Plasmon-Assisted Photoelectron Emission

• Plasmon-assisted photoelectron emission is the process where collective oscillations in metals enhance local fields, thereby modifying the probability, timing, and energy distribution of emitted electrons.
• It utilizes mechanisms like multiphoton absorption and field-assisted tunneling driven by localized surface plasmons, surface plasmon polaritons, and nanoparticle resonances.
• The phenomenon underpins advanced ultrafast spectroscopies and nanoelectronics, offering tunable device architectures through geometry-dependent field enhancements.

Searching arXiv for recent and foundational papers on plasmon-assisted photoelectron emission to ground the article in the current literature. Plasmon-assisted photoelectron emission denotes photoemission processes in which collective charge oscillations in metals—localized surface plasmons, surface plasmon polaritons, bulk plasmons, and surface plasmons generated either by optical pumping or by the emission event itself—modify the probability, timing, angular distribution, or energy spectrum of emitted electrons. In nanostructures, this assistance commonly appears as nanoscale field enhancement and confinement that drive multiphoton photoemission, photo-assisted tunneling, or optical-field emission; in photoelectron spectroscopy, it appears as plasmon satellites, loss and gain sidebands, and Floquet-like channels associated with coherent plasmon populations (Müller et al., 2015, Gonçalves et al., 2023, Caruso et al., 2018).

1. Electromagnetic basis: field confinement, nanofocusing, and plasmon localization

A central branch of the subject concerns the conversion of incident optical radiation into strongly enhanced and spatially confined near fields. In a gold nanotip, broadband few-cycle light can be coupled non-locally into a surface plasmon polariton (SPP) mode by a chirped grating milled on the shaft; the SPP wave packet then propagates toward the apex, where the conical taper compresses the mode in the “adiabatic nanofocusing” picture. In that description, the local SPP wavenumber $k_{\mathrm{SPP}}(z)$ increases as the tip radius decreases, so the apex field scales approximately as $E_a \simeq k \times E_{\mathrm{incident}}$, with field-enhancement factors $k$ of 5–20 for gold tapers. Over a 20 $\mu$m propagation path the group-velocity dispersion adds only $\sim 25\,\mathrm{fs}^2$ of chirp, preserving a sub-8 fs SPP pulse at the apex (Müller et al., 2015).

In sharply tapered silver tips, the collective response is localized even more strongly at the apex. Finite-element calculations for a 20 nm-diameter Ag cone give $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$–100, while a 1 nm asperity can raise $\beta \rightarrow 200$. The corresponding near field decays into vacuum over $\sim 4$ nm according to $\beta(z)\approx \beta(0)/(z+z_0)^3$ with $z_0\approx 15\,\mathrm{nm}$, creating the steep spatial gradients required for strong ponderomotive forces and non-local light–matter interaction (Lee et al., 8 Dec 2025).

A related near-field geometry appears around plasmonic nanoparticles. For isolated metal nanospheres illuminated by few-cycle pulses, the enhanced near field can be approximated by

$E_a \simeq k \times E_{\mathrm{incident}}$0

with decay length $E_a \simeq k \times E_{\mathrm{incident}}$1–40 a.u. The nonhomogeneous character of this field is not a minor correction: because $E_a \simeq k \times E_{\mathrm{incident}}$2 is comparable to the electron excursion length, the spatial gradient substantially alters above-threshold ionization dynamics (Ciappina et al., 2013).

For spherical nanoparticles in the quasistatic limit, the internal field is spatially uniform,

$E_a \simeq k \times E_{\mathrm{incident}}$3

and the localized-plasmon resonance occurs near $E_a \simeq k \times E_{\mathrm{incident}}$4. This quasistatic formulation is foundational for analyses that distinguish field enhancement from transport and emission physics at the interface (Uskov et al., 2013).

2. Microscopic emission channels

In nanotip emitters, the dominant emission mechanism often remains multiphoton photoemission even when the local field is plasmonically enhanced. In the gold-tip nanofocusing experiment, interferometric autocorrelation of the photocurrent gave an effective emission nonlinearity $E_a \simeq k \times E_{\mathrm{incident}}$5, with emitted current density obeying

$E_a \simeq k \times E_{\mathrm{incident}}$6

Depending on DC bias, the measured nonlinearity lay in the range $E_a \simeq k \times E_{\mathrm{incident}}$7–2.6; the Schottky effect reduced the work function, but the Keldysh parameter remained deep in the multiphoton regime, with $E_a \simeq k \times E_{\mathrm{incident}}$8 for $E_a \simeq k \times E_{\mathrm{incident}}$9 and $k$0 (Müller et al., 2015).

At higher local fields, the problem crosses into photo-assisted field emission and optical-field emission. Lee et al. model the instantaneous rate by a time-dependent Fowler–Nordheim form,

$k$1

and identify a crossover regime in which simultaneous multiphoton absorption and optical field emission occur. In their Ag-tip measurements, fitting the first- and second-order harmonics over seven decades of current yielded $k$2; with this enhancement, $k$3 moves from $k$4 without enhancement to $k$5, marking the onset of strong-field dynamics. At $k$6 and $k$7, the ponderomotive energy $k$8 exactly matched the measured kinetic-energy shift (Lee et al., 8 Dec 2025).

Fibre-optic nanotips provide a lower-power realization of plasmon-assisted emission. In gold-coated fibre tips with an apex radius of about 50 nm, the observed resonance at $k$9 coincided with a transition from $\mu$0 scaling with $\mu$1 at 500 nm to $\mu$2 at 660 nm, while typical count rates at $\mu$3 and 660 nm reached $\mu$4–$\mu$5 counts per second. The field-enhancement factor inferred from the power-law analysis was $\mu$6, and the physical interpretation advanced in that work was surface plasmon enhanced above-threshold photoemission rather than pure tunneling or thermionic emission (Keramati et al., 2020).

Plasmon-enhanced emission can also be combined with a static bias. In a resonant planar Au structure with 100 nm vacuum gaps, the barrier was modeled as

$\mu$7

Full-wave simulation gave $\mu$8, while fitting a time-dependent tunneling model to the measured currents required an effective enhancement $\mu$9 and $\sim 25\,\mathrm{fs}^2$0. With laser intensities of tens to hundreds of $\sim 25\,\mathrm{fs}^2$1 and DC voltages below 10 V, nA–$\sim 25\,\mathrm{fs}^2$2A currents were observed, and the measured log–log slope of current versus power lay between 1 and 2 (Piltan et al., 2017).

A distinct but related classification concerns surface and volume photoelectric mechanisms in nanoparticles. For a spherical plasmonic nanoparticle, the surface rate takes the form

$\sim 25\,\mathrm{fs}^2$3

whereas the volume channel is

$\sim 25\,\mathrm{fs}^2$4

The comparison concluded that surface photoeffect, at least, does not concede the volume one; a discontinuity of dielectric permittivity at the boundary can increase the surface-emission rate by 5 times, and realistic hot-carrier cooling tends to favor the surface channel, particularly for radii of about 10–50 nm (Uskov et al., 2013).

3. Geometries, device architectures, and morphology dependence

Plasmon-assisted emission is highly geometry dependent because the same plasmonic enhancement that raises the local field also controls the spatial mode profile, the boundary conditions at emission, and the accessible transport channel.

Platform	Plasmonic element	Reported characteristic
Gold nanotip	Chirped grating + nanofocused SPP	$\sim 25\,\mathrm{fs}^2$5 vs $\sim 25\,\mathrm{fs}^2$6 at $\sim 25\,\mathrm{fs}^2$7
Silver tip	Apex-localized plasmon	$\sim 25\,\mathrm{fs}^2$8–100, up to 200 with 1 nm asperity
Fibre-optic nanotip	Au-coated tapered fibre	Resonance at $\sim 25\,\mathrm{fs}^2$9
Nanoparticle array in semiconductor	LSPR + Rayleigh anomaly	5 to 20-times increase of photoemission on embedding
Planar resonant gap surface	Gap plasmon resonance	Emission at tens of $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$0 and $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$1 V DC

In the gold nanotip platform, geometry was tuned for non-local excitation: a 12-groove linearly chirped grating, centered at 1.5 $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$2m and positioned 20 $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$3m above the apex, phase-matched a bandwidth of about 300 THz and launched $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$4 fs SPP wave packets to a tip radius on the order of 10 nm (Müller et al., 2015). In the fibre-optic implementation, geometry served a different purpose: back-illumination through a metallized taper avoided a free-space focus at the apex, making positioning easier while still producing a resonant apex hot spot (Keramati et al., 2020).

Periodic nanoparticle arrays introduce an additional collective scale through lattice resonances. In Au nanodisk arrays partially embedded in GaAs, a rectangular lattice can be designed so that the lattice-induced Rayleigh anomaly overlaps the localized surface plasmon resonance. For a rectangular lattice with $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$5 under normal incidence, the Rayleigh-anomaly wavelength for the $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$6 order is $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$7. When $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$8, a narrow collective lattice-plasmon resonance with a Fano lineshape emerges; when the refractive indices above and below the array differ, the condition is disrupted and the narrow absorption peak is diminished (Zhukovsky et al., 2013).

Embedding depth is equally consequential. For the Au/GaAs arrays, an embedding fraction $\beta \equiv E_{\mathrm{local}}/E_0 \approx 80$9 maximized narrowing and field overlap in a homogeneous GaAs environment, while in the presence of an index step the enhancement saturated once $\beta \rightarrow 200$0. Across the spectrum, the ratio $\beta \rightarrow 200$1 could reach 5–20 at the lattice-resonance wavelength. This geometry dependence is not merely optical: because emission occurs through a Schottky barrier of height $\beta \rightarrow 200$2 into GaAs, the field normal to the metal–semiconductor interface directly controls the surface-photoeffect metric

$\beta \rightarrow 200$3

The resulting device concept is explicitly a sub-bandgap photodetector or photovoltaic structure driven by hot-electron injection rather than interband absorption in the semiconductor (Zhukovsky et al., 2013).

Morphology also controls plasmon-satellite photoemission in spectroscopic settings. A nonperturbative quantum-mechanical treatment of nanostructure core-level photoemission predicts that the probabilities of plasmon satellites depend dramatically on morphology and dimensionality but collapse onto universal curves when written as $\beta \rightarrow 200$4 with $\beta \rightarrow 200$5. Spheres, rods, shells, and 2D disks follow different shape-dependent $\beta \rightarrow 200$6 functions, while once $\beta \rightarrow 200$7 is known, the absolute coupling can be transferred across size and material within that morphology class (Gonçalves et al., 2023).

4. Spectroscopic manifestations: plasmon satellites, losses, gains, and shake-up structure

A second major branch of plasmon-assisted photoelectron emission concerns spectral sidebands generated by electron–plasmon coupling during photoemission. In the many-body picture, the sudden creation of a core hole and the outgoing photoelectron together act as an external charge density

$\beta \rightarrow 200$8

which drives a plasmon mode of frequency $\beta \rightarrow 200$9 through the coupling

$\sim 4$0

Because the driven mode is a harmonic oscillator, the final plasmon state is a displaced coherent state, and the probability to excite exactly $\sim 4$1 plasmons is Poissonian:

$\sim 4$2

The measured spectrum is then

$\sim 4$3

broadened in practice by the plasmon damping rate $\sim 4$4 (Gonçalves et al., 2023).

This spectral picture generalizes long-standing results for solids. In angle-resolved photoemission, plasmon satellites appear as shake-up peaks displaced from the quasiparticle line by $\sim 4$5. The GW self-energy contains the plasmon pole, but direct solution of Dyson’s equation often overestimates satellite binding energies and yields spurious plasmaron peaks. The cumulant expansion remedies this by writing

$\sim 4$6

with $\sim 4$7 constructed from $\sim 4$8. The resulting spectral function contains a quasiparticle line plus a realistic ladder of satellites, reproducing the Langreth series in the localized-electron limit and correctly generating multiple satellites in systems such as Na and Si (Caruso et al., 2018).

Hard x-ray photoelectron spectroscopy of nearly free-electron metals has resolved these contributions quantitatively. In Al and Mg, Balal et al. identified multiple bulk plasmons in core levels and valence bands and extracted nearly identical extrinsic and intrinsic probabilities for Al 1s, Al 2s, and Mg 2s:

• $\sim 4$9 for the extrinsic channel,
• $\beta(z)\approx \beta(0)/(z+z_0)^3$0 for the intrinsic channel,
• interference at $\beta(z)\approx \beta(0)/(z+z_0)^3$1 of about $\beta(z)\approx \beta(0)/(z+z_0)^3$2 to $\beta(z)\approx \beta(0)/(z+z_0)^3$3, rapidly tending to zero for $\beta(z)\approx \beta(0)/(z+z_0)^3$4.

The inferred intrinsic fraction diminished from 22% for $\beta(z)\approx \beta(0)/(z+z_0)^3$5 to 4.4% for $\beta(z)\approx \beta(0)/(z+z_0)^3$6 and to 0.5% for $\beta(z)\approx \beta(0)/(z+z_0)^3$7, while the bulk-plasmon spacing was about 15.3 eV in Al and about 10.6 eV in Mg. Surface-plasmon intensity increased strongly in grazing emission and was completely attenuated by oxide on Al, whereas the metal bulk plasmon remained nearly unchanged (Balal et al., 2024).

The same formalism predicts actively pumped gain and loss structures. If the plasmon mode is prepared in an initial coherent state with mean occupation $\beta(z)\approx \beta(0)/(z+z_0)^3$8, subsequent photoemission can produce both loss and gain sidebands at

$\beta(z)\approx \beta(0)/(z+z_0)^3$9

As the pump intensity increases, loss and gain peaks become comparable, and for $z_0\approx 15\,\mathrm{nm}$0 the zero-loss line becomes strongly depleted. This places pump–probe plasmon satellite spectroscopy in direct contact with ultrafast plasmon coherence and dephasing (Gonçalves et al., 2023).

5. Theoretical formalisms: from Floquet and cumulant methods to TDDFT and scattering matrices

Theoretical descriptions of plasmon-assisted photoemission span perturbative many-body approaches, explicitly time-dependent single-particle frameworks, and nonperturbative coherent-state formalisms.

A time-dependent density-functional approach combined with the time-dependent surface-flux method has been used to examine a one-dimensional model cluster. The time-dependent Kohn–Sham orbitals satisfy

$z_0\approx 15\,\mathrm{nm}$1

and the photoelectron spectrum is obtained from flux amplitudes $z_0\approx 15\,\mathrm{nm}$2 and $z_0\approx 15\,\mathrm{nm}$3 through

$z_0\approx 15\,\mathrm{nm}$4

In that model, two collective modes appeared at $z_0\approx 15\,\mathrm{nm}$5 and $z_0\approx 15\,\mathrm{nm}$6 a.u. Besides the usual ATI comb, the spectrum contained narrow plasmon spikes arising from long-lived post-pulse oscillations; an ATI peak at $z_0\approx 15\,\mathrm{nm}$7 a.u. scaled as $z_0\approx 15\,\mathrm{nm}$8, while a plasmon peak at $z_0\approx 15\,\mathrm{nm}$9 a.u. scaled as $E_a \simeq k \times E_{\mathrm{incident}}$00. The authors emphasized that the ALDA functional overestimates plasmon lifetimes because important damping channels are absent (Bednov et al., 30 Jul 2025).

A complementary route starts from optically pumped plasmons in metals and treats them as a supplementary, frequency-locked source for photoemission. In a quadratic-response formulation, the energy-resolved yield is decomposed as

$E_a \simeq k \times E_{\mathrm{incident}}$01

where $E_a \simeq k \times E_{\mathrm{incident}}$02 contains a Poisson series in the mean plasmon number $E_a \simeq k \times E_{\mathrm{incident}}$03 and yields peaks pinned to $E_a \simeq k \times E_{\mathrm{incident}}$04. In the nondispersive limit, the multiplasmon sector forms a truncated plasmonic coherent state,

$E_a \simeq k \times E_{\mathrm{incident}}$05

This framework was developed to interpret nonlinear Ag photoemission spectra that exhibit resonances pinned to multiples of the bulk plasmon energy rather than to the applied photon energy (Novko et al., 2021).

Floquet formulations make the time-periodic aspect explicit. In the prepumped coherent-state environment, a gauge-transformed electron–surface-plasmon Hamiltonian can be written as

$E_a \simeq k \times E_{\mathrm{incident}}$06

so that a surface-state wavefunction acquires a Floquet expansion

$E_a \simeq k \times E_{\mathrm{incident}}$07

with quasienergies shifted by the ponderomotive term $E_a \simeq k \times E_{\mathrm{incident}}$08. For Ag(111), the parameters quoted were $E_a \simeq k \times E_{\mathrm{incident}}$09, $E_a \simeq k \times E_{\mathrm{incident}}$10, and $E_a \simeq k \times E_{\mathrm{incident}}$11; the theory proposed relative sideband intensities as an in situ calibration of the plasmonic field (Gumhalter et al., 2022).

A more recent nonperturbative theory compared gauge-specific cumulant expansions with a Volkov ansatz over plasmonic coherent states. In the velocity gauge, the first-order cumulant contains a nonzero ponderomotive contribution,

$E_a \simeq k \times E_{\mathrm{incident}}$12

where $E_a \simeq k \times E_{\mathrm{incident}}$13 is quadratic in the coherent-state amplitude. Both the cumulant and Volkov routes lead to a generalized-Bessel-function structure for the $E_a \simeq k \times E_{\mathrm{incident}}$14-plasmon emission rate and interpolate smoothly between the Born limit and the semiclassical strong-field limit. In the Ag(111) application, even sub-single-mode occupations of the plasmonic coherent state were found sufficient to support multiplasmon electron emission (Gumhalter, 10 Apr 2025).

For spatially inhomogeneous time-periodic fields at nano-tips, stable numerical treatment has been achieved with a scattering-matrix reformulation of the Floquet matching problem. Starting from a one-dimensional Schrödinger equation with a space- and time-dependent vector potential $E_a \simeq k \times E_{\mathrm{incident}}$15, the method replaces unstable transfer-matrix products by an $E_a \simeq k \times E_{\mathrm{incident}}$16-matrix relation between incoming and outgoing amplitudes and preserves unitarity to about $E_a \simeq k \times E_{\mathrm{incident}}$17. Applied to plasmonic near fields at metal/vacuum interfaces, the method yields multiphoton peaks separated by $E_a \simeq k \times E_{\mathrm{incident}}$18, threshold shifts from the local ponderomotive energy, and strong enhancement of sub-threshold emission in the tunneling regime (Kaminski, 2016).

6. Temporal response, experimental performance, and applications

The temporal compression of plasmon-assisted emitters is among their most distinctive properties. In the gold nanotip source, interferometric autocorrelation of apex photocurrent generated through the grating yielded a sech$E_a \simeq k \times E_{\mathrm{incident}}$19 pulse of FWHM $E_a \simeq k \times E_{\mathrm{incident}}$20; direct apex illumination gave $E_a \simeq k \times E_{\mathrm{incident}}$21. Classical Runge–Kutta trajectory simulations predicted that for $E_a \simeq k \times E_{\mathrm{incident}}$22 and tip–sample distances up to $E_a \simeq k \times E_{\mathrm{incident}}$23, the electron arrival-time width remains below 10 fs, enabling femtosecond point-projection microscopy and low-energy holography (Müller et al., 2015).

That tip source was explicitly demonstrated in point-projection imaging of an individual InP p–n nanowire with diameter 30 nm at a tip–sample distance of $E_a \simeq k \times E_{\mathrm{incident}}$24, corresponding to a geometrical magnification of about 31,000. The same work argued that reducing $E_a \simeq k \times E_{\mathrm{incident}}$25 to $E_a \simeq k \times E_{\mathrm{incident}}$26 would convert the projection into a Gabor hologram, with an effective source size $E_a \simeq k \times E_{\mathrm{incident}}$27 and energy spread $E_a \simeq k \times E_{\mathrm{incident}}$28 supporting about 1 nm spatial resolution and few-cycle temporal resolution. A further stated prospect was femtosecond STM based on non-local SPP excitation, which would avoid direct laser heating of the tunneling junction (Müller et al., 2015).

Time-resolved measurements on silver tips reinforce the ultrafast character of the process. Cross-polarized double-beat interferometry resolved cross-correlation currents at the first through fourth harmonics of the beat frequency; the Fourier transforms retained the envelope width up to $E_a \simeq k \times E_{\mathrm{incident}}$29, which was interpreted as evidence that tunneling is fast, $E_a \simeq k \times E_{\mathrm{incident}}$30, and that the optical phase is coherently mapped into the emitted electron wavepacket (Lee et al., 8 Dec 2025).

Fibre-optic emitters sacrifice some temporal precision for operational simplicity. Their measured coincidence peak had a width of about 1 ns, which the authors identified as an instrumental upper limit rather than the intrinsic electron-pulse duration; no delayed thermionic tail was observed. Because the source is back-illuminated and requires only a few nJ per pulse, it was proposed for nanometrology, multisource electron lithography, and scanning probe microscopy (Keramati et al., 2020).

Plasmon-assisted photoemission also underpins photodetection and vacuum electronics. In nanoparticle arrays embedded in semiconductors, the relevant application is infrared photodetection below the semiconductor band gap, with photons between $E_a \simeq k \times E_{\mathrm{incident}}$31 and $E_a \simeq k \times E_{\mathrm{incident}}$32 generating hot electrons that cross the metal/semiconductor barrier (Zhukovsky et al., 2013). In the planar resonant Au surface, the application is a semiconductor-compatible vacuum device in which optical and electrical excitation together reduce the required laser power and DC bias (Piltan et al., 2017).

Outside solids, plasmonic near fields can also drive high-energy photoelectron emission from gases. In a reduced-dimensional TDSE treatment for xenon placed near a metal nanosphere, the nonhomogeneous enhanced field extended the rescattering cutoff far beyond the homogeneous-field value of $E_a \simeq k \times E_{\mathrm{incident}}$33: for $E_a \simeq k \times E_{\mathrm{incident}}$34 a.u. and $E_a \simeq k \times E_{\mathrm{incident}}$35 the cutoff shifted from about 10.5 eV to about 30 eV, and for $E_a \simeq k \times E_{\mathrm{incident}}$36 to several hundred eV (Ciappina et al., 2013).

7. Conceptual distinctions, recurring debates, and present limitations

The term plasmon-assisted photoelectron emission covers several physically distinct processes. One class is plasmon-enhanced emission, in which a near field generated by a localized plasmon or nanofocused SPP modifies the barrier or the local multiphoton probability. Another is plasmon-loss or plasmon-gain spectroscopy, in which the emission event itself creates or annihilates plasmons and leaves sidebands in the spectrum. A third is prepumped plasmon-mediated emission, in which a coherent plasmon population acts as a secondary drive field and produces Floquet-like or non-Einsteinian channels (Müller et al., 2015, Balal et al., 2024, Gumhalter, 10 Apr 2025). Conflation of these regimes is a common source of ambiguity.

A long-standing debate concerns the relative importance of intrinsic and extrinsic plasmon excitation in photoemission spectra. The HAXPES analysis of Al and Mg showed that neither channel can be neglected at keV kinetic energies, and that interference is appreciable for the first plasmon but becomes negligible for higher order. This directly contradicts simplified pictures in which high-energy photoemission is treated as purely extrinsic (Balal et al., 2024).

Another recurring issue is whether photoemission from plasmonic nanoparticles is mainly a volume hot-electron transport problem or a boundary-collision problem. The analytical comparison between surface and volume photoelectric effects concluded that the surface mechanism at least does not concede the volume one, and that dielectric discontinuity at the boundary can itself enhance the surface rate by a factor of 5. A plausible implication is that device optimization cannot be reduced to maximizing total absorption alone; the boundary field and carrier cooling length are equally decisive (Uskov et al., 2013).

Collective lattice effects are also more fragile than idealized designs might suggest. In nanoparticle arrays, the Rayleigh-anomaly-induced narrow resonance is strongly degraded by refractive-index mismatch between the surrounding media, even if each nanoparticle still supports a strong localized plasmon. This suggests that fabrication tolerances and encapsulation strategy are not secondary engineering details but part of the plasmonic emission physics itself (Zhukovsky et al., 2013).

Finally, several theoretical results presently depend on controlled approximations. The one-dimensional TDDFT+t-SURFF study explicitly attributes its long-lived plasmon peaks to the absence of realistic damping channels in ALDA, while nonperturbative coherent-state and Floquet approaches are most naturally formulated for idealized surface bands and prescribed plasmon occupations (Bednov et al., 30 Jul 2025, Gumhalter et al., 2022). The field therefore spans experimentally established ultrafast emitters, quantitatively mature many-body descriptions of satellites, and more exploratory coherent-plasmon theories whose experimental calibration remains an active problem.