Rotary Evaporation and Solvent Recovery

What a rotovap is for, and what it is not for

A rotary evaporator exists to take a large volume of solvent away from a small quantity of dissolved material quickly and gently, and to get most of that solvent back in a condition where it can be reused. It does this with three tricks at once. It reduces the pressure, so the solvent boils at a temperature the solute can tolerate. It rotates the flask, which spreads the liquid as a thin film over a large area of heated glass and continuously renews that film, so heat transfer is fast and there is no static superheated layer. And it condenses the vapour immediately into a separate receiving flask, so the solvent is recovered rather than exhausted. In a cannabinoid plant the rotovap sits after winterization, after a tincture or ethanol extraction, and after any chromatography or crystallisation step that left product in solvent. What it is not is a purification step for residual solvent. It removes solvent in bulk down to the point where the remaining solvent is held by the product rather than pooled in it, and then it effectively stops. Getting from there to a residual-solvent specification is a different operation with different equipment.

Sources: Armarego WLF 2017* · Green DW 2019* · Composite: extraction-equipment vendor documentation 2026*

The three temperatures, and the delta-20 rule industry practice, not published data

A rotovap is controlled by three temperatures and one pressure, and the relationship between them is the whole craft. The bath temperature supplies the heat of vaporisation. The vapour temperature is the temperature at which the solvent is actually boiling, which is set by the pressure, not by the bath. The condenser temperature must be low enough to condense that vapour completely. The standard heuristic — often called the delta-20 rule or the 20-40-60 rule — sets the working vacuum so that the solvent boils at around 40 °C, puts the bath at about 20 °C above that boiling point, and keeps the coolant at about 20 °C below it. For ethanol that is a bath near 60 °C, a working pressure that puts the boiling point near 40 °C, and a chiller at about 20 °C or lower. The 20 °C driving force on the bath side gives a usable evaporation rate without making the glass wall hot enough to damage a thermally sensitive solute; the 20 °C margin on the condenser side gives enough capacity that vapour does not slip past the condenser, because vapour that is not condensed is solvent you do not recover, solvent in your vacuum pump, and a vacuum reading that will not hold. A useful diagnostic follows from this: if the vapour temperature reading rises toward the bath temperature, the flask is running dry and the solvent is gone; if the condensation ring creeps up the condenser toward the top, the condenser is at its capacity and you are losing vapour to the pump.

• Pressure sets the boiling point. The bath only sets the rate.
• Bath about 20 °C above the boiling point at working vacuum; condenser about 20 °C below it.
• Watch where the condensation ring sits: creeping upward means the condenser is overloaded.
• Vapour temperature climbing toward bath temperature means the flask has run dry.

Sources: Armarego WLF 2017* · Composite: extraction-equipment vendor documentation 2026*

Boiling point against pressure for ethanol contested

The relationship a processor actually needs at the panel is the boiling point of their solvent at the pressure they are working at. The figures below are calculated from standard Antoine coefficients for ethanol and rounded; they are approximate and will differ by a degree or two between sources and with water content, which matters because an ethanol-water mixture does not behave like anhydrous ethanol and an azeotrope near 95 percent ethanol by volume sets a floor on what simple evaporation will concentrate to. Build the same table for any other solvent you run — pentane, heptane, and hydrocarbon blends all have their own curve and the delta-20 arithmetic is done against the curve, not against the atmospheric boiling point.

Contested — caveat. These values are computed from published Antoine coefficients for pure ethanol and rounded to the nearest degree; tabulated values differ slightly between sources and the presence of water shifts the curve and introduces the ethanol-water azeotrope, so treat the table as a working guide and verify against your own gauge and thermometer.

Sources: Armarego WLF 2017* · Green DW 2019*

Vacuum control: why a ramp beats slamming to full vacuum industry practice, not published data

Pulling full vacuum the moment the flask is on is the most common beginner error and it has three separate costs. First, it drops the boiling point below the bath temperature by a wide margin all at once, so the entire surface of a large volume of solvent flashes simultaneously and the flask foams over into the vapour duct. Second, it entrains liquid as fine droplets which travel up the duct and are collected in the receiving flask, so the product you were concentrating is now in your recovered solvent. Third, on a volatile solvent, the sudden latent-heat demand chills the liquid sharply and evaporation then stalls until the bath catches up, which is slower overall than a controlled ramp would have been. The correct approach is to bring the pressure down progressively — either with a manual needle valve trimmed by hand or, better, with a vacuum controller working to a setpoint or to a boiling-point-detection mode — so that the solvent establishes a steady, controlled boil and stays there. Production-scale practice makes the same point differently: at real volumes a rotovap is replaced by a falling-film evaporator, which is a continuous device with the ramp problem designed out, feeding steadily rather than charging a batch.

• Ramp with a needle valve or a controller; never slam to base pressure on a full flask.
• Foam-over and droplet entrainment both put your product in the recovered solvent.
• A vacuum controller holding a setpoint is the single best upgrade to a manual rotovap.
• Above a certain volume the answer is a falling-film evaporator, not a bigger rotovap.

Sources: Armarego WLF 2017* · Composite: extraction-equipment vendor documentation 2026*

Fill fraction and rotation speed industry practice, not published data

Fill the flask between about one third and one half of its nominal volume and no more. The headspace is not wasted space, it is the disengagement volume that lets foam collapse and droplets fall back before they reach the vapour duct, and an overfilled flask removes exactly that margin. Rotation speed sets the film: too slow and the liquid sits as a pool at the bottom with a small, static, superheated contact area; too fast and on a viscous charge the liquid climbs the wall, thins to the point where it stops wetting properly, and the drive works against a badly balanced load. A moderate speed that keeps a visible, continuously renewed film over the submerged portion of the wall is what you are looking for, and for most flask sizes and viscosities that is somewhere in the range of roughly 100 to 200 rpm, trimmed by eye. Rotation matters most on viscous, concentrated charges, which is precisely the end of the run where bumping is most likely and where a static pool would scorch, so as the charge concentrates it is normal to lift the rotation rather than lower it.

Sources: Armarego WLF 2017* · Composite: extraction-equipment vendor documentation 2026*

Bumping: what it is, why it happens, what it costs

Bumping is a sudden, violent, localised boil that throws liquid up the vapour duct instead of vapour. Its cause is superheating combined with nucleation delay: for a liquid to boil, a vapour bubble has to form, and bubble formation needs a nucleation site — a scratch, a dust particle, a dissolved gas pocket, a boiling stone. Smooth, clean glass under vacuum with a degassed liquid can be heated past the temperature at which it should boil without boiling at all, storing that superheat in the liquid. When nucleation finally occurs somewhere, the entire superheat is discharged at once and a large volume of vapour is generated in a fraction of a second. Under vacuum the pressure gradient carries the resulting slug of liquid straight up the duct. What it costs is immediate and expensive: product is thrown into the vapour path, into the condenser and into the receiving flask, contaminating recovered solvent that you intended to reuse; the batch has to be reconciled or written off; and if the material reaches the vacuum pump it contaminates the oil and degrades the pump. On a concentrated cannabinoid charge it also means a sticky, viscous deposit in the duct and condenser that is unpleasant and slow to clean. Prevention is a short list and all of it is routine. Keep the pressure ramp controlled so the liquid never gets far ahead of its boiling point. Keep the flask rotating, because rotation both renews the film and provides continuous mechanical nucleation. Keep the bath driving force modest — the delta-20 figure exists partly for this reason. Do not overfill. And watch the flask, particularly in the last third of the run when the charge has concentrated, its viscosity has risen, its boiling point has risen with it, and the temptation to raise the bath is strongest.

• Cause: superheat plus absence of nucleation sites, discharged all at once.
• Cost: product in the recovered solvent, a ruined mass balance, a contaminated condenser, and pump-oil damage.
• Prevention: controlled ramp, keep it rotating, modest bath driving force, fill to a half at most, and watch the concentrated end of the run.
• Never chase a stalled evaporation by raising the bath — deepen the vacuum instead.

Sources: Armarego WLF 2017* · National Research Council (US) 2011* · Composite: extraction-equipment vendor documentation 2026*

Recovery percentage as an economic driver industry practice, not published data

In an ethanol plant the solvent is one of the largest recurring costs, and it is consumed at a ratio measured in litres per kilogram of biomass for extraction and again at 5 to 10 litres per kilogram of crude for winterization. Every point of recovery is therefore money. Recovery losses have identifiable causes and each one is addressable: vapour passing an overloaded or under-chilled condenser; leaks on joints, seals and the bump trap, which both cost vacuum and let vapour out; solvent left behind in the product because the run was stopped early; solvent left wetting the vessel walls and transfer lines; and vapour lost through the pump exhaust, which on a rotary-vane pump also means solvent in the oil. Measuring recovery honestly requires weighing or metering the solvent in and the solvent out on every batch rather than estimating, because a recovery number nobody measures is always better in conversation than it is in the ledger. At production volumes the same argument is what justifies a falling-film evaporator with an efficient condenser and a heat-recovery loop over a bank of rotovaps: it is bought to recover solvent, and the throughput is a side benefit.

Sources: Composite: extraction-equipment vendor documentation 2026*

The limit of a rotovap: bulk solvent is not residual solvent

A rotovap removes solvent that is present as a bulk liquid phase. It does not remove the last fraction of a percent that is dissolved in, adsorbed onto, or occluded within a viscous product, because that solvent is no longer boiling out of a film — it has to diffuse through a thick, viscous, increasingly solvent-starved matrix, and the diffusion is slow at any temperature the product tolerates. This is why a batch can come off a rotovap apparently dry and still fail a residual-solvent panel by a wide margin. The operations that close that gap are a vacuum oven — thin the product out, hold it at a modest temperature under deep vacuum for hours, with a large surface-to-volume ratio so diffusion distances are short — or a purpose-built devolatilisation stage, and in the case of crystallised material a proper cold-solvent wash of the cake followed by vacuum drying. The specification you are working to comes from the residual-solvent frameworks: ICH Q3C classes solvents by toxic potential and sets permitted daily exposures, and USP General Chapter 467 is the compendial method and limit framework that laboratories work from. Class 1 solvents such as benzene are to be avoided outright, Class 2 solvents including hexane and methanol are limited to low parts-per-million figures, and Class 3 solvents such as ethanol, pentane and heptane carry much more generous limits. The practical consequence for a processor is that the solvent you choose upstream determines how hard this step is, and that the panel is the arbiter — not the appearance of the product and not how long it sat in the oven. Formulation/residual-solvent covers the limits, the panel and the arithmetic in full.

• Bulk solvent removal and residual-solvent specification are two different operations.
• The last fraction of a percent is a diffusion problem: thin films, long holds, deep vacuum.
• ICH Q3C classes and USP General Chapter 467 are the frameworks the panel is run against.
• Solvent choice upstream sets the difficulty of this step downstream.

Sources: International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use 2021* · United States Pharmacopeia 2023* · Composite: extraction-equipment vendor documentation 2026*

Safety, as part of the operation

Rotary evaporation combines evacuated glassware, a heated bath and a volume of flammable solvent vapour, and each of the three has a specific discipline. Inspect every flask and the condenser before each use for star cracks, chips at the joints and scratches, in good light, and retire damaged glass; a round-bottom flask under vacuum with a hot solvent charge fails inward and sprays. Use flask clips or a keck clip on every joint — a flask that drops off a rotating drive into a hot bath is a common and entirely avoidable incident — and never rely on the vacuum alone to hold a flask on. Do not evacuate a flask that is not rated for it, and never a flat-bottomed or Erlenmeyer flask. Hot water baths cause scald burns and, at bath temperatures near or above the boiling point of the solvent in use, the bath itself becomes an ignition-relevant hot surface, so keep the bath temperature as low as the delta-20 arithmetic allows rather than as high as the dial permits. The condenser and the vacuum pump exhaust both handle flammable vapour: route the pump exhaust out of the workspace or into an appropriate scrubber, use a solvent-resistant diaphragm pump where possible rather than a rotary-vane pump that will accumulate solvent in its oil, and do not let recovered solvent accumulate in an open receiving flask in an unventilated room. Bond and ground metal transfer containers when decanting recovered solvent, because static from pouring is a real ignition source at these volumes. Peroxide-forming solvents — diethyl ether, THF, diisopropyl ether, dioxane — must never be taken to dryness on a rotovap; peroxides concentrate in the residue and the residue is where they are most dangerous, and any such solvent should be date-marked, tested and disposed of within its recommended window. Finally, vent to atmosphere deliberately and slowly through a valve at the end of a run, with the flask lifted out of the bath first, rather than by breaking a joint under vacuum.

• Inspect glass for star cracks and chips before every run; clip every joint.
• Keep the bath as cool as delta 20 allows; a hot bath is both a scald and an ignition-relevant surface.
• Route the pump exhaust out; prefer a chemically resistant diaphragm pump for solvent duty.
• Bond and ground containers when decanting recovered solvent — static is sufficient ignition.
• Never take a peroxide-forming solvent to dryness; date-mark and test those solvents.
• Lift the flask out of the bath, then vent slowly through a valve.

Sources: National Research Council (US) 2011* · Kelly RJ 1996* · National Fire Protection Association 2024* · Armarego WLF 2017*