Relaxation

Relaxation components describe the decay of the muon-spin polarisation without coherent oscillation. They are used standalone for zero- and longitudinal-field relaxation, and multiplicatively as damping envelopes on oscillating components (Oscillatory * Exponential, FmuF_Linear * Exponential, …). This page covers the simple envelopes (Exponential, Gaussian, StretchedExponential), the dynamic-crossover functions (Abragam, Keren), and the 1D-transport function RischKehr. Relaxation from static field distributions with the characteristic \(1/3\) tail lives under Kubo–Toyabe.

Exponential

\[A(t) = A\,e^{-\lambda t}\]

The workhorse depolarisation function for any system in which the muon experiences a rapidly fluctuating local field. It is the Redfield motional-narrowing limit of slower-relaxation forms: when the fluctuation rate \(\nu\) of the local field is large compared with its static second moment \(\Delta\), the depolarisation collapses from a Kubo–Toyabe shape onto a pure exponential with rate \(\lambda \simeq 2\Delta^2/\nu\). In practice this covers paramagnetic spin fluctuations, electronic relaxation in metals and semiconductors where coherent dynamics are absent, and the high-temperature regime of essentially every diffusive system once the correlation time has dropped below the muon precession scale. The same shape also arises from a static but dilute (Lorentzian) field distribution in transverse field.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Lambda

\(\lambda\)

μs⁻¹

Exponential relaxation rate.

Both parameters are constrained non-negative by default. For a paramagnetic salt at room temperature \(\lambda \sim 0.1{-}1\;\mu s^{-1}\) is typical; above roughly \(\lambda \sim 10\;\mu s^{-1}\) the relaxation occurs almost entirely inside the instrumental deadtime window and Lambda becomes ill-conditioned — in that regime the signal is better described as missing initial asymmetry than as a fast exponential. When \(\lambda\) is recovered alongside a slow Gaussian or Kubo–Toyabe channel, expect correlation between the two rates; LF-decoupling data on the same sample is usually the cleanest way to break the degeneracy.

A standalone ExponentialRelaxation model (with explicit baseline) is available in the Python MODELS registry for scripted single-channel fits.

References

  • A. Yaouanc and P. Dalmas de Réotier, Muon Spin Rotation, Relaxation, and Resonance: Applications to Condensed Matter (Oxford University Press, Oxford, 2011).

Gaussian

\[A(t) = A\,e^{-(\sigma t)^2}\]

For static Gaussian frequency broadening of a transverse-field precession line, the envelope is \(\exp(-\Delta^2t^2/2)\), so this component uses \(\sigma=\Delta/\sqrt{2}\). Here \(\Delta=\gamma_\mu B_c\), where \(B_c\) is the standard deviation of the field component parallel to the applied field and \(\gamma_\mu\) is the angular gyromagnetic ratio. For an isotropic field distribution this is the high-field envelope, with transverse field components small compared with the applied field.

In zero field, the static Gaussian Kubo–Toyabe function (StaticGKT_ZF) instead has the short-time expansion

\[G_{\mathrm{KT}}(t)=1-\Delta^2t^2+\frac{5}{12}\Delta^4t^4+O(t^6).\]

Its early-time match to \(e^{-(\sigma t)^2}\) is therefore \(\sigma=\Delta\). The finite-time difference is \(G_{\mathrm{KT}}(t)-e^{-\Delta^2t^2}=-\Delta^4t^4/12+O(t^6)\); the Gaussian does not reproduce the full dip and recovery. In this isotropic zero-field model, \(B_c=\Delta/\gamma_\mu\) is the width of one Cartesian component, and \(\langle|\mathbf B|^2\rangle=3B_c^2\). See Gaussian widths for the conventions across related models.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

sigma

\(\sigma\)

μs⁻¹

Gaussian relaxation rate.

For polycrystalline samples with only nuclear moments, typical values lie in the range \(\sigma \sim 0.1{-}0.5\;\mu s^{-1}\), up to about \(1\;\mu s^{-1}\) when light, high-moment nuclei such as 1H or 19F are dense at the muon site. Larger widths can reflect short muon–nucleus distances, several inequivalent sites, or electronic moments. Resolved coupled F–μ–F oscillations call for the dedicated components in Nuclear dipolar. The \(1/3\) tail is a feature of the full static isotropic zero-field response, not of a transverse-field envelope. Use StaticGKT_ZF when the time window includes its dip and recovery; additional relaxation channels can modify the tail.

A standalone GaussianRelaxation 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).

StretchedExponential

\[A(t) = A\,e^{-(\lambda t)^{\beta}}\]

The Kohlrausch–Williams–Watts form, interpolating continuously between a simple exponential (\(\beta = 1\)) and a Gaussian (\(\beta = 2\)). It is the standard phenomenological relaxation function for systems in which the muon ensemble samples a distribution of relaxation rates rather than a single \(\lambda\) — spin glasses near and below freezing, dilute and concentrated magnetic alloys with broad RKKY-coupling distributions, frustrated magnets with quenched disorder.

The stretching exponent \(\beta\) is the physically informative parameter. \(\beta\) near 1 indicates a narrow distribution of dynamic rates; \(\beta = 1/2\) is the Walstedt–Walker form expected for dilute, broadly distributed static moments in the fast-fluctuation limit; values approaching 2 indicate a static near-Gaussian distribution better fitted with a Kubo–Toyabe form. A fitted \(\beta\) drifting from 1 toward \(\approx 1/3\) on cooling through a transition is one of the canonical μSR signatures of glassy freezing.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Lambda

\(\lambda\)

μs⁻¹

Relaxation rate scale.

beta

\(\beta\)

—

Stretching exponent.

Constrain \(\beta\) away from 0 (min=0.1 or similar) and cap it at 2. \(\lambda\) and \(\beta\) are strongly correlated: changing \(\beta\) rescales the effective rate, so the marginal uncertainty on \(\lambda\) from the Hessian usually understates the truth. Where this matters, fix \(\beta\) at a physically motivated value or quote the joint covariance. A standalone StretchedExponential model is available in the MODELS registry.

References

      1. Walstedt and L. R. Walker, Phys. Rev. B 9, 4857 (1974).

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

      1. Campbell et al., Phys. Rev. Lett. 72, 1291 (1994).

Abragam

\[A(t) = A\,\exp\!\left[-\frac{\Delta^2}{\nu^2} \left(e^{-\nu t} - 1 + \nu t\right)\right]\]

The Gaussian-to-exponential crossover function: a Gaussian static width \(\Delta\) fluctuating at rate \(\nu\), with the limits \(\nu \to 0:\ \exp(-\Delta^2 t^2/2)\) (static transverse-field Gaussian, equal to Gaussian at \(\sigma=\Delta/\sqrt{2}\)) and \(\nu \gg \Delta:\ \exp(-(\Delta^2/\nu)\,t)\) (motionally narrowed exponential). It is the classic model for extracting a hop or correlation rate from a transverse-field line shape — the textbook example being the Gaussian-to-Lorentzian change of the Cu line shape as muon diffusion sets in on warming. Numerically, the exponent is evaluated as \(\Delta^2|t|^2\varphi_2(-\nu|t|)\), where \(\varphi_2(z)=(e^z-1-z)/z^2\) and \(\varphi_2(0)=1/2\) (Hochbruck and Ostermann, 2010). A Taylor expansion around the origin avoids cancellation without substituting a static function below an absolute rate. The rate must be finite and non-negative; negative or non-finite values raise ValueError.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Delta

\(\Delta\)

μs⁻¹

Static Gaussian field-distribution width.

nu

\(\nu\)

MHz

Fluctuation (hop) rate.

References

Keren

\[A(t) = A\,e^{-\Gamma(t)},\qquad \Gamma(t)=\frac{2\Delta^2}{(\omega_0^2+\nu^2)^2} \Big[(\omega_0^2+\nu^2)\,\nu t +(\omega_0^2-\nu^2)(1-e^{-\nu t}\cos\omega_0 t) -2\nu\omega_0 e^{-\nu t}\sin\omega_0 t\Big]\]

with \(\omega_0 = \gamma_\mu B_L\). This is the exponent \(\Gamma(t)t\) in Keren’s Eq. (4). His perturbative generalisation of the Abragam function is an expansion in \(\Delta/\nu\), expected to approximate the dynamic result when \(\Delta/\nu<1\), or at sufficiently early times. It does not reproduce the static Kubo–Toyabe recovery. Prefer the full DynamicGaussianKT for slow fluctuations or a static \(1/3\) tail. At \(B_L=0\), its exponent is twice the transverse-field Abragam exponent, because both transverse field components contribute in longitudinal geometry.

The numerically equivalent expression is \(\Gamma(t)=2\Delta^2|t|^2\operatorname{Re}\varphi_2[-(\nu+i\omega_0)|t|]\). It remains continuous at zero rate and field and is even in time and field. At \(\nu=B_L=0\) it gives \(\exp(-\Delta^2t^2)\); at zero rate with finite field it gives \(\exp[-2\Delta^2(1-\cos\omega_0t)/\omega_0^2]\). These numerical limits do not extend the perturbative model’s physical validity. Negative or non-finite rates raise ValueError.

For unit amplitude, independent 70-digit exponential/trigonometric checks give absolute polarisation differences below \(10^{-14}\) on a grid spanning rates from zero to \(10^4\) μs⁻¹, widths 0.04–4 μs⁻¹, fields from zero to 100 G, and times 0–8 μs, including rates immediately around \(10^{-9}\) μs⁻¹. These are analytic checks, rather than measured agreement with another program.

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.

References

RischKehr

\[A(t) = A\, e^{\Gamma t}\,\mathrm{erfc}\!\left(\sqrt{\Gamma t}\right)\]

Relaxation of the muon (or muonium) polarisation by a spin carrier diffusing in one dimension — a polaron moving along a conducting-polymer chain, or an excitation confined to a structural channel. The 1D random walk keeps returning the carrier to the muon, so instead of an exponential the polarisation acquires a \((\pi\Gamma t)^{-1/2}\) long-time tail. A stretched-exponential fit drifting toward \(\beta \approx 1/2\) at early times is the usual hint to try this form; prefer it over a stretched exponential whenever 1D transport is physically motivated, since \(\Gamma\) then has a microscopic interpretation in terms of the intrachain diffusion rate and hyperfine coupling.

Name

Symbol

Unit

Description

A

\(A\)

%

Component asymmetry amplitude.

Gamma

\(\Gamma\)

μs⁻¹

Risch–Kehr relaxation rate.

\(\Gamma\) is constrained non-negative. The implementation evaluates the scaled complementary error function (erfcx), which is numerically stable for all \(\Gamma t\) — there is no asymptotic-branch switch (WiMDA changes form at \(\Gamma t = 20\)), and WiMDA’s mirrored branch for negative rates is intentionally not reproduced.

References

    1. Risch and K. W. Kehr, Phys. Rev. B 46, 5246 (1992).