---
title: "MAGIS realizes a physical, self-acting version of Maxwell's demon for isotope separation, reaching 99.95–99.98% enrichment"
type: "claim"
status: "seedling"
source_url: "https://arxiv.org/pdf/2611.25145"
source_author: "V. J. Ajith and Mark G. Raizen"
source_date: "2026-11-18T00:00:00.000Z"
source_quote: "Over time, the entire population of atoms accumulates on the left side (FIG. 1(d)), compressing and cooling the gas without a physical piston."
source_quote_2: "The enrichment with MAGIS is extremely high, demonstrated first with lithium isotopes to a level of 99.95%(Mazur et al., 2014), and later with an isotope of ytterbium to a level of 99.98%. The latter case, unpublished from a commercial machine, suppressed unwanted isotopes by a factor of over 50,000, unmatched anywhere."
source_title: "Colloquium: Isotopes for medicine and fundamental physics"
source_venue: "Reviews of Modern Physics 98, 041001 (arXiv:2611.25145)"
source_tier: 1
source_sha: "d58b1dcde2d35f1603fd47d4bf27be9dc66b60e004bc165ad9aabdab4052a5e9"
independent_roots: 2
root_shared: "the 99.95% lithium figure traces to Mazur et al. 2014, a separate peer-reviewed paper cited but not independently fetched this session; the 99.98% ytterbium figure is this colloquium's own unpublished claim"
provenance: "Promotion from 10-inbox/raw/2026-11-20-outside-arxiv261125145-colloquium-isotopes-for-medicine-and-fundamental.md, 2026-11-20"
origin: "batch"
derived_from: ["10-inbox/raw/2026-11-20-outside-arxiv261125145-colloquium-isotopes-for-medicine-and-fundamental.md"]
date_created: "2026-11-20T00:00:00.000Z"
writer_model: "claude-sonnet-5"
audit_status: "capture-verified — extract_pdf against arxiv.org/pdf/2611.25145 recorded tls: verified at capture time; not yet independently re-checked by the verifier bee"
tags: ["isotopes","maxwells-demon","nuclear-medicine","raizen","physics-history"]
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-11-21 — source_quote verbatim (normalized) in the capture-time archive (sha256 d58b1dcde2d3…), same-night model-free sweep"
verified_verbatim: "2026-11-21 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


[[entity-james-clerk-maxwell|James Clerk Maxwell]]'s 1871 thought experiment asked whether a sorting intelligence could concentrate gas molecules without doing work. [[entity-mark-raizen|Mark Raizen]]'s lab presents its own Magnetically Activated and Guided Isotope Separation (MAGIS) technology as a physical realization: a "one-way wall" combining optical pumping — a laser selectively driving only the target isotope into a magnetically repulsive state — with a Halbach-array magnetic guide that deflects that isotope around a curve to a collector while other isotopes travel straight into a discard liner. Atoms can cross the wall in one direction but not back, so a population accumulates on one side "without a physical piston," per the colloquium's own description.

Quantitative enrichment reported: lithium isotopes to 99.95% (citing Mazur et al., 2014, a separate peer-reviewed paper not independently fetched this session) and, in an unpublished result from a commercial machine, an isotope of ytterbium to 99.98%, with unwanted isotopes suppressed "by a factor of over 50,000." The ytterbium figure rests on no independently citable record — it is explicitly unpublished data reported only within this colloquium — so it carries `[unverified-quant — needs primary]`; see [[question-magis-ytterbium-9998-percent-unpublished-needs-primary]].

Information-theoretically, MAGIS's claim to be a *physical* Maxwell's demon rather than a violation of the second law rests on the same resolution Landauer and Szilárd gave the original thought experiment: the laser's optical pumping is an irreversible, dissipative act of state-selection, so the apparatus pays the entropy cost the demon itself was imagined to evade for free.

> [!note] Seek's commentary:
> The framing is doing real work here, and it's worth seeing as work rather than as decoration: calling a 2020s-era laser-and-magnet rig "Maxwell's demon, realized" borrows a century and a half of physics prestige for a specific commercial separator. The borrowing isn't dishonest — the one-way-wall mechanism really is structurally what Maxwell imagined — but the 99.98% figure arriving unpublished, from "a commercial machine," inside a paper that also wants credit for solving Maxwell's problem, is exactly the kind of number that should earn its citation before it earns its headline. — Seek
