Kubo–Toyabe

The Kubo–Toyabe family describes depolarisation by a random static (or stochastically fluctuating) distribution of local fields, with the characteristic zero-field dip and \(1/3\) tail. The notation follows Chapter 5 of Blundell, De Renzi, Lancaster and Pratt: the Gaussian width is \(\Delta\) (μs⁻¹), the Lorentzian half-width \(a_L\) (μs⁻¹), the fluctuation (hop) rate \(\nu\) (MHz ≡ μs⁻¹), and the applied longitudinal field \(B_L\) (Gauss). Two universal limits bracket every dynamic member of the family:

  • \(\nu \to 0\) recovers the static function (with its zero-field \(1/3\) tail);

  • \(\nu \gg \Delta\) gives motional narrowing — exponential decay with rate \(2\Delta^2/\nu\) (Gaussian, zero field), washing the tail away.

A longitudinal field adds a Larmor term \(\omega_0 = \gamma_\mu B_L\) that decouples the muon (\(G \to 1\) as \(B_L \to \infty\)); recovery under a field satisfying \(\gamma_\mu B_L\gg\Delta\) is consistent with static fields, but slow fluctuations (:math:` ullDelta`) can also decouple. Field recovery alone does not uniquely establish a static state. For an end-to-end walk-through see LF decoupling and static-vs-dynamic field distributions.

StaticGKT_ZF

\[A(t) = A\left[\tfrac{1}{3} + \tfrac{2}{3}\left(1-\Delta^2 t^2\right)e^{-\Delta^2 t^2/2}\right]\]

The zero-field static Gaussian Kubo–Toyabe function: the foundational response of a host in which a static, isotropic, Gaussian-distributed local field (typically randomly oriented nuclear moments) dominates. The \(1/3\) tail is the angular average of the squared spin projection along the local field; that component does not dephase. The complementary transverse projection precesses and dephases into the dip at \(t=\sqrt{3}/\Delta\). The width

\[\Delta = \gamma_\mu\sqrt{\langle B_x^2\rangle} = \gamma_\mu B_c\]

uses the standard deviation of one Cartesian component of the zero-mean isotropic field distribution; \(\langle|\mathbf B|^2\rangle=3\Delta^2/\gamma_\mu^2\). It sets both the dip position and the early-time response \(1-\Delta^2t^2+(5/12)\Delta^4t^4+O(t^6)\). For metals and diamagnetic hosts with only nuclear moments \(\Delta \sim 0.1{-}0.5\;\mu s^{-1}\); appreciably larger values can also reflect short muon–nucleus distances, several sites, or an electronic contribution.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Delta

\(\Delta\)

μs⁻¹

Static Gaussian field-distribution width.

A good seed follows directly from the data: the dip sits at \(t_{\min} = \sqrt{3}/\Delta\). For \(\Delta t\ll1\) it agrees to leading order with Gaussian at \(\sigma=\Delta\) (Gaussian), with a fourth-order difference \(-\Delta^4t^4/12\). A window ending before the dip can still resolve that difference if its precision is sufficient. Relaxation of the tail can indicate fluctuations or an additional relaxation channel; use DynamicGaussianKT when field dynamics are required by the data. Common composites are StaticGKT_ZF + Constant and StaticGKT_ZF * Exponential + Constant when an additional dynamic channel modulates the static recovery. A standalone StaticGKT_ZF model is available in the MODELS registry.

References

  • R. Kubo and T. Toyabe, in Magnetic Resonance and Relaxation, edited by R. Blinc (North-Holland, Amsterdam, 1967), p. 810.

  • R. S. Hayano, Y. J. Uemura, J. Imazato, N. Nishida, T. Yamazaki, and R. Kubo, Phys. Rev. B 20, 850 (1979).

LongitudinalFieldKT

Overlay of five Ag LF Kubo–Toyabe runs spanning the decoupling regime

Synthetic Ag polycrystal LF series with Δ ≈ 0.39 μs⁻¹ and B_L = 0, 5, 10, 25, 50 G. The 0 G run shows the characteristic 1/3 tail; as B_L grows the muon spins decouple from the nuclear dipolar field and the polarisation recovers toward unity.

\[A(t) = A\left\{1 - \frac{2\Delta^2}{\omega_0^2}\left[1 - e^{-\Delta^2 t^2/2} \cos(\omega_0 t)\right] + \frac{2\Delta^4}{\omega_0^3}\int_0^t e^{-\Delta^2\tau^2/2} \sin(\omega_0\tau)\,d\tau\right\}, \qquad \omega_0 = \gamma_\mu B_L\]

The static Gaussian Kubo–Toyabe function in a longitudinal field — the workhorse for magnetically disordered hosts where the local field is static on the muon time scale (frozen spin systems, dilute nuclear-dipole hosts) and the experiment sweeps \(B_L\) through the decoupling crossover \(\gamma_\mu B_L \sim \Delta\) to extract \(\Delta\). The \(B_L \to 0\) limit recovers StaticGKT_ZF exactly; at large \(B_L\) the polarisation decouples toward unity. If the polarisation does not recover with field, the local field is dynamic — use DynamicGaussianKT or Keren.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Delta

\(\Delta\)

μs⁻¹

Static Gaussian field-distribution width.

B_L

\(B_L\)

G

Applied longitudinal field.

When a dataset’s metadata carries a known applied field, the fit panel initialises B_L from it; fix B_L whenever it is not the quantity of interest. The fit wizard does that for you — it pins B_L from the recorded geometry and setpoint (at 0 on a zero-field run) and leaves it ticked Fix in the parameter table when you apply the candidate. \(\Delta\) is partially degenerate with the amplitude in a single run — pin it with a decoupling field sweep, or a global fit sharing \(\Delta\) across runs (Global fit wizard). The oscillatory integral is evaluated for all requested times at once by cumulative trapezoidal integration on a shared fine grid (accurate to better than 10⁻⁶); for zero-field-only data use the cheaper StaticGKT_ZF directly. A standalone LFKuboToyabe model is available in the MODELS registry.

References

  • R. S. Hayano, Y. J. Uemura, J. Imazato, N. Nishida, T. Yamazaki, and R. Kubo, Phys. Rev. B 20, 850 (1979).

      1. Hillier and R. Cywinski, Appl. Magn. Reson. 13, 95 (1997).

DynamicGaussianKT

The strong-collision (Markovian) dynamic generalisation of the Gaussian Kubo–Toyabe function: the static field of width \(\Delta\) reorients stochastically at rate \(\nu\) — muon hopping, ionic motion, or thermally fluctuating moments. The dynamic polarisation is obtained from the static function \(G^{\mathrm{s}}(t)\) by the strong-collision relation

\[G^{\mathrm{d}}(t) = G^{\mathrm{s}}(t)\,e^{-\nu t} + \nu\int_0^t G^{\mathrm{d}}(t-t')\,G^{\mathrm{s}}(t')\,e^{-\nu t'}\,dt' ,\]

solved on a uniform grid (trapezoidal rule) with the step chosen so the result is grid-independent to better than 0.5 %; solutions are cached per \((\Delta, \nu, B_L, t_{\max})\). This is the standard model for extracting a hop/fluctuation rate and its activation energy in metals (Cu) and ionic conductors. \(\nu \to 0\) recovers the static (LF) function; \(\nu \gg \Delta\) approaches exponential decay at rate \(2\Delta^2/\nu\).

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Delta

\(\Delta\)

μs⁻¹

Static Gaussian field-distribution width.

nu

\(\nu\)

MHz

Fluctuation (hop) rate.

B_L

\(B_L\)

G

Applied longitudinal field.

In the fast/intermediate regime the analytic Keren function is an excellent and cheaper alternative. A standalone DynamicGaussianKT model is available in the MODELS registry.

References

  • R. S. Hayano, Y. J. Uemura, J. Imazato, N. Nishida, T. Yamazaki, and R. Kubo, Phys. Rev. B 20, 850 (1979).

DynamicLorentzianKT

The Lorentzian analogue of DynamicGaussianKT, for dilute or randomly diluted moments (spin glasses, dilute-spin systems) whose local-field distribution is Lorentzian with half-width \(a_L\) rather than Gaussian. The zero-field static limit is the analytic Lorentzian Kubo–Toyabe,

\[G^{\mathrm{s}}(t) = \tfrac{1}{3} + \tfrac{2}{3}(1 - a_L t)\,e^{-a_L t},\]

and in zero field the strong-collision dynamics have an exact closed form: the static function is a three-state linear system, so the strong-collision integral equation reduces to a 3 × 3 eigenproblem and \(G^{\mathrm{d}}(t)\) is a sum of three exponentials, valid at every rate with no integration grid, cache, or fast-fluctuation crossover. Unlike the Gaussian case there is no motional narrowing — a Lorentzian distribution has no second moment — and the closed form shows the relaxation rate saturating at \(4a_L/3\) for \(\nu \gg a_L\).

In a longitudinal field the static line shape is evaluated from a closed-form spectral density: writing the stochastic field average over the isotropic Lorentzian distribution as a cosine transform in the total-field magnitude \(W\) and doing the magnitude integral analytically gives

\[S(W) = \frac{a_L}{2\pi\,\omega_0^{3}\,W}\left[ \left(a_L^{2} + W^{2} + \omega_0^{2}\right) \ln\frac{a_L^{2} + (W + \omega_0)^{2}}{a_L^{2} + (W - \omega_0)^{2}} - 4\,\omega_0 W\right],\]

with \(\omega_0 = \gamma_\mu B_L\), whose \(1/W^{2}\) tail is exactly the zero-field density (carrying the \(e^{-a_L t}\) cusp analytically) and whose remainder is transformed by one FFT with the frequency step chosen so its conjugate time step is the integration grid’s. The line shape is accurate to ~10⁻⁶ at any field and resolves the Larmor oscillation however long the window; the dynamic solution in a field is then the strong-collision solver of DynamicGaussianKT on that grid. The field path costs a few times the zero-field closed form, so fix \(B_L\) from the known applied field where it is not the quantity of interest.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

a_L

\(a_L\)

μs⁻¹

Lorentzian field-distribution half-width.

nu

\(\nu\)

MHz

Fluctuation (hop) rate.

B_L

\(B_L\)

G

Applied longitudinal field.

A standalone DynamicLorentzianKT model is available in the MODELS registry.

References

  • Y. J. Uemura, T. Yamazaki, D. R. Harshman, M. Senba, and E. J. Ansaldo, Phys. Rev. B 31, 546 (1985).

GaussianBroadenedKT

\[A(t) = A\int d\Delta'\,p(\Delta')\, G^{\mathrm{LF}}_{\mathrm{KT}}(t;\Delta',B_L), \qquad p = \mathcal{N}\!\left(\Delta,\,(w_\Delta\Delta)^2\right)\]

The static (longitudinal-field) Gaussian Kubo–Toyabe averaged over a Gaussian distribution of widths \(\Delta\) — for disordered hosts where a single-width KT fit is qualitatively right but the dip is too sharp and the \(1/3\)-tail recovery too pronounced: structurally disordered systems, dilute magnetic alloys, or several inequivalent muon sites. The fractional standard deviation \(w_\Delta\) is the broadening parameter (\(w_\Delta = 0\) reduces exactly and continuously to LongitudinalFieldKT); the average is evaluated by Gauss–Hermite quadrature directly at the requested times, vectorised over the nodes. WiMDA’s Gau broad KT “rel width” equals \(w_\Delta\sqrt{2}\).

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Delta

\(\Delta\)

μs⁻¹

Mean Gaussian field-distribution width.

B_L

\(B_L\)

G

Applied longitudinal field.

w_rel

\(w_\Delta\)

—

Fractional standard deviation of Δ.

Beware a fundamental ambiguity: width broadening and dynamics both fill in the dip and soften the tail, and a single spectrum rarely distinguishes them — vary temperature or field before preferring this model over DynamicGaussianKT.

References

      1. Noakes and G. M. Kalvius, Phys. Rev. B 56, 2352 (1997).