\textit{\textbf{First-principles}} description of pumped inelastic X-ray scattering: example of K-edge RIXS in graphite

This paper presents an *ab initio* framework combining the Bethe-Salpeter equation and real-time time-dependent density-functional theory to predict time-resolved resonant inelastic X-ray scattering (RIXS) in optically pumped materials, demonstrating its accuracy by successfully modeling angular-dependent K-edge RIXS spectra in graphite at various delay times.

The Gist

Imagine you are trying to understand how a complex machine works, like a grand piano. You could listen to it play a song (that's like normal spectroscopy), but Resonant Inelastic X-ray Scattering (RIXS) is like hitting a specific key with a hammer made of light, listening to the sound it makes, and then analyzing exactly how the internal strings and hammers vibrated in response. It tells you not just what the machine is made of, but how its parts move and interact.

This paper introduces a new, super-precise computer program that predicts exactly what this "light-hammer" experiment will look like, even when the machine is being shaken by a second, faster pulse of light (like a pump).

Here is the breakdown of their work using everyday analogies:

1. The Problem: Predicting the Unpredictable

Scientists have long been able to take "snapshots" of materials using X-rays. However, predicting exactly what those snapshots will look like—especially when the material is being "pumped" with a laser to wake it up from its sleep—is very hard.

• The Old Way: Previous computer models were like looking at a crowd of people and assuming everyone is standing still and acting alone. They missed how people (electrons) actually hold hands and move together (a phenomenon called "excitonic effects").
• The New Way: The authors built a new framework that acts like a high-speed, 3D movie simulator. It doesn't just look at individuals; it watches the whole crowd dance together, accounting for how they pull on each other.

2. The Method: A Two-Step Dance

The researchers combined two powerful tools to create their simulation:

• Step 1 (The "Pump"): They used a tool called RT-TDDFT to simulate what happens when a laser hits the material. Imagine shining a flashlight on a trampoline; this tool calculates how the trampoline bounces and how the people on it shift their weight immediately after the light hits. This gives them a "non-equilibrium" map of where the electrons are right after the laser pulse.
• Step 2 (The "Probe"): They then used the Bethe-Salpeter Equation (BSE). Think of this as a super-accurate rulebook for how X-rays interact with that bouncing trampoline. It calculates the complex dance between the electron that got kicked out and the "hole" (empty space) it left behind.

By combining these, they can predict the "echo" (the scattered X-ray) for any angle of light coming in and any angle of light going out.

3. The Test Case: Graphite (The Pencil Lead)

To prove their method works, they tested it on graphite (the stuff in pencil lead).

• Why Graphite? It's like a stack of paper sheets. The atoms inside each sheet are glued together tightly (like strong glue), but the sheets themselves are only loosely stuck together (like a stack of loose papers). This makes the material very "anisotropic," meaning it behaves very differently depending on whether you look at it from the side or from the top.
• The Result: The computer simulation successfully predicted two distinct types of "notes" the graphite would play:
  • $\pi$ (Pi) notes: These come from the electrons moving between the sheets (the loose paper).
  • $\sigma$ (Sigma) notes: These come from the electrons moving tightly within the sheets (the strong glue).
    The simulation showed that if you shine light from the side, you hear mostly the "glue" notes. If you shine it from the top, you hear the "paper" notes. This matched real-world experiments perfectly.

4. The "Pumped" Experiment: Shaking the Table

The most exciting part of the paper is what happens when they "pump" the graphite with a laser before hitting it with X-rays.

• The Analogy: Imagine the graphite is a calm pond. The laser pump is like throwing a rock into the pond, creating ripples. The X-ray is a sonar ping sent in to see how the ripples changed the water.
• The Finding: When the graphite was "pumped," the simulation showed that the "notes" changed slightly. New, faint sounds appeared in the low-energy range, and the volume of existing sounds shifted.
• The Takeaway: The computer predicted that even a short laser pulse changes the electronic "mood" of the material, creating a temporary state that is different from its resting state. The simulation matched experimental data so well that it could see these subtle shifts, proving the method works for "time-resolved" (movie-frame-by-frame) studies.

Summary

In simple terms, this paper says: "We built a new, highly accurate computer model that can predict exactly how a material will react to X-rays, even when that material is being jolted by a laser."

They tested it on graphite, and the computer's "prediction" matched the real-life "experiment" perfectly, correctly identifying how the material's internal structure (the tight sheets vs. the loose layers) responds to light from different angles and at different times. This gives scientists a powerful new tool to understand how materials behave in real-time, without needing to run expensive experiments for every single guess.

Technical Summary

Technical Summary: First-Principles Description of Pumped Inelastic X-ray Scattering

Problem and Context
Resonant Inelastic X-ray Scattering (RIXS) is a powerful spectroscopic technique for probing electronic excitations with element, orbital, and momentum resolution. While theoretical frameworks for RIXS have advanced from independent particle approximations to many-body perturbation theory (MBPT) using the Bethe-Salpeter equation (BSE) to capture excitonic effects, a significant gap remains in describing time-resolved (pumped) RIXS. Existing theoretical models often lack the capability to simultaneously account for:

1. Excitonic effects in both core and valence excitations, which are crucial for accurate spectral descriptions.
2. Non-equilibrium charge-carrier distributions induced by optical pumping.
3. Arbitrary polarization geometries for incoming and outgoing photons, which are essential for anisotropic materials.

Time-resolved RIXS (tr-RIXS) experiments are emerging but remain limited; theoretical modeling requires a framework that integrates electron-hole correlations with non-equilibrium dynamics.

Methodology
The authors present an ab initio framework that combines Real-Time Time-Dependent Density Functional Theory (RT-TDDFT) with the Bethe-Salpeter Equation (BSE) to predict polarization- and time-resolved RIXS spectra. The methodology proceeds as follows:

1. Theoretical Formalism: The approach is based on the Kramers-Heisenberg formula for the double-differential cross section (DDCS). The RIXS process is described as a two-step mechanism:

   • Absorption: An incident X-ray photon ($\omega_1$) creates a core-hole intermediate state ($|n\rangle$).
   • Emission: A valence electron fills the core hole, emitting a photon ($\omega_2$) and leaving a final electron-hole pair ($|f\rangle$).
     The DDCS is calculated using transition operators derived from the BSE eigenvectors and eigenvalues.

2. Integration of Non-Equilibrium States: To model the pumped case, the framework employs RT-TDDFT (specifically within the exciting code) to propagate the Kohn-Sham system under an optical pump laser. This yields time-dependent, non-equilibrium electronic occupation numbers.

3. Workflow:

   • Ground State: A DFT calculation provides the starting point.
   • BSE Calculations: Two separate BSE calculations are performed:
     • One for core-level excitations (to obtain core oscillator strengths $t^{(1)}$).
     • One for valence/optical excitations (to obtain the final state pathways $t^{(2)}$).
   • Non-Equilibrium Extension: For tr-RIXS, the non-equilibrium occupations from RT-TDDFT are used as constraints in the BSE calculations.
   • Polarization Generalization: The implementation was extended to handle arbitrary, independent polarization vectors ($\mathbf{e}_1$ and $\mathbf{e}_2$) for incoming and outgoing light, moving beyond previous limitations where $\mathbf{e}_1 = \mathbf{e}_2$.

4. Computational Details: Calculations were performed using the all-electron full-potential package exciting. The study utilized the carbon K-edge of graphite, employing a GGA-PBE functional with van der Waals corrections. The BSE included 8 occupied and 15 unoccupied bands for optical spectra, and 40 conduction bands for X-ray spectra.

Key Contributions

• Unified Framework: The paper establishes a fully parameter-free ab initio workflow that combines RT-TDDFT and BSE to simulate tr-RIXS, capturing both many-body excitonic effects and pump-induced non-equilibrium dynamics.
• Polarization Generalization: The authors generalized the RIXS implementation to support arbitrary polarization orientations for both incident and scattered photons, allowing for the investigation of complex scattering geometries and mixed polarizations.
• Implementation in exciting: The method is implemented within the full-potential code exciting, utilizing the pyBRIXS package for cross-section calculations.

Results: Graphite K-Edge RIXS
The framework was applied to graphite, a semi-metal with strong electronic anisotropy between in-plane ($\sigma$) and out-of-plane ($\pi$) states.

• Equilibrium Spectra:

  • The calculated absorption spectra (both optical and core-level) reproduced experimental features, confirming the necessity of BSE to capture excitonic effects, particularly for $\pi \to \sigma^*$ and $\sigma \to \pi^*$ transitions.
  • Anisotropy: Equilibrium RIXS spectra showed strong polarization dependence. In-plane polarization ($\mathbf{e} \parallel x$) highlighted high-energy $\sigma$-derived excitations ($>8$ eV), while out-of-plane polarization ($\mathbf{e} \parallel z$) emphasized low-energy $\pi$-derived excitations.
  • Scattering Geometry: Simulations of a $30^\circ$ incident angle (matching experimental setups) revealed non-linear mixing of $\pi$ and $\sigma$ contributions, showing interference effects and enhanced $\pi \to \pi^*$ transitions at low energy loss.

• Pumped (Non-Equilibrium) Spectra:

  • Simulations were performed for a system pumped by a 266 nm laser (45 fs duration) with a delay time of 150 fs.
  • Spectral Changes: The pumped spectra closely resembled the equilibrium case, indicating moderate modifications. However, a distinct additional low-energy-loss feature (a few eV) emerged for both polarizations, attributed to modifications in the electronic distribution near the Fermi level.
  • Mixed Polarizations: In mixed polarization setups, the non-equilibrium spectra showed a redistribution of spectral weight and broadened features compared to equilibrium.
  • Comparison with Experiment: The theoretical results reproduced experimental peak positions and general features well. Minor deviations (e.g., peak widths and specific shifts) were attributed to electron-phonon coupling, which is not included in the current static framework, and potential self-absorption effects in experiments.

Significance and Claims
The authors claim that their work provides a general and fully parameter-free pathway to compute RIXS spectra in complex materials, ranging from metals to large-gap semiconductors, for arbitrary scattering geometries. By successfully reproducing the characteristic anisotropy of graphite and its evolution under optical pumping, the method demonstrates its capability to:

1. Resolve distinct $\pi$- and $\sigma$-derived features.
2. Capture the interplay between electronic structure and polarization.
3. Model time-dependent phenomena induced by optical excitation.

The paper concludes that this approach offers a robust tool for interpreting tr-RIXS experiments and understanding non-equilibrium phenomena in solids, specifically highlighting the directional dependence of fundamental excitations. The authors remain modest regarding discrepancies with experiment, attributing them to known limitations such as the exclusion of electron-phonon coupling and self-absorption effects.