Refined or unrefined shea butter: what does refining take out?
Refining took free fatty acids from 12.2% to zero, while physical refining kept unsaponifiables at 5.6-6%. Tocopherols, colour and shelf life, measured.
Refining takes out what makes crude shea smell, look and keep like a raw product: almost all of its free fatty acids, most of its peroxides, its colour, its smoky odour, and a share of its minor antioxidants, chiefly the gamma and delta tocopherols, squalene and about a third of the flavonoids. It does not, on its own, strip the unsaponifiable fraction that shea is bought for. In a laboratory physical refining trial the unsaponifiables went from 5.6 percent in, to 5.8 to 6.0 percent out, and a commercial refined, bleached and deodorised shea lists 7 percent. The large losses that get blamed on refining come from fractionation, which splits shea into a hard stearin and a soft olein. This page puts the measured numbers for each step in one place.
Buy unrefined if you want the smell and the full set of minor components, and accept a fat that may arrive with high free fatty acids (12.2 percent in one measured crude). Buy refined whole shea for a neutral, pale, consistent butter: it keeps roughly the same unsaponifiables and fatty acids, loses the gamma and delta tocopherols, and in the one direct test lasted longer on the Rancimat. Avoid "refined shea" that is really shea stearin if you want the unsaponifiables: the stearin measured 0.2 percent.
What refining removes, measured
The first page of search results for this question is almost entirely shea sellers' blogs, a soap making forum thread and a copied page. They agree that unrefined shea keeps "vitamins A, E and F", that refining uses heat and "harsh chemicals", and that the refined butter loses most of its goodness. None of them gives a single measured figure. This site's own pages are not much better on this point: the grade table on shea butter and the refined row in butters compared describe the differences in words, not numbers. This page adds the numbers.
Three sources measure what refining does. Abdel-Razek and colleagues (2023, Foods) analysed a crude Ghanaian shea butter alongside a refined shea stearin, a refined shea olein, and a 1:1 mix of the two, all supplied by Bunge Loders Croklaan. Ruiz Méndez and colleagues at the Instituto de la Grasa in Seville (1991, Grasas y Aceites) physically refined two crude shea butters from Upper Volta, now Burkina Faso, step by step at laboratory scale, measuring after each step. And AAK's product data sheet for LIPEX Shea, a refined, bleached and deodorised cosmetic shea, states what one commercial refined grade is guaranteed to be at manufacture.
| Measure | Crude | Refined olein | Refined stearin | Refined 1:1 mix |
|---|---|---|---|---|
| Free fatty acids, % | 12.24 | 0.00 | 0.00 | 0.11 |
| Peroxide value, meq O2/kg | 5.91 | 0.63 | 0.24 | 0.30 |
| Unsaponifiable matter, % | 5.06 | 6.38 | 0.22 | 4.06 |
| Total tocopherols, mg/100 g | 10.06 | 9.25 | 1.27 | 4.94 |
| Alpha-tocopherol, mg/100 g | 8.52 | 8.28 | 1.06 | 4.58 |
| Gamma plus delta tocopherol, mg/100 g | 0.51 | Not detected | Not detected | Not detected |
| Squalene, mg/g | 0.20 | 0.05 | 0.03 | 0.04 |
| Squalene, lanosterol and amyrins, mg/g | 20.83 | 19.85 | 3.69 | 14.81 |
| Total phenolics, mg/g | 0.21 | 0.19 | 0.19 | 0.21 |
| Total flavonoids, mg catechin/100 g | 40.14 | 27.61 | 26.46 | 29.20 |
| Carotenoids, HPLC | Not detected | Not detected | Not detected | Not detected |
| Slip melting point, C | 32 | 25 | 36 | 31 |
| Rancimat induction period, h | 11.5 | 15.9 | 12.5 | 18.5 |
Read across the olein column first. Apart from the free fatty acids, peroxides, flavonoids and the two minor tocopherols, refined olein kept very nearly everything crude shea had: the paper's abstract says its unsaponifiables, phenolics, tocopherol and sterol content "was unchanged". Now read the stearin column. That is where the unsaponifiables fall from 5.06 to 0.22 percent and the tocopherols from 10.06 to 1.27 mg. The difference between those two columns is not refining, since both fractions were refined. It is fractionation, and the authors say so plainly: the minor components "selectively preferred to migrate to the liquid fraction (olein)".
What the four refining steps actually do
Abdel-Razek lists the four steps of making refined shea from crude: degumming, neutralisation, bleaching and deodorisation. Ruiz Méndez and colleagues ran three of them separately, leaving out neutralisation, which is what "physical refining" means: the free fatty acids are distilled off in the deodoriser instead of being washed out as soap. Their aim was the refining a cosmetic buyer of the time specified, with three limits: free acidity under 3 percent as oleic acid, colour no higher than 4 on the Gardner scale, and a total absence of foreign odours. They added a fourth aim of their own, to keep the unsaponifiable fraction at its starting level, because the market value of the butter rose with it.
| Stage and conditions | Colour, Gardner | Acidity, % | Unsaponifiables, % |
|---|---|---|---|
| Crude, as received | 9 to 10 | 4.7 | 5.6 |
| Degummed: 0.6% of 7M phosphoric acid, 70 C, up to 20 min, washed with 15% water | 6 to 7 | 4.7 | 5.6 |
| Bleached: 1% bleaching earth plus 0.5% activated carbon, 80 C, 15 min, under nitrogen | 2 to 3 | 5.0 | 5.5 |
| Deodorised: 220 C, 3 h, 5 Torr | 2 to 3 | 4.0 | 5.6 |
| Deodorised: 250 C, 1 h | 2 to 3 | 3.6 | 5.6 |
| Deodorised: 250 C, 2 h | 2 to 3 | 2.8 | 5.8 |
| Deodorised: 250 C, 3 h | 2 to 3 | 2.2 | 6.0 |
Degumming removes phosphorus compounds with dilute phosphoric acid: sample A went from 70 to under 6 ppm phosphorus with no measurable loss of unsaponifiables at 0.4 or 0.6 percent acid. At 0.8 percent acid, sample A dropped to 4.8 percent unsaponifiables and sample B from 7.6 to 7.0.
Bleaching is adsorption onto a clay, here a bleaching earth sold as Tierra de Gador type C, plus activated carbon for red tones, then filtration. Clay dose was the lever: 2 percent clay left sample A at 5.4 percent unsaponifiables against 5.6 percent with 1 percent, and nudged the acidity up, which the authors put down, tentatively, to a slight acid hydrolysis. They settled on 1 percent clay with 0.5 percent carbon, which took the colour to 2 to 3 on the Gardner scale.
Deodorisation is steam blown through the hot fat under vacuum. In the trial the flask ran at 5 Torr, with steam piped to the bottom and the vapours trapped in a series of condensers. Odour went first: at 220 C for 2 hours sample A was already completely deodorised, with no change in colour or unsaponifiables, but the acidity was still too high. It took 250 C for at least 2 hours to bring the free fatty acids under 3 percent. The authors cite earlier work showing that above 270 C, even briefly, fats come out darker, less stable and higher in free fatty acids, and they set 250 C as their maximum.
Neutralisation is the step the trial avoided. Sample B, at 7.0 percent acidity, came down only to 4.6 percent after 3 hours at 250 C, still over the limit, and the authors stopped refining it physically. Their conclusion sets the threshold: physical refining works on shea with under 5 percent free fatty acids; above that, a neutralisation is needed, and it "should be partial in order to reduce the loss of unsaponifiable matter". Chemical neutralisation turns free fatty acids into soap, and the paper notes that the soapstock it forms carries neutral oil away with it.
So the "harsh chemicals" on the seller pages are, in the one published shea refining trial, dilute phosphoric acid, a clay, activated carbon, steam and nitrogen, all washed, filtered or distilled back out. The solvent that seller pages often bring up, hexane, belongs to a different stage. Abdel-Razek lists the ways shea is taken out of the nut, "boiling, semi-mechanical, mechanical and solvent extraction methods", separately from the four refining steps. Extraction method and refining are two different questions about a butter, and a label answering one says nothing about the other. The wider problem with "chemical-free" wording is taken apart in chemical-free and clean beauty claims.
The unsaponifiables: refining keeps them, fractionation moves them
The unsaponifiable fraction, the triterpene alcohols, sterols, squalene and tocopherols that do not turn into soap with alkali, is the one strong argument for unrefined butter. The question is how much of it refining takes.
Ruiz Méndez's answer is: almost none, when the process is chosen to protect it. Sample A went in at 5.6 percent and came out at 5.8 percent after 2 hours at 250 C, and 6.0 percent after 3 hours. Sample B went from 7.6 to 7.8 percent. The rise is not new unsaponifiable matter. The authors' explanation is that the deodoriser distils off fatty acids, so what remains holds a slightly higher proportion of unsaponifiables. The commercial figure agrees in kind: AAK's data sheet calls LIPEX Shea refined, bleached and deodorised, says it "contains a relatively high amount of unsaponifiable matter", and gives a typical value of 7 percent by AOCS Ca 6a-40.
Abdel-Razek's figures show where the large losses come from. After refining and fractionation, the olein held 6.38 percent unsaponifiables, more than the crude's 5.06, and the stearin held 0.22 percent. The 1:1 mix of the two held 4.06 percent, a fifth less than the crude. The paper's conclusion puts it the same way: unsaponifiables were "markedly lower after refining in the case of the shea stearin fraction" and "increased in the case of the shea olein fraction", while in the mix they "decreased little".
That matters when buying, because stearin and olein are sold as shea. Shea stearin is the hard fraction used in chocolate fats, measured at 90.72 percent SOS and 60.37 percent stearic acid. A white, hard "refined shea" with a high melt may be stearin, and if it is, most of the unsaponifiable fraction is not in it. A whole refined shea specified at 31 to 35 C slip melt is a different material. The label names are explained in INCI names explained; the useful question for a supplier is not "refined or unrefined" but "whole butter or fraction, and what is the unsaponifiable figure on the certificate".
Unsaponifiable figures depend on the method as much as on the butter. Ruiz Méndez found the standard edible oil method neither exact nor reproducible for shea because it formed stable emulsions: it gave 4.4 and 6.1 percent on the two crude samples where their modified liquid-liquid extraction gave 5.6 and 7.6. Abdel-Razek used AOCS Ca 6b-53 and AAK uses AOCS Ca 6a-40. A difference of one point between two certificates can be the laboratory, not the shea.
The three crude samples on this page measured between 5.06 and 7.6 percent; Abdel-Razek, citing the Codex shea standard, gives a range of 1 to 19. The bigger swing is between butters, not between grades.
Vitamin E: what goes and what stays
Crude shea held 10.06 mg of total tocopherols per 100 g, 84.7 percent of it alpha-tocopherol. Refining removed the minor forms completely: gamma-tocopherol at 0.36 mg and delta-tocopherol at 0.15 mg were "wholly lost after the refining", in all three refined samples. The alpha and beta forms survived refining and then split between the fractions, with an 8.0 percent total loss in the olein against crude and an 87.4 percent loss in the stearin. The 1:1 mix held 4.94 mg, about half the crude figure.
Those numbers are real, and they are also small. A body balm with 32 percent shea, the dependable formula on the shea butter page, gets 3.2 mg of tocopherols per 100 g of balm from crude shea, or 1.6 mg if the shea behaved like the refined mix. The same formula adds tocopherol at 0.5 percent, which is 500 mg per 100 g: about 150 times what the crude shea brought with it. The added antioxidant, not the grade of the butter, sets the balm's vitamin E. Why it is added, and at what level it stops helping, is in vitamin E and antioxidants.
The other minor antioxidants follow the same pattern. Squalene fell from 0.20 mg per gram to 0.03 to 0.05, a loss the authors put at 76.9 to 87.2 percent. Flavonoids fell from 40.14 to 26.46 and 27.61 mg catechin per 100 g in the two fractions, about a third. Total phenolics barely moved, from 0.21 to 0.19 mg per gram, and the paper's results section says refining "did not cause a statistically significant decrease" in them. The triterpene alcohols that make up most of shea's unsaponifiable fraction, alpha- and beta-amyrin, showed "no change" in the olein and an 82 percent loss in the stearin, the fractionation pattern again.
Vitamins A and F
The vitamin A claim fails for both grades. Abdel-Razek looked for carotenoids, the plant source of vitamin A, by HPLC with a diode array detector in the crude butter and all three refined samples: "they were not detected." The paper adds that an earlier study reporting shea "pigments" measured at 295 nm, a wavelength "typical for polyphenols, but not for carotenoids". A yellow tint in unrefined shea is not, on this evidence, a carotene count. This site's tallow against shea butter page reaches the same conclusion from the other direction.
No source on this page measures anything called vitamin F, and none of the seller pages gives a quantity for it. What can be measured is the polyunsaturated fatty acids. Crude shea held 5.85 percent linoleic acid and 0.13 percent linolenic. Refining is not designed to change fatty acids, and the 1:1 refined mix came back at 5.13 percent linoleic, with linolenic below detection. The fractions differ more: 8.06 percent linoleic in the olein, 2.86 in the stearin.
Shelf life: the refined fractions outlasted the crude
The seller pages say refined shea keeps longer, and on this point they are right, though not for the reason usually given. Crude shea starts with the two things that shorten a fat's life already in it. Abdel-Razek's crude arrived with 12.24 percent free fatty acids and a peroxide value of 5.91 meq per kg; Ruiz Méndez's two samples had 4.7 and 7.0 percent acidity, and the authors note that most extracted shea has acidity above 4 percent, a dark colour and a very strong, unpleasant odour. Refining took the free fatty acids to zero and the peroxide value to 0.24 to 0.63.
On the Rancimat, at 110 C and an air flow of 20 L per hour, crude shea reached its induction point after 11.5 hours. Refined stearin lasted 12.5 hours, refined olein 15.9 and the 1:1 mix 18.5. The olein had the most polyunsaturated fat of the four, 8.06 percent, and still outlasted the crude by more than four hours; the authors suggest this "may be due to" the minor antioxidants concentrated in it. Those explanations are offered as possibilities; the induction periods are measured.
AAK's specification shows what a refined grade is held to. At manufacture LIPEX Shea must have an acid value of no more than 0.5 mg KOH per g, which is about 0.25 percent free fatty acids as oleic, and a peroxide value of no more than 1.0. Its typical oil stability index at 110 C is 45 hours, by AOCS Cd 12b-92. That is the same principle and temperature as Abdel-Razek's Rancimat but a different laboratory, method and material, so it shows the order of magnitude a commercial refined shea reaches rather than a figure to set beside 11.5 hours. AAK gives a maximum storage time of 3 years from production, unopened, stored dark and dry at 15 to 20 C or below, with a shelf life specification of peroxide value 5 and acid value 1.0.
No induction period converts into months in a jar, and a crude butter's free fatty acids vary with its storage and processing as much as with its grade. Abdel-Razek quotes published crude values from 9.0 to 23.9 percent. The site's working figure for unrefined shea is 12 to 24 months cool and dark; treat it as a starting point and run your own shelf life test.
The smell guide in rancid balm separates unrefined shea's normal smoky note from true rancidity, the chemistry is in rancidity and oxidation, and storage is covered in storing balms at home.
Texture and grain: refining is not fractionation
Refining a whole butter does not change its fatty acids, its triglycerides or its melt in any way these sources can detect. The textbook statement is in Abdel-Razek: "A properly conducted oil refining process should not affect the composition of fatty acids and triacylglycerols". The paper's own refined 1:1 mix came back close to the crude: 46.37 percent stearic against 43.53, 43.79 oleic against 45.77, and a slip melting point of 31 C against 32. AAK's refined whole shea gives a slip melting point of 31 to 35 C and an approximate melting point of 33 C.
The fractions are another matter. Refined stearin melted at 36 C and olein at 25 C, and the stearin was 90.72 percent SOS, the triglyceride that crystallises out of shea and makes it grainy. That is why grainy shea butter answers "does refined shea avoid graininess?" with a no: refining removes odour and colour, not the triglyceride mix, and a stearin-rich material is if anything more prone to visible crystals. The mechanism is set out in fat crystal polymorphism and the cure, a full hot melt then a fast chill, in pour temperatures.
The variation that does change texture is origin. Attikora and colleagues measured 122 selected shea trees in northern Côte d'Ivoire and found stearic acid at 38.76 plus or minus 4.67 percent and oleic at 51.26 plus or minus 4.21, with "a very high negative correlation coefficient (-0.98)" between the two. A point of stearic gained is a point of oleic lost, tree to tree. That swing is far larger than anything refining did to the fatty acids, so if a lot sets softer or harder than the last, look at its stearic figure before blaming the grade. When shea is the softening problem in a warm product, balm melting in the heat has the fixes.
Colour and smell: the reasons most people buy refined
Colour and odour are where refining does the most, and where most buyers make the choice. Ruiz Méndez's crude samples were reddish with a strong odour, Gardner 9 to 10; refined, they were 2 to 3 and odourless. AAK specifies LIPEX Shea at no more than 2.0 red on a 5 1/4 inch Lovibond cell and describes it as "white, creamy butter with a low vegetable butter odour".
Unrefined shea's smoky, nutty note fights fragrance and flavour and varies lot to lot; refined shea is close to a blank. The lip balm ratios work with either grade, so the choice is about taste and scent, not structure.
Which to choose for which product
| Product or constraint | Choose | Why |
|---|---|---|
| Unscented balm sold on shea's natural character | Unrefined, with a low acidity certificate | Keeps gamma and delta tocopherols, squalene and flavonoids; ask for free fatty acids, since one crude measured 12.2% |
| Fragranced balm or lip product | Refined whole shea | Deodorised to no odour, Gardner 2 to 3; unsaponifiables kept near the crude level |
| Product bought for the unsaponifiables | Unrefined or refined whole shea, with a stated figure | Physical refining kept 5.6% as 5.8 to 6.0%; a commercial refined grade lists 7% typical |
| Anything labelled shea stearin | Only for hardness | 0.22% unsaponifiables, 1.27 mg tocopherols, 36 C melt, 90.7% SOS |
| Soft balm or oil phase | Shea olein | Liquid-leaning at 25 C slip melt, and holds the minor components: 6.38% unsaponifiables |
| Longest unopened storage | Refined | Lower starting free fatty acids and peroxides; longer Rancimat time in the one direct test; 3 years quoted for one grade |
| Formula that keeps going grainy | Either grade, with a better cooling method | Refining does not change the SOS fraction; process control does |
| Product sold on vitamin A or vitamin F | Neither claim | Carotenoids not detected in crude or refined; no measurement of a vitamin F exists in these sources |
For unrefined shea, a free fatty acid figure is worth more than a grade letter, since the trade's A to F convention is not a legal standard. The documents to ask for, starting with a certificate of analysis, are listed in sourcing ingredients. Skin claims built on the minor components push a product toward the line covered in cosmetic against drug claims, whichever grade is in the jar.
What these numbers do not settle
The direct comparison has a gap that its abstract does not show. Abdel-Razek's crude butter and refined fractions were "kindly supplied" by the same company and the crude was imported from Ghana, but the paper does not say the refined stearin and olein were made from that crude lot. The table compares a crude and a refined product, not the same butter before and after. Its 1:1 mix is a reconstitution, not a refined whole shea: the paper does not give the proportions in which the crude split, so the mix is a guide to whole refined shea rather than a measurement of it.
Ruiz Méndez's trial is the reverse: a true before and after on the same two butters, but at laboratory scale on about 1 kg batches in 1991, and it measured acidity, colour, phosphorus and total unsaponifiables only. It does not say what deodorisation did to the tocopherols or squalene in shea. AAK's figures are a specification and typical values for one product, not an independent analysis. The Codex regional standard for unrefined shea butter, which sets the trade limits for the crude product, could not be read for this page, so its limits are not quoted here except where Abdel-Razek cites them.
And no source compares refined and unrefined shea on skin. Both are emollients and partial occlusives in the sense set out in occlusive, emollient and humectant; whether the minor components that refining removes change anything on skin at a balm's dose is not measured anywhere found for this page. What is measured is enough to decide most purchases, and it points one way: fractionation, not refining, is what strips shea of its unsaponifiables.
Frequently asked questions
Does refining remove the vitamins from shea butter?
Some vitamin E, not vitamin A. Crude shea measured 10.06 mg of tocopherols per 100 g; refined olein 9.25, a 1:1 refined mix 4.94 and refined stearin 1.27. Refining removed the gamma and delta forms entirely. Carotenoids, the plant source of vitamin A, were not detected in crude or refined shea. A balm's vitamin E comes mostly from added tocopherol.
Does refined shea butter still have unsaponifiables?
Yes, if it is whole shea. In a laboratory physical refining trial, unsaponifiables went from 5.6 percent to 5.8 to 6.0 percent, and a commercial refined, bleached and deodorised shea lists 7 percent typical. The big loss comes with fractionation: refined shea stearin measured 0.22 percent, while the olein held 6.38.
What chemicals are used to refine shea butter?
In the published physical refining trial: dilute phosphoric acid to remove phosphorus compounds, a bleaching clay and activated carbon to remove colour, then steam under vacuum at 250 C to remove odour and free fatty acids. Butters with over 5 percent free fatty acids need a partial chemical neutralisation as well. Hexane, when used, is an extraction solvent, not a refining step.
Does refined shea butter last longer than unrefined?
In the one direct test, yes. On the Rancimat at 110 C, crude shea lasted 11.5 hours and refined fractions 12.5 to 18.5 hours. The crude started with 12.2 percent free fatty acids and a peroxide value of 5.91. One commercial refined grade gives a 3 year unopened shelf life; unrefined is usually quoted at 12 to 24 months.
Is refined shea butter less grainy?
No. Refining does not change the fatty acids or triglycerides that cause grain; a refined 1:1 mix of stearin and olein came back close to crude shea in both, with a slip melt of 31 C against 32. Grain comes from slow cooling. Shea stearin, sometimes sold as refined shea, is 90.7 percent SOS and more prone to it.
Why is refined shea butter white and odourless?
Bleaching with clay and activated carbon took a crude shea from 9 to 10 on the Gardner colour scale to 2 to 3, and steam deodorisation under vacuum removed the odour completely, even at 220 C. Neither step needs to remove the unsaponifiable fraction, and in the published trial neither did.
Sources and further reading
- Abdel-Razek AG and colleagues, Effect of refining and fractionation processes on minor components, fatty acids, antioxidant and antimicrobial activities of shea butter, Foods 12(8): 1626, 2023.
- Ruiz Méndez MV, Huesa Lope J, Pereda J and Dobarganes MC, Assays for physical refining of shea butter, Grasas y Aceites 42(2): 121-126, 1991 (in Spanish, with English summary).
- AAK, Product Data Sheet: LIPEX Shea, product 8599, version 02, 2 October 2023.
- Attikora AJP and colleagues, Genome-wide association study of fat content and fatty acid composition of shea tree (Vitellaria paradoxa C.F. Gaertn subsp. paradoxa), BMC Genomics 26, 2025.
Reviewed and updated 10 October 2026. Spotted an error? Tell us and we will fix and log it.