UHV Bakeout & Outgassing Guide

Ultra-high vacuum systems succeed or fail based on two processes that happen long before an experiment begins: bakeout and outgassing control. Whether the chamber houses a trapped-ion quantum processor, a satellite payload in a thermal vacuum (TVAC) test, or a directed-energy test stand, the same physics governs how quickly a system reaches base pressure and how stable that pressure stays over years of service. This guide covers what is ultra-high vacuum and why it depends on bakeout, what happens at the molecular level inside metals and ceramics during outgassing, a repeatable bakeout workflow, and the temperature limits that keep UHV components (including ceramic-to-metal feedthroughs, viewports, and connectors) intact through the process.

Why Bakeout and Outgassing Control Matter

Even a chamber with a leak rate as low as 1×10⁻⁹ atm cc/sec can stall well above the UHV band, which spans roughly 1×10⁻⁹ to 1×10⁻¹² Torr, if its interior surfaces are still shedding water vapor and hydrocarbons — leak rate and pressure are different quantities, and a small leak doesn’t by itself guarantee low base pressure, since its actual contribution depends on the system’s pump speed. A single component leaking at that rate typically contributes only a small fraction of a system’s overall gas load once pump speed is factored in; outgassing, not leaks, is usually the dominant gas load in a freshly assembled system. Bakeout accelerates the desorption of adsorbed water and surface contaminants by raising the chamber and its internal components to an UHV Bakeout & Outgassing Guide for MPF Products Components

Ultra-high vacuum (UHV) systems succeed or fail based on two processes that happen long before an experiment begins: bakeout and outgassing control. Whether the chamber houses a trapped-ion quantum processor, a satellite payload in a thermal vacuum (TVAC) test, or a directed-energy test stand, the same physics governs how quickly a system reaches base pressure and how stable that pressure stays over years of service.

This guide covers why ultra-high vacuum depends on bakeout, what happens at the molecular level inside metals and ceramics during outgassing, a repeatable bakeout workflow, and the temperature limits that keep UHV components (including ceramic-to-metal feedthroughs, viewports, and connectors) intact through the process.

Why Bakeout and Outgassing Control Matter

The UHV band spans roughly 1×10⁻⁹ to 1×10⁻¹² Torr. A freshly assembled chamber can be leak-tight to 1×10⁻⁹ atm cc/sec and still stall well above that band, because its interior surfaces are still shedding water vapor and hydrocarbons. Leak rate and pressure are different quantities: a leak raises pressure by its gas flow divided by the pump speed. A 1×10⁻⁹ atm cc/sec leak is about 7.6×10⁻¹⁰ Torr·L/s, which adds less than 1×10⁻¹¹ Torr on a 100 L/s pump. In a freshly assembled system, outgassing from the walls, not leaks, is usually the dominant gas load.

Bakeout accelerates the desorption of adsorbed water and surface contaminants by raising the chamber and its internal components to an elevated temperature for a sustained period while actively pumping. The effect is large: Edwards Vacuum reports that a 30-hour bake at 250°C reduced stainless steel outgassing by a factor of more than 70,000, compared with a factor of 30 to 50 for electropolishing or mechanical polishing alone. A good bake lets the system settle into a stable UHV state instead of drifting for weeks after pump-down. Reaching extreme-high vacuum (XHV) usually also takes low-hydrogen materials and getter pumping.

The Physics of Outgassing in Metals and Ceramics

Two distinct outgassing mechanisms dominate UHV hardware: surface desorption and bulk diffusion. Water vapor adsorbed on stainless steel and ceramic surfaces desorbs readily once heated, so pressure rises sharply in the first hours of a bake as that water is released and pumped away. This rise is expected. The large drop in base pressure appears after the system cools, once most of the adsorbed water is gone.

Hydrogen dissolved in the bulk metal is a slower problem. Stainless steel absorbs hydrogen during manufacturing (melting, rolling, welding), and that hydrogen diffuses to the surface and releases over time at a rate that depends on temperature and material thickness. After a thorough bake, hydrogen is usually the largest remaining gas load. Reducing the hydrogen inventory takes heat. Long-term UHV and XHV systems benefit from vacuum-fired stainless steel (typically 950°C to 1050°C in a vacuum furnace), which drives hydrogen out of the bulk before the part ever reaches the chamber. Full vacuum firing gives the most complete hydrogen removal, but medium-temperature treatments in the 250°C to 650°C range, including air bakes, have also demonstrated meaningful, if less complete, outgassing reduction, and are sometimes used instead when the alloy or part geometry makes a full 950°C-plus fire impractical. Vacuum firing is done on bare stainless parts before any brazed components are attached, because firing temperatures are well above the 750°C to 850°C range at which ceramic-to-metal assemblies are typically brazed.

Electropolishing works differently. It smooths the surface and leaves a chromium-rich oxide layer, which reduces the area available for water to adsorb, but it does not remove hydrogen from the bulk metal.

Ceramics behave differently again. Alumina (aluminum oxide) used in ceramic-to-metal feedthroughs is far less permeable to hydrogen than stainless steel and outgasses primarily through surface-adsorbed water rather than bulk diffusion. The alumina is joined to the metal shell in one of two ways: molybdenum-manganese (Mo-Mn) metallization followed by brazing, most commonly with a 72/28 silver-copper eutectic filler, or active metal brazing, in which a titanium-bearing alloy wets the ceramic directly. Both the ceramic and its braze joints have thermal limits. Exceeding a feedthrough’s rated bake temperature, or heating it too quickly, introduces differential thermal expansion stress between the ceramic and the metal, a common cause of hairline leaks that appear only after multiple thermal cycles rather than immediately after bakeout.

A Practical Step-by-Step Bakeout Workflow

Pre-clean every component. Solvent-clean all parts. Where practical, vacuum-fire bare stainless chamber parts and flanges before any brazed feedthroughs or viewports are attached, to reduce hydrocarbon and hydrogen load before bakeout even starts.
Assemble under clean conditions. Handle CF and KF hardware with gloves. Tighten CF flange bolts in a star pattern in small increments so the copper gasket seats evenly, and use silver-plated bolts or a vacuum-compatible anti-seize to prevent galling. Vent any blind holes or unvented fasteners now, before the chamber is sealed. A trapped volume left here becomes a virtual leak that’s difficult to distinguish from a real one later; see our guide to vacuum feedthrough failure modes for how to diagnose one if it shows up during leak-checking.
Pump down and leak-check before heating. Bring the chamber down with a dry roughing pump, then switch to turbo pumping. Once the system reaches high vacuum, helium leak-check every flange, weld, and feedthrough. Baking does not repair a real leak, and heating a leaking system wastes the bake. Record the starting pressure and, where available, an RGA scan as a baseline.
Prepare the system for heat. Remove or protect anything that can’t take the bake temperature: gauge and RGA electronics, air-side connectors and cables, motors, and ion pump magnets if the bake will exceed their rating. Keep the turbo pump’s inlet flange within the manufacturer’s limit, commonly around 100°C to 120°C. Wrap the chamber so it heats evenly, cover viewports with several layers of aluminum foil, and place thermocouples at the coldest and most temperature-sensitive points, including viewports and feedthroughs. Set an over-temperature interlock. Water released from hot surfaces re-adsorbs on any cold spot, so every part of the vacuum envelope needs to reach temperature together.
Ramp temperature gradually. 1°C to 2°C per minute is a conservative whole-system ramp rate. Individual components may be rated faster (MPF Products viewports are rated for ramp rates up to 25°C per minute, but large chambers develop temperature gradients during heating, and the slowest-rated component in the assembly sets the limit.
Hold at target temperature. Maintain the bake temperature for 24 to 48 hours, depending on chamber size and prior history, while continuously pumping.
Monitor pressure and RGA data. Track total pressure and, where available, residual gas analyzer peaks: mass 18 for water, 28 for CO and N₂, 44 for CO₂, and mass 2 for hydrogen, which becomes more prominent later in the bake. Hydrocarbon contamination such as pump oil appears as a pattern of peaks at masses 39, 41, 43, 55, and 57. Mass 28 alone is ambiguous between CO and N₂. A real air leak also shows argon at mass 40, and usually oxygen at mass 32 at roughly one-quarter the height of the mass 28 peak, although the oxygen peak can be weak or missing in a UHV system. The bake is nearing completion when pressure at temperature has stopped falling.
Degas gauges and activate getters while hot. Near the end of the hold, degas ionization gauge, RGA, and titanium sublimation pump filaments, and activate any non-evaporable getter (NEG) pumps, so their own outgassing is pumped away before cooldown. If the ion pump runs during the bake, keep the temperature within its high-voltage cable rating.
Ramp down slowly. Cool the chamber at the same controlled rate used for ramp-up, again to protect brazed joints and windows from thermal shock.
Re-torque flanges after cooldown. Thermal cycling can relax bolt tension, so check and re-torque critical seals once the system returns to room temperature.
Confirm final leak rate and base pressure. Perform a final helium leak check and log the base pressure and an RGA scan as a baseline for future maintenance.

Component Bake Temperature Table

Different UHV components tolerate different maximum bakeout temperatures, and exceeding these limits is one of the most common causes of avoidable damage during commissioning. The values below are representative of typical conditions; always check the rating of the specific parts you are working with.

Component	Typical Max Bakeout Temp	Notes
Stainless steel chamber body	250°C typical; up to 450°C	The steel itself is rarely the limit. Seals and attached components are. With all-metal seals and rated components, bakes up to 450°C are possible.
CF flanges with OFHC copper gaskets	450°C	Copper gaskets are single-use once compressed.
Aluminum CF flanges (CrN-coated knife edges, aluminum gaskets)	200°C	Limited by the knife-edge coating. Check the flange manufacturer’s rating. MPF rates aluminum feedthroughs to 200°C.
KF elastomer-sealed flanges	150°C	Elastomer o-rings limit temperature. Check the o-ring material.
Alumina ceramic-to-metal feedthroughs	450°C (MPF Products rating)	Ratings vary by manufacturer and materials. Check the part’s data sheet.
Glass-ceramic sealed feedthroughs and connectors	250°C to 400°C	Depends on materials used. Check with the manufacturer.
Fused silica UHV viewports	200°C	MPF Products rating for fused silica seals. Follow the rated ramp rate.
Sapphire UHV viewports	450°C; 300°C if AR coated	MPF Products rating for sapphire seals. Follow the rated ramp rate.
Nano-etched viewports	200°C
CaF₂, MgF₂, ZnSe, crystal quartz, and germanium viewports	Typically 200°C. Lower than sapphire; check the part	Crystalline UV and IR window materials are often rated well below fused silica and sapphire. Some vendors rate CaF₂ viewports to 120°C.
UHV electrical connectors	Vacuum side: 200°C to 450°C. Air side: 125°C	Depends on materials used. Check insulator and pin material ratings. Remove air-side mating connectors and cables before baking.
Ion pumps	About 250°C to 350°C with magnets, depending on model; about 450°C with magnets removed	If the pump runs during the bake, its high-voltage cable rating can be the lower limit. Check the manufacturer’s data.
Turbo pump inlet flange	About 100°C to 120°C	Check the manufacturer’s limit.

Synchrotron UHV beamlines and XHV systems often push toward the higher end of these ranges, while quantum computing chambers with in-vacuum optics or wiring may be constrained to the lower end to protect sensitive components.

Where This Shows Up in the Field

Trapped-ion quantum computing chambers combine viewports, RF feedthroughs, and getter pumps in a single small volume, so bakeout planning has to respect the lowest-rated component in the assembly rather than the chamber body’s limit. In one published trapped-ion build, the final bake was limited to about 200°C by the optical viewports.

Satellite TVAC programs bake flight hardware to reduce its outgassing to levels the mission can tolerate, following practices such as ASTM E2900, which defines bakeout methods based on time and pressure or on quartz crystal microbalance (QCM) measurements. Hardware bakes often run far below UHV component limits (NASA box bakeouts commonly run around 80°C to 100°C), and the chamber itself is certified clean with a QCM before the hardware bake.

Directed-energy test stands run at high duty cycle and benefit from a thorough bake, since residual outgassing can show up as beam contamination during testing. Electric propulsion test facilities are a different case: they operate in high vacuum, around 10⁻⁵ to 10⁻⁶ Torr while a thruster fires, and rely on high-speed cryopumping rather than UHV bakeout.

Synchrotron light sources, with beam paths of hundreds of meters to kilometers and hundreds to thousands of CF joints, rely on in-situ bakes in the 150°C to 300°C range, often combined with NEG coatings. Bakes are staged so that one under-baked section does not become the limiting gas load for its part of the ring. For more, see vacuum components for synchrotrons and particle accelerators.

Bakeout-Day Checklist

Use the workflow above when building a bakeout profile from scratch. Keep this condensed version on hand as a printable quick reference for the bakeout day itself, once the temperature profile, hold time, and flange torque are set.

Confirm bare stainless parts were cleaned and, where specified, vacuum-fired before any brazed components were attached.
Confirm every component’s rated bake temperature and ramp rate, and set the profile by the lowest-rated part.
Torque flanges in a star pattern with calibrated tools.
Helium leak-check the system at base pressure before heating.
Remove or protect non-bakeable items, wrap the chamber for even heating, foil the viewports, and place thermocouples and an over-temperature interlock.
Ramp no faster than the profile allows; 1°C to 2°C per minute is a conservative default.
Hold for a minimum of 24 hours, longer for larger or previously atmosphere-exposed systems.
Track RGA or total pressure trends rather than relying on a fixed timer alone.
Degas gauge and pump filaments and activate getters before ending the bake.
Cool the system at a controlled rate matching the ramp-up.
Re-torque all flanges once the system reaches room temperature.
Perform and log a final helium leak check, base pressure, and RGA scan.

Bakeout and outgassing control are not a single step in a UHV build. They determine whether a system reaches and holds its design pressure at all. Getting the sequencing and temperature limits right, especially around ceramic-to-metal feedthroughs and viewports, is what separates a chamber that reaches its target pressure cleanly from one that fights a slow pump-down for months.

For guidance on selecting the right feedthrough style before you reach the bakeout stage, see Choosing the Right Vacuum Feedthroughs, Flanges & Hermetic Components for a breakdown of connector types and their tradeoffs. If your team is scoping a build and wants a second set of eyes on a bakeout plan or a feedthrough selection before it goes into hardware, talk with MPF Products engineering early.