Nuclear spin structure in dark matter search: The zero momentum transfer limit

This is a review, and it does the one thing a review in this niche should do: it puts every published calculation of ⟨Sp⟩ and ⟨Sn⟩ for dark matter target nuclei into one set of tables, from 1H through 209Bi, across nine nuclear models. If you work on spin-dependent direct detection you already know the pain of chasing these numbers through Ressell et al., Engel et al., Iachello et al., and the TFFS papers. Having them side by side, with the magnetic moments used as input marked in parentheses, is real value. The authors also did a small amount of original work: they converted the Ellis–Flores odd-group entries from λ²J(J+1) into ⟨Si⟩ and checked them, and they flag where models agree (39K, PT versus EOGM) and where they badly disagree (73Ge). That is honest bookkeeping, not a new result, and the abstract's claim [abstract:s2] that these values 'allow a conclusion about the event rate' is fair only in the sense that any tabulated nuclear input does. It changes no cross section and settles no physics question, which the paper does not pretend otherwise [abstract:s1]. The prose is candid about the main methodological weakness: the magnetic moment's spin operators come from the vector current and are hit by meson exchange currents, while the WIMP matrix element comes from the axial-vector current. They say so plainly, and that candor is a point in their favor. Where I'd push back is the ranking rule. The conclusions say the most reliable values are those from the approach that reproduces the measured magnetic moment, but Table VI headlines the unquenched Ressell et al. large-space 73Ge row at µ = −1.239 against µ_exp = −0.879, with the quenched variant demoted to a separate line, and §II then argues against quenching the spin matrix element at all. That is not a contradiction so much as a criterion applied loosely: µ agreement is a soft indicator, and the paper half-admits it. But a reader taking the stated rule literally would not pick the rows the tables headline, and the tables should say which rows are quenched and which are not. Two smaller things a referee should fix: several TFFS rows in Tables VIII–XI give a single number with no indication of which column it belongs to, and no entry carries an uncertainty. Neither breaks the compilation. This is a reference table for people doing spin-dependent direct detection or writing exclusion plots; it is not for anyone outside that community. It deserves a serious referee, mostly to tighten the table notes and the ranking language.

The review's practical payload is the guidance that 'the most reliable values of <Sp(n)> are considered to be those of the approach which reproduces well the experimental value of the magnetic moment' (Summary and Conclusions). That criterion is applied inconsistently in the tables, and the inconsistency is load-bearing because it determines which number an experimentalist actually takes from Tables I–XIII. For 73Ge, Table VI lists 'SM (large), Ressell et al. [8]' with <Sp>=0.011, <Sn>=0.468 and µ=-1.239 µN against µ_exp=-0.879. The model does not reproduce µ unquenched; the paper itself states that agreement requires an isovector quenching factor of 0.833. Yet the unquenched row is the headline entry and the quenched row (0.009, 0.372) is presented as a variant. The hybrid calculation of Dimitrov, Engel and Pittel [9], which reaches µ=-0.920 without an imposed factor, gives <Sn>=0.378, a roughly 20% reduction. Section II then argues the opposite direction: 'it is very doubtful that any extra quenching of the spin matrix elements (or equivalently the coupling constants a0 and a1) is desirable for these nuclei.' So the paper validates a model by a quantity that only agrees after quenching, and then declines to apply that same quenching to the quantity the tables report. A reader following the stated criterion literally would not select the rows the tables headline; a reader following the anti-quenching argument would not treat µ agreement as validation at all. The two positions cannot both be used to rank models. This is distinct from the meson-exchange-current caveat the paper already concedes (vector-current spin operators versus axial-vector), which it discusses openly; the unresolved issue is a consistency problem between the stated ranking rule and the tabulated entries, and it bears directly on the abstract's claim that these values 'allow a conclusion about the event rate.'

I read the paper as a compilation review whose central claim is that the tabulated <Sp> and <Sn> determine the spin-dependent event rate and that model spread, not force choice, dominates the uncertainty. I first checked whether the tables are internally consistent, since a compilation lives or dies on that. They largely are: the ISPSM rows reproduce Eq. (19), and the quoted magnetic moments follow µ = gs<Sz> + gl<Lz> with <Lz> = J - <Sz> in the stretched state (27Al gives 4.793, 209Bi gives 2.63, 13C gives 0.638, 15N gives -0.264, all matching the tables). The formalism is also self-consistent: Eq. (6) follows from Eqs. (14) and (15), and Eq. (24) matches Eq. (15). I found no arithmetic or algebraic error. I also considered and rejected the objection that the claim 'the interaction is not the primary uncertainty' rests on only two realistic potentials: the numbers in Tables VIII–X show force-induced differences of roughly 0.01–0.06 in <Sn>, well below the 0.1–0.23 spread across alternate models for the same nuclei, so the claim survives on its own data. The load-bearing soft spot is instead the reliability ranking. The reader located the same region, the magnetic-moment benchmark, but framed it as the criterion being 'softer than the prose implies.' I think it is sharper than that: it is an inconsistency between the stated criterion and the tabulated headline entries, since the rows the tables foreground for 73Ge are precisely the ones that fail the paper's own test unless a quenching factor is applied, and the paper then argues against applying that factor to the spin matrix element. On the reader's two presentation gaps I agree they are real but minor; I would add that the single TFFS numbers in Tables VIII–XI are ambiguous between the <Sp> and <Sn> columns, which matters for 131Xe (-0.186) and 133Xe (-0.201) because the sign of <Sp> controls the interference term in Eq. (7). None of this rises to a load-bearing error, so CONDITIONAL remains the right verdict and I would not move it.

As a reference compilation the paper has clear value: to my knowledge it is the broadest single-table collection of ⟨Sp⟩ and ⟨Sn⟩ values assembled for this purpose, and it would save an experimentalist or phenomenologist a substantial literature search. The manuscript is also more careful than a bare table dump. The ISPSM magnetic-moment column is internally consistent with the free g-factors of Eq. (17) in every case I checked (e.g. 1H, 13C, 15N, 27Al, 51V, 183W, 209Bi), and the OGM rows reproduce Eq. (20) with the quoted experimental moments in the large majority of entries (e.g. 7Li, 19F, 27Al, 29Si, 73Ge, 121Sb, 131Xe, 137La). The authors explicitly flag the weakness of the magnetic-moment benchmark in Section II, noting that the magnetic-moment spin operators descend from the vector current and are strongly affected by meson-exchange currents whereas the WIMP matrix element descends from the axial-vector current and is not. That is an honest statement of a limitation that many compilations of this kind would have left implicit, and it is the right thing to have written. The paper does not ship code or machine-checked derivations, but the arithmetic cross-checks above are the (

A weakly interacting massive particle that couples to nuclear spin scatters off a nucleus at a rate set, at zero momentum transfer, by a single combination of two quantities: the average proton spin $⟨S_p⟩$ and average neutron spin $⟨S_n⟩$ inside that nucleus. The paper's purpose is to gather, in one place, every published calculation of those two numbers across the periodic table, and to ask how much the answer depends on which nuclear model you trust. The authors argue that the spread between different models — the independent single-particle shell model, the odd-group model built from measured magnetic moments, large-basis shell-model diagonalizations, the interacting boson-fermion model, finite-Fermi-systems theory, the quasi Tamm-Dancoff approximation — is the dominant theoretical uncertainty, larger than the difference between two different realistic nucleon-nucleon forces inside the same model. If that is right, then any spin-dependent dark matter limit quoted from a single nuclear model is really a point inside a band, and the band's width is a nuclear-structure question rather than a particle-physics one. The practical payoff is a lookup table: pick a target nucleus, read off $⟨S_p⟩$ and $⟨S_n⟩$, and the spin-dependent event rate follows once the neutralino couplings are fixed.

What carries the argument. The load-bearing object is the zero-momentum-transfer spin structure function, which the authors write as $S_{SD}(0) = (2J+1)Λ^2/(π J(J+1))$ with $Λ = a_p⟨S_p⟩ + a_n⟨S_n⟩$, where $a_p$ and $a_n$ carry the particle-physics couplings and $⟨S_p⟩$, $⟨S_n⟩$ carry all the nuclear physics. Everything else in the review is in service of that factorization: because the nuclear and particle sides separate cleanly at $q=0$, the nuclear side can be compiled once and reused for any WIMP model. The second piece of machinery is the benchmark used to rank the models — the nuclear magnetic moment, $μ = ⟨g_s^n S_n + g_l^n L_n + g_s^p S_p + g_l^p L_p⟩$, which is built from the same spin operators, so a model that reproduces the measured $μ$ is taken as more credible. The third is the odd-group model's inversion of that relation, $⟨S_p⟩ = (μ - J)/4.586$ and $⟨S_n⟩ = -μ/3.826$, which converts a measured magnetic moment directly into a spin average.

Core claim. The paper's central claim is that the spin-dependent neutralino-nucleus cross section at zero momentum transfer, $S_{SD}(0) = (2J+1)/(4π) |(a_0+a_1)⟨S_p⟩ + (a_0-a_1)⟨S_n⟩|^2$ in the isospin language the authors use, is determined by two nuclear matrix elements that can be tabulated, and that a complete table of them now exists. The authors assemble the proton and neutron spin expectation values calculated within essentially every nuclear model applied to the problem and lay them side by side for nuclei from $^1$H through $^{209}$Bi. The review's substantive conclusion is a ranking: the phenomenological single-particle shell model overestimates the spin content almost everywhere and is inadequate; the odd-group model, which extracts the spin from the measured magnetic moment, is a clear improvement and agrees well with shell-model work for light nuclei but degrades as configuration mixing grows; and for open-shell medium-heavy and heavy nuclei the models of choice are the interacting boson-fermion model, finite-Fermi-systems theory, and the quasi Tamm-Dancoff approximation. The authors also report a specific and consequential comparison: for $^{127}$I, Ressell and Dean's shell-model calculation raises the spin-dependent sensitivity by roughly a factor of twenty over what had previously been assumed, which reshapes how much of a sodium-iodide detector's spin response comes from iodine rather than sodium.

The premise it rests on. The review ranks nuclear models by whether they reproduce the measured nuclear magnetic moment, but the magnetic moment's spin terms come from the vector current and are strongly renormalized by meson exchange, while the dark-matter matrix element comes from the axial-vector current and is not — so agreement on the magnetic moment does not by itself certify the tabulated spin averages.

Spin-dependent dark matter exclusion limits can be recomputed for any target in the tables by substituting the tabulated $⟨S_p⟩$ and $⟨S_n⟩$ into the cross-section formula, without redoing the nuclear calculation.

Because $^{19}$F combines an unquenched spin matrix element with isospin channels that add coherently, it is identified as the most favorable target for light spin-coupled dark matter.

Iodine's spin-dependent sensitivity is roughly twenty times higher than earlier estimates, so a sodium-iodide detector's spin response is still dominated by sodium but by a smaller margin than previously believed.

For neutralino masses above a few tens of GeV and medium-to-heavy targets, the zero-momentum approximation fails and the finite-momentum-transfer structure functions $S_{00}$, $S_{11}$, $S_{01}$ must be used instead.

If the model spread really dominates over the choice of nucleon-nucleon force, then a spin-dependent limit quoted from one nuclear model should be published with the envelope over models attached, in the same way scalar limits carry a form-factor uncertainty band.

The magnetic-moment benchmark is a proxy, not a calibration: an independent handle such as measured Gamow-Teller strength or polarized electron scattering on the same isotope would test whether magnetic-moment agreement actually transfers to the axial-vector matrix element.

For $^{73}$Ge the two most comprehensive calculations disagree by roughly a factor of two in $⟨S_n⟩$, so any germanium-based spin-dependent limit inherits that factor as a theory error rather than a statistical one.

The tables are indifferent to whether the WIMP is a neutralino, since the same axial-vector matrix elements govern any Majorana WIMP coupling to spin, so the compilation outlives the specific supersymmetric motivation.

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review