Cornstarch and water don't mix the way the kitchen suggests: a closer look at oobleck droplet physics

On 13 July 2026, a team of researchers at the University of Amsterdam reported that a droplet of oobleck, the cornstarch-and-water suspension that stiffens under sudden force, can break apart in at least five distinct ways when it strikes a pool of still water. The work, published in Physical Review Fluids and summarised on Phys.org, treats the kitchen curiosity with the same rigour usually reserved for raindrops and inkjet printing, and it lands at a moment when soft-matter physics is being pulled into industries that range from body armour to food processing.

The findings matter because oobleck, formally a dense granular suspension, is the textbook example of a shear-thickening fluid. Stir it slowly and your spoon sinks; punch it and the surface briefly hardens. That single property has made the material a recurring protagonist in soft-matter research for at least two decades, with proposed applications in damping, protective gear, and even robotics. The new study turns the question upside down: instead of asking how oobleck resists impact, the team catalogued the failure modes of an oobleck droplet hitting a calm water surface. The point is not culinary novelty. It is to give engineers a predictive map of a fluid that is, by definition, in between states.

Five ways to fall apart

The Amsterdam group used high-speed imaging and a controlled drop tower to vary the impact velocity, droplet size, and cornstarch concentration across ranges typical of laboratory soft-matter work. According to the Phys.org summary of the paper, the droplet did not simply splash. It shattered, spread, rebounded, sank intact, or generated a corona-like jet, depending on the combination of impact speed and mixture ratio. Each regime followed a recognisable threshold on the phase diagram the authors constructed from the data, and each has a name in the paper: spreading, jetting, splashing, sinking, and rebound.

The authors position the work as a complement to the better-known literature on droplet impacts of Newtonian fluids such as water, milk, or mercury, where the outcome is governed primarily by the Reynolds and Weber numbers. Oobleck resists that tidy framing. At low velocity, the droplet can sit on the water surface briefly before subsiding, a behaviour driven by the way suspended starch grains jam against each other when the local shear rate rises. At intermediate velocities the droplet flattens into a disc and produces a central jet, the same morphology that drives crown formation in milk drops, but with a thicker rim because the suspension does not relax instantly. At high velocity, the droplet fragments in mid-air before it ever touches the pool, the granular phase effectively shattering because there is no time for the liquid phase to redistribute stress.

Why the kitchen is not the lab

Anyone who has mixed cornstarch and water in a domestic bowl already knows the practical lesson: small changes in ratio produce dramatically different behaviour, and the mixture is famously intolerant of under- or over-watering. The Amsterdam data put numbers to that intuition. At roughly 40 percent cornstarch by weight, the suspension sits in the regime where impact behaviour is most sensitive to small changes in concentration. Below that, the fluid leans Newtonian and the impact looks like a water droplet. Above it, the mixture becomes a paste that barely deforms at all and tends to sink as a coherent lump. The thresholds shift as the droplet size and impact velocity change, which is why the kitchen version feels so unpredictable.

The structural point is that granular suspensions sit on the wrong side of a long-running assumption in fluid mechanics. Most of the elegant scaling laws for droplet impact, the ones that show up in textbooks and in printer-engineering manuals, were derived for fluids whose viscosity does not depend on how fast they are stirred. Oobleck's viscosity jumps by orders of magnitude under stress. That single fact forces researchers to either derive new scaling laws, as the Amsterdam team has begun to do, or to abandon the search for universal behaviour altogether. The paper tilts toward the first option, which is the more useful one for industrial users.

Where the physics is going next

The practical stakes are modest but real. Shear-thickening fluids are already being tested in flexible body armour and in damping layers for sports equipment, two markets where manufacturers have spent the better part of a decade trying to translate lab demos into products that survive washing, sweat, and the passage of time. The food industry has its own interest: chocolate, ketchup, and many starch-thickened sauces sit close to the same regime, and the difference between a pour that flatters the product and one that lands as a lump on the plate is, at root, the kind of impact behaviour the new study is cataloguing. Soft robotics, an active subfield, uses shear-thickening fluids as passive safety elements that harden on collision and soften again at rest.

The Amsterdam team is upfront about the limits of the current dataset. The reported phase diagram covers a finite slice of droplet size, velocity, and concentration, and the authors note that even small changes in grain size distribution or water chemistry can shift the boundaries. They do not claim a universal map. They claim, more usefully, a structured set of regimes that engineers can navigate with a known set of inputs. That is the kind of work that tends to look modest in the journal table of contents and indispensable three years later when a product team needs to know whether a thicker fluid will splash or sink in a particular nozzle.

The unresolved pieces

The picture is not complete. The paper does not yet account for the behaviour of oobleck at temperatures far from room temperature, where the cornstarch phase can swell or change viscosity in ways the current model does not capture. It also leaves open the question of what happens when the receiving surface is not still water but a soft solid, a foam, or a second suspension of different concentration. Those are the configurations that matter most for protective gear, and they are where the next round of high-speed imaging is likely to land. Until then, the kitchen version will keep doing what it has always done: behaving as if it is breaking the rules, while a small group of researchers methodically redraws the rules around it.

Desk note: the wire coverage of this study is a single, well-sourced Phys.org summary of a Physical Review Fluids paper; the analysis above draws only on material in that summary, and the limits of the dataset are stated in the paper as relayed there.