# Hyperpolarization (physics) Hyperpolarization is the deliberate manufacture of a nuclear [[Spin_(physics)|spin]] imbalance that [[Thermodynamic_equilibrium|thermal equilibrium]] would never grant — five orders of magnitude past the Boltzmann number — and for [[Helium-3|helium-3]] it is what turned an inert [[Noble_gas|noble gas]] into an [[Magnetic_resonance_imaging|MRI]] contrast agent that images the air in a lung rather than the tissue around it. The governing fact of this article is that the enhancement is a *stock*, not a *state*: nothing in the magnet renews it, [[Radioactive_decay|decay]]-like relaxation drains it, and every radio-frequency pulse spends a fixed fraction of it forever. ## Microsims — three.js <iframe src="https://wikitube-3d-microsims.netlify.app/Hyperpolarization_%28physics%29.html" width="100%" height="620" frameborder="0" loading="lazy" sandbox="allow-scripts allow-same-origin" title="Hyperpolarization (physics) — three.js microsim"></iframe> **`Hyperpolarization_(physics)` (three.js).** An ensemble of ³He nuclear spins precesses about B₀ with the net magnetisation drawn as a single arrow whose length is the honest vector sum of the individual spins — so at thermal equilibrium the arrow is invisible, and the HUD tells you why, evaluating P = tanh(γħB₀/2k_BT) live at whatever field and temperature you have dialled in and printing **7.53×10⁻⁶** at the default 3 T and 310 K. Switch the mode select to MEOP or SEOP and watch the enhancement bar open by about five decades, then set the pump laser to zero — because there is no laser inside a patient — and start firing RF pulses. T1 runs continuously and the oxygen slider shortens it; each pulse permanently spends cosine-of-the-flip-angle of what is left, implemented as a true Bloch rotation rather than a hard-wired cosine so that rapid pulse trains compose correctly. The small lung panel accumulates signal: it fills in a handful of pulses when the gas is hyperpolarized, stays black forever at thermal equilibrium, and dies into an **EXHAUSTED** state once the flip angle has eaten the budget. ## The number that makes the whole field necessary A [[Spin_(physics)|spin]]-½ nucleus in a static field B₀ is the plainest two-state system [[Quantum_mechanics|quantum mechanics]] offers: two Zeeman levels split by ΔE = 2μB₀, and a fractional population difference between them — the polarization — of **P = tanh(μB₀ / k_BT) = tanh(γħB₀ / 2k_BT)** For ³He at **B₀ = 3 T and T = 310 K** (body [[Thermodynamics|temperature]]), using the CODATA-2022 helion magnetic moment |μ_h| = 1.074 617 551 98×10⁻²⁶ J/T and k_B = 1.380 649×10⁻²³ J/K, this evaluates to **P = 7.53×10⁻⁶**. Seven and a half parts per million. [[Proton|Protons]], with a larger moment, manage **9.89×10⁻⁶** in the same [[Superconducting_magnet|magnet]] at the same [[Thermodynamic_equilibrium|equilibrium]]; at 1.5 T the ³He figure halves to **3.77×10⁻⁶**. The hyperbolic tangent is not decoration, but neither is it doing much work: the argument is a hundred-thousandth of unity, so P is linear in B₀ and inverse in [[Thermodynamic_system|T]] to eight digits. That is why the [[Second_law_of_thermodynamics|thermodynamic]] price of a bigger magnet buys so little [[Signal-to-noise_ratio|signal]], and why [[Nuclear_magnetic_resonance|NMR]] has always been a [[Detection_theory|detection]]-limited [[Science|science]]. Conventional [[Magnetic_resonance_imaging|MRI]] survives this only by [[Density|counting]]. A litre of tissue water holds **6.69×10²⁵** [[Proton|protons]], so even at 9.89×10⁻⁶ there are **6.6×10²⁰** net [[Spin_(physics)|spins]] per litre for the receive coil to find. A [[Kinetic_theory_of_gases|gas]] has no such reserve: a litre at 1 bar and 310 K holds only **2.34×10²²** atoms. The round "roughly a thousandfold" that circulates in the review literature is the *molecular* ratio — closer to **1430** — while the ratio that governs the [[Signal|signal]], resonant nucleus for resonant nucleus, is **about 2900**, because every water molecule carries two [[Hydrogen|hydrogen]] nuclei and every [[Helium|helium]] atom carries one. Multiply an already-marginal 10⁻⁵ by a further 3×10⁻³ and thermal airspace imaging is not a hard [[Engineering|engineering]] problem. It is a dead one. **Brute force does not rescue it.** Push to a 7 T [[Superconducting_magnet|superconducting magnet]] *and* cool the sample to 4 K — [[Cryogenics|cryogenic]] conditions no [[Medicine|patient]] survives, and the working regime of [[Liquid_helium|liquid helium]] rather than of [[Medical_ultrasound|clinical imaging]] — and P is still only **1.36×10⁻³**. The Boltzmann factor is not something you out-engineer; it is something you go around. Hyperpolarization goes around it by driving the [[Spin_(physics)|spin]] system far out of [[Thermodynamic_equilibrium|equilibrium]] against an external [[Entropy|entropy]] sink — angular momentum carried in on a [[Photon|photon]] stream — and then simply *not letting it come back* long enough to take a picture. At 78% polarization that same litre of gas carries **1.8×10²²** net spins, about **28 times more** than a litre of thermally polarized tissue water at 3 T. That inversion is the whole subject: you image the [[Breathing_gas|air]], not the [[Tissue_engineering|tissue]]. ## Two ways to pump a helium nucleus Hyperpolarization is a family, not a technique. Dynamic nuclear polarization (DNP) transfers [[Electron|electron]] polarization to nuclei by [[Microwave_engineering|microwave]] irradiation and dominates ¹³C metabolic imaging; parahydrogen-induced polarization and its SABRE variant exploit a chemically prepared [[Hydrogen|hydrogen]] singlet; brute-force polarization simply applies the previous section's [[Thermodynamics|thermodynamic]] arithmetic at extreme fields and the millikelvin [[Thermodynamic_system|temperatures]] a [[Dilution_refrigerator|dilution refrigerator]] reaches. This article's centre is the [[Noble_gas|noble-gas]] branch — optical pumping — which for [[Helium-3|³He]] comes in two forms a reader should be able to tell apart. **MEOP — metastability-exchange optical pumping** is the ³He-specific route, because it climbs a ladder only [[Helium|helium]] has. A weak RF discharge sustains a small [[Plasma_(physics)|plasma]] population in the metastable 2³S₁ [[Atomic_orbital|state]]; circularly polarized light at **1083 nm** pumps that state's *electronic* angular momentum; and metastability-exchange collisions hand the momentum down to ground-state ³He **nuclei**. It runs at low pressure — **~1 mbar** — and it is fast: seconds to minutes, set by cell volume and [[Optical_engineering|laser]] power. Steady-state polarizations of **80–86% in fields of 2–5 T** are measured in ~1.2 mbar sealed cells, **89% at 1.33 mbar and 3 T**, and the best sub-millibar cells approach **90%**; Gentile et al. give the honest time-averaged working figure for either method as **55–85%, depending on conditions**. MEOP's price is that the product leaves the cell at a [[Density|density]] three orders of magnitude below breathing pressure and must be mechanically compressed, a [[Mechanical_engineering|mechanical]] stage with its own [[Leak|leak]] and depolarization budget. **SEOP — spin-exchange optical pumping** goes through an [[Ion|alkali]] vapour, usually [[Rubidium|rubidium]], sometimes [[Potassium|potassium]] or [[Caesium|caesium]]. Light at 795 nm on the Rb D1 line polarizes the alkali valence [[Electron|electron]], and Rb–³He collisions transfer that polarization to the helium nucleus. SEOP works at bar-scale pressure, so no compressor is needed — but the ³He spin-exchange rate is minuscule, and that is the entire difference. The spin-up [[Half-life|time constant]] is **hours**: 4–8 h for the best near-litre cells under 100 W of volume-holographic-grating-narrowed [[Optical_engineering|optical]] pumping, with Gentile et al. noting that in general "a day is required to approach the maximum polarization." Production cells commonly sit in the 5–20 h band, ~7 h being a serviceable working number, for an achievable **75–85%**. The timescale gap is not a footnote — it decides how each machine is used, in the way a [[Doubling_time|time constant]] usually does. A MEOP polarizer is a batch device that fills a bag beside the scanner while the [[Medicine|patient]] is being positioned. A SEOP cell is a thing you switch on and leave, which is why SEOP is the standard route for [[Xenon|¹²⁹Xe]] — whose spin-exchange rate is far more favourable — and why polarized-³He [[Neutron_detection|neutron]] spin filters, which live in a beamline for weeks, are content to be pumped in situ. ## The gyromagnetic ratio, stated correctly Two values circulate and they are not interchangeable. The **shielded** helion ratio — the one measured on real ³He gas, with the atom's own two [[Electron|electrons]] and their [[Molecular_orbital|orbital]] currents screening the nucleus — is **γ′_h = 2.037 894 6078(18)×10⁸ s⁻¹ T⁻¹ = 32.434 100 033(28) MHz/T** (CODATA 2022) while the **bare** nuclear value, back-derived from the helion g-factor, is **32.436 05 MHz/T**. The gap is the shielding shift σ = 5.996 7029(23)×10⁻⁵ — 60 ppm. That is irrelevant to the polarization above, where it moves P in the eighth significant figure, and entirely relevant to [[Nuclear_magnetic_resonance|NMR]] practice, where 60 ppm is **5.85 kHz** of offset at 3 T, several linewidths in the [[Frequency_domain|frequency domain]] and a visible artefact after the [[Fast_Fourier_transform|FFT]]. **The shielded value is the one used in [[Magnetic_resonance_imaging|MRI]] and NMR**, and quoting the bare one is the commonest [[Accuracy_and_precision|precision]] error in this corner of the literature. From it, ³He resonates at **48.65 MHz at 1.5 T** and **97.30 MHz at 3 T** — [[Angular_frequency|frequencies]] about 76% of the [[Proton|proton]]'s in the same magnet, which is why ³He imaging needs a dedicated broadband [[Radio-frequency_engineering|radio-frequency]] transmit chain and a [[Transducer|resonant coil]] rewound for the [[Signal|band]], not merely a different [[Sensor|probe]]. ## A consumable, not an equilibrium Thermal magnetisation is a *state*: perturb it and it returns, because the lattice restores it. Hyperpolarized magnetisation is a *stock*, and that single distinction reorganises everything downstream. It decays toward the thermal value with T1 — an [[Ordinary_differential_equation|exponential]] with no source term — and every excitation pulse spends part of it irrecoverably. For a train of hard pulses of flip angle α, the longitudinal magnetisation before the *n*-th pulse is **M_n = M₀ cosⁿ(α) · exp(−t_n / T1)** The cosine is a ratchet, and geometric [[Doubling_time|decay]] is unforgiving. Fire 40 pulses at 45° and **9.5×10⁻⁷** of the magnetisation remains — the transverse component each pulse creates dephases within milliseconds and is gone, lost to [[Damping|damping]] rather than returned. The same 40 pulses at 10° leave **54%**. For an N-line acquisition the constant flip angle that spreads the budget evenly is approximately **atan(1/√(N−1))** — **9.1° for 40 lines** — and variable-flip ramps do better still, an [[Mathematical_optimization|optimisation]] under a hard [[Necessity_and_sufficiency|constraint]] rather than a [[Filter_design|filter-design]] problem. Hence hyperpolarized [[Sampling_(signal_processing)|acquisition]] looks nothing like conventional MRI: no steady state to reach, no [[Repeatability|signal averaging]] to lean on, no second pass if the first is spoiled by [[Oscillation|motion]] or [[Noise_(electronics)|noise]]. It is a one-breath, one-shot technique whose flip-angle schedule is a budgeting problem before it is a contrast problem — nearer [[Operations_research|operations research]] than [[Signal_processing|signal processing]]. T1 in a well-prepared storage cell is enormous — **hundreds of hours**, with ~100 h routine in fully blown aluminosilicate (GE-180 class) glass and the practical band running roughly **50–500 h** depending on wall quality and [[Accuracy_and_precision|field uniformity]]. Wall relaxation at the glass [[Surface_engineering|surface]], not the [[Kinetic_theory_of_gases|gas]] itself, sets that limit, which is why cell fabrication is a [[Materials_science|materials]] discipline. In vivo the same number collapses by four orders of magnitude, and the culprit is unambiguous. ## Oxygen: the problem that is also the measurement Molecular [[Oxygen|oxygen]] is paramagnetic — one of the few common [[Chemical_element|elements]] whose ground state carries unpaired [[Spin_(physics)|spin]], and the reason [[Allotropes_of_oxygen|O₂]] behaves so differently from [[Nitrogen|nitrogen]] in a field — and in the lung it is the dominant ³He relaxation channel. The relationship is quantitative and clean: **p(O₂) = ξ / T1, with ξ = 2.61 bar·s at 37 °C** A normal alveolar oxygen tension of **0.13 bar** (≈100 mmHg) therefore gives **T1 ≈ 20 s**. That is the central operational problem of in-vivo ³He [[Magnetic_resonance_imaging|MRI]]: the [[Breathing_gas|inhaled gas]] begins dying the instant it meets alveolar air — the same [[Diffusion|diffusive]] mixing that carries it into the [[Porous_medium|acini]] is what kills it — and the whole acquisition must finish inside a breath-hold. The elegant consequence is that the problem is also the instrument. Because T1 is *set* by p(O₂), fitting the regional decay rate maps alveolar oxygen tension voxel by voxel: a functional measurement of respiratory [[Homeostasis|gas exchange]] that no other modality delivers directly, applied to emphysema in the MESA COPD study and complementary to what [[Hemoglobin|haemoglobin]] saturation reports downstream. That p(O₂) can be *read off* T1 at all is itself the proof that O₂ dominates the relaxation — the [[Estimation_theory|estimator]] and the interference are the same physics. A [[Sensor|sensor]] built out of its own worst [[Noise_(electronics)|noise source]] is a recurring pattern in [[Physics|physics]]; here it is exact, and it is the one thing [[Auscultation|listening to a chest]] can never do. ## The clinic moved to xenon, and it moved for money Hyperpolarized ³He lung MRI was the original demonstration of [[Noble_gas|noble-gas]] imaging and stayed the reference technique through the 2000s, yielding ventilation maps and — via apparent-diffusion-coefficient (ADC) imaging, which turns restricted [[Diffusion|diffusion]] into a length scale — [[Porous_medium|airspace]] microstructure finer than any chest radiograph or [[Medical_ultrasound|ultrasound]] resolves. The present tense, however, is [[Xenon|xenon]]. Stewart et al. are unambiguous: the hyperpolarized-gas [[Magnetic_resonance_imaging|MRI]] field **"has generally transitioned to the use of ¹²⁹Xe over the last 5–10 years."** Their stated reason is not [[Physics|physics]]. ³He, they write, **"is not naturally abundant, its availability has become severely regulated, and costs have risen to the point that it is not an economically viable agent for widespread clinical use."** The prices they quote settle it: **³He at ~£500/L**, against **enriched ¹²⁹Xe at ~£150/L** and **natural-abundance xenon at ~£25/L** — and xenon, unlike [[Helium-3|³He]], can be re-enriched from air by [[Fractional_distillation|fractional distillation]] whenever someone is willing to pay the [[Energy|energy]] bill. The regulatory record follows the economics. **XENOVIEW (xenon Xe 129 hyperpolarized, Polarean) was approved by the FDA on 23 December 2022**, for evaluation of lung ventilation with MRI in adults and paediatric patients aged 12 and over, and it is **the first and only approved hyperpolarized MRI contrast agent**; CMS granted a reimbursement code in **August 2023**, and the 2025 review literature confirms the status is current. **There is no FDA-approved hyperpolarized ³He agent.** Any source presenting ³He lung MRI as a live clinical modality — and there are many — is describing 2008 in the present tense. ## What ³He keeps The displacement is commercial, not physical, and it is worth being equally precise about what ³He still wins. Its gyromagnetic ratio is **about 2.8 times** that of ¹²⁹Xe, so at equal polarization and equal [[Density|spin density]] it simply returns more [[Signal|signal]] and a better [[Signal-to-noise_ratio|signal-to-noise ratio]]. Its achievable polarization is higher. And its gas-phase [[Diffusion|diffusivity]] is far larger — [[Helium|helium]] is the lighter and smaller atom, and by the [[Kinetic_theory_of_gases|kinetic theory]] the [[Velocity|mean speed]] scales as the inverse square root of [[Atomic_mass|atomic mass]] — which makes ADC microstructure work more sensitive to acinar dimensions than the xenon equivalent. ³He also does not dissolve appreciably in tissue, which is at once a limitation (xenon's dissolved-phase [[Hemoglobin|red-blood-cell]] signal reports gas exchange that ³He cannot) and a simplification (the ³He map is purely a ventilation map, with no tissue-partition term to model). Outside [[Medicine|medicine]], polarized ³He is a working instrument rather than a legacy one. Because the enormous ³He(n,p)T absorption cross-section — **5316 b** at 0.0253 eV as an international [[Accuracy_and_precision|standard]], **5333 ± 7 b** on the Sears/NIST evaluated table — is concentrated almost entirely in the singlet channel, a polarized cell absorbs one [[Neutron|neutron]] spin state and transmits the other. That makes it a broadband neutron **spin filter**, used across [[Neutron_diffraction|neutron scattering]] instrumentation and in [[Nuclear_engineering|reactor]] and spallation beamlines, and it is the application that keeps SEOP and MEOP development funded. The same nuclear property underwrites ³He magnetometry and precision tests of fundamental symmetries in [[Physics|physics]]. The sentence that ties this article to the rest of the portal is this: **³He was displaced from the clinic by its own supply crisis, not by physics.** The gas that made airspace [[Magnetic_resonance_imaging|MRI]] possible is the same gas that fills [[Neutron_detection|neutron proportional counters]], and post-2001 demand for radiation portal monitors drew down a stockpile that only ever grew by tritium [[Beta_decay|beta decay]] at ~5.5% a year — a [[Decay_product|decay product]] harvested from weapons reservoirs, with no terrestrial [[Natural_gas|natural-gas]] source of any commercial consequence. The [[Cold_War|Cold-War]] complex was the sole supplier; when [[Neutron_detection|detection]] demand roughly tripled, the price went from tens of dollars a litre into the hundreds, and lung imaging — a small, price-taking customer with a [[Xenon|xenon]] substitute available — was simply outbid. For where that crisis came from see [[Neutron_detection]]; for the supply picture entire, see [[Helium-3]], [[National_Helium_Reserve]] and [[Helium_production_in_the_United_States]]. ## Sources - **Gentile, T. R.; Nacher, P. J.; Saam, B.; Walker, T. G. (2017).** "Optically polarized ³He." *Reviews of Modern Physics* **89**(4), 045004. [10.1103/RevModPhys.89.045004](https://doi.org/10.1103/RevModPhys.89.045004). The authoritative review of MEOP and SEOP physics; source for the 55–85% time-averaged working range, the 85% SEOP ceiling in near-litre neutron-spin-filter cells, the 4–8 h spin-up constants, the "a day is required" statement, and the ~100 h room-temperature cell relaxation times. Use this, not clinical reviews, for polarization and T1 figures. - **CODATA 2022 fundamental constants**, NIST. [physics.nist.gov/cuu/Constants](https://physics.nist.gov/cuu/Constants/Table/allascii.txt). Source for the shielded helion γ′ = 32.434 100 033(28) MHz/T, the shielding shift σ = 5.996 7029(23)×10⁻⁵, and |μ_h| = 1.074 617 551 98(93)×10⁻²⁶ J/T. The bare 32.436 05 MHz/T is derived from the tabulated helion g-factor; the thermal polarizations quoted above are computed from P = tanh(μB/kT) with these values. - **Stewart, N. J.; Smith, L. J.; Chan, H.-F.; et al. (2022).** "Lung MRI with hyperpolarised gases: current & future clinical perspectives." *British Journal of Radiology* **95**(1132), 20210207. [10.1259/bjr.20210207](https://doi.org/10.1259/bjr.20210207). The current authoritative clinical review and the source of both direct quotations above, of the ³He/¹²⁹Xe/natural-Xe cost figures, and of the "of the order of hours" T1 for O₂-free gas. - **Taskiran, N. P.; Hiura, G. T.; Zhang, X.; et al. (2022).** "Mapping Alveolar Oxygen Partial Pressure in COPD Using Hyperpolarized Helium-3: The MESA COPD Study." *Tomography* **8**(5), 2268–2284. [10.3390/tomography8050190](https://doi.org/10.3390/tomography8050190). Establishes p_AO₂ = ξ/T1 with ξ = 2.61 bar·s at 37 °C — Eq. (2) — from which the ≈20 s in-vivo T1 at 0.13 bar alveolar oxygen follows directly. - **Polarean Imaging plc (2022).** *FDA Approves XENOVIEW (xenon Xe 129 hyperpolarized)*, press release, 28 December 2022. [polarean.com](https://polarean.com/wp-content/uploads/2022/12/POLX-FDA-Approves-XENOVIEW%E2%84%A2-hyperpolarized-Xe-129.pdf). Approval date 23 December 2022; "the first and only hyperpolarized MRI contrast agent." - **MacLeod, J. L.; Khan, H. M.; Franklin, A.; Myc, L.; Shim, Y. M. (2025).** "Hyperpolarized Xenon-129 MRI: Narrative Review of Clinical Studies, Testing, and Implementation." *Diagnostics* **15**(4), 474. [10.3390/diagnostics15040474](https://doi.org/10.3390/diagnostics15040474). Confirms that the clinical status described above is current. - **Fain, S.; Schiebler, M. L.; McCormack, D. G.; Parraga, G. (2010).** "Imaging of lung function using hyperpolarized helium-3 magnetic resonance imaging." *Journal of Magnetic Resonance Imaging* **32**(6), 1398–1408. [10.1002/jmri.22375](https://doi.org/10.1002/jmri.22375). The best retrospective on the ³He era; source of the "up to five orders of magnitude" enhancement statement and of the 32.4 MHz/T working value. - **Maxwell, J. D.; et al. (2019).** "Enhanced polarization of low pressure ³He through metastability exchange optical pumping at high field." *Nuclear Instruments and Methods in Physics Research A*. [OSTI 1607711](https://www.osti.gov/servlets/purl/1607711). Reports 89% at 1 torr (1.33 mbar) and 3 T, and cites 85% at 0.3 torr under low-field pumping — the basis for the "approaching 90%" ceiling quoted above. - **Compact ³He gas polarizers based on metastability-exchange optical pumping (2025).** *Review of Scientific Instruments* **96**(4), 045110; preprint [arXiv:2505.15876](https://arxiv.org/abs/2505.15876). Source for "steady-state polarizations up to 86% in magnetic fields between 2 and 5 T" in a 1.2 mbar sealed cell, and for the weak field dependence across 2–5 T. - **Carlson, A. D. (2011).** "The neutron cross section standards, evaluations and applications." *Metrologia* **48**(6), S328–S345. [10.1088/0026-1394/48/6/S09](https://doi.org/10.1088/0026-1394/48/6/S09). Background for the ³He(n,p)T cross section that makes polarized ³He a neutron spin filter; σ(0.0253 eV) = 5316 b as a standard, against the 5333 ± 7 b Sears/NIST evaluated value. *Numbers computed in this article — the thermal polarizations, the spin-density ratios, the cosⁿ(α) depletion figures, the 5.85 kHz shielding offset, the 9.1° optimal flip angle and the ≈20 s in-vivo T1 — are `[DERIVED]` from the constants and parameterisations cited above and are reproducible from them. The 5–20 h SEOP production band and the 50–500 h cell-T1 band are engineering ranges consistent with, but not individually itemised in, Gentile et al.* ## Wikipedia : Wikitube **Strict pair:** [Wikipedia](https://en.wikipedia.org/wiki/Hyperpolarization_(physics)) : [Wikitube](https://en.wikitube.io/wiki/Hyperpolarization_(physics)) ## Previous hub tags Hubs: `Life_Physics`. Portals: [[PORTAL_Helium-3]].