Vacuum Chamber Machining: Billet Quality, Sealing Faces and Virtual Leaks
A vacuum chamber is judged by a number nobody can see: the leak rate at the flange, measured in mbar·L/s of helium, hours after the part was machined and cleaned. A chamber that holds 1 × 10⁻⁶ mbar·L/s will reach 10⁻³ mbar with a rotary pump. One that holds 1 × 10⁻⁹ mbar·L/s can be baked and reach 10⁻⁹ mbar. The difference between those two parts is rarely the design. It is the billet, the blind holes, and how flat the sealing face was when it left the machine.
If you are buying vacuum chamber machining for a deposition tool, an electron microscope column, or a load lock, the drawing will specify material, flange types and a surface finish. It will probably not specify billet inspection, virtual leak paths, or the order in which faces are cut after welding. Those are the things that decide whether the chamber passes its helium test the first time.
Billet, plate or bar
Chamber bodies up to about 400 mm are usually cut from 6061-T651 plate. Plate is preferred over extruded bar because extrusion leaves a seam and a coarser grain, and both open up when you machine a deep pocket across them. A 100 mm thick 6061-T651 plate is roughly 10 to 15 percent more expensive per kilogram than bar and worth it for any part that will be helium tested.
Porosity is the failure you cannot machine out. A void 0.3 mm across that intersects a sealing face becomes a leak path under the O-ring, and it will not show up until the chamber is assembled and pumped. Ask for ultrasonic inspection of the blank before cutting, and specify that material within 10 mm of any sealing face be free of voids larger than 0.5 mm. For UHV parts, use vacuum-degassed 6061 or switch to 316L stainless, which is weldable, bakeable to 250 °C, and has no porosity issue in rolled plate.
Wall thickness drives distortion more than alloy choice. A 6 mm wall on a 300 mm chamber moves 0.1 to 0.3 mm after deep pocketing, so rough with 1.5 mm of stock and relieve at 180 °C for 4 hours.
Sealing faces: flatness before roughness
The most common mistake on these drawings is a tight Ra with no flatness callout. An O-ring seals on a contact band about 0.5 mm wide. If the face is wavy by 0.05 mm across a 100 mm flange, the O-ring cannot follow the wave, and no amount of polishing to Ra 0.4 µm fixes it. Flatness comes first, roughness second.
| Flange type | Seal | Face flatness | Surface finish | Groove or edge detail |
|---|---|---|---|---|
| ISO-KF (NW25, NW40) | Elastomer O-ring on centering ring | 0.05 mm across face | Ra 1.6 µm | Flat face, no groove; O-ring sits in the centering ring |
| ISO-K (ISO 63, ISO 100) | Elastomer O-ring in groove | 0.05 mm, groove depth ±0.05 | Ra 1.6 µm groove, Ra 3.2 elsewhere | Groove 5.33 mm wide × 3.80 mm deep for 5.33 mm cord |
| CF (DN 63 CF) | Oxygen-free copper gasket, knife edge | 0.025 mm on the knife edge plane | Ra 0.8 µm on the 70° knife edge | Edge land 0.2 mm, no tool marks across the sealing line |
| Baked UHV door | Metal C-ring | 0.02 mm | Ra 0.4 µm on the flat | 0.8 mm flat land, no interrupted cuts |
| Weld flange | None, welded | 0.1 mm before weld prep | Ra 3.2 µm | 60° included bevel, 1.5 mm land |
Two practical notes. The CF knife edge is cut in one continuous pass with a sharp, positive-geometry tool; a tool change mid-cut leaves a step a copper gasket will not bridge. And measure flatness on the bench at 20 °C, not on the machine — a 150 mm aluminum flange reads differently once the clamps come off.
On a DN 63 CF flange, six M8 holes sit on a 130 mm bolt circle. Keep 4 mm of material between each hole and the knife edge relief, or the flange will never seal however flat the face is.
See also: Drone Frame Machining: Billet Center Plates, Ribs and Motor Patterns and Battery Housing Machining: Gasket Grooves, Flatness and Leak Test Prep.
Virtual leaks and the holes you cannot see
A virtual leak is trapped gas that bleeds out slowly, making a perfect chamber look like it has a real leak. Blind tapped holes are the usual culprit. An M6 hole 12 mm deep with a screw 8 mm into it traps about 0.07 cm³ of air at the bottom, and that volume bleeds through the thread clearance for days.
The fixes are simple and worth building into the drawing. Drill a 0.5 mm vent hole from the bottom of the blind hole through to a region that will be pumped, or use a through-hole with a sealing washer on the atmosphere side. On internal assemblies, avoid blind holes altogether: a through-hole with a screw and a captured nut is easier to pump than a blind hole with a vent. For slots and pockets, add a 1 mm drain to the nearest pumped volume.
The same logic applies to any two plates bolted together with a gasket. Air trapped between them cannot escape and shows up in a rate-of-rise test as a slow, endless climb.
Machining after welding
Most chamber bodies are welded from plate or tube, and welding moves the flanges. A 300 mm 6061 body with four welded ports typically distorts 0.2 to 0.4 mm at the flange faces, and 316L moves less in magnitude but more unpredictably because of shrink on the last pass. Leave 1.0 mm on all flange faces and port bores after welding, then machine the datum face, all sealing faces, and the port bores in one setup on a 5-axis machining center.
Weld prep should be cut, not ground. A 60° included bevel with a 1.5 mm land gives a consistent root and keeps the weld from pulling the flange. Remove oxide and cutting fluid from the bevel within 8 hours of welding; 6061 oxidizes fast enough to affect arc stability.
Port threads on a chamber are usually the last operation and the one most likely to leak. Cut them with a thread mill rather than a tap so the crest is clean and the thread does not tear, and keep a 1.5 mm counterbore at the entrance for an O-ring or bonded washer.
Cleaning before bagging
A machined chamber that is rinsed with tap water and bagged will outgas for weeks. The sequence that works: alkaline degrease at 60 °C, ultrasonic clean at 40 kHz for 15 minutes, DI water rinse at 18 MΩ·cm, nitrogen dry, then bag in an ISO Class 7 area with all ports capped.
Do not anodize a chamber that will see UHV. Anodic coating is porous, holds water, and outgasses; if you need corrosion resistance on the outside, mask every sealing face and every internal surface. Specify that no silicone-based lubricant touches the part.
Frequently asked questions
What leak rate should I specify for a chamber?
Specify the rate at the test condition. For rough vacuum, 1 × 10⁻⁶ mbar·L/s helium is typical. For UHV, 1 × 10⁻⁹ mbar·L/s after a 24-hour bake at 150 °C. A shop that quotes without asking which one you need is guessing.
Should the chamber be machined before or after welding?
Weld first, then machine, with 1.0 mm of stock left on every sealing face and port bore. Welding moves flanges 0.2 to 0.4 mm on aluminum, which is far more than the flatness tolerance, so any face finished before welding has to be recut.
Why does my chamber show a leak that helium testing cannot find?
That is the signature of a virtual leak. Check every blind tapped hole for a vent path, and check any joint where two sealed surfaces trap a volume. A rate-of-rise test will show a slow climb that never reaches a stable pressure, which is different from a real leak.
Can 6061 be used for a baked UHV chamber?
It can, but only with vacuum-degassed plate, a proper cleaning sequence, and no anodize on internal surfaces. Most UHV chambers that bake above 150 °C are 316L or 304L. Above 200 °C, specify stainless.
Related reading
Send a drawing for a quote and DFM review.
STEP, IGES, X_T, PDF or DWG up to 50 MB. Tell us quantity and finish, and the quote comes back with tolerance and process notes, not just a price.