Structural Scattering: Revealing Hidden Material Order

• Structural scattering is a family of techniques that translates reciprocal-space observables into direct insights on material order, connectivity, and structural heterogeneity.
• It employs diverse modalities—including single-molecule X-ray, neutron, and optical scattering—and advanced inversion methods to extract real-space features from averaged data.
• Applications range from probing local order in liquids and glasses to determining conformational states in biomolecules and engineering materials with tunable optical and terahertz properties.

Structural scattering denotes a family of scattering measurements and theories in which reciprocal-space observables are interpreted as direct signatures of structure, structural heterogeneity, and structural change. Across current work, the term encompasses single-molecule X-ray scattering in the extreme Poisson regime, diffuse and wide-angle X-ray scattering from crystals and biomolecules, small-angle X-ray and neutron scattering from soft matter, coherent speckle analysis in amorphous systems, polarization-resolved optical scattering in correlated disorder, nonlinear second-harmonic scattering from liquids, and terahertz surface scattering from structured materials. The common principle is that measured intensity, angular distribution, speckle statistics, or polarization response can encode pair correlations, connectivity, strain, conformational multiplicity, or phase information that is not evident in an averaged real-space picture (Schultze et al., 2023, Petersen et al., 2023, Tocci et al., 2016).

1. Reciprocal-space observables and structural content

A central formal quantity is the structure factor. For liquid water, it is written as

$$S(\mathbf{q}) = \frac{1}{N} \left\langle \rho_\mathbf{q} \rho_{-\mathbf{q}} \right\rangle, \qquad \rho_\mathbf{q} = \sum_{j=1}^N e^{-i \mathbf{q}\cdot \mathbf{r}_j},$$

and the apparent first diffraction peak can itself contain multiple structural contributions (Shi et al., 2019). In small-angle scattering of DNA nanostars, the measured intensity is expressed as

$$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$

with the experimentally useful factorization $I(q)=S(q)P(q)$ when orientational and positional degrees of freedom are decoupled (Fernandez-Castanon et al., 2016). In WAXS of nucleic acids, the scattering variable is

$$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$

and the accessible structural scale is $d = 2\pi/q$, which places the $0.4 < q < 0.95\, \text{\AA}^{-1}$ regime in the range of helix radius and groove-spacing features (Pabit et al., 2015).

In photonic and thin-film settings, the same reciprocal-space logic appears through the first-order Born approximation and the Ewald sphere construction. There the scattered field in $k$-space is proportional to the Fourier transform of the permittivity perturbation, and the Ewald sphere selects the Fourier components that contribute to a given scattering direction. The formulation is explicitly vectorial and includes polarization factors as well as Fresnel transmission and reflection coefficients for thin films on substrates (Lang et al., 2018). This suggests that “structural scattering” is less a single technique than a common inverse-language for connecting Fourier content to morphology, order, and disorder.

2. Structural inversion and ensemble reconstruction

One major branch of structural scattering is the explicit solution of inverse problems. In single-molecule XFEL scattering, the measured image is a sparse set of photon positions on the detector, acquired in an extreme Poisson regime with random, unknown molecular orientation and potentially unknown conformational state. The Bayesian formulation for a structural ensemble $\{\rho_i\}$ with weights $\{w_i\}$ marginalizes over both conformer index and orientation,

$$P(\text{image} \mid \{\rho_i\}, \{w_i\}) = \sum_{i=1}^n w_i \int_{\mathrm{SO}(3)} P(k_1, ..., k_l \mid R \rho_i) \, dR,$$

and samples the posterior with MCMC using hierarchical simulated annealing, resolution staging, and dynamic filtering of informative images and photons (Schultze et al., 2023). On synthetic data, this framework recovered a weighted ensemble of 8 conformers of alanine dipeptide at a resolution of $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$0–$$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$1, determined folded and unfolded chignolin conformers at $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$2 and $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$3, and achieved $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$4 for single-structure crambin using $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$5 images at 15 photons per image (Schultze et al., 2023). A striking scaling result is that the required number of images scales empirically as $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$6 for an ensemble of $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$7 conformers of size $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$8, rather than as $$I(\mathbf{q}) \equiv \frac{1}{N_S} \left\langle \sum_{l=1}^{N_S} \sum_{m=1}^{N_S} e^{i\mathbf{q} \cdot (\mathbf{r}^{l}_{CM} - \mathbf{r}^{m}_{CM})} F_l(\mathbf{q}) F_m^*(\mathbf{q}) \right\rangle,$$9 for a single structure with the same total number of atoms (Schultze et al., 2023).

A different inversion strategy appears in small-angle scattering of soft materials through Kolmogorov-Arnold Networks. There the learned forward map $I(q)=S(q)P(q)$0 predicts reciprocal-space intensity from structural descriptors, and inversion is performed by minimizing

$I(q)=S(q)P(q)$1

The framework is explicitly designed for cases where closed-form scattering functions are unavailable, and it is demonstrated on lyotropic lamellar phases and charged colloidal suspensions, including irregular experimental $I(q)=S(q)P(q)$2-sampling (Tung et al., 2024). Relative to analytical models, the KAN approach is model-independent, accepts arbitrary $I(q)=S(q)P(q)$3, and uses spline-based smoothness for robust optimization (Tung et al., 2024). Taken together, these studies establish structural scattering as an inversion discipline in which the central challenge is not merely forward simulation of intensity but statistically stable recovery of real-space ensembles or descriptors from incomplete or highly averaged reciprocal-space data.

3. Heterogeneity, averaging, and hidden structural order

A recurring result across the literature is that conventional averaging can erase the very signatures one seeks. In supercooled liquids and glasses, the weak temperature dependence of the traditional structure factor $I(q)=S(q)P(q)$4 is attributed to averaging of the scattering intensity either through incoherent radiation or explicit angular averaging. When static coherent radiation is used without angular averaging, the full speckle pattern at individual $I(q)=S(q)P(q)$5-vectors exhibits strong temperature dependence. The restricted average

$I(q)=S(q)P(q)$6

defines a slope $I(q)=S(q)P(q)$7 that acts as an amorphous Debye-Waller factor and an intercept $I(q)=S(q)P(q)$8 associated with variance of local restraint (Petersen et al., 2023). In the cited simulations, $I(q)=S(q)P(q)$9 drops from 1 in the glassy state to 0 in the liquid, while $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$0 tracks the amplitude of structural heterogeneity (Petersen et al., 2023).

In liquid water, what appears as a single broad first diffraction peak in $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$1 is decomposed into two overlapping peaks: a Lorentzian component identified with a first sharp diffraction peak from the locally favored tetrahedral structure and a Gaussian component associated with the disordered normal-liquid structure. The result is presented as direct evidence for coexistence of two local structural motifs, with motif-resolved scattering functions $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$2 linking local order to reciprocal-space features (Shi et al., 2019). In second-harmonic scattering of water, the conventional assumption that bulk SHS is dominated by incoherent single-molecule terms is revised by a simulation-based decomposition into incoherent and coherent contributions. The coherent term is shown to arise from radial and angular correlations on a length scale of less than $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$3, concentrated within the first and second solvation shells (Tocci et al., 2016).

Diffuse scattering provides a complementary route to heterogeneity in solids. In morphotropic PZT single crystals with $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$4, highly anisotropic quasielastic diffuse scattering is observed for $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$5 and disappears above $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$6. Its $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$7 intensity profile and directional dependence are reproduced by a Huang-scattering model of inhomogeneous lattice deformations caused by tetragonal inclusions embedded in a rhombohedral or monoclinic host phase (Burkovsky et al., 2012). The broader significance is that structural scattering often depends less on stronger signals than on avoiding destructive averages that collapse heterogeneity into a misleadingly featureless mean.

4. Structural change under flow, excitation, and phase transition

Structural scattering is also a method for following nonequilibrium transformations. In colloidal glasses, simultaneous rheology and SAXS resolve shear-induced changes in the nearest-neighbor configuration through the first peak of the angle-averaged structure factor. Under shear, the first peak position $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$8 increases, the width $$q = \frac{4\pi}{\lambda} \sin\left(\frac{\theta}{2}\right),$$9 decreases, and the height $d = 2\pi/q$0 increases, while at $d = 2\pi/q$1 the angular dependence $d = 2\pi/q$2 develops a two-fold “p-wave” symmetry quantified by

$d = 2\pi/q$3

At the shear-banding plateau near $d = 2\pi/q$4, two distinct structural states coexist at the same stress, and the anisotropy disappears in the faster-flowing liquid-like band (Denisov et al., 2013).

In monolayer WSe$d = 2\pi/q$5, femtosecond surface X-ray diffraction on the $d = 2\pi/q$6 crystal truncation rod resolves anisotropic ultrafast lattice dynamics. The diffraction intensity follows the Debye-Waller form

$d = 2\pi/q$7

and the time-resolved analysis shows that the in-plane RMSD increases by 12% within 2 ps, while the out-of-plane RMSD remains virtually unchanged during the first 10 ps (Tung et al., 2018). Model refinement further suggests an asymmetric intralayer spacing change upon excitation, with $d = 2\pi/q$8 and $d = 2\pi/q$9, a behavior attributed to substrate-induced symmetry breaking in the monolayer (Tung et al., 2018).

High-resolution X-ray scattering in Cr-doped Ba(Fe$0.4 < q < 0.95\, \text{\AA}^{-1}$0Cr$0.4 < q < 0.95\, \text{\AA}^{-1}$1)$0.4 < q < 0.95\, \text{\AA}^{-1}$2As$0.4 < q < 0.95\, \text{\AA}^{-1}$3 uses Bragg peak splitting to quantify the tetragonal-to-orthorhombic transition through the strain

$0.4 < q < 0.95\, \text{\AA}^{-1}$4

Increasing Cr concentration suppresses the structural transition temperature $0.4 < q < 0.95\, \text{\AA}^{-1}$5 and reduces $0.4 < q < 0.95\, \text{\AA}^{-1}$6, with a magnetostructural crossover at $0.4 < q < 0.95\, \text{\AA}^{-1}$7 separating strong orthorhombicity and predominantly itinerant behavior from weak orthorhombicity and localized magnetic behavior (Clancy et al., 2011). These cases illustrate that structural scattering is not restricted to static structure determination; it is equally a framework for measuring order-parameter evolution, anisotropic energy flow, and mechanically coupled reorganizations.

5. Soft matter, biomolecular systems, and hierarchical architectures

In soft and biological matter, structural scattering often depends on separating intraparticle form from interparticle organization. For tetravalent DNA nanostars, SANS measurements at dilute concentration determine the form factor $0.4 < q < 0.95\, \text{\AA}^{-1}$8, while temperature-dependent measurements at $0.4 < q < 0.95\, \text{\AA}^{-1}$9 follow aggregation and equilibrium gelation through the effective structure factor. Molecular dynamics with the oxDNA2 model reproduces the form factor without an approximate shape, yields $k$0, and supports an essentially temperature-independent $k$1. On cooling from $k$2C to $k$3C, a structural peak develops near $k$4, corresponding to $k$5–$k$6, and saturates when the fully bonded equilibrium gel forms (Fernandez-Castanon et al., 2016).

In engineered 3D cardiac microtissues, SAXS is used as a non-destructive probe of myofilament lattice spacing. The first-order peak position $k$7 gives

$k$8

and grid-based mapping shows that the myofilament lattice spacing monotonically decreases over ten days of maturation. The 3×3 grid protocol reduces the standard deviation of measured $k$9-spacings across samples from $\{\rho_i\}$0 to $\{\rho_i\}$1, while spatial maps show greater heterogeneity early and increased uniformity by day 7–10 (Dover et al., 2023). In nucleic acids, WAXS combined with all-atom MD uses absolute profiles and the difference profile

$\{\rho_i\}$2

to distinguish DNA and RNA responses to cobalt(III) hexammine. The cited study reports that MD captures the RNA structural change induced by CoHex, whereas DNA remains largely B-form-like (Pabit et al., 2015).

At a more abstract level, formal treatments of composite structures show how scattering from branched or acyclic assemblies can be built from sub-unit quantities. The core objects are the form factor $\{\rho_i\}$3, the form-factor amplitude $\{\rho_i\}$4, and the phase factor $\{\rho_i\}$5, which combine recursively to describe stars, pom-poms, bottle-brushes, dendrimers, micelles, and chains (Svaneborg et al., 2011, Svaneborg et al., 2011). The generalization to distributed reference points allows linkage positions to be probabilistic rather than fixed, which is essential for random grafting or micellar tethering (Svaneborg et al., 2011). This formalism makes explicit that structural scattering can encode architecture as much as geometry: the connectivity graph and the internal sub-unit correlations enter separately and can therefore be modeled modularly.

6. Optical, polarization-resolved, and terahertz regimes

In optical and electromagnetic settings, structural scattering often hinges on coherence, polarization, and dependent-scattering effects. Mutual scattering with two properly phased incident beams is proposed as a way to probe opaque media more sensitively than one-beam differential cross-sections. For ensembles of up to $\{\rho_i\}$6 point scatterers, the mutual-scattering signal is proportional to the imaginary part of the off-forward scattering amplitude, and the “susceptivity”

$\{\rho_i\}$7

is used to quantify how strongly the signal responds to displacement of a single dipole. The reported angular sensitivity is at least 10 times higher than in traditional one-beam techniques, and the depth of the moved scatterer can be inferred from the dependence of susceptivity on the initial position (Truong et al., 2022).

For polarized light in correlated disorder, a vector radiative transfer equation derived from the Dyson and Bethe-Salpeter equations shows that structural correlations, encoded through the phase function

$\{\rho_i\}$8

affect different polarization eigenmodes in distinct ways. Only the scalar intensity mode obeys the familiar $\{\rho_i\}$9; non-scalar modes have their own diffusion constants and attenuation lengths (Vynck et al., 2016). In dual-dipolar random media, dependent scattering is likewise controlled by the structure factor $\{w_i\}$0 and by effective exciting-field amplitudes $\{w_i\}$1 and $\{w_i\}$2. Increasing surface stickiness in a sticky-hard-sphere system increases both the scattering coefficient and the asymmetry factor, with far-field interference identified as the dominant mechanism and local-field modification as a subtler correction (Wang et al., 2018).

Applications extend to coloration and propagation. In disordered colloids in the Rayleigh regime, Monte Carlo light-transport simulations map the structural color palette as a function of particle diameter, refractive index contrast, density, thickness, and absorption. The parameter $\{w_i\}$3 emerges as the key scaling variable, and the study identifies a narrow range in which the same blue color appears in diffuse reflection and transmission. It also shows that adding black absorbents to a white multiply scattering material can produce blue diffuse reflection through the interplay of multiple scattering and absorption (Vynck et al., 17 Dec 2025). At terahertz frequencies, structure-induced surface scattering from indoor materials is measured at 113, 140, and 170 GHz. Pine wood exhibits angular side lobes attributable to quasi-periodic earlywood-latewood structure and reproduced by a beam-propagation model, while folded curtains generate wide-angle scattered components spanning $\{w_i\}$4 relative to the specular direction in conference-room bistatic measurements (Li et al., 27 Feb 2026). A plausible implication is that, across optical and THz regimes alike, structural scattering is increasingly treated not merely as a diagnostic artifact of disorder but as a controllable channel property or functional design variable.