Charge-Transfer Satellites in Correlated Solids

• Charge-transfer satellites are distinct spectral features arising from local and nonlocal electron transfer processes in correlated materials.
• LDA+DMFT accurately models these satellites by incorporating material-specific hybridization and dynamical screening effects.
• These spectral markers provide insight into screening mechanisms and covalency in transition-metal oxides, rare-earth, and actinide compounds.

Charge-transfer satellites are distinct spectral features observed in the core-level and valence photoemission, X-ray absorption, and resonant inelastic X-ray scattering (RIXS) spectra of correlated materials. These satellites originate from electronic excitations involving intersite (nonlocal) and local charge-transfer processes. Their quantitative description is crucial for a microscopic understanding of screening, covalency, and dynamical correlations in transition-metal oxides, rare-earth and actinide compounds, and other strongly correlated systems. State-of-the-art ab initio methods, notably LDA+DMFT (Local Density Approximation plus Dynamical Mean-Field Theory), have enabled predictive, material-specific modeling of these features, linking them directly to dynamical screening and hybridization effects in real solids.

1. Physical Origin and Spectroscopic Manifestation

Charge-transfer satellites occur when core-level or valence photoemission spectroscopies—such as hard X-ray photoemission (XPS), soft X-ray absorption (XAS), or RIXS—probe many-body excitations of the system that involve transferring an electron between a transition metal and its ligands (typically O 2p for oxides), or between neighboring correlated sites. In the photoemission process, creation of a deep core hole (e.g., 1s or 2p) can be accompanied by local or nonlocal screening:

• Local screening: An electron hops from a ligand site to the photoemission site (e.g., O 2p → TM 3d) to screen the core-hole potential. This forms well-known "ct satellites" at fixed energy loss relative to the main line.
• Nonlocal screening: Electrons from more distant sites (e.g., other TM sites) participate, resulting in additional spectral features—main-line asymmetries, shoulders, or continuum satellites—arising from the hybridization and extended character of the screening processes (Ghiasi et al., 2018).

Experimentally, these satellites appear as sidebands or shoulders on core-level spectra (both in binding energy and RIXS energy-loss scales), or as broad continua at fixed energy loss associated with fluorescence-like features in RIXS (Hariki et al., 2019, Winder et al., 2020). Their spectral weight, energy separation from the main peaks, and detailed lineshapes encode the strength of charge-transfer, the degree of covalency, and the effectiveness of local versus nonlocal screening channels.

2. Theoretical Framework: LDA+DMFT and Extended Anderson Impurity Models

Predictive modelling of charge-transfer satellites requires inclusion of strong local correlations and dynamical screening, as well as realistic material-dependent hybridization. Here, the LDA+DMFT framework provides a powerful ab initio scheme:

• Hamiltonian construction: Starting from LDA (or GGA) one constructs a material-specific tight-binding Hamiltonian involving transition-metal 3d and ligand 2p Wannier orbitals. Local Hubbard-Kanamori interactions (U, J) are added to the correlated shells, and a double-counting term subtracts the static mean-field energy (Pavarini, 2014, Ghiasi et al., 2018, Hariki et al., 2017).
• DMFT mapping: The lattice problem is mapped onto a quantum impurity (Anderson impurity model, AIM) dynamically coupled to a self-consistently determined bath hybridization Δ(ω). The local DMFT self-energy Σ(ω) encodes all many-body corrections relevant for satellites (Hariki et al., 2017, Ghiasi et al., 2018, Hariki et al., 2019).
• Core-level spectroscopy: For XPS/XAS/RIXS, the impurity Hamiltonian is extended to include explicit core orbitals (e.g., 1s—no multiplets or spin–orbit, or 2p—full multiplet and core–valence interactions) and core-hole–valence interactions U_dc. The resulting spectral functions and cross-sections are obtained via configuration interaction (CI), Lanczos, or similar solvers (Ghiasi et al., 2018, Winder et al., 2020).

This methodology systematically includes both local and nonlocal screening contributions through the hybridization function Δ(ω), which—unlike in finite cluster models—naturally incorporates the full bandstructure, continuum of screening states, and essential physics for CT satellites.

3. Charge-Transfer Satellites in Core-Level and RIXS Spectra

The DMFT-based approach explains and quantifies the formation of charge-transfer satellites in several classes of correlated materials:

• 1s and 2p XPS in late TM oxides: Nonlocal screening is directly evidenced in the asymmetric main line and lower-binding-energy shoulders of the 1s spectra. The 1s satellites arise from screening by both local ligand holes and more extended oxygen or TM states, whose spectral weight and energy separation are set by the hybridization density of the DMFT bath (Ghiasi et al., 2018, Hariki et al., 2017).
• 2p XPS and XAS multiplet structure: The inclusion of realistic core-level multiplets and the DMFT hybridization enables accurate reproduction of experimental 2p line splittings, satellite positions, and relative intensities for NiO, CoO, Fe₂O₃, and other systems (Hariki et al., 2017). The fine structure—including main line doublets and high-binding energy satellites—directly encodes the material-specific CT energy, U, J, and hybridization parameters.
• RIXS fluorescence and charge-transfer excitations: In RIXS, both bound excitons (Raman-like d–d and charge-transfer peaks) and unbound electron–hole-pair fluorescence-like features are captured. The latter require a continuum of (unbound) bath states and their presence (or absence) is highly sensitive to lattice geometry and hybridization (e.g., strong in perovskite cobaltites, nearly absent in edge-sharing geometries) (Hariki et al., 2019, Winder et al., 2020). Material-dependent features in the RIXS intensity can be traced to the low- and high-energy structure of the DMFT hybridization function.

The ability of LDA+DMFT+CI to simultaneously predict all these spectral features—without ad hoc parameters beyond the ab initio band structure, U, and J—establishes its reliability for charge-transfer satellites.

4. Dynamical Screening, Plasmon Satellites, and Frequency-Dependent U

Dynamical screening effects, namely the frequency dependence of the effective interaction U(ω), are central to the quantitative description of charge-transfer satellites and related plasmon sidebands. Within the LDA+DMFT extension to dynamical interactions (LDA+U(ω)+DMFT), or the GW+DMFT hybrid approaches, the following picture emerges (Biermann, 2014):

• Plasmon satellites: Dynamical screening gives rise to sidebands (plasmon satellites) at energies equal to collective bosonic modes (e.g., 15 eV in SrVO₃), observable as features well separated from the main quasiparticle and Hubbard bands in both XPS and RIXS.
• Renormalization and spectral weight transfer: The dynamical reduction of the effective U at low frequency, and the coupling to bosonic (plasmon, charge-transfer) modes, leads to mass renormalization (Z_B < 1), spectral broadening, and transfer of weight from coherent quasiparticles to incoherent satellites.
• Impurity action and hybrid models: Frequency-dependent U(ω) can be treated either by introducing retarded interactions in the impurity action, or by coupling to explicit boson fields (Hubbard–Holstein–Anderson models). In the antiadiabatic limit, this reduces to a static model with renormalized hoppings and static U₀, but for real materials, full dynamical treatment is necessary to correctly obtain satellite spectra (Biermann, 2014).

The experimental confirmation of plasmon and CT satellites, with energies, intensities, and dispersion matching LDA+U(ω)+DMFT and GW+DMFT predictions, validates these frameworks (Biermann, 2014, Winder et al., 2020).

5. Material-Specific Results and Trends

Comprehensive LDA+DMFT studies across transition-metal oxides, rare-earth nickelates, actinide dioxides, and their core-level spectra demonstrate:

• NiO, CoO, Fe₂O₃, FeTiO₃: Well-separated CT satellites in 1s/2p XPS; modulation of main-line/asymmetry and satellite weight with increasing covalency and hybridization (U_dd ∼ 6–7 eV, Δ ∼ 3.5–4.4 eV, V_T2g/Eg ∼ 1.2–2.5 eV) (Ghiasi et al., 2018, Hariki et al., 2017).
• Actinide dioxides (UO₂, NpO₂, PuO₂): Satellite peaks in 4f core spectra directly track increasing 5f–O 2p hybridization, with occupation numbers n_f ∼ 2.5–4.4 and satellite energies matching experiments (Kolorenc et al., 2015).
• Nickelates and cobaltites: In LuNiO₃ and SrCoO₃, rich RIXS structure arises from coexisting bound d–d excitations and fluorescence-like unbound charge-transfer continua, controlled by site (dis)proportionation and correlated bath structure (Winder et al., 2020, Hariki et al., 2019).

Key systematic trends: as hybridization V_eff increases, nonlocal screening and satellite intensity grow; as U or CT energy decreases, satellites shift toward the main line. The theoretical predictions are not only qualitative but quantitative, with absolute satellite energies, spectral weights, and lineshape asymmetries matching high-resolution experiment.

6. Methodological Distinctions, Controversies, and Practical Considerations

Compared to traditional cluster models, the LDA+DMFT (and LDA+DMFT+CI) frameworks provide several critical advances:

	Cluster Model	LDA+DMFT+CI
Screening channels	Local only (finite ligand set)	Local + nonlocal (continuum bath from full bandstructure)
Satellite types	Only bound CT	Bound and unbound (fluorescence-like) CT satellites
Material specificity	Parametric, limited	Quantitative, ab initio (Wannier, cRPA, …)

• Nonlocal screening: The hybridization function Δ(ω) in DMFT includes continuum contributions from all bands and neighbors, allowing for accurate nonlocal screening; cluster models lack this key ingredient (Hariki et al., 2017, Ghiasi et al., 2018).
• Double-counting corrections: Substantial efforts address ambiguities in subtracting the LDA mean-field interaction; methods include FLL, AMF, and parameter-free schemes as in LDA'+DMFT (Nekrasov et al., 2012, Lee et al., 2014).
• Improvements and limitations: Full account of dynamical U(ω) is computationally expensive, often requiring CI, Lanczos, or CT-QMC impurity solvers; charge self-consistent schemes (Grånäs et al., 2011) further improve fidelity but increase cost.

Major open challenges remain in consistently including nonlocal correlations and fully dynamical screening in large, multiorbital systems, but the LDA+DMFT+CI methodology represents the current state-of-the-art for predicting and understanding charge-transfer satellites in correlated materials.

7. Significance and Outlook

Charge-transfer satellites are fundamental markers of the interplay between strong local correlations, screening, and hybridization in quantum materials. Their accurate modeling within LDA+DMFT frameworks has provided in-depth insight into the microscopic origin of satellites, their universality (across transition-metal oxides, rare-earths, actinides), and their material dependence. These methods have also clarified the role of CT processes in renormalizing effective masses, opening bandgaps, and generating high-energy incoherent excitations, directly impacting the understanding and design of functional correlated electron systems.

As experimental resolution and complexity increase, ab initio treatment of charge-transfer satellites remains an essential benchmark for any first-principles methodology targeting the full excitation spectrum of correlated solids (Ghiasi et al., 2018, Hariki et al., 2019, Biermann, 2014, Kolorenc et al., 2015, Winder et al., 2020).