Industries / Consumer goods / Cosmetics and personal care
Formulation
Formulation is deciding what goes into a cream, shampoo, dye, sunscreen or perfume, and in what amounts, so that it works, stays stable on the shelf and is safe to use.
7
Milestones achieved, from c. 200 CE to 2026.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
c. 200 CE
Galen's cold cream
Galen of Pergamum, a physician who lived from 129 to about 216 CE, described a cold cream that medical texts named after him for centuries; a 4th-century medical text by Oribasius records a mixture of wax and oil of roses based on his formula. Cold creams today use other ingredients, but the idea, an emulsion (a stable mix) of oil and water, is the same, according to a 2023 paper in Clinics in Dermatology.
Credit
Galen of Pergamum; history traced by Lipsker, Nwabudike, Parish and Hoenig
1883
Dye made inside hair
The chemist August Wilhelm von Hofmann described in 1863 how p-phenylenediamine (PPD), a colorless compound, turns brown when oxidized. In 1883 Monnet patented hair dyeing by oxidation with PPD and a related compound, the first such patent according to a 2025 review in ACS Omega, and between 1888 and 1897 the brothers Hugo and Ernst Erdmann added more dye compounds and hydrogen peroxide as the oxidizer. The review says PPD is in practically all hair colors sold today. In 1909 the chemist Eugène Schueller founded the hair dye company that became L'Oréal.
Credit
Monnet; Hugo and Ernst Erdmann; Eugène Schueller
1949
Numbers for emulsifiers
William C. Griffin of the Atlas Powder Company in Wilmington, Delaware, presented the HLB system at an October 1949 meeting of the Society of Cosmetic Chemists in Chicago and published it in the society's journal. HLB, for hydrophile-lipophile balance, gives each emulsifier a number for the balance between the part of its molecule that likes water and the part that likes oil. Griffin wrote that the number predicts whether an emulsifier makes an oil-in-water or a water-in-oil emulsion and "reduces the amount of work" in choosing one.
Credit
William C. Griffin, Atlas Powder Company
1988
Silicone held in shampoo
A Procter & Gamble patent by Mark Grote and Glen Russell, granted in May 1988, claims a shampoo that carries droplets of an insoluble silicone conditioner, kept from rising out by a suspending agent such as ethylene glycol distearate; in the patent's examples, versions without that agent separated into a clear silicone layer. P&G introduced Pert Plus, a shampoo and conditioner in one bottle, in 1987.
LessonAn ingredient chosen for how it feels is later judged by where it goes after use: the EU stopped two volatile silicones, D4 and D5, in wash-off products from 31 January 2020 because they persist and build up in the environment, and extends the limit to D4, D5 and D6 in all other cosmetics from 6 June 2027.
patents.google.compackagingstrategies.comeur-lex.europa.eueur-lex.europa.eu
Credit
Mark B. Grote and Glen D. Russell, Procter & Gamble
2014
Preservatives judged again
Preservatives keep microbes from growing in a product once it is opened. A 2004 study by P. D. Darbre and colleagues reported parabens, preservatives used in many thousands of products, in samples of 20 human breast tumors; Denmark later banned two parabens in products for children under three, and the EU's scientific committee reviewed them in 2010, 2011 and 2013. In April 2014 the EU banned five less common parabens because industry had sent "limited or no information" for their safety review, while its scientific committee found methyl- and ethylparaben safe at the permitted levels. Meanwhile methylisothiazolinone (MI), allowed alone as a preservative in EU cosmetics from 2005, caused what dermatologists call an epidemic of contact allergy, and the EU banned it from leave-on products from 12 February 2017.
LessonAn ingredient that nobody defends with safety data is lost, and the one that replaces it needs evidence of its own.
Credit
European Commission and its Scientific Committee on Consumer Safety (SCCS); Darbre and colleagues; European dermatology clinics that tracked MI allergy
2015
Plastic scrub beads banned
In July 2012 Marcus Eriksen, Sherri Mason and colleagues trawled the surface of the US Great Lakes and found an average of about 43,000 plastic particles per square kilometer, over 466,000 at one station; many were colored spheres they suspected were microbeads from face scrubs and other products. The US Microbead-Free Waters Act of 28 December 2015 banned making rinse-off cosmetics with added plastic microbeads, solid plastic particles under 5 mm used to scrub or cleanse, from 1 July 2017, and selling them from 1 July 2018.
LessonA rinse-off ingredient goes down the drain, so it is judged by what it does in rivers and lakes.
Credit
Eriksen, Mason and colleagues; US Congress
2026
A new US sunscreen filter
In the US a sunscreen is an over-the-counter drug, and most are sold under an FDA list, called a monograph, of permitted UV filters. On 10 June 2026 the FDA issued a final order adding bemotrizinol, a filter widely used outside the US, at up to 6% of the product. The testing firm SGS calls it the first change to that list in more than 20 years; the FDA had proposed it on 12 December 2025.
Credit
US Food and Drug Administration
Watch for
AI simulates sunscreen actives
L'Oréal says that building NVIDIA's ALCHEMI machine-learning software into its research gives it an engine that predicts how molecules behave and simulates ingredient performance and texture before lab work, for "a discovery process that is 100x faster than traditional methods". The work starts with sun protection and skin tone products. The release gives no baseline and cites no outside check.
Achieved whenL'Oréal or an outside party publishes development times for comparable projects run with and without the engine, or a product designed with it is sold with that record.
Claimed by
L'Oréal and NVIDIA, March 2026
Fewer formulation rounds
Unilever reported that AI has cut formulation cycles in its Beauty & Wellbeing division from up to six rounds of lab trials to one or two, sped up writing product claims by 75% and cut time spent analyzing consumer insight by about 60%. The trade report that carries the figures cites no outside check.
Achieved whenUnilever or an outside party publishes round counts for a defined set of projects before and after the tools, in enough detail to compare.
Claimed by
Unilever, April 2026
Open
01
Silicone-free conditioners, combed
A public, repeatable test of whether shop-bought conditioners sold as silicone-free detangle as well as those with silicones, measured by machine on locks of cut human hair (called tresses) rather than on anyone's head. Each tress is washed, treated, and pulled through a comb while a load cell records the force; the drop in combing work against untreated tresses is the score. Publish the rig design, every force trace and the code, so anyone can rerun a product.
The briefCredit · unclaimed02
Silicone swaps, measured
A neutral, published comparison of D5, the volatile silicone that gives many serums and dry oils their light, fast-drying feel, against the replacements sold for it, all run under one method: how fast each evaporates at skin temperature, how far it spreads and what residue it leaves. Then test blends that suppliers say match D5.
The briefCredit · unclaimed
Open 01 · the brief
Silicone-free conditioners, combed
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public, repeatable test of whether shop-bought conditioners sold as silicone-free detangle as well as those with silicones, measured by machine on locks of cut human hair (called tresses) rather than on anyone's head. Each tress is washed, treated, and pulled through a comb while a load cell records the force; the drop in combing work against untreated tresses is the score. Publish the rig design, every force trace and the code, so anyone can rerun a product.
Why it is still open
The machine test itself is not new. Stiftung Warentest, the German consumer testing foundation, measured the wet-combing force of 17 shop-bought conditioners for damaged hair in 2019 with a tensile tester, on chemically bleached natural hair strands, five strands per product, each combed five times; its results are sold as grades, not force data, and are not split by silicone content. A 2026 paper adapted the method for tightly curled hair. Published comparisons of silicone-free conditioning use formulas made in the lab, such as a 2022 study on bleached hair. Suppliers say their silicone-free conditioning agents match silicones, from tests on their own prototypes. No public set of combing forces for retail conditioners was found.
Built with what exists
Human hair tresses sold for lab testing, both untreated and bleached; a universal tensile tester or texture analyzer fitted with a comb, or a home-built rig from a load cell, an HX711 amplifier board, a stepper motor and an Arduino; Python with pandas and SciPy; a public repository such as Zenodo.
First step
Build the load-cell rig and comb 5 untreated tresses and 5 treated with one silicone conditioner, 10 times each, to see how much the readings vary.
Done looks like
At least 20 retail conditioners, 10 with silicones and 10 without, including any from Stiftung Warentest's 2019 test still on sale so the two rankings can be compared, each tested on at least 8 tresses, with the mean drop in combing work and its 95% confidence interval; the home-built rig checked against a commercial tensile tester on the same tresses, agreeing within 10%; rig drawings, raw force traces and code published under an open license.
Open 02 · the brief
Silicone swaps, measured
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A neutral, published comparison of D5, the volatile silicone that gives many serums and dry oils their light, fast-drying feel, against the replacements sold for it, all run under one method: how fast each evaporates at skin temperature, how far it spreads and what residue it leaves. Then test blends that suppliers say match D5.
Why it is still open
From 6 June 2027 the EU limits D4, D5 and D6 to below 0.1% in leave-on cosmetics, so formulators must replace D5 in products that stay on skin and hair. A supplier's page gives evaporation rates for isododecane, hemisqualane, C13-15 alkane and polydecene relative to D5, with no test method and no source for the numbers. I found no independent comparison that tests D5 and its hydrocarbon and ester replacements side by side under one stated protocol.
Built with what exists
An analytical balance reading to 0.1 mg inside a draft shield, with a hot plate held at 32 °C, or a thermogravimetric analyzer, which records a sample's weight as it heats; filter paper or a synthetic skin substrate; a camera above the plate to measure spreading area; Python with pandas. Work with small amounts in a ventilated lab, away from flames, since some replacements are flammable.
First step
Buy small amounts of D5, isododecane and C13-15 alkane and record the weight loss of 0.2 g spots on filter paper every 5 minutes for 2 hours.
Done looks like
Mass-loss curves at 32 °C for D5 and at least 11 replacements or blends, 5 repeats each, with time to 50% and 90% loss and spreading area at 1 minute; a second lab repeats three of the materials and agrees within 10%; method and data published openly.