Carriers and Diluents, Organised by Route

Why this page is organised by route human data

Every safety evaluation a food or cosmetic ingredient has ever passed was an evaluation of a specific exposure route at a specific level. Generally-recognised-as-safe status in the United States is a determination about ingestion. A cosmetic ingredient review is a determination about dermal contact. Neither is a determination about what happens when the material is heated on a metal coil to several hundred degrees and drawn into an alveolus. The lung is not a modified stomach: it has no acid stage, no bile, no first-pass hepatic clearance, a surface area of the order of 70 square metres, a lining a single cell thick, and a clearance mechanism designed for particles and not for oils. Material that the gut would emulsify and metabolise sits in the lung and gets phagocytosed, or does not. This is the reason a diluent table must be indexed by route, and the reason the most common and most expensive formulation error in this industry is a category error rather than a chemistry error.

• Ingestion, sublingual absorption, dermal application and inhalation are four different toxicological questions about the same molecule.
• A supplier certificate saying food grade answers the ingestion question and is silent on the other three.
• The burden runs the right way round: for an inhaled product, absence of inhalation data is a reason not to use a material, not a neutral state.

Sources: U.S. Food 2020* · Rowe RC 2020* · Marchiori E 2011*

MCT oil (fractionated coconut oil) human data

Medium-chain triglyceride oil is the default oral and sublingual carrier for cannabinoids and deserves that position. It is a fractionated coconut or palm kernel product consisting predominantly of the C8 (caprylic) and C10 (capric) triglycerides, with the longer chains removed, which gives it three useful properties: it is liquid and stays liquid at refrigerator temperature unlike whole coconut oil, it is far more oxidatively stable than a polyunsaturated seed oil because it is essentially saturated, and it is close to colourless and neutral in taste. Pharmacokinetically it does real work rather than just filling volume: cannabinoids are highly lipophilic and their oral absorption is strongly food- and lipid-dependent, and co-administration with lipid excipients raises systemic exposure substantially. Medium-chain triglycerides are also handled differently from long-chain fats, being absorbed more directly, which is the rationale offered for preferring them. The hazard is route-specific and well documented: aspirated or inhaled lipid causes exogenous lipoid pneumonia, a recognised clinical and radiological entity, so MCT is a good oral carrier and is not an inhalation diluent regardless of how cleanly it mixes with distillate.

• Composition: predominantly C8 and C10 triglycerides; specification usually states the C8 to C10 ratio.
• Oral and sublingual: appropriate, and lipid co-administration meaningfully increases cannabinoid exposure.
• Inhalation: not appropriate. Exogenous lipoid pneumonia is the documented outcome of lipid in the lung.
• Practical note: MCT is a solvent for many plastics and elastomers over time; check container and gasket compatibility.
• Oxidative stability is good but not infinite; it is still stored cool, dark and with minimal headspace.

Sources: Rowe RC 2020* · Zgair A 2016* · Millar SA 2018* · Marchiori E 2011*

Propylene glycol and vegetable glycerin: the e-liquid heritage and its thermal chemistry contested in vitro

Propylene glycol and glycerin arrived in cannabinoid products by inheritance: they are the base of nicotine e-liquid, the hardware was designed around their viscosity and wicking behaviour, and the supply chain already existed. They are not cannabinoid solvents in the way a lipid is — cannabinoids have limited solubility in glycerin especially — so in a cannabinoid vape they function as thinning and aerosol-forming agents rather than as the carrier proper, and heavily PG- or VG-cut cannabinoid fills tend to separate. Both are hygroscopic, glycerin strongly so, which changes the water content of a stored fill and with it the aerosol. The important body of evidence is thermal. Heating these two compounds on a coil produces carbonyls, and the chemistry is not obscure: propylene glycol and glycerin dehydrate and fragment to formaldehyde, acetaldehyde and acrolein, glycerin being the principal acrolein source, and the yields rise sharply with coil temperature, power and dry or low-liquid operation. This has been measured repeatedly in the e-cigarette aerosol literature, including a widely discussed report of formaldehyde-releasing species in aerosol at high voltage and systematic work showing carbonyl output as a function of solvent and battery output.

• PG: lower viscosity, carries flavour, more throat hit, and the more cannabinoid-tolerant of the two.
• VG: higher viscosity, more visible aerosol, strongly hygroscopic, poor cannabinoid solvent.
• Documented thermal degradation products: formaldehyde, acetaldehyde, acrolein. Acrolein output is associated particularly with glycerin.
• Yields are power- and temperature-dependent, and worst under dry-wick or low-liquid conditions. Device behaviour is part of the product hazard.
• PG has documented airway-irritant properties on inhalation in occupational settings; that is separate from the degradation chemistry.
• A PG or VG cut also dilutes the cannabinoid mass per unit volume, which the label arithmetic has to account for.

Contested — caveat. The existence and identity of the carbonyl degradation products is well established. The quantitative yields reported across the e-cigarette literature vary widely with device, power, coil condition and puff regime, and some early high-yield reports were criticised for using operating conditions a user would find unpalatable. Treat the qualitative chemistry as settled and any specific yield figure as device- and condition-dependent.

Sources: Jensen RP 2015* · Sleiman M 2016* · Kosmider L 2014* · Rowe RC 2020* · Meehan-Atrash J 2019*

Ethanol human data

Ethanol is the traditional carrier for a tincture and remains the correct choice for one. It dissolves cannabinoids readily, it is self-preserving above roughly 20 percent by volume, it permits sublingual absorption, and it evaporates cleanly, which is why it doubles as the extraction and winterisation solvent upstream. Its formulation limits are its taste and mucosal burn at high strength, the mouthfeel of a high-ethanol dose, its incompatibility with anyone avoiding alcohol, and its regulatory treatment as an alcoholic beverage ingredient in some jurisdictions once the product is consumable. In a vape it is not a diluent: it is volatile enough to change the aerosol dramatically, it is an airway irritant, and the practice of thinning distillate with ethanol produces a harsh product and an unquantified residual-solvent question. Ethanol also appears on this page in a second role — as the reason the 10-ethoxy cannabitriol ethers may be extraction artefacts, which is a reminder that a solvent is a reagent whenever the conditions allow.

• Oral and sublingual tincture: appropriate, well established, self-preserving at sufficient strength.
• Topical: used as a penetration aid and as a carrier in sprays; drying and irritating at high load.
• Inhalation: not appropriate as a formulation diluent, and residual ethanol from processing is a specification limit rather than an ingredient.
• Residual-solvent limits for ethanol in a finished product come from the pharmacopoeial residual-solvent framework and from state cannabis regulation; see the residual-solvent page.

Sources: Rowe RC 2020* · Hazekamp A 2007* · ElSohly MA 2005*

Olive oil and hemp seed oil human data

Olive oil is the carrier with the longest documented pharmaceutical history for cannabis preparations and has a genuine evidential advantage: the standardised olive-oil preparations used in medical-cannabis research were characterised and their cannabinoid content quantified, so the behaviour of the matrix is known. It contributes flavour, it is nutritionally unobjectionable, and its mostly monounsaturated fatty-acid profile is reasonably stable. Hemp seed oil is chosen mostly for narrative reasons — a hemp product in a hemp oil — and it has a real formulation drawback: it is high in polyunsaturated fatty acids, principally linoleic and alpha-linolenic, which makes it the least oxidatively stable of the common carriers, prone to rancidity, and demanding of cool dark storage and honest shelf-life dating. It also carries a labelling confusion the industry has never fixed, because hemp seed oil contains no meaningful cannabinoid content of its own and consumers routinely read the name as meaning it does. Neither oil belongs in an inhaled product, for the lipid reason that applies to all of them.

• Olive oil: documented pharmaceutical carrier for cannabis preparations, moderate stability, strong flavour contribution.
• Hemp seed oil: poor oxidative stability from its polyunsaturated profile; requires the shortest shelf life of the common carriers.
• Hemp seed oil contains no significant cannabinoids; any cannabinoid in the product was added.
• Both: oral only. Neither is an inhalation diluent.
• Antioxidant strategy for either: minimise headspace, store cool and dark, and date the product from fill rather than from sale.

Sources: Hazekamp A 2007* · Rowe RC 2020* · Marchiori E 2011*

Vitamin E acetate and EVALI: the central cautionary case, told properly contested human data

In mid-2019 hospitals in the United States began reporting a severe acute lung injury in otherwise healthy people who vaped, and by the time the outbreak investigation closed the CDC surveillance summary recorded on the order of 2800 hospitalised cases and at least 68 deaths. The case-finding pointed overwhelmingly at informally supplied THC-containing cartridges. The chemical answer came from bronchoalveolar-lavage fluid: vitamin E acetate, alpha-tocopheryl acetate, was identified in lavage samples from the large majority of sampled patients and was not found in the comparison samples, and it was found in product samples associated with cases. The reason it was in the cartridges is the part a formulator needs to understand, because it is a formulation decision and not an accident. Vitamin E acetate is cheap, food- and cosmetic-legal, and its colour, viscosity and refractive behaviour are close enough to cannabinoid distillate that a cartridge cut heavily with it still looked, poured and hung on the glass like uncut distillate — it passed the only quality tests an informal supply chain applied, which were visual. Subsequent analytical work found that it hydrogen-bonds with tetrahydrocannabinol in the liquid, which helps explain why cut fills behaved like genuine ones, and pyrolysis chemistry showed that heating vitamin E acetate can release ketene, a potently toxic gas, giving a plausible mechanism for the acute injury. Every property that made it attractive was a property of the liquid; none was a property of the aerosol.

• Scale: approximately 2800 hospitalisations and at least 68 deaths in the CDC final surveillance summary; cases concentrated in informally sourced THC cartridges.
• Marker: vitamin E acetate in bronchoalveolar-lavage fluid in the great majority of sampled cases, and in associated product samples.
• Why it was used: cheap, food- and cosmetic-legal, and visually and rheologically indistinguishable from distillate at typical cut rates.
• Why that mattered: the supply chain tested by eye, and the adulterant was designed, in effect, to pass a visual test.
• Proposed mechanism: thermal release of ketene on vaping pyrolysis, alongside the general lipid-in-lung mechanism.
• The generalisable lesson is not about one molecule. It is that rheological and visual equivalence to distillate is exactly the property a dangerous cut will have.

Contested — caveat. Vitamin E acetate is the dominant and best-supported cause and the association with lavage fluid is strong, but the outbreak investigation did not establish that it was the sole agent in every case, and other contributors including other lipids and coating or metal contaminants were raised in the literature. The ketene mechanism is chemically plausible and demonstrated in pyrolysis experiments rather than proven as the in-vivo mechanism in patients. Case and death figures are the CDC surveillance totals at the close of the investigation and are subject to the limits of case-based surveillance.

Sources: Blount BC 2020* · Krishnasamy VP 2020 · U.S. Centers for Disease Control 2020* · Lanzarotta A 2020* · Wu D 2020*

Emulsifiers and surfactants industry practice, not published data

An emulsifier is not a carrier; it is the interfacial agent that lets a lipophilic active exist as a dispersed phase in an aqueous one, which is what any cannabinoid beverage, water-soluble powder or fast-onset edible depends on. Three families dominate. Lecithin, a phospholipid mixture from soy or sunflower, is a food-heritage emulsifier that is cheap, label-friendly, and moderately effective; it produces relatively coarse emulsions on its own and contributes flavour. The polysorbates, principally polysorbate 80 and polysorbate 20, are non-ionic ethoxylated sorbitan esters with high emulsifying power at low inclusion and are the workhorse of commercial nanoemulsion systems; their limitations are a soapy taste at higher inclusion, a consumer-perception problem because of the ethoxylated character, and — the important one — no inhalation qualification whatever. Gum acacia and the other hydrocolloids, along with modified starches, work by a different mechanism: they are stabilisers that raise continuous-phase viscosity and provide steric protection at the droplet surface rather than dropping interfacial tension much, so they hold an emulsion that something else created. In practice a commercial system uses a high-power surfactant to form the droplets and a hydrocolloid to keep them from coalescing.

Sources: Rowe RC 2020* · McClements DJ 2012* · Barrus DG 2016*

Nanoemulsion and liposomal claims against the evidence contested human data

The underlying science is real and should not be dismissed along with the marketing. A cannabinoid taken orally in an oil has low and highly variable bioavailability, limited by poor aqueous solubility, by food effects, and by extensive first-pass metabolism; the human pharmacokinetic literature on cannabidiol documents exactly that variability. Reducing the dispersed lipid phase to droplets in the tens to low hundreds of nanometres increases interfacial area enormously, keeps the active presented to the intestinal wall in a solubilised state, and in self-emulsifying and pro-nanoliposphere systems has been shown in human volunteers to raise exposure and shorten time to peak relative to a simple oil. That is a genuine pharmacokinetic rationale with human data behind specific systems. The marketing claims are a different matter. Label statements of particle size are frequently unverifiable by the purchaser, because measuring droplet size requires dynamic light scattering on the diluted product under controlled conditions and the number on the carton is rarely traceable to a method. The words nanoemulsion, liposomal and water-soluble are used loosely and often interchangeably for systems that are none of those things — a true liposome is a phospholipid bilayer vesicle, not a surfactant-stabilised oil droplet, and a genuinely water-soluble cannabinoid is a chemically modified molecule rather than a dispersed one. And bioavailability multipliers quoted on packaging are almost never traceable to a human pharmacokinetic study of that product. The fair summary: the mechanism is sound, some specific formulations have human data, and the claim printed on a given commercial label usually does not.

• Sound: smaller droplets, larger interfacial area, better solubilisation, faster and higher absorption for a lipophilic active. Demonstrated for specific self-emulsifying systems in human volunteers.
• Sound: onset time genuinely shortens with a fine emulsion, which is the property beverage products are actually selling.
• Unverifiable as usually presented: a stated droplet size with no method, and a stated bioavailability multiple with no study of that product.
• Terminology abuse: liposomal, nanoemulsion and water-soluble describe different things and are used as synonyms in the trade.
• Stability is the hidden problem: fine emulsions coarsen by Ostwald ripening and coalescence over shelf life, so the size at fill is not the size at sale unless someone measured it at sale.
• What to ask a supplier: the measurement method, the instrument, the dilution protocol, the time point, and whether there is human pharmacokinetic data on this formulation rather than on the concept.

Contested — caveat. The pharmacokinetic mechanism is well supported and human data exist for particular self-emulsifying and pro-nanoliposphere formulations. Specific commercial claims — a named particle size, a bioavailability multiple, a liposomal designation — are generally not supported by published data on the product making the claim, and should be treated as marketing until a method and a study are produced. Nothing here says a given product does not work; it says the published basis for the printed number is usually absent.

Sources: McClements DJ 2012* · Cherniakov I 2017* · Izgelov D 2020* · Millar SA 2018* · Zgair A 2016*

The route table human data

The summary table. Read it as a route qualification, not as a ranking of materials: nearly every entry is appropriate somewhere and hazardous somewhere else, which is the entire point of the page. Where the hazard column says no inhalation data, that is a statement about the record and it is a reason for caution rather than a finding of harm — but for an inhaled product it is a reason not to use the material.

Sources: Marchiori E 2011* · Rowe RC 2020* · Hazekamp A 2007* · Jensen RP 2015* · Sleiman M 2016* · Kosmider L 2014* · Blount BC 2020* · Krishnasamy VP 2020 · Wu D 2020* · McClements DJ 2012* · ElSohly MA 2005* · Meehan-Atrash J 2017* · Meehan-Atrash J 2019* · MELEK hemp-science shelf 2026*