Condensed Matter Ep 7: Scaling and the Renormalization Group

You can buy the demonstration from a catalogue: a thick-walled cell of sulphur hexafluoride, liquid below, gas above, a flat line between them. Warm it towards forty-five degrees and the line goes soft, wavers, and stops existing — no boiling, no bubbles. And in the half-minute before it goes, the whole cell turns milky white. Two centimetres of it will not pass the light of a lamp. To scatter light that hard, a clear fluid needs structures the size of the wavelength of light — about a thousand molecules across — and it has just produced them out of nothing but being at a transition. That whiteness is the subject. Every strange thing about critical points follows from one fact: approaching the transition, the size of the largest correlated patch grows without bound, and at the transition there is no characteristic size left in the problem at all. The method is to squint. Kadanoff's move in nineteen sixty-six: replace each block of spins by a single spin carrying the block's majority, rescale, and ask what the couplings look like now. Repeat. You are not solving the system; you are watching where it flows. A system with no characteristic size is one the squint cannot change — a fixed point — and everything the transition does is set by which fixed point it flows to, not by what the material is made of. Wilson turned that into a calculation and took the nineteen eighty-two Nobel for it. That is the answer to the scandal the previous conversation left open. Uniaxial magnets, the liquid-gas critical point of every simple fluid, binary alloys ordering, two liquids unmixing — nothing in that list resembles anything else in it, and they share critical exponents to the measurable digit, because they flow to the same fixed point. Guggenheim's eight liquefied gases falling on one curve is the same statement, drawn in nineteen forty-five. Also: what "irrelevant" means physically — two samples differing only in that feature become not merely similar but indistinguishable under enough squinting, which is why the microscopic details are allowed to not matter; why four dimensions is the dividing line and what changes below it; why the exponents are not independent of one another, so pinning down two decides the third; the Harris criterion and what happens when you deliberately dirty a sample; and the small-number expansion that made the whole thing computable. And a reversal worth the hour: Landau's construction was declared broken last time, and it comes back — what he wrote down was always the coarse-grained free energy stopped at a short distance, which is analytic and fine. The non-analyticity everyone objects to belongs to the exact free energy with every scale still in it. The flaw was never in what he wrote. It was in the step he did not take. One limit belongs beside all of it, because it is the price of the method: nothing here predicts a critical temperature. The flow tells you which destination a system reaches and nothing about how far it had to travel, so every number a chemist would actually want is still measured rather than derived. ——— Series: Physics - Tutorial · Grad level. Every equation spoken in full, no chalkboard required. Built by Fermi AI.