Reflection Ultrafast Electron Diffraction

• Reflection Ultrafast Electron Diffraction is a pump–probe method that employs femtosecond lasers and grazing-incidence electron beams to capture surface dynamics at atomic layer depth.
• It uses a unique reflection geometry to isolate surface scattering from bulk effects, enabling quantitative analysis of nanoscale heat transport and structural transitions.
• Advances in electron gun technologies like fs RHEED and THz-compressed DC guns have enhanced its temporal resolution and sensitivity to subtle surface phenomena.

Searching arXiv for recent and foundational work on reflection ultrafast electron diffraction to ground the article in published sources. Reflection ultrafast electron diffraction is a pump–probe diffraction method in which a femtosecond laser pulse excites a surface and a delayed electron pulse probes the transient structure in a grazing-incidence reflection geometry, commonly implemented as time-resolved RHEED. Its defining feature is surface specificity: at 20–30 keV and incidence angles of $1^\circ$–$6^\circ$, the probe interacts predominantly with the topmost atomic layers, and for Au(111) and Pt(111) at the (444) Bragg condition the effective probe depths are $2.33$ Å and $2.13$ Å, respectively (Hoegen, 2023, He et al., 2024).

1. Geometry, surface sensitivity, and reciprocal-space description

In reflection geometry, the surface is probed by electrons incident at a small angle to the sample plane. At 20–30 keV, a normally incident beam would penetrate tens of nanometers and be dominated by forward scattering, whereas grazing incidence produces strong surface sensitivity and the characteristic RHEED pattern of streaks and Laue circles. The vertical momentum transfer is

$$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$

and for $E = 30$ keV and $\theta = 2.5^\circ$ the value is $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$, comparable to the momentum transfer of a $\sim 57$ eV LEED experiment at normal incidence (Hoegen, 2023).

A complementary formal description treats the surface as the plane $z=0$, with the crystal extending to $6^\circ$0, and assumes two-dimensional periodicity parallel to the surface. The stationary Schrödinger equation,

$6^\circ$1

is then solved with Bloch-periodic boundary conditions in the surface plane and reflection boundary conditions at the vacuum interface, yielding beam intensities $6^\circ$2 as functions of incident altitude and azimuth. In this formulation, the experimentally relevant observables are the rocking curves of reflected beams as functions of $6^\circ$3 and $6^\circ$4 (Kudo et al., 2023).

This geometry makes reflection UED fundamentally distinct from transmission UED. Transmission averages through the thickness of a thin membrane, whereas reflection directly interrogates near-surface order, reconstructions, adsorbates, and interface-proximal thermal transport. The difference is not only spatial but also reciprocal-space: transmission samples bulk reciprocal lattice points, while reflection primarily intersects surface rods or truncated reciprocal-lattice features (Hoegen, 2023, Kudo et al., 2023).

2. Source architectures and pulse-format requirements

Established reflection-UED implementations use tens-keV electron guns in RHEED geometry. A mature femtosecond implementation employs UV photoemission from a back-illuminated 10 nm Au photocathode, an extraction field of $6^\circ$5 kV/mm, electron energies of 26–30 keV, and pump pulse-front tilting to compensate the velocity mismatch inherent to grazing incidence. In that configuration, the measured temporal response reaches $6^\circ$6 fs FWHM for pulses containing $6^\circ$7 electrons (Hoegen, 2023).

Several source developments outside canonical reflection geometry are directly relevant to reflection UED because they address the same constraints of pulse duration, coherence, charge, and timing jitter. THz-driven compression has been demonstrated with a 53 keV DC gun and a dielectric-lined-waveguide buncher, compressing $6^\circ$8 electrons per pulse by a factor of $6^\circ$9 to $2.33$0 fs FWHM at 1 kHz, with $2.33$1 fs RMS relative timing jitter between the THz field and the electron beam (Zhang et al., 2021). A cold atom electron source has already demonstrated single-shot RHEED at 8 keV with $2.33$2 electrons per pulse and a $2.33$3 ns pulse duration, showing that low-temperature sources can support reflection diffraction directly (Speirs et al., 2015). Ultracold-source transmission work further shows that source temperatures near $2.33$4 K enable high spatial coherence even for micron-scale probe spots, which is a desirable regime for localized surface diffraction (Mourik et al., 2014). An all-optical MeV LWFA source has not been used in reflection in the cited experiment, but the beamline design, energy filtering, compression, and isochronicity were described as directly relevant to reflection UED, with only the sample and detector geometry needing reconfiguration (Fang et al., 2022).

Architecture	Reported parameters	Reflection-UED significance
fs RHEED gun	26–30 keV; $2.33$5 fs FWHM; $2.33$6 electrons	Direct surface-sensitive implementation (Hoegen, 2023)
THz-compressed DC gun	53 keV; $2.33$7 fs FWHM; $2.33$8 electrons; 1 kHz	Compact, low-jitter source directly relevant to RUED (Zhang et al., 2021)
Cold atom source	8 keV; $2.33$9 electrons; $2.13$0 ns; single-shot RHEED	Demonstrated reflection diffraction with high coherence (Speirs et al., 2015)
LWFA + isochronous beamline	4.27 MeV; $2.13$1 FWHM spread; $2.13$2 fC; $2.13$3 fs rms	Transmission experiment, but beam physics and timing scheme described as directly relevant to reflection UED (Fang et al., 2022)

These results indicate that reflection UED is not tied to a single gun technology. The operative requirements are a short bunch, low timing jitter, sufficient transverse coherence, and enough charge to record weak surface diffraction. Different source classes satisfy these requirements in different parts of parameter space (Hoegen, 2023, Zhang et al., 2021).

3. Diffraction observables, Debye–Waller analysis, and quantitative interpretation

For ultrafast surface heating experiments that do not cross a structural phase boundary, the primary observable is the pump-induced reduction of Bragg intensity through the Debye–Waller effect. In reflection UED on cubic metal surfaces, the analysis used for surface-normal motion is

$2.13$4

where $2.13$5 is the scalar momentum transfer along the surface-normal direction, $2.13$6 is the out-of-plane displacement, and $2.13$7 is the transient surface lattice temperature (He et al., 2024). In the more general isotropic form used for RHEED Debye–Waller analysis,

$2.13$8

with $2.13$9 related to temperature by a Debye-type expression, and for small relative intensity changes the exponential can be linearized (Hoegen, 2023).

A notable strength of reflection UED is that the diffuse background and the Bragg-intensity loss can be compared directly. In Au(111) and Pt(111), the diffuse scattering background increases with the same temporal dependence as the Bragg intensity decrease, supporting a Debye–Waller interpretation of the surface structural dynamics (He et al., 2024). In monolayer adsorbate systems, the vector Debye–Waller form,

$$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$0

permits separation of in-plane and out-of-plane vibrational contributions by exploiting the different $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$1 and $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$2 sensitivities of different diffraction features (Hoegen, 2023).

Quantitative reflection diffraction is complicated by excitation errors and dynamical scattering. A recent computation framework for RHEED and TRHEPD therefore reformulates the stationary reflection problem as a matrix ODE and accelerates it substantially relative to conventional multi-slice implementations. In benchmark tests, the reported method is up to $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$3 times faster than the conventional method, which is important because inverse structural analysis and time-resolved fitting require very large numbers of forward simulations (Kudo et al., 2023). A related development in ultrafast four-dimensional precession electron diffraction showed that angular averaging can suppress sensitivity to excitation errors and dynamical effects in transmission UED; this suggests that analogous angle-averaged strategies may be useful in reflection geometries when quantitative refinement is limited by local tilt, rocking-curve sharpness, or multiple scattering (Shiratori et al., 2024).

4. Nanoscale heat transport and vibrational dynamics at surfaces

Reflection UED has been used to quantify heat flow across interfaces, size-dependent cooling of nanostructures, damping of adsorbate vibrations, and surface-specific metal dynamics. In ultrathin epitaxial Bi(111) films on Si, the transient intensity of the Bi(00) RHEED spot was converted to temperature using a stationary $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$4 calibration, and for a 6 nm film the cooling was well fit by a mono-exponential with $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$5 ps, corresponding to an effective thermal boundary conductance $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$6 (Hoegen, 2023).

On Ge/Si(001), spot-profile analysis separated the responses of huts, domes, and relaxed islands within the same diffraction movie. The extracted cooling time constants were $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$7 ps, $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$8 ps, and $$|k_\perp| = \frac{4\pi \sin\theta}{\lambda_{\mathrm{el}}},$$9 ps, demonstrating that reflection UED can resolve size- and strain-dependent nanoscale heat transport without spatially isolating each nanostructure beforehand (Hoegen, 2023).

For monolayer Pb on Si(111)-$E = 30$0, vector Debye–Waller analysis separated a fast in-plane vibrational response from a slow out-of-plane one. The reported amplitudes were $E = 30$1 and $E = 30$2, with relaxation times $E = 30$3 ps and $E = 30$4 ps, respectively (Hoegen, 2023).

Reflection UED on Au(111) and Pt(111) extends this program to bulk single-crystal metal surfaces. The effective probe depth at the (444) Bragg condition is only about one interplanar spacing, so the method measures true surface lattice dynamics rather than thickness averages. In Pt(111), the observed surface Debye–Waller dynamics are reproduced by a depth-resolved two-temperature model using accepted material constants, including $E = 30$5. In Au(111), by contrast, the measured surface mean-square displacements are about $E = 30$6 larger than the bulk-like two-temperature-model prediction and the excess decays with a time constant of about $E = 30$7 ps, which was interpreted in terms of weaker binding and enhanced vibrational amplitudes of surface atoms (He et al., 2024).

A central result of the Au/Pt work is that hot-electron diffusion strongly reduces surface excitation. For Au at 10.5 mJ/cm$E = 30$8, a naive absorption-profile estimate gives $E = 30$9 K, but including rapid hot-electron diffusion reduces the effective surface temperature jump to about $\theta = 2.5^\circ$0 K; for Pt at 12.8 mJ/cm$\theta = 2.5^\circ$1, the corresponding reduction is from $\theta = 2.5^\circ$2 K to $\theta = 2.5^\circ$3 K (He et al., 2024). This overturns the common assumption that surface recovery can be inferred directly from the optical absorption profile.

5. Driven surface phase transitions and nonthermal pathways

Reflection UED has also been used to resolve strongly driven nonequilibrium phase transitions. A prominent example is the In-induced $\theta = 2.5^\circ$4 reconstruction on Si(111), a Peierls-distorted surface charge-density-wave system with a discontinuous transition at $\theta = 2.5^\circ$5 K from a $\theta = 2.5^\circ$6 insulating ground state to a $\theta = 2.5^\circ$7 metallic excited state. The energy barrier separating the two structures is about $\theta = 2.5^\circ$8 meV (Hoegen, 2023).

Under femtosecond optical excitation, the $\theta = 2.5^\circ$9 transition is nonthermal and extremely fast. The reported excitation time constants are $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$0–$|k_\perp| \approx 7.8\ \text{\AA}^{-1}$1 fs, and the structural transformation is essentially complete within about $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$2 fs. At higher pump fluence, $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$3 saturates near $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$4 fs, which the cited work interprets as the quantum limit of a critically damped, directed structural motion on the excited-state potential-energy surface (Hoegen, 2023).

The structural conversion precedes significant lattice heating. Debye–Waller analysis of the $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$5 state gives a heating time $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$6 ps, a cooling time $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$7 ps, and a maximum transient temperature $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$8 K for a base temperature of $|k_\perp| \approx 7.8\ \text{\AA}^{-1}$9 K, still below $\sim 57$0. The phase transition is therefore not driven by raising the lattice above the equilibrium transition temperature (Hoegen, 2023).

The return to the $\sim 57$1 state is much slower because the $\sim 57$2 state becomes metastable. Clean samples show recovery on nanosecond timescales, whereas adsorbates and steps act as nucleation centers that accelerate the reverse transformation. From the adsorbate-density dependence and terrace-width analysis, the reported phase-front velocities are $\sim 57$3 m/s and $\sim 57$4 m/s (Hoegen, 2023). Reflection UED thereby accesses not only sub-picosecond switching but also mesoscale recovery kinetics of surface domains.

6. Methodological issues, controversies, and current directions

A persistent concern in reflection UED is the role of laser-induced transient electric fields. Photoemission can generate near-surface charge separation and deflect the probe, raising the possibility that apparent diffraction changes are electrostatic rather than structural. The Au/Pt surface study addressed this directly by comparing Bragg dynamics with intensity changes near the shadow edge, where TEF effects are strongest. The TEF-induced signals showed position-dependent temporal behavior distinct from the Debye–Waller-like structural traces, and the authors concluded that surface structural dynamics can be reliably obtained by reflection UED even in the presence of transient electric fields (He et al., 2024).

Another methodological issue is that reflection diffraction intensities are highly sensitive to excitation errors and dynamical effects. Fast forward solvers for RHEED/TRHEPD therefore matter not only for static inverse problems but also for time-resolved studies that require repeated simulation over many delays, angles, and structural hypotheses (Kudo et al., 2023). For low-symmetry or imperfect crystals, indexing and geometry refinement remain difficult. GARFIELD was introduced for transmission UED of imperfect quasi-single crystals, but its Ewald-sphere and reciprocal-space formulation was described as conceptually well suited to porting to reflection UED once detector geometry and rod-like reciprocal-space features are treated appropriately. This suggests that future reflection-UED analysis pipelines will likely combine explicit reflection solvers with interactive reciprocal-space geometry optimization (Marx et al., 2024).

Beamline development is likewise converging across geometries. THz-driven compression, ultracold and cold-atom sources, and all-optical LWFA platforms were all described as relevant to reflection UED because they address the same core constraints of pulse duration, coherence, charge, and timing jitter (Zhang et al., 2021, Speirs et al., 2015, Fang et al., 2022). The field is therefore moving toward an overview in which surface-sensitive reflection geometries inherit increasingly sophisticated source and compression technologies originally validated in transmission UED, while retaining the distinctive capability to probe the structural dynamics of the topmost atomic layers.