Dense suspensions shear-thicken: stir them faster and they suddenly resist more. The standard explanation involves friction — particles jam against each other, forming force chains that transmit stress. Kikuchi, Kobayashi, and Takada show that a discontinuous viscosity jump can happen without friction, without jamming, without any dense-suspension mechanism at all.
Their system is a granular gas — particles so dilute they interact only through binary collisions. The key ingredient is a velocity-dependent restitution coefficient: gentle collisions are nearly elastic, but hard collisions dissipate more energy. When the restitution coefficient switches sharply between two values at a threshold collision velocity, the viscosity develops an S-shaped dependence on shear rate.
Reader note (2026-11-17): Checked against the full text of arXiv:2603.19605 (Kikuchi, Kobayashi and Takada, a Letter in J. Phys. Soc. Jpn.; v1 of March 2026), which I had earlier held as matching on a partial read. The authors, the kinetic-theory setting, the S-shaped viscosity and the two Bagnold-type regimes match. The direction of the restitution contrast is reversed in the essay. The paper states that the S-shaped curve and the discontinuous viscosity change appear only when the restitution coefficient at low collision velocity is smaller than at high collision velocity (e1 less than e2; its example is e1 = 0.80 and e2 = 0.99), notes that this is opposite to the trend commonly observed in dry granular matter, and reports that for e1 greater than e2 no S-shape and no shear thinning appear. The essay says gentle collisions are nearly elastic and hard collisions dissipate more, and that the high-shear regime has lower restitution and higher effective viscosity; the paper's high-shear regime is the more elastic one. The paper also states its limits: a dilute gas with volume fraction fixed at 0.01, a stepwise restitution law, no rotational degrees of freedom, and whether the transition persists in denser systems is left as a future problem. The essay's closing point that the same macroscopic behavior can come from different microscopic causes follows the paper's own comparison with the Wyart-Cates picture for dense suspensions.
In the low-shear regime, most collisions are gentle, restitution is high, and the gas behaves as a nearly elastic granular fluid with Bagnold scaling — viscosity proportional to shear rate. In the high-shear regime, collisions are violent, restitution drops, dissipation increases, and the gas settles into a different Bagnold regime with higher effective viscosity. Between the two, the S-curve means increasing shear rate can produce a discontinuous jump in viscosity.
The physics is purely kinetic. No contact networks, no friction coefficients, no microstructural rearrangement. Just particles bouncing off each other with a collision-energy-dependent bounciness. The transition looks like shear thickening but operates through a completely different mechanism — energy dissipation at the collision level rather than geometric constraint at the packing level.
The same macroscopic behavior can emerge from fundamentally different microscopic causes. The viscosity curve doesn't tell you what the particles are doing.