Skip to content

Rayleigh surface phonons and other characteristic terahertz dynamics in diamond nanocrystals

FieldValue
JournalApplied Physics Reviews
Published2026-09-01
AuthorsCaleb Stamper, Matteo Baggioli, Pablo Galaviz, Roger A. Lewis, K. C. Rule
DOI10.1063/5.0325718
Citations3

Free-surface Rayleigh phonons explain the previously unexplained linear low-energy vibrational density of states in nanodiamond. Their normalized surface displacement exceeds that of the largest recorded earthquakes by a factor of 10^(5 ± 1). Atomistic simulations and inelastic neutron spectroscopy connect these motions to modes across 0 to 40 meV.

Rayleigh waves explain the linear density of states

Section titled “Rayleigh waves explain the linear density of states”

At 300 K, the 5.3 nm spherical nanocrystal produced g(E) proportional to E between 0 and 40 meV. The periodic bulk-diamond calculation produced g(E) proportional to E squared, consistent with Debye theory. Its calculated neutron dynamic structure factor contained a linearly dispersive mode below the bulk acoustic branches. The mode had low group velocity and strong transverse polarization, matching the expected signature of a Rayleigh surface acoustic wave. The authors therefore assign the excess low-energy states to free-surface Rayleigh phonons. Their model treats the surface as a two-dimensional propagation plane with radial exponential decay. That confinement gives a linear density of states without requiring liquid-like dynamics.

Classical molecular dynamics used LAMMPS and the AIREBO potential for carbon-carbon interactions. Isolated particles ran in a 140 nm vacuum box under an NVT ensemble at 300 K. The timestep was 1 fs. Each 10 ps run contained 2 ps of equilibration and 8 ps of production.

The central 5.3 nm particle was compared with a periodic bulk supercell and a graphitic or amorphous core-shell particle. The shell was produced by annealing at 850 K for 2 ns, then quenching to 300 K for 1 ns.

Velocity autocorrelation functions gave the density of states, while MDANSE calculated dynamic structure factors with 1 THz smoothing. The authors separated the inner 40%, outer 30%, and outer 10% of the particle diameter. Core atoms retained quadratic Debye scaling, whereas outer shells developed linear scaling.

Fixing surface atoms under Dirichlet conditions removed the excess low-energy density of states and restored quadratic scaling. Atom displacement decayed bi-exponentially from the surface, consistent with coupled Rayleigh and Love polarizations. The simulated Rayleigh speed ratio reached about 0.73. The classical continuum range is 0.87 to 0.96 for a Poisson ratio between 0 and 0.5. The authors suggest finite-size breakdown causes this extra softening.

Neutron spectroscopy confirms the prediction

Section titled “Neutron spectroscopy confirms the prediction”

Commercial detonation nanodiamond powder with an average diameter of 5 nm served as the nanodiamond sample. A synthetic monocrystalline diamond powder with 1 μm particles served as the reference. The authors heat-treated both powders at 300 °C for about 20 h because the nanodiamond was highly hygroscopic.

Time-of-flight measurements on Pelican used 3.7 meV incident neutrons, a 4.69 Å wavelength, and 583 K. The elastic-line energy resolution was 0.13 meV. Triple-axis scans on Taipan used Q = 4.5 Å⁻¹ and 300 K.

ResultValueSample and condition
Low-energy density of statesLinear scaling over 0 to 10 meV5 nm detonation nanodiamond at 583 K
Linear contributionDominant up to 40 meVSame nanodiamond powder at 583 K
Reference density of statesQuadratic energy dependence1 μm microdiamond at 583 K
Boson-like featurePeak near 31 meVNanodiamond at 583 K
High-order Lamb mode19 meVTriple-axis scan at Q = 4.5 Å⁻¹ and 300 K

The neutron density of states agreed with most MD features after energy scaling for the MD force constants. The linear term was absent from the microdiamond reference. The boson-like feature corresponded to the MD mode, while the assigned Lamb mode lay below the transverse acoustic branch. Triple-axis spectra also showed transverse acoustic broadening and softening in nanodiamond. The microdiamond linewidth was instrument-resolution-limited.

MD resolved seven characteristic energy regions in the nanodiamond density of states. An acoustic gap arose because wavelengths longer than the particle diameter were forbidden in an isolated particle. Spheroidal Lamb resonances included the particle breathing mode and higher angular momentum orders. The MD boson-like feature appeared near 50 meV, while neutron data placed the corresponding feature near 31 meV.

Bulk acoustic and optical modes broadened because confinement and disorder reduced phonon lifetimes. The authors also observed mode softening, with radial tensile stress explaining the acoustic shift. High-energy rattling modes came from low-coordination surface carbon with sp2 or sp3 bonding.

Surface disorder broadened the Rayleigh and Lamb resonances. It also increased the boson-like peak and the density of high-energy rattling modes. These overlapping features make the vibrational spectrum of disordered nanodiamond broader than that of an ideal particle.

Size and particle contacts change the spectrum

Section titled “Size and particle contacts change the spectrum”

Eight spherical particles covered diameters from 1.8 to 8.8 nm. Smaller particles strengthened the linear density-of-states contribution and softened the optical modes. The acoustic gap, Lamb resonance energy, and linewidth broadening each scaled approximately as 1/d. The authors attributed the optical softening to increasing radial tensile stress.

Clusters placed 2, 3, 5, 10, or 15 particles of 1.2 nm diameter within less than 10 Å and ran for about 20 ns. Interparticle coupling progressively closed the acoustic gap and suppressed Lamb resonances. It also produced diffusive and rotational modes below 2 meV in some cluster calculations. These results show that free-particle predictions do not transfer directly to powders, embedded particles, or contacting assemblies.

The linear low-energy density of states had been reported for several nanocrystals by neutron spectroscopy, Raman spectroscopy, and simulation. Earlier studies associated it with surface phonons but did not identify a Rayleigh mechanism. This work supplies that mechanism and identifies Rayleigh surface waves in a nanocrystal for the first time. The same quadratic-to-linear size transition had also been reported for isolated gold particles, although the present experiments concern diamond.

Prior Raman work on 2.6 nm diamond powder reported a 16.2 meV fundamental Lamb resonance. MD of isolated 2.8 nm diamond reported 22 meV. The present 19 meV signal was assigned to a higher-order Lamb mode, so it is not a direct fundamental-mode comparison.

Classical MD omits quantum statistics, so mode populations and low-temperature linewidths are not fully quantitative. AIREBO force constants were stiff enough to require energy scaling. The simulated Lamb resonance intensity changed with minimization and equilibration, raising questions about equilibrium population. The authors could not determine whether aggressive equilibration damped the resonance or whether the original setup produced non-equilibrium behavior.

Instantaneous normal mode analysis found only 0.7% unstable modes in the core-shell particle, too few to explain the linear density of states. In the powder experiment, core-shell morphology, interparticle coupling, and size inhomogeneity likely suppressed and broadened the fundamental Lamb mode. A QENS-like signal also resembled interparticle and diffusional motion. The authors judged these contributions minor above a few meV but did not exclude them.

The boson-like peak remained unexplained. The authors considered resonant coupling between the Rayleigh wave and a higher-order Lamb or transverse disorder mode. They also considered intermediate-phase surface carbon. The paper did not measure changes in thermal conductivity, heat capacity, catalytic rate, or quantum coherence. It also did not isolate one contribution to the 31 meV peak.

Implications for diamond device engineering

Section titled “Implications for diamond device engineering”

Free-boundary condition should be treated as a design variable in nanodiamond models. Rigid constraints remove the Rayleigh contribution, smaller free particles increase the surface contribution, and contacts suppress Lamb modes while closing the acoustic gap. Models for nanodiamond thermometry, sensing, or optomechanics should therefore specify particle size, boundary condition, surface disorder, and local environment.

The authors suggest that large surface displacements can change energy transfer between particles and surroundings. A compliant environment may add available modes and improve thermal transfer, while stronger scattering or localization would reduce it. They also expect surface vibrations to influence catalysis, heat capacity, vibrational entropy, and thermal expansion, but none of those effects was measured.

For quantum computing and levitated nanodiamond, the large surface motion may alter optomechanical coupling. This remains a proposed consequence, not an experimental result of the paper. When selecting diamond for a nanoscale thermal or quantum device, use size-dependent, surface-aware vibrational data rather than bulk Debye values. For free nanodiamond below 40 meV, a Debye density of states alone omits the measured Rayleigh contribution.

View Original Abstract

The free boundaries of nanocrystals introduce new surface wave solutions, analogous to the seismic waves on Earth, yet the implications of these surface waves on nanocrystals have remained largely unexplored. Here, we use atomistic molecular dynamics simulations and experimental neutron spectroscopy to elucidate mechanical terahertz-scale features in diamond nanocrystals. Our key insight is the observation and assignment of thermally induced Rayleigh surface phonons, which have low group velocity and amplitude that decays exponentially away from the surface. These states are shown to result in a deviation from the standard three-dimensional form of Debye’s law, giving a characteristic linear scaling of the low-energy vibrational density of states in nanocrystals. Normalized to the radius, the surface displacement ratio in diamond nanocrystals exceeds that of the largest recorded earthquakes by a factor of 105±1. We explain how these Rayleigh waves coexist with other distinctive features, including confined lattice phonons, the acoustic gap, and Lamb modes, thereby offering a comprehensive framework for the vibrational dynamics of nanocrystals which can be leveraged for quantum sensing, thermal management, and catalysis. The mechanical vibrations of atoms in nano-sized crystals vary greatly from bulk crystals due to their large surface-area-to-volume ratios and confinement effects. This work characterizes the terahertz vibrations of nanodiamonds, revealing the existence of Rayleigh surface phonons, analogous to surface earthquakes. The unique vibrations observed in nanodiamond are likely universal among nanocrystals and critical for understanding their exceptional physical properties.