Starches and silica: how to take the shine off without wrecking the glide

Arrowroot, tapioca, maize starch and silica compared on oil absorption, particle size and use at 1 to 8 percent, plus the sliminess and microbial risks of starch.

Powders go into a balm for one reason: to absorb oil at the skin surface so the film stops looking wet. They do it at a cost in glide, and the cost rises faster than the benefit above about 5 percent. This page compares arrowroot, tapioca, maize starch and the two useful silicas on oil absorption and particle size, sets out what each level actually delivers, and covers the two failures people hit first, sliminess on damp skin and mould where water has got in.

• Starch use 1-8%
• Fumed silica 0.5-2%
• Oil absorption 60-350 g/100 g
• Starch carries 10-14% water

Five powders, and what each one is

Arrowroot (INCI Maranta Arundinacea Root Powder) is starch milled from the rhizome of a tropical perennial. It is the powder most craft recipes name, largely because it sounds gentler than cornflour, and it has the largest granules of the common starches.

Tapioca starch (Manihot Esculenta Root Starch, or Tapioca Starch on some lists) comes from cassava root. Its granules are smaller and more uniform than arrowroot's, which is why it reads silkier at the same dose and why it has largely replaced arrowroot in commercial dry-touch formulas.

Maize starch (Zea Mays Starch, cornflour in a British kitchen) is the cheapest and the most consistent, because it is a commodity with pharmacopoeial monographs behind it. It performs almost identically to tapioca and is disliked mainly for the association.

Fumed silica (INCI Silica, or Silica Dimethyl Silylate for the hydrophobic grade) is amorphous silicon dioxide made by flame hydrolysis. Its primary particles are nanoscale but they fuse into branched aggregates during manufacture and clump into much larger agglomerates in the bag, which is why the number that matters in practice is the aggregate size, not the primary particle size.

Spherical silica (also INCI Silica) is a different material sold under the same name: solid or porous microbeads, typically 3 to 12 micrometres, made by sol gel or spray routes. Porous grades absorb oil; solid ones mainly roll, giving slip and a soft focus optical effect rather than mattifying.

Oil absorption and particle size compared

Read the two columns together. Oil absorption tells you how much shine a given weight can remove; particle size tells you what it will feel like doing it. Arrowroot at 70 micrometres is large enough that a fingertip can detect individual granules on a thin film, which is the source of the gritty complaint. Tapioca and maize at 5 to 25 micrometres sit below the threshold at which most people perceive individual particles, roughly 20 to 25 micrometres, which is why they read as smooth at the same loading. Silica absorbs three to four times as much oil per gram as any starch, so the effective dose is far lower, and the reason for the low use level is that it thickens hard well before it mattifies.

What 1, 3, 5 and 8 percent actually do

The curve flattens between 5 and 8 percent while the sensory penalty keeps climbing, which is why 8 percent is the practical ceiling for native starch in anything applied to a large area. Drag is the complaint that arrives first, and it is worth separating from the other causes listed in drag and poor glide, because powder drag cannot be fixed by changing the wax. It is fixed by using less powder, by moving to a finer or modified grade, or by adding a fast spreading ester to carry the film further before the powder starts to bite. The mechanisms behind that trade are set out in balm texture science.

Powder also changes the optics of the product in the tin as well as on the skin, lightening a tint and dulling a gloss. If the finish has gone flat when you wanted shine, powder is the first suspect, ahead of the wax, and the diagnosis runs as in matte instead of glossy balm.

Why starch turns slimy on damp skin

This is the complaint that surprises people, because the powder was added to make things drier. Starch granules are semi-crystalline packages of amylose and amylopectin held together by hydrogen bonding. Add water and heat and they gelatinise, swelling irreversibly and releasing amylose into solution; maize starch gelatinises at roughly 62 to 72 C, tapioca at about 58 to 70 C, arrowroot at around 65 to 70 C, all in excess water.

Skin never gets that hot, but it does not need to. Surface moisture, a humid armpit or a hand that was not fully dry hydrates the outer layer of each granule enough to make it swell and soften, and the amylose that leaches out is a thickening polymer. The result is the slippery, pasty, faintly gluey film that people describe as slimy, gummy or "rolling into little balls". The pilling has the same cause: partially hydrated starch plus the oil phase forms a soft mass that shears into rolls under a sleeve or a second application.

Three things reduce it. Use a smaller granule starch, since surface area to volume works in your favour for perception but the swelling itself is proportionally similar. Use a hydrophobically modified starch, which is the real answer and is covered below. Or cut the total loading, which is the answer people resist because they added the powder for a reason. What does not help is more wax, which traps the hydrated starch in a stiffer film and makes the pilling worse.

Fumed silica is a thickener that happens to mattify

Fumed silica earns its place in anhydrous formulas for rheology, not for shine control. Its surface carries silanol groups that hydrogen bond to each other across a non-polar oil, building a three-dimensional network at loadings under 2 percent. That network gives the oil phase a yield stress, so the product holds its shape and suspends particles at rest, and it breaks down under shear so the product still spreads. It rebuilds when the shear stops, which is what thixotropic means, and the framework for reading that behaviour is in rheology and yield stress.

Practical uses at 0.5 to 2 percent: stopping pigment settling in a tinted pour, which is the fix for pigment sinking to the bottom; stopping oil bleed from a soft balm, one route out of balm separated in layers; and giving a low-wax formula body without hardness. Two details decide whether it works. Hydrophilic grades depend on hydrogen bonding, so a polar oil phase heavy in castor or in polar esters competes for those silanols and you need more silica to reach the same thickening; hydrophobic grades such as silica dimethyl silylate behave the opposite way. And the network only forms if the agglomerates have been broken up by shear, so a spoon will not do it.

Starch, water and the microbial question

An anhydrous balm resists microbial growth because there is no free water in it, not because oil is hostile. Water activity, not water content, is the controlling number: most bacteria need a water activity above about 0.90 to grow, most yeasts above about 0.88, common moulds above about 0.80, and the most xerophilic moulds can manage down to roughly 0.61. A properly made anhydrous balm sits far below all of those, which is the whole argument set out in do balms need preservatives.

Starch does not raise water activity on its own, but it changes what happens when water arrives. Native starch as supplied is not dry: pharmacopoeial monographs for maize starch allow a loss on drying of roughly 14 to 15 percent, and typical commercial material carries 10 to 13 percent water bound in the granule. At 8 percent starch in a formula carrying 12 percent moisture, you have added just under 1 percent water to a product being sold as anhydrous. Bound water in a starch granule has a low activity and is not available to microbes, so that alone does not spoil anything. The problem is that starch is also a carbohydrate substrate, so if free water does enter, from a wet fingertip, a splash on the bench or condensation in a cooling tin, the local water activity at that spot rises and there is now food waiting there. A starch-free balm contaminated the same way usually just goes cloudy. A starch-loaded one grows something, which is the pattern behind most of the cases in mould in balm.

Modified starches, and what the modification buys

Aluminium starch octenylsuccinate (INCI Aluminum Starch Octenylsuccinate, often shortened to ASO) is maize starch reacted with octenyl succinic anhydride and then aluminium salt. The octenyl group is a hydrophobic tail grafted onto the polymer, and it changes three things: the granule no longer hydrates and swells the way native starch does, so the slimy failure largely disappears; oil absorption rises to roughly 100 to 150 g per 100 g; and the surface feels silkier, closer to talc than to flour. It has been reviewed for cosmetic safety and is used widely in dry-touch products. The objections to it are positioning rather than performance: it is chemically modified and it contains aluminium, and some natural certification standards exclude it on both counts.

Other grades you will meet: distarch phosphate and hydroxypropyl starch phosphate, cross-linked to resist swelling and used more in emulsions than in balms; tapioca starch polymethylsilsesquioxane, tapioca coated with a silicone resin for a soft focus optical effect and a very dry finish, which is excellent and expensive; and dextrin palmitate, an oil-soluble starch ester that thickens rather than absorbs. If you want absorbency with a mineral rather than a starch, clays are the parallel option and are covered in kaolin and cosmetic clays.

Deodorants, foot balms and the bicarbonate problem

Starch earns its keep in two product types. In a foot or heel balm it stops an occlusive-heavy formula from leaving a slick sole, which matters because the alternative is someone slipping on a bathroom floor; 3 to 6 percent is the usual band, and the rest of the structure is worked through in the foot and heel balm formula. In a deodorant it absorbs moisture and carries the dry finish that makes the format tolerable, at 5 to 15 percent in a stick where the higher loading is acceptable because the application area is small and rarely rubbed.

The recurring failure in that category is not the starch, it is the bicarbonate beside it. Sodium bicarbonate neutralises the odour-forming fatty acids effectively and a saturated solution sits at about pH 8.3, well above the skin surface pH of roughly 4.5 to 5.5. In a proportion of users, and particularly under a freshly shaved arm, that alkalinity produces the red, itchy, sometimes burning rash that the format is notorious for. It is an irritant response related to dose and to skin condition rather than a true allergy, so it can appear after weeks of trouble-free use. Lower loadings, magnesium hydroxide, zinc ricinoleate or arrowroot alone are the usual answers, and a worked version is at bicarbonate free deodorant balm. The comparison against a conventional bicarbonate build is in the solid deodorant stick.

Dust, weighing and getting it in without lumps

Handle the powders with more care than their kitchen reputations suggest. Amorphous synthetic silica is not crystalline silica, and the distinction is the one drawn by the IARC monograph programme, which classifies crystalline silica as carcinogenic to humans and amorphous silica as not classifiable. That removes the silicosis question but not the dust question: fumed silica has a bulk density around 0.03 to 0.06 g/mL, so a gram is a startling volume of material that becomes airborne from any draught, and NIOSH sets a recommended exposure limit of 6 mg per cubic metre for amorphous silica. Wear a fitted FFP2 or N95 mask when weighing it, work away from any fan, and do not tip it from a height. Starch dust is a nuisance at bench scale and a genuine explosion hazard at industrial scale, which is worth knowing before scaling a formula up.

1. Weigh in a tall vessel. Tare a beaker rather than weighing onto a pan. Fumed silica is light enough that a balance draught shield changes the reading, so let it settle before recording, using the practice in weighing and calibration.
2. Sieve, do not scoop. Pass starch through a fine kitchen sieve into the vessel. Agglomerates that go in whole come out as lumps that no amount of stirring will break.
3. Make a slurry. Mix the powder with three to four times its weight of cool liquid oil to a smooth paste before it meets anything hot. Powder added dry to molten fat forms a skin around each lump.
4. Add on the cool side. Stir the slurry into the melted phase at 55 to 65 C, once the waxes are fully liquid and before the mixture starts to thicken.
5. Shear the silica. A stick blender for one minute, or a mini homogeniser for thirty seconds. Starch needs stirring only; silica needs shear to build its network.
6. Pour promptly and stir until it thickens. Powders settle in a thin melt, so keep the mixture moving until it is too viscous to separate, then pour without delay.

Choosing, and where powders cannot help

If you want less shine, use tapioca or maize starch at 3 to 5 percent and accept that arrowroot offers nothing the other two do not, apart from a better ingredient list. If the product will meet damp skin, spend the money on aluminium starch octenylsuccinate or a modified tapioca instead, because the slimy failure is what customers actually complain about. If the problem is that the balm slumps, bleeds oil or drops its pigment, reach for fumed silica at 0.5 to 2 percent and stop thinking of it as a mattifier. If the problem is a greasy afterfeel rather than visible shine, powder is the wrong tool: the fix is redistributing the oil phase towards faster spreading materials, as set out in greasy heavy afterfeel.

Two limits are worth stating plainly. Powders reduce how a film looks, not how much oil is present, so a heavy balm with 6 percent starch is still a heavy balm; you have changed the optics and the initial touch, not the load. And any starch in a formula moves it from a product that tolerates a small contamination event to one that does not, which is a real change to the safety case rather than a footnote to it. If a formula cannot guarantee dry hands throughout its life, that argues for a modified starch, for a mineral absorbent, or for accepting the shine.

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