A deodorant balm without bicarbonate, and why the rash usually stops

A bicarb free deodorant balm with magnesium hydroxide at 10 percent and zinc ricinoleate at 3, in a 60 ml jar poured at 64 to 70 C, and the antiperspirant line.

The bicarbonate rash is the commonest reason people abandon natural deodorant, and it is almost always blamed on a detox period rather than on what it is: an irritant reaction to a soluble alkali held against thin, occluded, frequently shaved skin by a layer of fat. This page replaces the bicarbonate with two materials that work by different mechanisms, then deals with the problem that replaces it, which is keeping 28 percent of insoluble powder off the bottom of the jar.

• INCI: Magnesium Hydroxide
• INCI: Zinc Ricinoleate
• 28% total powder
• Cosmetic, not an antiperspirant
• Pour 64-70 C

Why bicarbonate causes the classic underarm rash

Three things act together, which is why the single-cause arguments, "too alkaline" or "too abrasive", both fail to predict who reacts.

Alkalinity, but not as usually stated. A 5 percent solution of sodium bicarbonate sits near pH 8.3, which is not extreme, and a saturated suspension of magnesium hydroxide is nominally more alkaline at around pH 10. On pH alone the replacement looks worse. The number that matters is delivered dose. Sodium bicarbonate dissolves at roughly 9 to 10 g per 100 mL of water at 20 C, so the moment it meets sweat it goes fully into solution and presents its entire buffer capacity at once. Skin surface pH normally runs between about 4.1 and 5.8, and the enzymes that maintain the barrier lipids have acidic optima. An alkaline challenge activates serine proteases that break down corneodesmosomes and slows lipid processing until the surface reacidifies over some hours. Applied every morning, the surface never gets a full recovery interval.

Particle shape. In an anhydrous balm the bicarbonate is not dissolved. It is angular milled crystal suspended in fat, rubbed into a skin fold that rubs against itself all day, on a site often shaved within the last 24 hours. That is mechanical irritation, not a metaphor.

Occlusion. The fat base is the aggravating factor nobody suspects. It holds the alkali in continuous contact with the stratum corneum, stops it being diluted or wiped away, and raises local hydration and temperature. Occlusion increases the irritant response to a given material, which is why bicarbonate rinsed off in the shower does nothing and the same bicarbonate in a balm produces a symmetrical, itching, sometimes burning eruption confined to the application area.

Where the smell actually comes from

Fresh sweat is close to odourless. Eccrine sweat is dilute salt water, and the apocrine glands of the axilla secrete lipids, proteins and amino acid conjugates that barely smell as they leave the gland. The odour is made afterwards by resident skin bacteria. Branched medium-chain acids, principally 3-methyl-2-hexenoic acid and 3-hydroxy-3-methylhexanoic acid, are released from glutamine conjugates by an aminoacylase found in axillary corynebacteria; volatile thioalcohols, above all 3-methyl-3-sulfanylhexan-1-ol, are released from a cysteinylglycine conjugate by a carbon-sulphur lyase, with particular staphylococci identified as the main producers. The thioalcohols have odour thresholds low enough to matter at trace concentrations.

That gives four points of attack: reduce the sweat, remove the bacteria, block the enzymes, or capture the odorants once formed. Only the first is an antiperspirant action, and it is the one this formula deliberately does not attempt.

What each powder does, and what it does not

Magnesium hydroxide at 10 percent is mid-band: below about 5 percent the effect is hard to distinguish from the base, above about 15 percent it reads chalky and white without adding much. Its whole advantage over bicarbonate is the solubility difference in the second column, roughly four orders of magnitude. A particle in a fat film cannot deliver more alkali than it can dissolve, so exposure is self-limiting. It is the material sold as milk of magnesia, which is why plain milk of magnesia has a following as a deodorant.

Zinc ricinoleate is the zinc salt of ricinoleic acid, the hydroxy fatty acid making up 85 to 90 percent of castor oil. It is not an antibacterial, not a fragrance and not a masking agent: it traps the odorants themselves, which is why it leaves the skin flora alone and works in a fragrance-free product. The published efficacy work is largely supplier-generated, so treat 3 percent as an established craft level rather than a validated dose.

Starch at 15 percent is sensory work, not deodorant work: it absorbs the water and sebum reaching the surface and turns a greasy film into something that feels dry within a minute. Tapioca has a finer, silkier particle than most arrowroot and is the better choice if you can get it. The absorbent group is compared in starches and silica and kaolin and cosmetic clays; swapping up to a third of the starch for kaolin buys absorbency and a matter finish at the cost of drag.

The formula: 180 g, three 60 ml jars

All percentages are weight percent of the finished batch and total exactly 100. A 60 ml jar is the right format because the product goes on with two fingertips, which lets the powder load stay high without the drag a stick would produce.

Twelve percent beeswax looks high for a finger-applied jar, and in a plain balm it would be. Twenty-eight percent powder pulls both ways: the powders absorb free oil and stiffen the product, but they also dilute the wax network that has to hold everything in suspension while the jar cools. The 12 percent answers the second requirement, not the first. Why filler load and wax level fight each other is covered in balm texture science, and the batch calculator resizes the table for a different number of jars.

The real problem: suspending 28 percent of insoluble powder

Nothing here dissolves in the fat, and magnesium hydroxide has a density around 2.35 g/cm3 against roughly 0.9 for the molten oil phase. Stokes' law sets the terms: settling velocity rises with the square of particle diameter and falls with the viscosity of the fluid around it. Both are levers you control.

Primary particles are not the problem; agglomerates are, and each one you fail to break up becomes both a grain on the skin and a sediment in the jar. Four things fix it.

1. Wet the powder before it meets the bulk. Slurry the magnesium hydroxide, zinc ricinoleate and starch into the castor oil and work it smooth. The hydroxyl group on ricinoleic acid makes castor unusually good at wetting polar mineral surfaces, which is why it is also the standard pigment dispersant.
2. Raise the viscosity you pour at. Pouring at 80 C gives you the bottom row of the table. Pouring at 64 to 70 C, close to where the beeswax network starts to build, cuts the fall rate several fold.
3. Keep it moving through the set. The step craft recipes omit. Stir the jug continuously from the moment it leaves the heat, then stir each filled jar with a wooden skewer until the surface goes cloudy. Sixty seconds a jar.
4. Give the network something to grip. Hydrogenated castor oil at 1 to 2 percent crystallises early and fine, holding particles while the beeswax catches up. Ozokerite at the same level does a similar job with better oil binding: see mineral waxes.

Method

1. Weigh in two parts. Powders and castor oil into one bowl, everything else except tocopherol and essential oil into a heatproof jug, on a 0.1 g scale.
2. Make the slurry. Work the powders into the castor oil until it is a smooth thick paste with no dry pockets and no lumps against the bowl.
3. Melt the fat phase at 75 to 80 C. In a water bath, holding 75 to 80 C (167 to 176 F) for five minutes after the last wax fleck goes, so no shea crystals survive. Do not go past 85 C.
4. Combine and disperse. Stir the slurry into the melt in three additions, then run a stick blender for 20 seconds. That is worth more than any amount of spatula work, because it breaks the agglomerates in the bottom two rows of the settling table.
5. Cool to 70 C, stirring without a break. Add the tocopherol and any essential oil between 65 and 70 C. The moment the jug is still, the clock in the settling table starts.
6. Pour at 64 to 70 C. Into dry jars on a level tray, then stir each with a skewer until it thickens and loses its gloss. Pour windows for other formats are in pour temperatures.
7. Cool fast, then cure 24 hours. Move the jars somewhere cool so the shea passes quickly through its crystallisation range. Cap once opaque and leave a day before judging texture, which reads firmer at 24 hours than at two.

Deodorant or antiperspirant: where the line sits

In the United States, a product intended to reduce underarm wetness is an over-the-counter drug. The antiperspirant monograph at 21 CFR part 350 sets out the permitted aluminium and aluminium zirconium actives, their maximum concentrations and the labelling a compliant product carries. Making one to sell means drug facility registration, product listing and drug good manufacturing practice, none of which applies to a deodorant. Label wording decides the class as surely as the ingredient list does: "keeps you dry", "stops sweat" and "reduces wetness" are antiperspirant claims even on a product containing no aluminium at all.

In the European Union and the United Kingdom, antiperspirants are cosmetics. Aluminium salts are permitted, and the Scientific Committee on Consumer Safety reassessed systemic exposure in its 2020 opinion, concluding that use is safe up to stated concentrations for each product type, non-spray deodorants and antiperspirants among them. Adding an aluminium salt in the EU does not change the product class, but it does change what your safety assessment must cover.

Which jurisdiction you sell into is the one thing this page cannot decide for you, and both frameworks apply to a market stall as much as a factory. Read cosmetic versus drug claims before writing label copy, and see labelling cosmetics in the US for what a compliant cosmetic panel looks like. One more trap: "aluminium free" and "chemical free" carry their own problems, examined in chemical free and clean beauty claims.

What to expect across a working day

This balm does not reduce sweat volume, so shirts will be as damp as they ever were. It slows the formation of odour and captures some of what forms. Applied to clean, dry skin in the morning, six to ten hours is realistic in a temperate office. Exercise, heat, stress sweating and a heavy apocrine output all shorten that, and reapplying at midday is normal rather than a failure. Apply before the bacteria have done their work: on skin that already smells you are adding a fat film over the odorants.

People switching from a conventional antiperspirant often report a fortnight of worse odour before things settle. The plausible explanation is a shift in the axillary bacterial population rather than anything leaving the body, and either way the consequence is the same: judge a new deodorant over two to four weeks, not two days.

Failure modes

If you want it as a stick instead

A twist-up barrel changes two things. The product must be self-supporting rather than scoopable, so total wax rises from around 13.5 percent to roughly 18 to 24 percent, part of it a harder wax. And it is dragged across skin under pressure instead of patted on with a fingertip, which makes every abrasive particle count for more. Both changes pull the powder load down, not up.

A stick carrying 28 percent powder on top of 20 percent wax has very little liquid oil left to carry it, and the result drags, crumbles at the edge of the barrel and deposits in flakes. Eighteen to 22 percent total powder is the working ceiling, which usually means magnesium hydroxide at 8 percent and starch at 10 to 12. The full stick formula and the barrel-filling method are at solid deodorant stick. If you want the higher active load, keep the jar.

Two smaller points for either format. Keep essential oil at the low end: axillary products have their own tightly restricted IFRA category because the site is occluded, thin-skinned and often freshly shaved, so check the standard per material rather than assuming a body-lotion limit, as set out in essential oils in balms. And use a gasketed or wadded closure, because a slide-lid tin lets a high-surface-area powder blend pick up humidity: closures are compared in the packaging guide.

The honest limits

Removing the bicarbonate removes the commonest cause of the rash, but guarantees nothing: magnesium hydroxide is still an alkali, fragrance is still a sensitiser, and some people react to something in every deodorant they try. If this version still causes trouble, drop the fragrance, then halve the magnesium hydroxide, then remove it and run on zinc ricinoleate and starch alone, which is milder and noticeably less effective.

The second limit is efficacy. No home test will tell you how well a batch performs, and the wear times above are experience rather than measurement. That matters if you sell, because a deodorant claim has to be substantiated like any other and "lasts all day" is not a phrase you can support from a kitchen.

The decision rule: choose this jar if the bicarbonate rash is what brought you here, because the high powder load is what makes it work and the jar is what makes the high powder load tolerable. Choose the stick if convenience matters more than active level. Choose a conventional antiperspirant if what you actually want is dry clothing, because nothing on this page will give you that.

Reviewed and updated 6 September 2026. Spotted an error? Tell us and we will fix and log it.