Dr.Vakuum
Все применения

High vacuum for coating and analytical instruments: turbo, cryo and the backing pump

8 мин чтения

Above about 10⁻³ mbar a single pump does the job. Below it, gas no longer behaves as a fluid — molecules travel in straight lines between wall collisions, the molecular flow regime — and the machines that work there, turbomolecular pumps and cryopumps, cannot compress gas all the way to atmosphere. They need a backing pump behind them, and the pairing is what you are actually sizing. This note covers the pairing for the two duties where Dr.Vakuum sees it most: physical vapour deposition coaters and analytical instruments.

1. Why a high-vacuum pump needs backing

A turbomolecular pump is a bladed rotor spinning at 25,000–60,000 rpm. Each blade row gives a passing molecule a kick toward the exhaust; the pump works only when molecules are far enough apart to be kicked individually, which means an exhaust pressure below roughly 1 mbar (the pump's maximum foreline pressure). Above that, the blades are pushing on a fluid, the motor overloads and the rotor heats.

Two figures on a turbo datasheet matter for the backing pump:

  • Compression ratio K — the exhaust-to-inlet pressure ratio the pump can sustain with zero throughput. For nitrogen it is 10⁸–10⁹; for hydrogen only 10³–10⁴, which is why the ultimate of a turbo system is usually set by hydrogen and water, not nitrogen.
  • Maximum foreline pressure — typically 0.5–5 mbar for a conventional turbo; some hybrid designs tolerate 10–20 mbar.

The backing pump has to do two things: keep the foreline below the maximum while removing everything the turbo passes, and — on a system without a separate roughing line — pull the chamber from atmosphere to the pressure where the turbo can start.

2. Sizing the backing pump

The gas throughput is conserved: what passes through the turbo inlet at 10⁻⁵ mbar leaves its exhaust at a much higher pressure and much lower volume flow. If the turbo is pumping S_turbo at inlet pressure p_in, the throughput is Q = S_turbo × p_in, and the backing pump must provide at least

S_back ≥ Q / p_fore = S_turbo × p_in / p_fore

A T1600 turbo at 1,600 L/s (5,760 m³/h) with a sputter process gas load holding the chamber at 5 × 10⁻³ mbar: Q = 5,760 × 0.005 = 28.8 mbar·m³/h. To hold the foreline at 0.5 mbar:

S_back ≥ 28.8 / 0.5 ≈ 58 m³/h

That is a surprisingly small number — a DV60 two-stage vane pump would do — and it is the first of two requirements. The second is roughing: pulling the chamber from atmosphere to the turbo's start pressure (0.1–1 mbar) in an acceptable time. For a 0.5 m³ coater chamber to 0.1 mbar in 5 minutes:

S = (0.5 / 0.083 h) × ln(1,013 / 0.1) = 6 × 9.2 ≈ 55 m³/h — plus the knee correction, say 90 m³/h

So the backing pump for this coater is a DV90 at 90 m³/h, 5 × 10⁻³ mbar ultimate, and the roughing requirement, not the backing requirement, decided it. On production coaters the two jobs are often split: a large roughing pump (or Roots booster set) brings the chamber down quickly, then a small backing pump holds the foreline while the roughing set serves the next chamber.

Why does the backing pump's ultimate matter when it only has to reach 0.5 mbar?

Because of the compression ratio. The turbo's ultimate inlet pressure is approximately the foreline pressure divided by K. With K = 10⁸ for nitrogen and a foreline at 0.5 mbar, nitrogen is pumped to 5 × 10⁻⁹ mbar — fine. With K = 10³ for hydrogen and the same foreline, hydrogen only reaches 5 × 10⁻⁴ mbar. Every decade you lower the foreline pressure lowers the light-gas ultimate a decade too. A backing pump with a 5 × 10⁻³ mbar ultimate running at 10⁻² mbar in the foreline gives a system two decades better on hydrogen and water than one running at 1 mbar. That is why two-stage pumps back turbos and single-stage ones do not, and why the foreline is kept short and wide.

3. Dry or oil-sealed backing

Oil-sealed two-stage vane pumps (DV Series) are the traditional backing pump: 5 × 10⁻³ mbar ultimate, quiet, cheap to run. Their weakness is oil back-streaming into the foreline when the pressure falls below about 10⁻² mbar, where there is no longer a viscous flow of gas to hold the oil vapour back. A turbo's compression ratio for heavy oil molecules is enormous, so little reaches the chamber — but on a coater producing optical films or a mass spectrometer with a hydrocarbon background problem, "little" is still too much. A foreline trap (molecular sieve or activated alumina) handles it; so does switching to a dry pump.

Dry screw pumps (DSI Series, 0.05 mbar ultimate) give a hydrocarbon-free foreline and tolerate the process gases and dust that coaters produce — reactive sputtering with oxygen or nitrogen degrades vane pump oil within weeks. Their 0.05 mbar ultimate is a decade higher than a DV, which costs a decade on the light-gas ultimate; for most PVD and all industrial coating that is irrelevant, and the oil-free advantage wins. For analytical instruments where the last decade matters, a DV with a trap, or a DV behind a small Roots booster, remains the right answer.

4. Cryopumps: capacity, not speed

A cryopump has no exhaust. It freezes gas onto surfaces at 10–20 K: water on the first-stage array at 65–80 K, nitrogen, argon and oxygen on the second stage, hydrogen and helium adsorbed in charcoal. Pumping speed is very high — the CR300 pumps 1,500 L/s of nitrogen, 2,200 L/s of water vapour and 5,000 L/s of hydrogen — and the ultimate is below 10⁻⁸ mbar with a completely hydrocarbon-free chamber, which is why cryopumps dominate sputter deposition of metals and semiconductor implant.

What a cryopump has instead of a foreline requirement is capacity: the CR300 holds 2.1 × 10⁸ Pa·L of argon before the arrays are saturated and the pump must be regenerated — warmed to room temperature, the frozen gas pumped away by a roughing pump, and cooled again, a cycle of 2–4 hours. A sputter tool running 100 sccm of argon for 16 hours a day accumulates about 10⁷ Pa·L per day and regenerates monthly; a tool running 1,000 sccm regenerates every few days and should have two cryopumps alternating.

The roughing pump for a cryopump system does the chamber pump-down (to the 10⁻¹–10⁻² mbar where the cryopump's gate valve may open) and the regeneration. Both are oil-sensitive — oil in a cryopump's charcoal ruins it — so dry roughing is the norm: a DSI150 or DSI300. Each cryopump needs a compressor; the CR300 pairs with the HC80W helium compressor, and the compressor's cooling water (3–6 L/min at ≤ 28 °C) is the one utility that gets forgotten in a Thai plant room.

5. Two worked systems

Batch sputter coater for decorative and tool coatings. Chamber 1.2 m³, cycle target: 10⁻⁵ mbar in 20 minutes from atmosphere, process at 3 × 10⁻³ mbar with 200 sccm argon.

  • Roughing to 0.1 mbar in 6 minutes: S ≈ (1.2/0.1) × 9.2 ≈ 110 m³/h, with knee → a DSI150 (150 m³/h, dry)
  • High vacuum: a T1300Z turbo at 1,300 L/s reaches 10⁻⁵ mbar from 0.1 mbar in roughly 10 minutes on a clean chamber; a CR500 cryopump at 3,500 L/s does it faster and gives a cleaner film, regenerating weekly at this gas load
  • Backing the turbo during process: Q = 4,680 × 0.003 = 14 mbar·m³/h → 28 m³/h at 0.5 mbar foreline; the DSI150 covers it with the chamber roughing valve closed
  • Add a Roots booster (MB150, 500 m³/h) ahead of the DSI if the 6-minute roughing has to become 2

Scanning electron microscope / mass spectrometer. Column volume a few litres, 10⁻⁶ mbar, hydrocarbon-sensitive.

  • Turbo: a T600 at 600 L/s is far more than the throughput needs; it is chosen for compression ratio and the ultimate
  • Backing: a DV16 (14 m³/h, 5 × 10⁻³ mbar) with a foreline trap, or a small dry pump; the foreline is kept at 10⁻² mbar so hydrogen and water reach the 10⁻⁷ range
  • The whole set is often run from one controller that interlocks the turbo start on foreline pressure and vents the turbo on power failure

6. Installation points that decide the ultimate

  • Short, wide foreline. In molecular flow conductance goes with d³/L; 1 m of DN25 has about 10 L/s, which is nothing against a 600 L/s turbo but is plenty against a 90 m³/h (25 L/s) backing pump. Keep it under 2 m, DN40 where possible.
  • Bake-out and leak check before blaming the pumps. A chamber that will not get below 10⁻⁵ mbar after a long pump-down almost always has a leak or an outgassing load; a helium leak detector finds it in an hour.
  • Water vapour is the dominant residual below 10⁻⁵ mbar on any chamber opened in Thai humidity. A Meissner coil (LN₂ or refrigerant) in the chamber, or a cryopump's first stage, removes it far faster than any turbo.
  • Vent with dry nitrogen, not room air, and from the turbo's own vent port so it slows evenly.

Recommended Dr.Vakuum machines

Turbomolecular (T Series) — 600 to 4,200 L/s, < 10⁻⁷ mbar: T600, T1300Z, T1600, T4200.

Cryopumps (CR Series) and helium compressors (HC) — 1,500 to 10,000+ L/s, < 10⁻⁸ mbar, hydrocarbon-free: CR300, CR500, CR1000, with HC80W / HC100W.

Backing and roughing

Role Oil-sealed Dry
Backing a turbo, small instrument DV16, DV30 —
Backing + roughing, 0.3–1 m³ coater DV90, DV175 DSI150, DSI300
Roughing large chambers, cryopump regeneration DV275 DSI630
Fast roughing MB150/MB300 booster + any of the above same, with DSI backing

System sizing. Tell us the chamber volume, the base and process pressures, the process gas flow and whether hydrocarbons matter. We will size the high-vacuum pump and the backing set together, with the foreline pressure and roughing time shown. Talk to an engineer.