Short-Path Distillation

What distillation is actually doing

Distillation separates a mixture by exploiting differences in vapour pressure. At any temperature, each component in a liquid mixture exerts its own partial vapour pressure; the component with the higher vapour pressure at that temperature is enriched in the vapour phase, and if you condense that vapour separately you have separated it from the rest. Boiling point is not a property of a molecule in isolation, it is the temperature at which that molecule exerts a vapour pressure equal to the pressure above the liquid. That single sentence is the whole basis of vacuum distillation: lower the pressure above the liquid and the temperature at which the liquid boils falls with it. In a cannabinoid distillation the components you are separating span an enormous range of volatility — monoterpenes that flash off almost immediately, residual solvent and water below them, then sesquiterpenes, then the cannabinoids themselves in a fairly narrow band, then sterols, waxes that survived winterization, pigments and finally material that will not distil at all at any temperature you can safely apply and stays behind as residue.

Sources: Green DW 2019*

Pressure, boiling point and thermal load contested

The relationship between vapour pressure and temperature is exponential, not linear. The Clausius-Clapeyron relation captures the intuition: the logarithm of vapour pressure falls roughly linearly with the reciprocal of absolute temperature, with the slope set by the enthalpy of vaporisation. The practical consequence is that the first decade of vacuum buys you a large drop in boiling point, and each further decade buys you a smaller but still substantial drop — for a heavy, low-volatility compound, dropping from atmospheric pressure to around 1 mbar typically pulls the boiling point down by well over a hundred degrees. That is not a convenience, it is the enabling condition. Cannabinoids are thermally labile: at the temperature at which they would boil under one atmosphere they degrade appreciably before and while they boil, so an atmospheric-pressure distillation of THC or CBD does not produce distillate, it produces decomposition products. Under deep vacuum the same molecules move at a jacket temperature in the 130 to 180 °C band, where the rate of thermal degradation is slow enough that the material can be got over the head and condensed before very much of it is lost. The widely reproduced table of cannabinoid boiling points — THC at 157 °C, CBD at 160 to 180 °C, CBN around 185 °C and so on — comes originally from vaporiser research rather than from a distillation determination, and the figures are best read as approximate volatilisation temperatures rather than as thermodynamic boiling points measured at a stated pressure.

• Boiling point is a function of the pressure above the liquid, not an intrinsic constant.
• Reducing pressure reduces the temperature required, therefore reduces the thermal load, therefore reduces degradation.
• The reason to buy a better pump is not speed, it is product quality.
• Degradation is not a threshold effect at some magic temperature; it is a rate that rises steeply with temperature and accumulates with time.

Contested — caveat. The 157 °C figure for THC and the associated cannabinoid boiling-point table are quoted throughout the trade without a stated pressure and trace back to vaporiser volatilisation work (Gieringer and colleagues), not to a controlled boiling-point determination. Published values for the same cannabinoid differ by tens of degrees between sources. Treat them as indicative of relative volatility only, and set your own cut points from what you observe at your own vacuum.

Sources: Green DW 2019* · Gieringer D 2004*

Vacuum depth: the units, and what each decade buys industry practice, not published data

Three unit systems are in daily use and a processor needs all three because gauges, pumps and papers do not agree on one. One standard atmosphere is 1013 mbar, 760 torr (equivalently 760 mmHg) and 760000 micron, where a micron is one thousandth of a torr. So 1 mbar is about 0.75 torr, which is about 750 micron; 0.1 mbar is about 75 micron; 0.01 mbar is about 7.5 micron; and 0.001 mbar is about 0.75 micron. A single-stage rotary-vane pump in good condition with fresh oil will reach the low tens of micron at the pump; a two-stage pump will reach the single digits or below. What you get at the boiling flask is always worse than what the pump can do, because every fitting, every length of hose, every bend and the conductance of the glassware itself costs you, and because anything volatile still dissolved in the charge is actively fighting the pump. This is why the gauge belongs as close to the still head as practical rather than at the pump inlet: a gauge at the pump reports the pump, not the process. Each decade of additional vacuum lowers the required vapour temperature further, sharpens the separation between adjacent fractions because relative volatility differences become easier to exploit at lower temperature, and shortens the time the material spends hot. The diminishing return is real but the direction never reverses: deeper is better, always, for product quality.

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

The apparatus, part by part industry practice, not published data

Every component of a short-path train has one job, and knowing which job it has is what lets you diagnose a bad run instead of guessing. The defining feature of the geometry is in the name: the distance from the evaporating surface to the condensing surface is short, so a molecule that leaves the liquid has a high probability of reaching the condenser without colliding its way back, and the pressure drop across the path is small enough that the deep vacuum you paid for actually exists at the boiling surface.

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

Why vapour temperature is the control variable industry practice, not published data

The mantle temperature tells you what the heater is doing. The vapour temperature at the head tells you what is distilling. Those are different facts and only the second one is actionable. The heat path runs mantle to glass to liquid to vapour, and each step has a lag and a gradient, so at any moment the mantle is substantially hotter than the bulk liquid and the bulk liquid is hotter than the vapour arriving at the head. A processor who drives mantle setpoint to a number read off someone else’s run is controlling the wrong end of the system: the same mantle setpoint on a different flask, a different fill level, a different vacuum or a different crude produces a completely different cut. The vapour probe must sit in the vapour stream at the head, below the point where the vapour turns toward the condenser, so it reads the temperature of what is actually passing. Run the process by watching that number stabilise, plateau and then climb: a plateau means a fraction is coming over at a consistent composition, and a climb means that fraction is exhausted and the next, heavier one is beginning. The mantle is then adjusted to sustain a steady rate of take-off rather than to hit a temperature.

• Plateau in vapour temperature means a fraction is running; a rise means the fraction is ending.
• Two runs at identical mantle setpoints can produce entirely different products; two runs at identical vapour temperature and vacuum will not.
• Rate of take-off, not temperature, is what you actually trim with the mantle.

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

The fraction cuts, in order industry practice, not published data

A short-path run is a sequence of cuts taken in ascending order of boiling point. What follows is the order in which they arrive and what is chemically in each. The cut points are decided by watching the vapour temperature and the appearance of the distillate, not by a clock, and they shift with vacuum depth and with the crude.

Sources: Green DW 2019* · Composite: extraction-equipment vendor documentation 2026* · Namdar D 2018*

First pass versus second pass industry practice, not published data

A first pass on properly prepared feed does the bulk of the work: it separates the cannabinoid body from the volatiles below it and from the pigments, sterols and polymerised material above and outside it. What it does not do reliably is deliver colour. A first-pass distillate from good feed is usually amber to light gold; from mediocre feed it is dark. A second pass, run on the collected body cut alone, buys three specific things: colour, because the small quantity of coloured and coloured-precursor material that survived the first pass is left behind; potency, typically a few percentage points of total cannabinoids, because the remaining non-cannabinoid mass is removed; and the last of the volatiles and heads material, which is what actually fixes taste. What a second pass does not buy is a rescue for bad feed. Every pass adds heat history, and heat history is cumulative and irreversible, so a crude that needs three passes to look acceptable has been degraded three times to get there and its Δ8 and CBN numbers will say so. The correct response to a dark first pass is almost always to fix the upstream step — better winterization, a harder polish, a carbon or bentonite treatment of the crude, cleaner extraction — rather than to distil the same material again.

• Pass one: bulk separation. Pass two: colour, a few points of potency, and taste.
• Colour remediation belongs upstream, in winterization and polishing, not in a third distillation pass.
• Each pass is another full thermal exposure of the entire body cut.

Sources: Composite: extraction-equipment vendor documentation 2026*

Realistic potency outcomes contested industry practice, not published data

A well-run pass on crude that has been correctly decarboxylated and winterized typically lands in the high-80s to mid-90s percent total cannabinoids by mass, and a clean second pass on a good first-pass body cut can sit in the mid-90s. Those numbers are achievable and ordinary; they are not remarkable. The honest caveat is that input quality dominates the outcome far more than technique does. Feed that was extracted from poorly stored or poorly dried biomass, feed that was not fully decarboxylated, feed that still carries wax because winterization was rushed, and feed with residual solvent in it all cap the achievable potency and colour regardless of how well the still is run, because the non-cannabinoid mass either distils with the target, stays behind and holds product with it, or degrades and colours the distillate. The corollary is that a potency number quoted without the input specification tells you nothing about the process. The honest way to report a distillation is input assay, output assay, mass balance and the degradation markers, together.

Contested — caveat. Single-figure potency claims for short-path output vary widely across the trade and are usually quoted without the input assay, the number of passes or the mass balance. The high-80s to mid-90s band reflects ordinary competent practice on good decarboxylated, winterized feed; it is not a specification and it is not achievable from poor feed at any skill level.

Sources: Composite: extraction-equipment vendor documentation 2026*

Thermal degradation management: total heat history, not peak temperature

The single most useful idea in cannabinoid distillation is that degradation is driven by the integral of temperature over time, not by the peak temperature reached. Ten minutes at 170 °C can cost less product than four hours at 140 °C. That reframes every decision on the still: fill level, because a deeper charge takes longer to work through; vacuum depth, because deeper vacuum means lower temperature for the same rate; rate of take-off, because a slow run is a long run; pass count, because each pass re-exposes the whole cut; and whether to hold a finished flask hot while you deal with something else, because that is heat history with no separation being achieved in exchange. Two degradation markers will appear on the COA and both are diagnostic. Δ8-THC, or in a CBD-dominant material other isomerisation products, indicates acid- or heat-driven rearrangement — an isomer that was not in the plant appearing in the product is a process readout. CBN indicates oxidation of Δ9-THC, and it accumulates with heat, with time and with exposure to air and light; the storage literature going back to Fairbairn and colleagues established the same progression in stored cannabis and preparations, and Trofin and colleagues followed it in oils over long-term storage. If your product shows CBN that the feed assay did not, your process oxidised it. Limiting both markers is the same short list every time.

• Lower the temperature by deepening the vacuum before you raise the temperature to hit a rate.
• Shorten residence: smaller charges, steady take-off, do not hold a hot flask idle.
• Degas the feed properly before the body cut so you are not fighting volatiles at temperature.
• Exclude air. Oxidation to CBN needs oxygen; a leak is both a vacuum problem and a chemistry problem.
• Do not blend tails back into a fresh pass without accounting for it — you are importing an existing degradation history into a clean batch and it will show on the COA as CBN and colour.
• Assay input and output and keep the mass balance. Degradation you do not measure is degradation you will ship.

Sources: Fairbairn JW 1976* · Trofin IG 2012* · Perrotin-Brunel H 2011* · Composite: extraction-equipment vendor documentation 2026*

Safety, as part of the operation

Vacuum distillation hazards are mechanical and thermal rather than chemical, and they are entirely manageable by a person who was taught them as part of learning the operation. Evacuated glassware stores no energy in the way a pressure vessel does, but it fails inward, and a flask that implodes under vacuum while holding two litres of 160 °C oil produces a spray of hot product and glass. Inspect every piece of glass before it goes under vacuum, in good light, for star cracks, scratches, chips at the joints and any cloudiness or strain mark; a star crack is a radiating set of fine lines from a point of impact and it is the classic precursor to an implosion. Retire damaged glass rather than using it one more time. Never apply vacuum to flat-walled or thin-walled vessels not rated for it. Heating mantles fail closed and can run away, so a mantle is used with a controller and a temperature limit, never bare on a variable transformer and unattended. Hot oil and hot glass cause the majority of actual injuries: heat-resistant gloves, a face shield when breaking a hot joint, and never a hot flask lifted by the neck. If the feed still holds solvent, the first cut is a flammable vapour being pulled through a hot system toward a pump, so solvent stripping happens with ventilation, with the pump exhaust routed out, and with the recognition that a rotary-vane pump exhaust is not a place to put hydrocarbon. Cold traps are a cryogenic hazard: dry ice and slush baths at −78 °C and liquid nitrogen at −196 °C cause contact burns through ordinary gloves in seconds, containers must be vented so they cannot pressurise, and a liquid-nitrogen trap left open to air condenses liquid oxygen, which with organic residue is an oxidiser hazard. Peroxide-forming solvents — diethyl ether, tetrahydrofuran, diisopropyl ether, dioxane — must never be concentrated to dryness anywhere on this equipment: peroxides accumulate in the residue and the residue is where the energy ends up. Finally, every vacuum system needs a defined way to be brought back to atmosphere deliberately, slowly and at a controlled point, so nobody ever vents a hot flask by pulling a hose off.

• Inspect for star cracks, chips and scratches every time, before vacuum, in good light.
• Controller and limit on every mantle; never unattended on a bare transformer.
• Face shield and heat gloves for hot joints; never lift a hot flask by the neck.
• Route pump exhaust outside; keep flammable first cuts ventilated.
• Cryogenic traps: vent the container, never seal it, treat −78 °C and −196 °C as burn hazards, and do not leave a liquid-nitrogen trap open to air.
• Never evaporate a peroxide-forming solvent to dryness.
• Have a defined, slow, controlled path back to atmospheric pressure.

Sources: National Research Council (US) 2011* · Kelly RJ 1996* · Armarego WLF 2017* · Composite: extraction-equipment vendor documentation 2026*