Electron Energy-Loss Function (ELF)

• ELF is defined as the imaginary part of –1/ε(q, ω), representing the probability for a fast electron to lose energy via inelastic scattering.
• Advanced computational methods such as TDDFT and model dielectric functions precisely calculate ELF to capture plasmonic, interband, and finite-size effects.
• Analysis of ELF through EELS techniques enables accurate extraction of material properties, aiding in stopping power calculations and nano-optical applications.

The electron energy-loss function (ELF), typically denoted as Im[–1/ε(q, ω)], is a fundamental response function of condensed-matter systems that encodes the probability for a fast charged particle (e.g., electron) to lose energy ℏω and transfer momentum q through inelastic interactions with a target material. The ELF directly connects dielectric theory to experimentally accessible observables in electron energy-loss spectroscopy (EELS), reflects the spectrum of collective and single-particle electronic excitations, and serves as a central theoretical quantity in numerous applications from plasmonics to stopping power calculations.

1. Formal Definition and Physical Interpretation

The ELF is defined in terms of the frequency- and momentum-dependent dielectric function ε(q, ω):

$$\mathrm{ELF}(q, \omega) \equiv -\operatorname{Im}\left[\frac{1}{\epsilon(q, \omega)}\right]$$

where ε(q, ω) = ε₁(q, ω) + i ε₂(q, ω) is the complex dielectric function. In the optical (q→0) limit, this simplifies to ELF(ω) ≡ Im–1/ε(ω). Experimentally, the ELF determines the intensity in EELS measurements, where the double-differential cross section for inelastic scattering is:

$$\frac{d^2\sigma}{d\Omega\,d\omega} \propto \mathrm{ELF}(q, \omega)$$

A peak in the ELF indicates a resonant energy loss, typically associated with a collective excitation (plasmon), strong interband transition, or, in some regimes, even phonon or intra-band processes (Roth et al., 2014, Pomarico et al., 2019). Physically, ELF(q, ω) quantifies the imaginary part of the longitudinal dielectric response: the degree to which an incident field (external charge fluctuation) is absorbed via electronic polarization modes.

2. Theoretical Formalism and Computational Schemes

The ELF's computation proceeds from first-principles or model dielectric approaches, with varying levels of sophistication:

• Ab initio linear response: Time-dependent density functional theory (TDDFT) or density functional perturbation theory (DFPT) yields ε(q, ω) via Kohn-Sham response functions and self-consistent screening (Zubizarreta et al., 2013, Nazarov et al., 2015, Tao et al., 2015, Simsek et al., 2012, Timrov et al., 2021, Timrov et al., 2013).
• Model dielectric functions: Free-electron Lindhard, Mermin (adds phenomenological damping), or Drude–Lindhard–Mermin oscillator models parameterize ε(q, ω) for metals and semiconductors (Knapen et al., 2021, Xu et al., 2016, Cheng et al., 2024).
• Experimental parametrization: Reflectivity, optical absorption, and EELS spectra are inverted via Kramers–Kronig analysis or fit to oscillator expansions (e.g., MELF-GOS, Drude–Lindhard), in some cases with reverse Monte Carlo (RMC) or simulated annealing to refine ELF for optimum agreement with measured spectra (Xu et al., 2016, Cheng et al., 2024).
• Surface and slab response: The surface loss function:

$$L_s(\omega) = -\operatorname{Im}\left[\frac{1}{\epsilon(\omega)+1}\right]$$

describes surface excitations, notably surface plasmons, and is relevant in REELS and thin-film EELS, as rigorously developed for Pb(111) thin films (Zubizarreta et al., 2013).

• Nonlocal response and local-field effects: Full microscopic dielectric matrices ε_{GG'}(q,ω), including off-diagonal terms and reciprocal-lattice vectors, are necessary for periodic solids and anisotropic or nanostructured materials (Timrov et al., 2021, Timrov et al., 2013).

Numerically, advanced schemes such as Liouville–Lanczos recursion within TDDFPT enable efficient and scalable computations of ELF(q, ω) over broad frequency and momentum ranges without explicit construction of virtual orbitals or large dielectric matrices (Timrov et al., 2013, Timrov et al., 2021).

3. Experimental Determination and Data Analysis

EELS and Related Techniques:

• Transmission EELS: Raw spectra (counts vs. energy loss) are related to ELF(q, ω) after background subtraction, ZLP (zero-loss peak) removal, multiple-scattering deconvolution, and geometric corrections (Roth et al., 2014, Pomarico et al., 2019, Brokkelkamp et al., 2022).
• Reflection EELS (REELS): Provides increased surface sensitivity; however, interpretation requires full modeling of surface excitation and multiple scattering. RMC approaches enable absolute extraction of the ELF by direct comparison of simulated and measured spectra (Xu et al., 2016).
• Spectral image processing: For spatially resolved EELS, machine-learning classifiers (e.g., K-means and deep learning models for ZLP) enable semi-automated extraction of ELF with spatial resolution down to nanometers (Brokkelkamp et al., 2022).
• Kramers–Kronig (KK) transformations: From single-scattering loss spectra S₁(E) (proportional to Im[–1/ε(ω)]), the real part ε₁(ω) is reconstructed via the KK relation, enabling full complex ε(ω) and thus ELF (Pomarico et al., 2019, Brokkelkamp et al., 2022).

Parametrization:

• Drude–Lindhard–Mermin oscillator fits: The ELF is modeled as a weighted sum of damped oscillators, with parameters (frequency, damping, strength) optimized for spectral reproduction and consistency with sum rules (f-sum, ps-sum) (Xu et al., 2016, Cheng et al., 2024).
• Semi-empirical fitting: For molecular targets (e.g., H₂ isotopologs in KATRIN), ELF is parameterized as a sum of Gaussians (for excitations) plus model ionization tails, with parameters fitted directly to EELS and differential data (Aker et al., 2021).

4. Quantum Size, Surface, and Dimension Effects

ELF exhibits pronounced finite size, surface, and dimensionality dependencies:

• Thin films: Quantum-size effects (QSE) in ELF of Pb(111) were resolved via ab initio RPA calculations. Monolayer and bilayer slabs lack the classical plasmon modes seen in bulk; instead, quantized interband transitions and non-dispersing collective modes dominate (Zubizarreta et al., 2013).
• 2D Materials (Graphene, TMDs): ELF in quasi-2D systems is sensitive to both electronic response and elastic scattering resonances. Unified theories demonstrate that conventional ELF formalism (neglecting detailed probe kinematics and lattice diffraction) fails for Q2D crystals; elastic scattering modulates or even dominates the observed EEL features (Nazarov et al., 2017). Practical computational treatments employ Coulomb kernel truncation and vacuum padding to eliminate spurious periodic-image interactions and recover experimental ELF in graphene (Mowbray, 2014).
• Surface plasmons: Thin film and surface-sensitive experiments (REELS) necessitate inclusion of the surface ELF L_s(ω). Quantum confinement can induce non-dispersive (bulk-like) modes and modify the classic symmetric/asymmetric surface plasmon branches (Zubizarreta et al., 2013).

5. Applications and Impact Across Disciplines

Electronic Excitation Spectra and Stopping Quantities:

• ELF is central in calculating stopping power and inelastic mean free paths; direct integration of the ELF and its extensions (to finite q via MELF-GOS, or to surface modes) yields physical quantities relevant for charged particle transport, radiation damage, and detector response (Cheng et al., 2024, Xu et al., 2016).

Material Characterization and Band Structure:

• ELF encodes both plasmonic and interband transitions, enabling extraction of plasmon energies, band gaps, and identification of collective excitations. In ABO₃ ferroelectrics, ELF analysis tracks subtle symmetry changes across phase transitions, with observable effects on plasmon peaks (Simsek et al., 2012).

Plasmonics, Nanooptics, and Ultrafast Spectroscopy:

• Mapping of ELF at the nanoscale (e.g., via ultrafast EELS) reveals spatial and spectral inhomogeneities in plasmon mode structure, lifetime, and coupling to phonons or electronic degrees of freedom (Pomarico et al., 2019). ELF variations under photoexcitation or in nonequilibrium states probe electronic and lattice dynamics with combined energy, spatial, and temporal resolution.

Dark Matter Detection and Fundamental Interactions:

• In dark-matter direct detection, ELF serves as the key screening input in predicting electron and phonon excitation rates via mediator exchange, with first-principles or experimental ELF data facilitating reliable rate calculations for a variety of dielectric targets (Knapen et al., 2021).

6. Limitations, Extensions, and Best Practices

• Breakdown of the ELF-Only Formalism: For quasi-2D crystals and in the presence of strong elastic scattering, the standard proportionality between EEL spectrum and Im[–1/ε(q, ω)] is quantitatively and even qualitatively invalid. Fully quantum-mechanical treatments incorporating probe kinematics and lattice effects are required (Nazarov et al., 2015, Nazarov et al., 2017).
• Sum-Rule Validation: Consistency of the extracted ELF must be confirmed against the optical f-sum and KK-sum rules (e.g., requiring Z_eff ≈ Z and ps-sum ≈1 for the total oscillator strength) (Xu et al., 2016, Cheng et al., 2024).
• Role of “Semi-Core” and Inner-Shell Bands: For transition and rare-earth metals, accurate representation of ELF above 20 eV requires explicit identification and fitting of semi-core shells, as these states contribute disproportionately to stopping and IMFP integrals (Cheng et al., 2024).
• Computational Efficiency: Liouville–Lanczos recursion and batch representation of the density response enable the ELF to be computed for large, complex systems at moderate computational cost (Timrov et al., 2013, Timrov et al., 2021).
• Experimental Uncertainties: Reliable ELF extraction demands careful deconvolution of multiple scattering, precise intensity normalization, and robust parametrization, with absolute and relative accuracy validated via independent measurements (e.g., stopping power, IMFP) (Xu et al., 2016, Aker et al., 2021, Cheng et al., 2024).
• Future Prospects: Integration of ultrafast, multimodal experimental schemes with advanced first-principles theory will further expand the capability to map ELF with combined high spatial, spectral, and temporal resolution (Pomarico et al., 2019, Brokkelkamp et al., 2022).

7. Representative Results and Benchmark Systems

System	Plasmon Peak(s) (eV)	Notable ELF Features	Reference
Pb(111) Thin Film	10.58 (surface); 7.0 (bulk-like mode)	QSE-induced suppression of classical modes, discrete QWS transitions	(Zubizarreta et al., 2013)
β-Si₃N₄	24–27	High-energy plasmon, low-energy suppression in perfect crystal	(Tao et al., 2015)
KNbO₃ (cubic)	6.5, 15.0, 22.6, 28.1, 41.5	Phase-fingerprint structure dominated by BO₆ octahedra	(Simsek et al., 2012)
Graphene	4.8 (π), 15.0 (σ+π)	Direction-dependent dispersion, suppressed image-coupling artifacts	(Mowbray, 2014)
Fe	19.8 (plasmon), 6.0 (shoulder), 55–60 (M₂,₃ edge)	Absolute ELF from RMC, validated to <1% f-sum error	(Xu et al., 2016)
H₂ isotopolog gas	11.9, 12.8, 15.0	Sub-eV resolved inelastic peaks, semi-empirical fit to KATRIN data	(Aker et al., 2021)
Cr	27 (plasma), 49 (semi-core 3p)	Importance of semi-core states in stopping cross section	(Cheng et al., 2024)

Peaks reflect dominant inelastic channels accessible in the ELF, with band-structure, dimensionality, and microstructure effects encoded in the detailed lineshape. ELF fingerprints are thus diagnostic of both electronic structure and dynamic response properties across a wide range of materials classes.

---

The consistent, experimentally and theoretically validated extraction of ELF remains essential for accurate modeling of energy transfer phenomena, interpretation of EELS and related spectroscopies, and the design of materials for electronic, photonic, and quantum-technology applications. Advanced methodologies continue to refine both the calculation and the measurement of ELF, especially in the context of nanostructured, low-dimensional, and strongly correlated systems.