baffle, in a 📝tokamak, is the structure that closes the mouth of a 📝divertor against the main chamber, trapping neutral gas in the divertor volume so it can be radiated and pumped rather than escaping back to the core.
The word carries its ordinary engineering sense — a plate or wall that obstructs a flow — applied to neutral particles rather than to fluid or sound. Ions arriving at a divertor target neutralise on the surface and re-emerge as atoms and molecules that no longer follow field lines; without an obstruction they stream back into the main chamber, where they cool the edge, dilute the fuel and are lost to the pumps. A baffle raises the neutral pressure in the divertor by an order of magnitude or more, which is what makes 📝divertor detachment accessible, keeps 📝impurity seeding concentrated where it should radiate instead of contaminating the core, and gives the pumps a dense gas to remove. Baffling is therefore inseparable from 📝fuel recycling control: how tightly a divertor is closed sets how much of the recycled flux the machine gets to choose the fate of.
"Tightly baffled" is the design term for a divertor closed so completely that neutrals are effectively confined to the leg. Paired with a long leg — a divertor channel extended well beyond the 📝X-point, sometimes with a secondary X-point in it — the combination is the long-legged tightly-baffled geometry that Wigram and colleagues modelled for 📝ARC in 2019 (DOI 10.1088/1741-4326/ab394f), finding passively stable detachment across the reactor's projected exhaust power. The 2026 ARC exhaust paper (DOI 10.1017/S002237782610155X) carries the same principle into an up–down-symmetric design predicting divertor neutral pressures near 20 pascals. Both are simulation results; ARC has not been built, and the closure that a real divertor achieves against erosion, assembly tolerance and diagnostic access is the part experiments on 📝SPARC are meant to settle.