🔬 mesoscale physics

Temporal structures of the X-ray photoemission problem

This paper investigates X-ray photoemission in simple metals within the time domain to reveal coherent interference effects and methodological advances, specifically comparing complementary time-dependent eNRG algorithms against analytical and numerical benchmarks to address strong core-hole potentials.

Original authors: F. D. Picoli, G. Diniz, M. P. Lenzarini, I. D'Amico, L. N. Oliveira

Published 2026-07-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: F. D. Picoli, G. Diniz, M. P. Lenzarini, I. D'Amico, L. N. Oliveira

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you're watching a crowded dance floor (the metal) where everyone is moving in perfect sync. Suddenly, a spotlight hits one dancer (the core electron) and yanks them off the stage. This leaves a sudden, empty spot—a "hole"—that acts like a giant magnet, pulling the other dancers toward it.

For decades, scientists have studied what happens to the music (the spectrum) after this event, looking at the frequencies of the sound. But in this paper, the authors decided to watch the movie of the event in real-time, frame by frame. They wanted to see how the dance floor settles down after the shock.

The Main Discovery: A Wobbly Decay
The authors found that the "fidelity"—a fancy word for how much the dance floor still looks like it did before the dancer left—doesn't just fade away smoothly. Instead, it does something much more interesting: it wobbles.

Think of the decay like a swing slowing down. The swing doesn't just stop; it swings back and forth while getting smaller. The authors discovered that the photoemission rate (the dance floor's reaction) follows a specific mathematical curve (a power law) but is wrapped in a weakly damped harmonic oscillation.

Why does it wobble? The authors explain that the dancers (electrons) are actually splitting into two rival groups based on how they react to the empty spot:

The "Plugged" Group: Some dancers jump into the empty spot to fill the hole.
The "Unplugged" Group: Others stay in their original spots, leaving the hole empty.

These two groups are like two different bands playing slightly different tunes. When they play together, they create an interference pattern—a beat. The "Plugged" group plays a tune that fades slowly, while the "Unplugged" group plays a tune that fades much faster. Because one group dies out quicker than the other, the "beat" (the wobble) gets quieter over time. This explains the damped oscillation the authors saw in their simulations.

The Tools: A New Way to Count
To figure this out, the authors had to build a new kind of calculator. Traditional methods for studying these systems (like the standard NRG) are great for static snapshots but get messy when trying to watch the movie in motion. They often introduce "ghosts"—fake wiggles and ripples in the data that aren't real physics, just math errors.

The authors used a clever, flexible version of the calculator called eNRG (real-space numerical renormalization group). They tested two specific ways to use it:

The "Big Offset" Method: They kept a large chunk of the system untouched (like keeping the first 100 dancers in their original formation) and only simplified the rest. This gave them essentially exact results for a specific time window, but it gets too expensive to run if they want to watch the movie for too long.
The "Smoothing" Method: They ran the simulation many times with slightly different settings and averaged the results. This washed out the fake "ghost" wiggles caused by the math, but it couldn't fix the initial "transitory" errors that happen right at the start.

What They Ruled Out
The paper explicitly argues against the idea that the decay is just a simple, smooth slide. They show that if you only look at the frequency (the sound), you miss the wobble entirely. The wobble is a real physical feature caused by the interference of those two groups of electrons, not a glitch in the math.

They also demonstrate that standard, brute-force calculations (trying to solve the whole dance floor at once) become impossible for large systems because the computer time explodes. Their new methods are the only way to get accurate results for long periods without needing a supercomputer the size of a city.

How Sure Are They?
The authors are very confident in their findings, but they are careful to say how they know.

The Physics: They proved the wobble exists by comparing their new eNRG simulations against two "gold standard" benchmarks: a direct, brute-force calculation (which is exact but slow) and a new analytical formula they derived (which is exact for long times). The results matched perfectly.
The Numbers: In their simulations with a potential strength of K = -2τ, they saw the fidelity decay following a power law with an exponent α (defined by the phase shift δ). They noted that for a potential of K = -5τ, the damping was barely perceptible, while for K = -τ, the wobbles died out very quickly.
The Limits: They admit that their "smoothing" method still has small errors at the very beginning of the simulation (the "transitory" phase) because of how the math handles the very first moments. However, for the long-term behavior, their method is precise.

The Takeaway
This paper didn't just solve a math problem; it revealed a hidden layer of physics. By watching the "movie" of the electron dance, they showed that the system doesn't just calm down; it vibrates with a specific rhythm caused by the clash between electrons that fill the hole and those that don't. They also provided a new, more flexible toolkit (the eNRG methods) that other scientists can use to study similar "dance floors" in the future, especially those where the dancers interact with each other in more complex ways.

In short: The dance floor wobbles, and thanks to a new way of counting the dancers, we finally know why.

Technical Summary: Temporal Structures of the X-ray Photoemission Problem

Problem Statement
Traditional theoretical studies of X-ray photoemission from simple metals have predominantly operated in the frequency domain to reproduce experimental spectra. This paper investigates the same problem in the time domain to uncover physical insights and methodological advances that are inconspicuous in frequency-domain analyses. The specific focus is on the time dependence of the photoemission current, mathematically equivalent to the fidelity $F(t)$ —the squared projection of the time-evolved quantum state $|\Psi(t)\rangle$ onto the initial ground state $|\Psi(0)\rangle$ —following the sudden creation of a core hole. While the frequency-domain behavior is well-characterized by power laws (Doniach-Sunjic for XPS, Nozières-De Dominicis for XAS), the temporal evolution reveals complex interference patterns and damping mechanisms that require rigorous numerical treatment.

Methodology
The study employs a spinless tight-binding Hamiltonian to model a half-filled conduction band interacting with a core state via an attractive potential $K$ . The authors utilize two complementary approaches to calculate the time-dependent fidelity:

Benchmarking via Exact and Analytical Methods:
- Numerical Diagonalization: For a tight-binding lattice of size $2L$ , the authors perform brute-force diagonalization of the initial and final Hamiltonians. Using Wick's theorem, the fidelity is computed as the determinant of an $L \times L$ matrix. This provides essentially exact results for short to moderately long times.
- Analytical Derivation: An analytical expression for $F(t)$ is derived based on the partitioning of final many-body eigenstates into "plugged" (bound state occupied) and "unplugged" (bound state vacant) classes. This derivation is highly accurate for moderately long times ( $\omega_\tau t \gg 1$ ).
- These two methods serve as a "reference" to gauge the precision of renormalization-group calculations.
Time-Dependent eNRG (exponential Numerical Renormalization Group):
The authors apply the eNRG method, a real-space variant of Wilson's NRG, to the problem. The eNRG construction is parameterized by:
- $\lambda > 1$ : The discretization parameter (analogous to Wilson's $\Lambda$ ).
- $\theta \in [-1, 1]$ : The twist parameter (analogous to Wilson's $z$ ).
- $\zeta$ : An integer offset representing the number of sites in the un-discretized "offset region" of the Wilson chain.
Two specific eNRG procedures are explored to mitigate discretization artifacts:
- Large- $\zeta$ Procedure: Setting $\zeta$ large enough to cover the time interval of interest ( $\zeta \approx \omega_\tau t_{max}$ ). This avoids iterative diagonalization for quadratic Hamiltonians and yields exact results within the interval.
- Smoothing Procedure: For cases where large $\zeta$ is computationally prohibitive (e.g., in interacting models), the authors average raw fidelities over a set of offsets $\zeta$ and integrate over the twist parameter $\theta$ . This aims to suppress persistent oscillations caused by the discretization of the conduction band.

Key Results and Physical Insights

Damped Harmonic Oscillations: The calculated fidelity $F(t)$ $F (t)$ exhibits a weakly damped harmonic oscillation modulating the Doniach-Sunjic power-law decay. This behavior arises from the coherent interference between two classes of excitations:
- Plugged States: Final states where the bound state created by the core hole is occupied. These follow the Doniach-Sunjic power law ( $F_P \sim t^{-\alpha}$ ).
- Unplugged States: Final states where the bound state is vacant (an electron is promoted to the conduction band). These follow the Nozières-De Dominicis power law ( $F_U \sim t^{-\beta}$ ), with $\beta > \alpha$ .
- The oscillation frequency is determined by the energy difference $\epsilon_B$ between the plugged and unplugged states. The damping occurs because the unplugged component decays faster than the plugged component.
Analytical Validation: The derived analytical expression (Eq. 53) accurately reproduces the numerical data for $\omega_\tau t > 15$ , confirming the physical interpretation of the interference between plugged and unplugged channels.
eNRG Performance and Artifacts:
- Transitory Deviations: At short times, deviations arise from "irrelevant operators" introduced by the discretization. These decay slowly, governed by the uncertainty principle rather than thermal factors.
- Persistent Deviations: At longer times, the discretization of the continuous band introduces artificial, polychromatic oscillations (analogous to de Haas-van Alphen oscillations in thermodynamics).
- Large- $\zeta$ Efficacy: For quadratic Hamiltonians, the large- $\zeta$ procedure yields essentially exact results up to $t \approx T_\zeta$ (where $\omega_\tau T_\zeta = \zeta$ ) with minimal computational cost compared to brute-force diagonalization.
- Smoothing Efficacy: The smoothing algorithm (averaging over $\zeta$ and integrating over $\theta$ ) successfully eliminates persistent fluctuations but cannot remove transitory deviations. It is effective for medium to long times but leaves residual errors at short times.

Significance and Claims
The paper claims to advance the understanding of X-ray photoemission by revealing rich temporal physics—specifically the interference between distinct many-body excitation classes—that is obscured in frequency-domain spectra. Methodologically, the work establishes the eNRG method as a viable tool for time-dependent problems in condensed matter physics.

The authors modestly position their findings as a necessary step toward handling more complex, interacting systems. They note that while the large- $\zeta$ method is highly efficient for quadratic Hamiltonians, its application to correlated impurity models (requiring iterative diagonalization) is limited by computational cost. Consequently, the smoothing algorithm is presented as a promising, albeit imperfect, alternative for accessing long-time dynamics in interacting systems where standard ultraviolet truncation schemes are inadequate. The study serves as a benchmark and a testing ground for developing time-dependent numerical procedures based on the eNRG framework.

More like this