shows up on a helium leak detector — I found this out on a batch of 24 flanges where one insert had a slight wear flat and produced micro-grooves just barely visible under 10x magnification. The knife-edge finish needs to be Ra 0.8 or better with no circumferential marks.
For internal surfaces facing vacuum, mechanical polish to Ra 0.4 followed by electropolishing removes the Beilby layer — the micro-cracked, smeared metal left by machining — and reduces real surface area by a factor of two to three. Faster pump-down, lower ultimate pressure, and less virtual leak from trapped contaminants.
Electropolishing and the cleanliness chain
Electropolishing removes 20-50 microns of surface material electrochemically, stripping away embedded iron particles, tool steel smearing, and grinding residue. It leaves a uniform chromium-rich passive layer. A mechanically polished surface can look shiny but still have folded-over metal flaps that trap contamination. An electropolished surface exposes the true grain structure.
Post-EP handling is where shops lose the game. Once a UHV component leaves the EP tank, it goes through DI water rinses (resistivity above 15 megaohm-cm), nitrogen blow-dry, and double-bagging in clean polyethylene. One fingerprint on a sealing surface deposits enough skin oil to create a virtual leak that requires hours of bake-out to remove. Final assembly should happen in a clean area with nitrile gloves. Tapped holes must be solvent-flushed to remove residual cutting fluid.
- Target Ra 0.2-0.4 on electropolished internal surfaces; Ra 0.8 minimum on CF knife edges
- DI rinse >15 megaohm-cm, nitrogen dry, double-bag immediately after final clean
- No silicone-based lubricants anywhere in the process chain — silicone outgasses for weeks
- Label every bag with part number, lot, date, and clean-process status
CF knife edges: geometry that does not forgive
ConFlat knife edges have a 20-degree included angle (70 degrees from horizontal per side) with a flat land of roughly 0.1-0.2 mm at the tip. The edge must be concentric to the bolt circle within 0.05 mm. A nick or ding of 0.05 mm on the knife edge prevents the copper gasket from fully seating, creating an intermittent, temperature-dependent leak. I use a form tool ground to the exact angle for the finish pass, at 0.05 mm depth of cut and slow feed, with the part staying fixtured between rough and finish to preserve concentricity.
Virtual leaks: the hidden enemy
A virtual leak is a trapped gas volume connected to vacuum by a small diffusion path. It behaves like a real leak but cannot be fixed by tightening bolts. Blind tapped holes are the most common source — cross-drill a vent hole to the thread root, or use through-bolts. Welds must be full penetration ground flush internally; double O-ring designs need inter-seal vent holes.
| Feature | UHV solution | Reason |
|---|---|---|
| Blind tapped holes | Cross-drilled vent to thread root | Eliminates trapped gas volume |
| Welds | Full penetration, ground flush, internal EP | No crevices for gas trapping |
| Double O-ring grooves | Vent hole between grooves | Pump the inter-seal volume |
| Bolted joints | Through-bolts preferred over blind taps | No thread volume facing vacuum |
| Castings | Avoid; use wrought bar or forging | Castings contain internal porosity |
Tolerance and inspection reality check
UHV components are not exotic-tolerance parts. CF knife-edge flatness needs to be within 0.02 mm across the seal, bolt holes at +/-0.1 mm true position, bore-to-knife-edge concentricity at 0.05 mm. We hold +/-0.005 mm on critical dimensions using Zeiss CMM inspection, with Cpk reports on production runs and free FAI reports on new parts. What separates competent UHV shops from the rest is process discipline: material verification, clean handling, EP quality control, and vent holes that get designed in from the start rather than added after a leak-test failure.
Frequently Asked Questions
What surface finish do I need on CF knife-edge seals?
Ra 0.8 um or better with no circumferential feed marks. Axial marks (along the bolt circle) are less harmful because they do not cross the seal. Circumferential marks create continuous leak channels across the copper gasket.
Can 304 stainless be used for UHV instead of 316L?
304L works for HV above 1e-7 mbar and for components not aggressively baked. For UHV below 1e-9 mbar, 316L (preferably VAR or ESR with low sulfur) is standard due to better corrosion resistance, lower magnetic permeability after welding, and more consistent electropolish results.
Why do blind tapped holes cause virtual leaks?
The void at the bottom of a blind thread traps gas that diffuses slowly out through thread clearances, mimicking a real leak. Cross-drill vent holes to the thread root, or use through-bolts wherever possible.
Is electropolishing required for UHV?
For HV above 1e-7 mbar, a good mechanical polish to Ra 0.8 may suffice. For UHV below 1e-9 mbar, electropolishing is standard because it removes the machining-damaged surface layer, reduces real surface area, and produces a uniform passive layer that minimizes outgassing.
What tolerance should CF flange bolt holes be held to?
+/-0.1 mm true position relative to the bore and knife edge, per ISO 3669 / ConFlat standards. Over-tolerance holes cause uneven gasket compression and leaks.