From soybean to bottle: how industrial seed oils are actually made

The seed oil debate is one of the most polarized food arguments online. One side calls them “industrial poison” and recommends tallow and butter for everything. The other side calls the criticism pseudoscience and points to decades of cardiovascular research using polyunsaturated vegetable oils. Both sides overclaim.

This piece does not try to settle that debate. It does something more useful: it walks through, in detail, what actually happens to a soybean between the field and the bottle on a grocery store shelf. Most people arguing about seed oils online — including a lot of people selling alternatives — have never seen the process described accurately. Once you have, you can decide for yourself whether the process bothers you.

The two oils covered here are soybean and canola, because between them they account for the overwhelming majority of US seed oil consumption through both bottled product and the “vegetable oil” used in restaurant kitchens and packaged food. The other industrial seed oils (corn, cottonseed, sunflower, safflower, grapeseed, rice bran) are made via essentially the same process with minor variations noted at the end.

The crops

Soybean is the largest US oilseed crop by a wide margin. USDA NASS reports recent annual production of roughly 4.2 billion bushels, grown primarily in Illinois, Iowa, Indiana, Minnesota, and Nebraska. Roughly 96% of US soybean acres are planted with herbicide-tolerant genetically engineered varieties (predominantly glyphosate-tolerant, with a growing share of dicamba-tolerant and stacked-trait cultivars), per USDA Economic Research Service adoption data. The majority of the US crop is now crushed domestically — the USDA projects about 57% of the 2025/26 crop will be crushed in the US, driven by renewable diesel demand for soybean oil — with the remainder exported. The crushing yields two products: oil (roughly 18-20% of seed weight) and high-protein meal (roughly 78-80%) used predominantly for animal feed.

Canola is a younger crop with a specific origin story. It was developed in the 1960s and 1970s by Keith Downey at Agriculture Canada (Saskatoon Research Station) and Baldur Stefansson at the University of Manitoba, who bred rapeseed cultivars with both low erucic acid (under 2%) and low glucosinolates. The name “canola” was originally registered as a trademark in 1978 (combining “Canadian” and “oil”) and is now treated largely as a generic term with a regulated technical definition tied to those compositional thresholds. Canola seed contains roughly 40-45% oil by weight — a notably higher oil yield than soybean. Most North American canola is grown on the Canadian prairies (Saskatchewan, Alberta, Manitoba) and in North Dakota.

Both crops arrive at a crush plant as cleaned, dried seeds. Everything that follows happens in a single industrial facility, usually attached by rail to the growing region.

Cleaning, cracking, and flaking

The first stage is mechanical preparation, and it’s the simplest part of the process. Soybeans are screened for stones, metal, and crop residue, then dried to a target moisture content of around 10-11%. The seeds are then cracked between large fluted rollers into roughly four to six pieces each. Cracking exposes the cotyledon (the oil-bearing tissue) and separates it from the hull, which is removed via aspiration. Dehulling matters because the hulls contain very little oil and would dilute extraction yield.

The cracked, dehulled cotyledon pieces are then conditioned by heating with steam to roughly 60-70°C and softened, then passed through smooth-surfaced flaking rollers that produce flakes about 0.25-0.40 mm thick. The flakes are the form the next step requires: thin enough for solvent to penetrate, large enough not to clog the extractor.

Canola seeds are smaller, contain more oil, and aren’t dehulled. Instead, they typically go through a pre-press step: a screw press (expeller) that mechanically squeezes out roughly 60-70% of the seed’s oil before solvent extraction handles the rest. This two-stage approach is standard for high-oil seeds. Soybeans, with their lower oil content, go straight to solvent extraction without a pre-press.

Hexane extraction

This is the step the seed oil critics focus on, and it’s worth understanding precisely.

The solvent used is n-hexane, a six-carbon hydrocarbon refined from petroleum. It’s specifically chosen because it dissolves triglycerides (fat) efficiently but not phospholipids or other polar compounds, has a low boiling point (about 69°C) that allows easy recovery, and is significantly more efficient than mechanical pressing — extracting roughly 99% of available oil versus 60-70% for an expeller press alone. This extraction efficiency is the reason hexane is used industry-wide for soybean and the secondary extraction stage for canola.

The flakes (or pre-pressed canola cake) enter a large countercurrent extractor — typically a rotary basket or sliding-cell design — where they’re repeatedly washed with hexane. The result is a mixture called miscella: roughly 25-30% oil dissolved in hexane. The extracted solids (meal) and the miscella exit the extractor as separate streams.

Hexane is regulated at multiple federal levels:

• FDA permits n-hexane as a food-additive extraction solvent under 21 CFR 173.270. That regulation sets explicit residual limits for hexane only in spice oleoresins (25 ppm) and hops extract (2.2%); it does not set a vegetable-oil-specific residual limit. Residuals in finished vegetable oils are governed by general food-safety practice and in international trade are bounded by the European Union maximum residue limit of 1 mg/kg (1 ppm) in fats and oils.
• OSHA sets a permissible workplace exposure limit (PEL) for n-hexane of 500 ppm as an 8-hour time-weighted average, under 29 CFR 1910.1000. OSHA itself notes this PEL is outdated; NIOSH’s recommended exposure limit and the ACGIH threshold limit value are both 50 ppm (ten times lower), and well-run modern plants operate well below the OSHA number.
• EPA lists n-hexane as a hazardous air pollutant under the Clean Air Act. Crush plants are required to report hexane emissions to the Toxics Release Inventory (TRI) and operate emissions controls.

Residual hexane in finished, fully refined oil is typically measured in the low single-digit parts per million or below — well under the FDA-permitted limit. Peer-reviewed measurement studies in JAOCS (the Journal of the American Oil Chemists’ Society) place typical residuals around 0.1-1 ppm in fully refined soybean oil after the downstream steps that follow.

Whether very low residual hexane in finished oil matters as a consumer health question is the kind of debate this article isn’t entering. What’s beyond debate is that the hexane is there, the regulation exists because the solvent is real, and the workplace and environmental controls exist for the same reason.

Desolventizing and meal recovery

After extraction, the two streams diverge.

The miscella (oil + hexane) is heated in evaporators to drive off the hexane, which condenses and is recovered for reuse. A final steam-stripping stage in a vacuum stripper removes the last traces of solvent. What remains is crude soybean (or canola) oil — dark, strong-smelling, and not yet suitable for food use.

The defatted solids (the meal) go to a desolventizer-toaster (DT), a vertical vessel where steam is injected through the meal to evaporate residual hexane and recover it. The “toasting” step also heat-treats the meal to deactivate trypsin inhibitors and other anti-nutritional factors so the meal can be used in animal feed. The resulting soybean meal is what feeds most of the US livestock industry. Hexane recovery rates are very high — typically over 99% — but the small fraction not recovered is part of what the EPA emissions reporting tracks.

Degumming

Crude oil from extraction contains 1-3% phospholipids (lecithin), which would cause cloudiness, off-flavors, and shortened shelf life if left in. Degumming removes them.

In water degumming, the crude oil is mixed with about 1-2% water at 60-90°C. Phospholipids are hydrophilic and hydrate, becoming insoluble in the oil and separable by centrifugation. The recovered gum is dried and sold as commercial lecithin — soy lecithin is a common food additive and emulsifier (FDA GRAS, generally recognized as safe), used in everything from chocolate bars to non-stick spray.

For tighter specifications, water degumming is followed by acid degumming using phosphoric or citric acid, which removes non-hydratable phospholipids that water alone can’t. By the end of degumming, the oil is clearer but still contains free fatty acids, color compounds, and other impurities.

Neutralization (caustic refining)

Crude oil typically contains 0.5-3% free fatty acids — fatty acids that have hydrolyzed off the triglyceride backbone. These shorten shelf life, generate off-flavors, and smoke at lower temperatures than triglycerides. Neutralization removes them.

The standard method is alkali refining: sodium hydroxide (lye) is added in a controlled stoichiometric amount, reacting with the free fatty acids to form soap (soapstock). The mixture is heated, agitated, and then separated by centrifuge — the soapstock falls out as an aqueous layer and the cleaned oil floats. The oil is then water-washed to remove residual soap.

The soapstock byproduct is not waste. It’s typically acidulated (acid-treated to split the soap back into free fatty acids), dried, and sold as feed-grade acidulated soapstock or processed into fatty acid distillates used in industrial applications like biodiesel feedstock, animal feed energy supplements, and soap manufacture.

After neutralization, the oil is lighter in color, lower in free fatty acids, and ready for the next step.

Bleaching

“Bleaching” in oil refining doesn’t mean chlorine or peroxide; it’s an adsorption step. The oil is mixed with bleaching clay — typically activated bentonite or montmorillonite — at 90-110°C under partial vacuum. The clay adsorbs color compounds (chlorophyll, carotenoids), residual phospholipids, soap traces, trace metals, and oxidation products like peroxides.

The clay dose is typically 0.5-2% of oil weight. After roughly 20-40 minutes of contact under vacuum, the mixture is filtered, recovering the clarified oil and the spent clay. The spent clay, now saturated with the adsorbed material, is disposed of or sent for further processing — it’s also part of the residual material stream tracked under environmental reporting.

A second function of bleaching that gets less attention: it also adsorbs much of the residual hexane that survived the earlier evaporation steps, contributing to the very low final hexane levels measured in deodorized oil.

Deodorization

Deodorization is the highest-temperature step in the process and the one with the most consequential side effects.

The bleached oil is heated under high vacuum (typically 2-6 millibar absolute pressure) and treated with direct steam injection at temperatures of 230-260°C for 30-60 minutes. The high temperature and low pressure together allow volatile compounds — the remaining free fatty acids, aldehydes, ketones, peroxide breakdown products, and the compounds responsible for the strong odor of crude oil — to vaporize and be carried off in the steam.

This is the step that transforms a dark, fishy, beany-smelling crude oil into the neutral, clear, shelf-stable bottle product consumers recognize.

It also has two consequences worth understanding, both well documented in peer-reviewed processing chemistry:

Trans fatty acid generation. The high temperatures and long residence times of deodorization cause partial cis-trans isomerization of polyunsaturated fatty acids. Even in oils with no industrially hydrogenated component (the kind banned under FDA’s partially hydrogenated oils rule), deodorized vegetable oils typically contain 0.5-3% trans fats as a process byproduct, measured in studies published in JAOCS and Food Chemistry. Lower deodorization temperatures and shorter times reduce this; some refiners have moved to “soft deodorization” protocols specifically to minimize trans fat formation. The trans fats here are not the same isomers as those from industrial partial hydrogenation, but they are trans fats by definition.

3-MCPD and glycidyl esters. Heating refined oils at high temperatures generates 3-monochloropropane-1,2-diol (3-MCPD) esters and glycidyl esters — process contaminants that the European Food Safety Authority (EFSA) and the WHO/FAO Joint Expert Committee on Food Additives (JECFA) have flagged as health concerns. EFSA established a tolerable daily intake for 3-MCPD in 2018. The European Commission has set maximum levels for these compounds in refined oils (Regulation (EU) 2020/1322). The US has not set parallel limits, but the chemistry happens in US-refined oils the same way it happens in European ones. Mitigation involves controlling chloride precursors earlier in the process and using lower-temperature deodorization.

These are not contested findings. They appear in industry technical literature, in EFSA scientific opinions, and in JAOCS papers on deodorization optimization. The frame is the same as with hexane: the process produces these compounds, regulators have characterized them, and what you do with that information is your decision.

Winterization

The final optional step. Some oils — particularly sunflower, but also some canola batches — contain higher-melting-point waxes or saturated triglycerides that crystallize at refrigerator temperatures and make the oil cloudy. Winterization solves this by cooling the oil slowly to 0-7°C, holding it for hours to days while crystals form, then filtering them out.

For soybean oil, winterization is uncommon — soybean oil’s fatty acid profile keeps it clear at refrigerator temperatures without dewaxing. For canola, dewaxing is done when targeting clarity specs for retail bottles.

After winterization (or, more commonly for soybean, straight after deodorization), the oil is bottled. The product on the shelf is a neutral, clear, near-tasteless liquid with a shelf life typically rated at 12-24 months.

Canola: the parallel pass

Canola processing follows the same flow with minor differences already noted:

• Smaller seeds, no dehulling — canola seeds are too small to hull individually, so the hulls go through extraction with the rest.
• Pre-pressing first — mechanical expeller press takes out roughly 50-70% of the oil before solvent extraction handles the remainder. This is why you sometimes see “expeller-pressed canola” labeled as a real distinction: that oil came from the pre-press stage only and skipped the hexane step. Most bottled canola is a blend of pre-press and solvent-extracted oil unless explicitly labeled otherwise.
• Identical refining — once you have crude canola oil, the degumming, neutralization, bleaching, deodorization, and (sometimes) winterization stages are essentially the same as soybean.
• Higher monounsaturated content — canola is roughly 60% oleic acid, versus soybean’s roughly 23%. This higher monounsaturated content makes canola somewhat more oxidatively stable than soybean oil in cooking applications, which is part of why it dominates restaurant fryer use.

The other industrial seed oils (corn, sunflower, safflower, cottonseed, grapeseed, rice bran) follow the same general template. Cottonseed has an additional historical step — gossypol removal, since the cotton plant produces this naturally occurring toxin in the seed — but the refining cascade after that is the same.

Label-reading addendum

The terminology on a bottle or ingredient list has specific definitional meaning that’s worth knowing:

• “Vegetable oil” on a US ingredient label is a specific FDA-regulated term. Under 21 CFR 101.4(b)(14), a single, unblended oil ingredient must be declared by its specific common name (“soybean oil,” “canola oil,” etc.) — not as generic “vegetable oil.” The umbrella term is reserved for blends, which must still name the specific oils used (e.g., “vegetable oil shortening (soybean and cottonseed oil)” or, for variable blends, “contains one or more of the following: cottonseed oil, palm oil, soybean oil”). Because soybean oil is the highest-volume edible oil in the US food supply, it’s the dominant component in most blended “vegetable oil” products, but the label itself shouldn’t be read as a synonym for it.
• “Refined” means the oil went through the full degumming → neutralization → bleaching → deodorization cascade described above. Almost all bottled cooking oils sold in US grocery stores are refined unless explicitly labeled otherwise.
• “Expeller-pressed” means the oil was extracted mechanically with a screw press only, without a hexane solvent step. This is a real, meaningful label distinction. Note that expeller-pressed oils are typically still refined afterward (deodorized at high temperature), so they still go through most of the downstream processing — they just skip the solvent step.
• “Cold-pressed” means the oil was mechanically expressed without applying external heat. There’s no FDA definition of “cold-pressed” for US-sold oils, but the International Olive Council and EU regulations set temperature thresholds (typically under 27°C) for the term as used on olive oil. For other oils, “cold-pressed” is largely a marketing term without a binding standard.
• “Unrefined” means the oil skipped the degumming/neutralization/bleaching/deodorization stages. Unrefined oils are typically darker, have stronger flavor, lower smoke points, and shorter shelf lives. Unrefined seed oils are uncommon at supermarket scale and more common in specialty channels.
• “High-oleic” refers to genetically or conventionally bred varieties (high-oleic soybean, high-oleic sunflower, high-oleic safflower) bred for elevated monounsaturated fat content and higher oxidative stability. High-oleic varieties are widely used in restaurant frying because they last longer in the fryer before degrading. The processing is the same as conventional varieties; only the starting seed composition differs.

None of this is a recommendation to seek or avoid any particular label term. They’re just definitions, and reading a bottle accurately requires knowing them.

For contrast: how other fats are made

For perspective, here’s what the process looks like for the fats and oils that don’t go through the industrial cascade above:

• Extra virgin olive oil. Olives are washed, crushed into paste, the paste is malaxed (slowly stirred) for 20-40 minutes at temperatures the International Olive Council requires to stay below 27°C for the “cold extracted” designation, then the oil is separated by centrifugation. No solvent, no degumming, no bleaching, no deodorization. The total process is mechanical and thermal-low.
• Butter. Milk is separated into cream and skim by centrifugation, the cream is churned until the fat globules coalesce and separate from buttermilk, the butter is washed and salted (if salted butter). No solvent, no high-temperature stages.
• Tallow and lard. Beef or pork fat trim is heated (rendered) until the fat melts and separates from connective tissue, then strained. Traditional rendering is done at relatively low temperatures (under 80°C for “leaf lard” or “wet-rendered” tallow); commercial dry rendering operates higher. No solvent.
• Virgin coconut oil. Coconut meat is pressed (cold-pressed varieties stay under ~50°C; expeller-pressed varieties run higher) and the oil is separated. Refined coconut oil goes through deodorization and sometimes bleaching, but virgin coconut oil skips this.

The contrast isn’t being offered as an argument for one fat over another. It’s just the comparison: the industrial seed oils on a US grocery shelf went through a process with roughly 8-10 distinct stages including a petroleum solvent, multiple high-temperature treatments, and chemical refining. The traditional fats listed above went through 1-3 mechanical or thermal stages. Whether that matters to you is a separate question.

What’s hopefully clearer now is what the process actually is — not what either side of the seed oil argument claims it is.