Vacuum pumps for process tools: rotary vane, dry scroll or turbo
How the process decides the pump, what backstreaming really does to a chamber, why corrosive chemistry eats a pump from the inside, and the service interval that sets how much of the year your tool is available.
An etch tool at a Bangladeshi university has been down for eleven weeks. The fault is a backing pump that seized because nobody changed the oil, and the replacement is quoted with a lead time that assumes air freight and a customs clearance nobody has started. The tool cost a large fraction of an equipment grant. The pump that stopped it costs a small fraction of that, and the oil that would have saved it costs almost nothing.
Vacuum pumps are treated as an accessory at purchase and as an emergency afterwards. They deserve to be specified as carefully as the chamber they serve, because tool availability is mostly a pump question.
What the process decides
Ultimate pressure is the specification everybody compares, and it is rarely the one that matters. What matters is the pumping speed available at the pressure the process actually runs at, with the gas load the process actually produces.
A microscope column or an analytical chamber has almost no gas load once it is down. The requirement is a low base pressure, a clean vacuum and reliability, and the duty is gentle.
A plasma etch or PECVD tool runs continuously at a process pressure with a real gas flow through it. The requirement is throughput at that pressure, and tolerance of what the process produces.
A load lock cycles from atmosphere repeatedly. The requirement is fast roughing and tolerance of the water load that comes in with every cycle.
An evaporator or sputter chamber that is vented regularly spends much of its life pumping water off the chamber walls, and that determines how long the pump down takes far more than the pump's ultimate pressure does.
Ask the tool supplier for the pump specification they validated the process on, then confirm the local supply and service position for that pump before agreeing to it. Accepting a pump nobody in the region supports is a decision that costs you every year the tool is in service.
Primary pumps: rotary vane, dry scroll, dry screw
Pump type
Suits
Consumable
What kills it
Oil sealed rotary vane
Clean applications, roughing, backing a turbo on an analytical tool
Oil, oil mist filter element, vanes
Condensables emulsifying the oil, running without gas ballast, process gas attacking the oil
Dry scroll
Clean processes where no oil is acceptable, backing on deposition and analytical tools
Tip seals, bearings
Particulate, running at high inlet pressure for long periods, ignored tip seal intervals
Diaphragm
Backing small turbos, low throughput clean service
Diaphragms and valves
Continuous operation near its pressure limit, solvents attacking the diaphragm material
Dry screw or multistage roots
Process tools with real gas loads, corrosive and powder-forming chemistry
Bearings, seals, purge nitrogen
Powder accumulation without purge, loss of cooling water, loss of nitrogen purge
Two habits determine how long a rotary vane pump lasts. The first is running gas ballast when pumping anything condensable, which keeps water and solvent from condensing into the oil and turning it into an emulsion that no longer seals. The second is not leaving the pump running against a closed valve for hours, which is where oil migrates back into the chamber.
Dry scroll pumps remove the oil question but introduce a different one: tip seals wear, and the wear produces dust. Change them on the interval the manufacturer states rather than when performance drops, because by the time the ultimate pressure has visibly degraded the debris has already been through the mechanism.
Turbo pumps, and what sits behind them
A turbomolecular pump does not work alone. It compresses gas into a backing pump, and it has a maximum backing pressure above which it cannot function. Most turbo failures in research labs trace back to the backing pump rather than the turbo.
Bearing type is the main lifetime decision. Grease or oil lubricated bearings need scheduled replacement, and skipping that interval is how a rotor fails at full speed. Magnetically levitated rotors avoid the scheduled bearing change and cost more to buy, which is often the right trade for a tool that runs continuously.
Never vent a spinning turbo through a flange. Venting has to go through the vent valve, at a controlled rate, with dry nitrogen, and only when the rotor has slowed to the speed the manufacturer specifies. A sudden vent at full speed can destroy the pump and, in the worst case, the chamber it is attached to.
Check the vent behaviour on power loss. This is specific to Bangladesh. Load shedding and generator changeovers drop power mid-run routinely, and a control scheme that opens the vent valve while the rotor is at speed will eventually wreck a pump. Put the pump controller and the vent logic on a UPS, then test the behaviour deliberately at commissioning by pulling the supply.
Cooling is an interlock, not a convenience. Many turbos and most process pumps need water or forced air. If the chilled water loop stops during a power event and the interlock is not wired, the pump runs hot until it fails.
Corrosive service needs a purge and the right coatings. A standard turbo on a chlorine or fluorine process has a short life. Purge gas, coated rotors and a heated foreline are the difference between an annual replacement and a working tool.
Backstreaming, and the trap that becomes a source
Hydrocarbon contamination in a chamber shows up as poor film adhesion, as a carbon signal in analysis, as a contamination square in a microscope image, and as gauges that read the wrong pressure. It comes from the pumping system.
Backstreaming is worst at the extremes. A rotary vane pump backstreams most when it is running at low pressure with little gas flow, and a diffusion pump backstreams during crossover. The transitions are where contamination happens, not steady state operation.
A foreline trap has a finite capacity. Molecular sieve and zeolite traps adsorb oil until they are saturated, after which they release it. A trap that has never been regenerated or replaced is not protecting anything, and in an old installation it is frequently the largest single source of contamination.
Fouled gauges lie. A cold cathode gauge with oil on its electrodes reads low, which conceals the problem it is caused by. When a chamber's readings stop making sense, clean the gauge before believing them.
The cheapest fix is a dry primary pump where the process allows it. On a tool where cleanliness matters and the gas load is modest, the price difference against a rotary vane pump is repaid in avoided contamination investigations.
Corrosive gases and powder
Etch and deposition chemistry attacks the pump from the inside, and the two failure modes are corrosion and blockage.
Use the oil the process requires. Fluorinated chemistry needs perfluoropolyether oil rather than hydrocarbon oil. Never mix the two, and never top up a PFPE-filled pump with mineral oil, which has ruined more pumps than corrosion has.
Nitrogen purge dilutes the process gas inside the pump and keeps the internal surfaces above the condensation point of the reaction products. If the pump has a purge port, it is there for a reason and should be plumbed and flowing.
Powder in the foreline is the classic etch and PECVD failure. Reaction products condense into solids that pack the line and eventually stop the pump. A powder trap before the pump, a heated foreline, and a cleaning schedule on the trap are what prevent it. Cleaning that trap is a routine job, and it is normally the one nobody has been trained to do.
Treat a pump off a toxic process as contaminated. It has to be purged and, if it is going anywhere, accompanied by a decontamination declaration. No service agent will open a pump without one, and no freight agent should carry it.
Service intervals and spares in Bangladesh
The interval is stated by the manufacturer in running hours, and the honest way to manage it is to log hours rather than dates. A tool that runs continuously reaches its service point far sooner than a calendar suggests.
Hold the service kit locally from day one. Oil, oil filters, tip seals, gaskets and the small parts in the manufacturer's kit are inexpensive relative to their effect on uptime, and they are exactly what is never in stock when needed. Buy them with the tool, while the purchase order is still open.
Hold a spare primary pump for the tool you cannot afford to lose. A pump is the item most likely to fail and the item with the longest replacement path. For a shared facility supporting several research groups, a spare pump is cheaper than a term of lost work.
Understand the exchange programme before you need it. Manufacturer rebuild and exchange schemes require the pump to travel both ways, with dangerous goods and decontamination paperwork if it has been on process. Ask for the round trip time to Bangladesh, in writing, before you rely on it.
Protect the motor. Voltage fluctuation is normal on many campus supplies, and pump motors are not tolerant of it. Check the motor's acceptable voltage window against what the building actually delivers, measured rather than assumed, and fit a stabiliser where it does not match.
Train two people, not one. An oil change and a tip seal replacement are straightforward tasks. Whether they happen depends on whether more than one person in the building knows how, and whether they are permitted to do it without waiting for a visiting engineer.
Keep a maintenance log on the tool itself. Hours at last service, what was done, what was fitted. It is the document a service engineer asks for first and the one that is almost never there.