Freeze drying: why the condenser, not the pump, sets the rate
Freeze drying is the one vacuum application where buying a bigger pump reliably achieves nothing. The reason is arithmetic, and it is worth doing before specifying anything.
1. The number that decides the whole design
Water vapour at low pressure occupies an enormous volume. From the ideal gas law, with the specific gas constant for water vapour R = 461.5 J/(kg·K):
V̇ = ṁ · R · T / p
For one kilogram per hour of ice subliming, with vapour at −40 °C:
| Chamber pressure | Vapour volume |
|---|---|
| 0.05 mbar | 21,500 m³/h |
| 0.1 mbar | 10,800 m³/h |
| 0.5 mbar | 2,150 m³/h |
| 1.0 mbar | 1,080 m³/h |
| 6.11 mbar (triple point) | 176 m³/h |
At a typical primary-drying pressure of 0.1 mbar, a single kilogram of ice per hour generates around 10,800 m³/h of vapour. The largest rotary vane pumps are rated near 800 m³/h. The pump is not within two orders of magnitude of the duty, and no realistic pump is.
The condenser does the pumping. It removes the water by freezing it out, turning ~10,800 m³ of vapour into a litre of ice. The vacuum pump's job is to evacuate the chamber at the start and then remove non-condensable gas — air in-leakage, dissolved gases, elastomer outgassing — for the rest of the cycle. Size the pump for that load, not for the water.
Note also the 1/p scaling: halving chamber pressure doubles the volumetric load. Deep vacuum is expensive here in a way it is not elsewhere.
2. Why the process must sit below the triple point
The triple point of water is 611.657 Pa (6.117 mbar) at 0.01 °C. Below that pressure liquid water cannot exist: heat applied to ice takes it directly to vapour. Above it, ice melts first and then boils under vacuum, which destroys the cake structure the process exists to preserve.
That is the entire reason for the vacuum. Everything else is a consequence.
3. Sublimation, and what limits it
The mass flow off the ice front is driven by the pressure difference between the ice and the chamber, against the resistance of the already-dried layer above it:
ṁ = A · (P_ice − P_chamber) / R_p
R_p is the dried-layer resistance, and it grows as the cake gets thicker — which is why sublimation slows through primary drying even at constant conditions.
The energy has to arrive as well. At steady state:
ṁ · ΔH_sub = K_v · A · (T_shelf − T_product)
with ΔH_sub ≈ 2,830–2,840 kJ/kg. Usefully, 1 kg/h of ice needs about 790 W delivered through the shelf, across a temperature difference of only a few kelvin.
4. Lower pressure is not faster — the result that surprises people
The instinct is that deeper vacuum dries faster. In a heat-transfer-limited process it does the opposite.
Heat reaches the product mainly by gas conduction across the small gap between shelf and vial bottom, and that conduction depends on pressure:
K_v = K_contact + K_radiation + K_gas(P)
As chamber pressure falls, there is less gas in the gap to carry heat, so K_v falls with it. Published work measured the effect directly: raising chamber pressure from 0.04 to 1.3 mmHg at constant shelf temperature increased the primary drying rate by about 160%.
So there is a genuine optimum. Below it, you gain driving force (P_ice − P_chamber) but lose heat input, and since most cycles are heat-limited the net result is slower drying at a colder product. This is why practitioners sometimes bleed nitrogen into the chamber to raise pressure deliberately — making the vacuum worse on purpose to dry faster.
Typical primary drying therefore runs 0.05–0.3 mbar (50–300 mTorr), most commonly 100–150 mTorr, rather than as low as the equipment can reach.
5. Three ceilings the pump cannot lift
The condenser's capacity is a flat plateau in kg/h. It does not improve with more pumping speed. Exceed it and vapour backs up, chamber pressure climbs, the product warms, and the cake collapses.
The duct chokes. Flow from chamber to condenser reaches Mach 1 at the duct exit, and the maximum mass flow through a choked duct falls with pressure. Pushing chamber pressure down to go faster can drive the duct into choking, at which point additional vapour raises chamber pressure uncontrollably and pressure control is lost entirely. This sets the minimum controllable chamber pressure — typically 0.03–0.15 Torr depending on dryer scale — and it is a classic lab-to-production scale-up trap.
Ice distributes unevenly on the coils. Modelling over a 24-hour cycle found 3.2 cm of ice on the coil nearest the duct but only 4 mm on distant coils, with the duct increasing non-uniformity by 65%. Ice can bridge and block flow well before the condenser volume is nominally full. Counter-intuitively, a moderately higher condenser set point such as −60 °C distributes ice better than an extreme −110 °C.
6. Choosing a condenser temperature
The vapour pressure of ice sets the driving force:
| Ice temperature | Vapour pressure |
|---|---|
| 0 °C | 6.108 mbar |
| −20 °C | 1.032 mbar |
| −40 °C | 0.1283 mbar |
| −50 °C | 0.03935 mbar |
| −60 °C | 0.0108 mbar |
| −80 °C | 0.000547 mbar |
Read the differences, not the values. Going 0 → −10 °C buys 3.5 mbar of driving force; going −50 → −60 °C buys only 0.029 mbar. Colder condensers give sharply diminishing returns, which is why the industry sits at −50 to −85 °C and reserves −105 °C for solvents rather than for drying water faster.
7. Protecting the pump
Even with a good condenser, some water reaches the pump. An oil-sealed rotary vane pump has a water-vapour tolerance of roughly 60 mbar at the inlet; above that, vapour condenses inside the pump and emulsifies the oil. The oil then loses its sealing film, ultimate pressure degrades, and recovery is slow because the condensed water must be re-evaporated before it can be pumped out.
Gas ballast prevents it, at the cost of slightly worse ultimate pressure. Run it through the wet phase and close it for the final pressure-holding phase.
For pharmaceutical work the stronger answer is a dry pump. Sterile duty requires clean-in-place and steam sterilisation, which an oil-sealed pump cannot meet, and an oil-free machine removes any possibility of hydrocarbon back-streaming onto a high-value product.
8. Sizing the pump properly
Four separate limits, all of which must be satisfied:
- Pump-down time — from chamber volume, t = (V/S)·ln(p₀/p₁)
- Leak and non-condensable load at operating pressure — the load the pump actually carries through the cycle
- Condenser capacity in kg/h — the real throughput ceiling
- Duct choked flow — the minimum controllable pressure
Sizing on chamber volume alone is the common error. It produces a pump that reaches working pressure on an empty dryer and cannot hold it under load.
One further caution: the pressure-rise test used to detect the end of primary drying infers sublimation from dP/dt with the chamber isolated. A real leak produces an identical-looking ramp, so a leaky dryer reads as "still subliming" and never shows an endpoint. Characterise the leak rate on the empty machine and subtract it. Typical acceptance is ≤0.02 mbar/min.
Recommended Dr.Vakuum machines
Dry screw pumps (DSI Series) — oil-free, 0.05 mbar ultimate, the right choice for pharmaceutical duty where CIP and SIP are required and hydrocarbon contamination is unacceptable.
| Model | Flow (50/60 Hz) |
|---|---|
| DSI150 | 150 / 180 m³/h |
| DSI300 | 300 / 360 m³/h |
| DSI630 | 630 / 756 m³/h |
| DSI1200 | 1,200 / 1,440 m³/h |
Oil-sealed rotary vane (V and DV Series) with gas ballast remain cost-effective for industrial and food freeze drying, where sterile requirements do not apply.
Sizing help. Send us chamber volume, batch size in kg of ice, target primary-drying pressure and cycle time, and condenser capacity and set point, and our engineers will size against all four limits rather than the one that is easiest to calculate. Talk to an engineer.
See also: the shallower, warmer cousin of this duty in Vacuum drying ovens and tray dryers for pharma and chemicals.