Evidence for Many-Body States in NiPS$_3$ Revealed by Angle-Resolved Photoelectron Spectroscopy

Using $\mu$-ARPES, this study demonstrates that NiPS$_3$ exhibits many-body multiplet structures at the valence-band edge that cannot be explained by mean-field DFT+$U$ calculations, proving that the material's electronic properties are driven by complex local Ni-S quantum correlations.

The Gist

The Mystery of the "Ghost Band" in NiPS3

Imagine you are a detective trying to map out the layout of a massive, high-tech skyscraper (this is the NiPS3 crystal). To understand how the building works, you use a specialized drone (this is ARPES, the scientific tool) that flies through the halls and reports back on where the people (the electrons) are standing.

For years, architects have used a set of blueprints called DFT+U (a mathematical model) to predict where the people should be. These blueprints are usually incredibly accurate. But when the detectives flew the drone into the NiPS3 building, they saw something impossible: a "Ghost Band."

There was a whole crowd of people standing in a hallway that, according to every blueprint in existence, shouldn't even exist.

---

The Problem: The Blueprint vs. Reality

In most materials, the "blueprints" (the math models) work fine because the electrons behave like predictable commuters on a train—they follow set tracks and stay in their lanes.

However, NiPS3 is a "Strongly Correlated" material. In our skyscraper analogy, this means the people aren't just commuters; they are members of highly intense, synchronized dance troupes. They don't just walk; they react to every move their neighbors make. If one person jumps, the whole room reacts.

Because the math models (DFT+U) assume people act mostly independently, they completely missed the "dance" happening in the hallways. They only saw the "tracks," not the "dance."

---

The Discovery: The "Dance" of the Multiplets

The researchers realized that the "Ghost Band" wasn't a mistake or a glitch. It was a Many-Body State.

To solve the mystery, they stopped looking at the building as a collection of individual rooms and started looking at it as a collection of "Dance Studios" (these are the NiS6 clusters).

They used a different, much more complex math method called Exact Diagonalization. Instead of trying to map the whole city, they zoomed in on one single dance studio and simulated every possible interaction between the dancers.

They found the answer: When an electron is removed from the material (which is what the ARPES drone does), it’s like a dancer suddenly leaving the floor. In a normal material, the other dancers wouldn't care. But in NiPS3, the remaining dancers immediately rearrange themselves into specific, complex formations called "Multiplets."

These formations—these specific "poses" the electrons take after one is gone—create new energy signatures. The "Ghost Band" is actually the visual signature of these electron dance formations.

---

Why Does This Matter?

This paper is a big deal for two reasons:

1. It proves the drone works: It shows that ARPES isn't just a tool to check blueprints; it’s a tool that can reveal entirely new "rooms" and "dances" that the blueprints were too simple to see.
2. It redefines the material: It confirms that NiPS3 is a playground for "quantum many-body physics." It’s a place where you can't understand the whole by just looking at the parts; you have to understand the relationships between the parts.

In short: The scientists found a "ghost" in the machine, and instead of ignoring it, they proved that the ghost is actually a beautiful, complex dance that only happens when electrons work together in perfect, quantum harmony.

Technical Summary

Technical Summary: Evidence for Many-Body States in $\text{NiPS}_3$ Revealed by ARPES

1. Problem Statement

The study addresses a fundamental discrepancy in the electronic structure of $\text{NiPS}_3$, a van der Waals Mott-insulating antiferromagnet. While conventional mean-field theories, such as Density Functional Theory plus Hubbard $U$ (DFT+$U$), successfully describe the band structures of related transition-metal thiophosphates ($\text{MnPS}_3$, $\text{FePS}_3$, and $\text{CoPS}_3$), they fail to account for a specific spectral feature in $\text{NiPS}_3$.

Specifically, $\mu$-ARPES measurements reveal a weakly dispersive, additional spectral feature (a "shoulder") located approximately $700\text{ meV}$ above the upper valence band edge. This feature is absent in DFT+$U$ and hybrid functional calculations, suggesting that the electronic physics of $\text{NiPS}_3$ involves strong many-body correlations that transcend the single-particle/mean-field approximation.

2. Methodology

The authors employ a multi-pronged approach combining experimental spectroscopy with two distinct theoretical frameworks to bridge the gap between itinerant band theory and local correlation physics:

• Experimental ($\mu$-ARPES): High-resolution micro-angle-resolved photoelectron spectroscopy was used to map the valence band dispersion. The study performed temperature-dependent measurements to observe the feature across the Néel transition ($T_N$) and photon-energy-dependent measurements to determine the $k_z$ dispersion (confirming its bulk nature).
• Mean-Field Theory (DFT+$U$): Used to establish a "single-particle baseline." The authors analyzed the dependence of the density of states (DOS) on the Hubbard $U$ and Hund’s exchange $J_H$ to understand the bonding-antibonding splitting of $e_g$ states and the spin-splitting of $t_{2g}$ states.
• Many-Body Theory (Exact Diagonalization - ED): To capture the missing physics, the authors modeled a local $\text{NiS}_6$ cluster using a Single Impurity Anderson Model (SIAM). This approach treats the system as a collection of interacting transition-metal-ligand clusters, allowing for the calculation of multiplet final-state configurations (e.g., $d^7$ and $d^8L$ states) following electron removal.

3. Key Results

• Failure of Mean-Field Theory: DFT+$U$ correctly predicts the $t_{2g}$ and $e_g$ manifolds and their respective spin-splitting/hybridization effects but cannot generate the additional shoulder. The authors proved that the $e_g$ splitting in DFT+$U$ is a result of Ni–S covalency (bonding/antibonding), which is a single-particle effect.
• Identification of Multiplet States: The ED calculations revealed that the "missing" ARPES feature corresponds to multiplet-split final-state transitions. Specifically, the removal of an electron from the $\text{Ni}^{2+}$ ($3d^8$) ground state creates a manifold of $\text{Ni}^{3+}$ ($3d^7$) and ligand-hole ($d^8L$) configurations.
• Energy Matching: The energy separation between the observed ARPES shoulder and the valence band maximum matches the energy differences between the low-energy multiplet eigenstates calculated via ED.
• Robustness to Magnetism: The feature's insensitivity to the antiferromagnetic-to-paramagnetic transition confirms that it originates from local Ni–S multiplet physics rather than long-range magnetic ordering or collective excitations like spin polarons.

4. Key Contributions & Significance

• Direct Observation of Many-Body Physics in ARPES: The paper provides rare evidence that ARPES—often used to map "bands"—can directly probe local many-body multiplet structures in strongly correlated Mott insulators.
• Validation of the Cluster Approach: It demonstrates that in charge-transfer insulators like $\text{NiPS}_3$, the "cluster" (local atomic-like physics) is a more appropriate degree of freedom than the "Bloch wave" (itinerant band physics) for describing the $e_g$ sector.
• Theoretical Synthesis: The work establishes a hierarchy of applicability:
  • DFT+$U$ is sufficient for itinerant, weakly correlated states (e.g., $t_{2g}$ and ligand-dominated bands).
  • Cluster/ED methods are required for highly localized, strongly correlated states (e.g., $e_g$ multiplets).
• Broader Impact: This study underscores the necessity of using genuinely quantum many-body descriptions (like Cluster-DMFT) for 2D quantum materials where covalency and reduced dimensionality stabilize complex entangled states.