Time-Resolved Elastic Neutron Scattering

• Time-resolved elastic neutron scattering is a technique that interprets cumulative elastic counts to reveal the underlying time-domain diffusion dynamics.
• It employs precise instrumental resolution and differentiation methods to recover the van Hove self-distribution function and characteristic relaxation times.
• Applications span spin-echo, MIEZE, and SANS implementations, enabling in situ strain imaging, adaptive measurement sufficiency, and probing heterogeneous dynamics.

In the cited literature, time-resolved elastic neutron scattering appears as a set of related methodologies in which temporal information is carried by elastic or effectively elastic neutron signals rather than obtained only from a scanned energy-exchange spectrum. The most explicit formulation recasts elastic intensity measured as a function of instrumental resolution time as the running time-integral of the van Hove self-distribution function; spin-echo-derived implementations encode comparable dynamical information through intensity modulation or depolarization at the elastic echo; time-resolved Small-Angle Neutron Scattering (SANS) treats the evolving elastic profile $I(Q)$ itself as the central observable; and energy-resolved transmission methods use coherent elastic Bragg-edge shifts to reconstruct internal strain fields in situ (Benedetto et al., 2019, Georgii et al., 2011, Hendriks et al., 2017, Tung et al., 17 May 2025).

1. Time-domain meaning of the elastic signal

The foundational dynamical quantity in the direct time-domain formulation is the van Hove self-distribution function $G(r,t)$, or equivalently its spatial Fourier transform $I(Q,t)$, the intermediate scattering function. In the 2019 formulation, $G(r,t)$ is described as the probability that a species has diffused a distance $r$ over time $t$, while $I(Q,t)$ is its Fourier counterpart in momentum-transfer space, representing the probability that a species is within a characteristic spatial volume after time $t$. The key point is that $I(Q,t)$ fully encodes the single-particle dynamics (Benedetto et al., 2019).

The central reinterpretation is that an experimentally counted elastic intensity is not simply “elastic scattering” in the intuitive sense. When one spectrometer is made much narrower than the other and the experiment is operated in the purely elastic regime, the detected intensity becomes a weighted time integral of $I(Q,t)$, and in the ideal limit a running integral:

$G(r,t)$0

Differentiation then returns the dynamical correlator,

$G(r,t)$1

so the elastic count rate as a function of observation time is interpreted as the cumulative probability that motion has not yet been resolved (Benedetto et al., 2019).

A related but conceptually distinct elastic-resolution viewpoint appears in Elastic Scattering Spectroscopy (ESS). There the measured elastic intensity at zero energy transfer is written as the convolution of the instrumental resolution with the system intermediate scattering function,

$G(r,t)$2

and the overall relaxation time is identified from the inflection point of the elastic-intensity-versus-resolution curve, equivalently from the maximum in its second derivative (Benedetto et al., 2017). This suggests two elastic routes to dynamics: one aimed at recovering $G(r,t)$3 itself by differentiation of a cumulative signal, and one aimed at extracting a characteristic relaxation time from the curvature of that signal.

2. Direct measurement of the van Hove integral

The direct method starts from the general neutron-scattering experiment in which a primary spectrometer prepares an incoming distribution $G(r,t)$4, the sample contributes $G(r,t)$5 or equivalently $G(r,t)$6, and a secondary analyzer or filter transmits according to $G(r,t)$7. In time representation, the detected neutron number is the time integral of $G(r,t)$8 weighted by the two resolution functions. When $G(r,t)$9 and $I(Q,t)$0, the filter factor is effectively unity over the support of the primary resolution, so the detected elastic intensity reduces to the primary-resolution-weighted integral of $I(Q,t)$1. In the ideal step-function limit, the elastic measurement becomes exactly $I(Q,t)$2. The observation time is tied to the energy resolution by

$I(Q,t)$3

with an analogous relation for the filter, $I(Q,t)$4 (Benedetto et al., 2019).

This framework is sharply distinguished from standard quasielastic neutron scattering. In the conventional route, one scans energy exchange $I(Q,t)$5, measures $I(Q,t)$6, deconvolves the instrumental resolution, and then Fourier transforms to obtain $I(Q,t)$7. The direct elastic-resolution method avoids both steps: it stays at $I(Q,t)$8, varies the observation time through the instrumental energy resolution, and interprets the elastic fraction as the cumulative count of particles that remain effectively stationary within the chosen time window. A common misconception addressed explicitly in the paper is that elastic intensity versus resolution measures $I(Q,t)$9 directly; the stated correction is that it measures the running integral of $G(r,t)$0, not $G(r,t)$1 itself (Benedetto et al., 2019).

Practical feasibility was tested numerically using a single exponential, a sum of two exponentials, and a stretched exponential as representative inputs for simple diffusion, multiple relaxation processes, and heterogeneous dynamics. After Fourier transformation to $G(r,t)$2, convolution with a Gaussian primary resolution, multiplication by a narrow Gaussian filter, and integration, the synthetic $G(r,t)$3 curves were numerically differentiated. The recovered $G(r,t)$4 matched the inputs closely, with discrepancies attributed mainly to Gaussian rather than step-like resolution functions. When counting statistics were added, direct differentiation became unstable, and two smoothing or differentiation strategies were tested: polynomial fitting followed by analytic differentiation, and a Monte Carlo/Gaussian-error reduction scheme. Both recovered the input dynamics reasonably well. McStas simulations of a proposed backscattering spectrometer then generated twenty-five observation times from about $G(r,t)$5 to $G(r,t)$6 ps; vanadium served to characterize instrumental effects, and a synthetic sample with single-exponential relaxation of $G(r,t)$7 ps yielded $G(r,t)$8 whose derivative lay close to the expected exponential decay. A practical conclusion was that the original mismatch between primary and secondary resolutions, previously seen as a drawback in earlier ESS designs, is exactly what enables access to $G(r,t)$9 (Benedetto et al., 2019).

3. Spin-echo implementations: ESS-NSE, MIEZE, and MISANS

ESS-NSE was proposed as a third ESS implementation, alongside continuous-wavelength and time-of-flight ESS, by using neutron spin echo at or near the elastic echo condition while systematically changing the instrumental resolution. In standard NSE, the first magnetic precession region produces a phase

$r$0

and an ideal elastic scatterer yields zero net depolarization at the echo after the second region. ESS-NSE repurposes that configuration: one measures depolarization at the elastic echo as field strength, field length, and/or wavelength are varied, and one reads the system’s overall relaxation time from the inflection point of the depolarization-versus-resolution curve. The paper estimates that a favorable configuration with $r$1, $r$2, and $r$3 yields $r$4, corresponding to $r$5, while a coarser configuration with $r$6, $r$7, and $r$8 gives $r$9, corresponding to tens of picoseconds. The proposed accessible span is about five orders of magnitude in energy resolution, probing relaxation times from nanoseconds down to tens of picoseconds. Because biophysical samples are often small, the proposal also discusses a curved, high-mosaic monochromator and shaped magnetic fields so that different wavelengths satisfy a common echo at the detector rather than at the sample (Benedetto et al., 2017).

MIEZE, by contrast, is a neutron resonance spin-echo-derived technique in which all beam manipulations are completed before the sample. A compact “MIEZE box” built from two phase-locked neutron resonance spin flippers, together with polarizer/analyzer optics and a fast detector, was reported as a turn-key route to quasi-elastic neutron scattering in the low sub-$t$0eV range. The second flipper reverses the energy splitting introduced by the first, and the two spin components re-overlap at

$t$1

A polarizing analyzer converts the spin interference into a detector intensity beat, with contrast $t$2. The MIEZE time is

$t$3

which the paper states is equivalent to the spin-echo time in NSE/NRSE. At MIRA, with $t$4, $t$5, $t$6, and flipper frequencies in the approximate ranges $t$7 and $t$8, the accessed interval was $t$9. A direct-beam test showed that the modulation contrast was unaffected by a magnetic field up to $I(Q,t)$0 T, and measurements on MnSi demonstrated operation below $I(Q,t)$1 K, above $I(Q,t)$2, and in the A-phase at $I(Q,t)$3 T (Georgii et al., 2011).

For SANS, the same upstream-modulation principle was developed into a MISANS concept that would add high energy resolution to a structural small-angle instrument. The detector intensity is written as

$I(Q,t)$4

with the MIEZE spin-echo analogue

$I(Q,t)$5

The paper emphasizes that MIEZE is sensitive to flight-path differences $I(Q,t)$6, which reduce contrast through sample geometry, scattering angle, coil imperfections, and detector thickness. Analytical reduction factors were derived for spheres, cylinders, cuboids, and disks, and for plate-like samples a counter-rotation by $I(Q,t)$7 minimizes the $I(Q,t)$8-dependence so that the sample thickness becomes the dominant geometrical parameter. A proposed ESS SANS beamline with a $I(Q,t)$9 m collimation section, $t$0 m, $t$1 m, and $t$2 m, operated at $t$3, was presented as capable of $t$4 using $t$5 and $t$6. Together with TISANE and stroboscopic SANS, the proposed accessible window extends from nanoseconds to minutes, while MIEZE itself was stated to reach time scales of $t$7s at momentum transfers up to $t$8 (Brandl et al., 2011).

4. Time-resolved SANS and measurement sufficiency

In time-resolved SANS, the primary observable is the elastically scattered intensity $t$9 as a function of momentum transfer, and the central experimental problem becomes how long to measure at each time point so that the profile is statistically reliable without blurring the underlying kinetics. A 2025 study addressed this question by combining Gaussian Process Regression (GPR) with a dimensionless convergence metric derived directly from the measured profiles. In that framework,

$I(Q,t)$0

so the profile is reconstructed nonparametrically with a posterior mean or MAP estimate and uncertainty bands. The paper emphasizes that this is model-independent and does not require assuming a sphere form factor, Guinier law, or other specific structural model (Tung et al., 17 May 2025).

The convergence metric compares adjacent time-resolved profiles, normalizes the change by the measurement uncertainty, and further normalizes by the typical signal magnitude. Its $I(Q,t)$1-averaged form, the mean relative variation, is plotted against total measurement time in log-log coordinates. Demonstrated on the EQSANS instrument at the Spallation Neutron Source using a D$I(Q,t)$2O solution of CTAB/NaSal at cumulative exposure times of $I(Q,t)$3 s, $I(Q,t)$4 s, $I(Q,t)$5 s, and $I(Q,t)$6 s, the method showed that time-normalized convergence curves collapse when the time axis is scaled by a system-specific characteristic time $I(Q,t)$7. Across multiple soft-matter systems, the scaling exponent lay between $I(Q,t)$8 and $I(Q,t)$9. The statistical interpretation given in the paper is that the $I(Q,t)$0 regime corresponds to very low counts with many zero-count pixels, while the $I(Q,t)$1 regime corresponds to high-count Poisson behavior and is connected to the Central Limit Theorem (Tung et al., 17 May 2025).

The most practically significant claim is that the scaling trend becomes apparent very early: after only the first ten time steps, the regression lines for $I(Q,t)$2 versus $I(Q,t)$3 stabilized into a narrow envelope. This supports early forecasting of measurement sufficiency and real-time adaptive optimization, particularly in low-flux environments such as compact accelerator-based neutron sources. Within the scope stated in the paper, the current validation is centered on isotropic 1D profiles obtained from 2D detector data and on soft-matter systems, with future directions including extension to 2D and anisotropic scattering and integration into real-time feedback workflows (Tung et al., 17 May 2025).

5. Transmission geometries, Bragg edges, and detector temporal response

A distinct branch of time-resolved elastic neutron work uses transmission rather than diffraction geometry. In Bragg-edge elastic strain tomography, pulsed-source time-of-flight neutron transmission provides wavelength-resolved images in which coherent elastic scattering from lattice planes produces sharp Bragg edges. The strain observable is the path-averaged normal strain,

$I(Q,t)$4

with the underlying Bragg relation

$I(Q,t)$5

The measurement is modeled by the Longitudinal Ray Transform, so a single ray reports a line integral of the projection of the strain tensor along the beam direction. The inverse problem is explicitly stated to be ill-posed and non-unique in general, following work by Lionheart, Withers, and Sharafutdinov. For in situ loaded systems with no residual stress, no eigenstrain, and no body forces, however, Wensrich et al. showed that the measurement can be reformulated in terms of boundary displacements, leading to a tractable inversion followed by a Dirichlet boundary-value problem solved by finite element analysis. An experimental proof-of-concept on RADEN at J-PARC used a 2D C-shaped EN-26 steel sample under about $I(Q,t)$6 kN compressive load, an MCP detector with $I(Q,t)$7 pixels, about $I(Q,t)$8m spatial resolution during the experiment, nano-second time resolution, and $I(Q,t)$9 projection angles collected by golden-angle increments. Using $G(r,t)$00 boundary nodes, the reconstruction recovered $G(r,t)$01, $G(r,t)$02, and $G(r,t)$03, with validation by Digital Image Correlation and constant-wavelength neutron strain scans on KOWARI (Hendriks et al., 2017).

Detector physics also becomes central when time resolution is encoded by gated transmission imaging rather than by event-by-event spectroscopy. The TRION detector for Fast Neutron Resonance Radiography operates in integrative TOF imaging mode: instead of recording each neutron arrival time, it integrates the detector signal during a chosen neutron TOF window corresponding to a selected energy bin in the $G(r,t)$04–$G(r,t)$05 MeV range. The paper identifies the principal temporal-resolution contributors as accelerator beam burst duration, source-to-detector distance, gate and delay jitter, detector thickness, scintillator decay time constants, neutron lifetime within the scintillator, and minimal achievable image-intensifier gate width, with the detector-focused analysis emphasizing scintillator decay, gate width, and neutron residence or scattering time in the scintillator. For the NE102 plastic scintillator, the decay model used was

$G(r,t)$06

and the paper reported substantial contrast losses at narrow resonances. Minimum effective gate opening was reduced from about $G(r,t)$07 ns in TRION Gen. 1 to $G(r,t)$08 ns in TRION Gen. 2. The practical deconvolution strategy used the measured gamma-ray TOF peak from the $G(r,t)$09Be(d,n) source as a system temporal PSF and applied the Lucy–Richardson algorithm to recover lost contrast in graphite transmission spectra. The stated conclusion was that scintillator decay dominates for gate widths shorter than about $G(r,t)$10 ns, while gate width dominates at around $G(r,t)$11 ns or longer (Mor et al., 2013).

6. Scope, misconceptions, and adjacent methods

One persistent misconception is that any elastic-resolution scan directly yields the intermediate scattering function. The direct time-domain formulation rejects that identification: when one spectrometer is much narrower than the other, the elastic intensity is the running integral $G(r,t)$12, and recovering $G(r,t)$13 requires differentiation. Earlier attempts were said either to identify the elastic-resolution profile itself with $G(r,t)$14 or, in the case of RENS, to recognize dynamical content in elastic intensity versus resolution without identifying it specifically as the running integral of $G(r,t)$15 (Benedetto et al., 2019).

A second misconception concerns simple Fourier inversion of a measured inelastic spectrum. A 2021 study on four-dimensional imaging of lattice dynamics argued that, for inelastic scattering data $G(r,t)$16, a direct Fourier transform into a quantity such as

$G(r,t)$17

is not the true time evolution measured in a pump-probe experiment because it violates causality. Instead, the authors reconstruct the causal response function $G(r,t)$18 through the fluctuation-dissipation relation, detailed balance, and the Kramers–Kronig relation, and they state explicitly that the framework is general enough for inelastic neutron scattering. This places the method adjacent to, but not identical with, time-resolved elastic neutron scattering: it is a route from inelastic neutron data to a causal time-domain picture when direct pump-probe neutron diffraction is not feasible (Rana et al., 2021).

A third boundary concerns experiments that are time-resolved only in the beam-timing sense. Measurements of neutron–deuteron elastic scattering at LANSCE used a pulsed 800 MeV proton-driven spallation source with about $G(r,t)$19 ns pulse width and $G(r,t)$20s spacing so that time-of-flight could determine incident neutron energy. The reported observable was the elastic differential cross section $G(r,t)$21 over $G(r,t)$22–$G(r,t)$23 MeV and $G(r,t)$24–$G(r,t)$25 in the center-of-mass frame, with normalization by simultaneously measured $G(r,t)$26 elastic scattering and comparison to Faddeev calculations with and without explicit three-nucleon-force contributions. This is an elastic neutron measurement with time-of-flight resolution, but its primary target is reaction kinematics and few-body dynamics rather than the time-domain correlators, modulation contrasts, or evolving elastic profiles that define the methods discussed above (Ertan et al., 2012).

Taken together, these works delineate a field whose common thread is not a single instrument class but a shared use of elastic or effectively elastic neutron observables to access slow dynamics, evolving structure, or internal strain. Depending on geometry and instrumentation, the relevant temporal variable may be an instrumental observation time, a spin-echo or MIEZE time, a cumulative exposure in time-resolved SANS, or a wavelength-resolved transmission gate. The literature therefore treats time-resolved elastic neutron scattering not as a single technique, but as a technically heterogeneous domain spanning direct dynamical correlation measurements, high-resolution SANS extensions, in situ elastic imaging, and detector-limited time-gated transmission methods (Georgii et al., 2011, Brandl et al., 2011, Hendriks et al., 2017, Tung et al., 17 May 2025).