First-principles Newns-Anderson Hamiltonian Construction for Chemisorbed Hydrogen at Metal Surfaces

This paper presents a first-principles method for constructing Newns-Anderson Hamiltonians via projection operator diabatisation of Kohn-Sham DFT data to accurately model chemisorbed hydrogen on Al, Cu, and Pt surfaces, revealing that the widely used wideband limit approximation is valid for Al but insufficient for Cu and Pt.

The Gist

Here is an explanation of the paper, translated into everyday language using analogies.

The Big Picture: Building a "Simplified Map" of a Complex World

Imagine you are trying to understand how a single drop of water (a hydrogen atom) behaves when it lands on a giant, bustling dance floor (a metal surface).

In the real world, the dance floor is crowded with millions of dancers (electrons) moving in complex, chaotic patterns. The drop of water interacts with all of them at once. To simulate this perfectly on a computer, you would need to track every single dancer and the drop simultaneously. This is computationally impossible for many real-world problems, like designing better catalysts for making fuel or cleaning pollution.

The Solution: Scientists use a "simplified map" called the Newns-Anderson Model. Think of this model as a cheat sheet. Instead of tracking millions of dancers, it simplifies the whole dance floor into a single "average" crowd, and it treats the water drop as a single person interacting with that crowd.

The Problem: For decades, scientists have built these cheat sheets using a big assumption: they assumed the dance floor was perfectly flat and uniform (the "wideband limit"). They assumed the water drop felt the same amount of resistance no matter where it was on the floor.

The Discovery: This paper says, "Wait a minute. That assumption isn't always true." Sometimes the dance floor has bumps, pits, and different textures (like different types of metals). If you use the flat-floor assumption on a bumpy floor, your predictions will be wrong.

The New Method: The "Projection" Camera

The authors developed a new, high-tech way to build these simplified maps directly from the laws of quantum physics (specifically, Density Functional Theory or DFT).

Imagine you have a high-resolution 8K video of the chaotic dance floor. You want to turn it into a simple cartoon animation.

1. Old Way: You guess what the cartoon should look like based on general rules (assuming the floor is flat).
2. New Way (POD): You use a special "projection camera" (Projection-Operator Diabatization). This camera looks at the complex 8K video and mathematically extracts the exact interaction between the water drop and the crowd, preserving the specific "texture" of the floor.

The Experiment: Testing on Three Different Dance Floors

The team tested their new method on three different metal surfaces, which act like three different types of dance floors:

1. Aluminum (Al): A floor made mostly of smooth, simple steps (s-orbitals).
2. Copper (Cu): A floor with a mix of smooth steps and some complex spins (s and d-orbitals).
3. Platinum (Pt): A very complex floor with lots of intricate, heavy spins (d-orbitals).

They placed a hydrogen atom on each and watched how it "chemisorbed" (stuck) to the surface.

Key Findings: The "Bumpy Floor" Surprise

1. The "Flat Floor" Assumption is Broken for Some Metals
For Aluminum, the old assumption worked fine. The "dance floor" was smooth enough that the simplified map was accurate.
However, for Copper and Platinum, the floor was not flat. The interaction between the hydrogen and the metal changed drastically depending on the energy level.

• Analogy: Imagine driving a car. On a highway (Aluminum), you can assume the road is flat and drive at a constant speed. But in a city with hills and potholes (Copper/Platinum), assuming the road is flat will get you stuck or crash. The authors showed that for transition metals, you must use their new, detailed map to get the physics right.

2. The "Zoom" Problem (Basis Sets)
The paper also discovered a tricky technical issue. When they tried to build the map using too much detail (a large "basis set" of mathematical functions), the math got messy.

• Analogy: Imagine trying to take a photo of a single flower in a garden. If you use a low-resolution camera (small basis set), you get a clear picture of the flower. If you use a super-high-resolution camera that captures every leaf on every tree in the background, the computer gets confused trying to isolate just the flower, and the image gets distorted.
• The Fix: They found a "Goldilocks" setting (a specific medium-sized basis set) that was detailed enough to be accurate but simple enough to keep the math clean.

3. Predicting How Long Things Last
Using their new, accurate maps, they calculated two important things:

• Electronic Tunnelling: How fast an electron jumps off the hydrogen atom.
• Vibrational Lifetimes: How long the hydrogen atom "shakes" before it stops vibrating.
  Their results matched other high-level computer simulations perfectly, proving their new method works.

Why Does This Matter?

This paper is a toolkit upgrade for scientists.

• For Chemists: It helps them design better catalysts (chemicals that speed up reactions) for things like making green hydrogen fuel or creating new medicines.
• For Physicists: It proves that we can no longer rely on "lazy" assumptions (the wideband limit) when dealing with complex metals like Platinum. We need the full, detailed picture.

In a nutshell: The authors built a better, more accurate "cheat sheet" for how atoms stick to metal surfaces. They proved that while the old, simple cheat sheet works for some metals, it fails for the complex ones we use most often in technology. Their new method ensures we don't make mistakes when predicting how these materials behave.

Technical Summary

Here is a detailed technical summary of the paper "First-principles Newns-Anderson Hamiltonian Construction for Chemisorbed Hydrogen at Metal Surfaces."

1. Problem Statement

The Newns-Anderson Hamiltonian (NAH) is a fundamental model used to describe adsorption at gas-solid interfaces, characterizing the coupling between an adsorbate's electronic state and the continuum of metal states. It is essential for simulating charge transfer, non-adiabatic dynamics, and vibrational relaxation. However, constructing NAHs from first principles typically relies on two major simplifying assumptions that limit their accuracy:

1. The Wideband Limit (WBL) Approximation: Assumes the molecule-metal coupling is energy-independent (constant). This is often valid for $sp$-bands but questionable for transition metals with $d$-bands.
2. Parameterization Methods: Traditional methods often use constrained DFT (cDFT) or analytical functions, which can be sensitive to constraints or fail to capture strong hybridization in chemisorbed systems.

There is a lack of a robust, first-principles method to construct high-dimensional NAHs that accurately capture energy-dependent couplings and diabatic potential energy surfaces (PESs) for strongly chemisorbed systems without relying on the WBL approximation.

2. Methodology

The authors present a systematic first-principles approach to construct NAHs using Projection-Operator Diabatisation (POD) applied to Kohn-Sham (KS) Hamiltonian matrices obtained from Density Functional Theory (DFT).

• System: Chemisorbed hydrogen (H) on three fcc metal(111) surfaces: Aluminum (Al), Copper (Cu), and Platinum (Pt). These represent systems dominated by $p$-, $s$-, and $d$-states, respectively.
• POD2GS Approach:
  • The full KS Hamiltonian ($H$) and overlap matrix ($S$) are partitioned into adsorbate ($aa$), substrate ($ss$), and coupling ($as/sa$) blocks.
  • The method employs Löwdin orthogonalization within the fragment blocks followed by a Gram-Schmidt (GS) orthogonalization step. Specifically, the authors orthogonalize all adsorbate states with respect to all substrate states to ensure the projected density of states (PDOS) matches the original DFT results.
  • This transforms the Hamiltonian into a block-diagonal form where the off-diagonal blocks represent the diabatic couplings ($V_s$) between the adsorbate and the metal.
• Basis Set Sensitivity Analysis: A critical part of the methodology involves testing the sensitivity of the POD results to the size of the atom-centered basis set (ranging from a minimal 1s basis to Tier2 settings in FHI-aims). The authors found that large basis sets lead to information loss when projecting a multi-dimensional adsorbate subspace onto a single effective state.
• State Following: To ensure smooth potential energy curves, the authors implemented a "state-following" algorithm that tracks the specific diabatic adsorbate state with the highest overlap to a reference state as the adsorbate-surface distance changes.
• Validation Metrics: The constructed NAHs were validated by computing:
  • Projected Density of States (PDOS).
  • Electronic tunnelling lifetimes ($\tau_{el}$).
  • Vibrational lifetimes ($\tau_{vib}$) due to electron-phonon coupling, calculated using both the full energy-dependent chemisorption function and the WBL approximation.

3. Key Contributions

• First-Principles NAH Construction: Demonstrated a robust workflow to generate NAHs directly from DFT without empirical fitting, specifically tailored for strongly chemisorbed systems.
• Basis Set Dependency Discovery: Identified a critical limitation in POD-based NAH construction: using large, diffuse basis sets causes the orthogonalization procedure to shift diabatic states to unrealistic energies outside the metal's valence band. The authors determined that a Tier1-s basis set (removing higher principal $s$-orbitals) offers the best compromise between accurate ground-state energies and reliable diabatic couplings.
• Refutation of the Wideband Limit: Provided quantitative evidence that the WBL approximation is invalid for transition metals (Cu, Pt) even near the Fermi level, as the chemisorption function exhibits strong energy dependence.
• Shift Function Analysis: Calculated the real part of the self-energy (shift function, $H(\epsilon)$), showing it is non-negligible for transition metals, further invalidating the WBL assumption.

4. Key Results

• Basis Set Impact:
  • Large basis sets (Tier2) caused the diabatic adsorbate state energy to drop significantly (due to orthogonalization shifts), moving it outside the metal's density of states (DOS). This led to artificially long electronic tunnelling lifetimes and incorrect PDOS reconstruction.
  • The Tier1-s basis set successfully reconstructed the PDOS and provided lifetimes consistent with reference Time-Dependent Perturbation Theory (TDPT) calculations.
• Chemisorption Functions ($\Delta(\epsilon)$):
  • H/Al(111): The chemisorption function is nearly constant over a wide energy range, validating the WBL approximation for this $p$-band metal.
  • H/Cu(111) & H/Pt(111): The chemisorption function shows strong energy dependence, particularly near the Fermi level and above. This is attributed to the varying coupling strengths with $d$-states and the non-constant DOS.
• Shift Functions ($H(\epsilon)$):
  • For H/Al(111), the shift function is small (consistent with WBL).
  • For H/Cu(111) and H/Pt(111), the shift function exhibits large fluctuations and non-zero values, indicating significant deviations from the WBL.
• Lifetimes:
  • Electronic Tunnelling ($\tau_{el}$): The POD-based lifetimes agree well with literature values for H/Al and H/Cu. For H/Pt, discrepancies were traced back to the basis set shifting the state out of the DOS.
  • Vibrational Lifetimes ($\tau_{vib}$): POD-based lifetimes (using the full energy-dependent function) agreed qualitatively and quantitatively (for Cu) with TDPT reference calculations. The WBL approximation overestimated lifetimes for Pt and Al, highlighting the necessity of the full energy-dependent approach for accurate dynamics.

5. Significance

This work establishes a rigorous, first-principles framework for constructing Newns-Anderson Hamiltonians that goes beyond the widely used but often inaccurate Wideband Limit approximation.

• For Dynamics Simulations: The constructed NAHs can be directly used in mixed quantum-classical dynamics (e.g., Surface Hopping) or Hierarchical Equations of Motion (HEOM) to simulate non-adiabatic effects with higher fidelity, particularly for transition metal catalysts.
• Methodological Insight: The study highlights that "black-box" application of projection methods to large basis sets can yield unphysical results. It provides specific guidelines (using Tier1-s basis sets and state-following) for researchers aiming to model chemisorption.
• Physical Understanding: By quantifying the energy dependence of couplings and shift functions, the paper clarifies why the WBL fails for transition metals, offering a more nuanced understanding of molecule-surface charge transfer and vibrational relaxation.

In conclusion, the authors provide a validated pathway to move from standard DFT calculations to accurate, energy-dependent Newns-Anderson models, enabling more realistic simulations of surface chemistry and catalysis.