Optical properties of a diamond NV color center from capped embedded multiconfigurational correlated wavefunction theory

This paper demonstrates that capped density functional embedding theory combined with multiconfigurational perturbation theory accurately reproduces the optical excitation energies of diamond NV centers with minimal errors, while offering robust predictions that are largely independent of both embedded cluster and supercell sizes.

The Gist

The Big Picture: Finding the Perfect "Quantum Lightbulb"

Imagine you are trying to build a super-fast computer that uses the laws of quantum physics (the weird rules that govern tiny particles) instead of regular electricity. To do this, you need tiny, stable "bits" of information called qubits.

One of the most promising candidates for these qubits is a tiny flaw inside a diamond. Specifically, it's a spot where a Nitrogen atom has swapped places with a Carbon atom, and right next to it, a Carbon atom is missing entirely. This is called an NV center (Nitrogen-Vacancy).

Think of this defect like a tiny, glowing "lightbulb" trapped inside a giant, clear diamond. When you shine a laser on it, it glows. Scientists want to use this glow to read and write information. But to build a reliable computer, they need to know exactly how this lightbulb works, how much energy it takes to turn it on, and what color it glows.

The Problem: The Diamond is Too Big to Simulate

To understand how this "lightbulb" works, scientists use powerful computers to run simulations. They try to model the diamond atom-by-atom.

However, there is a major problem: Diamonds are huge.

• If you try to simulate the whole diamond, the computer gets overwhelmed. It's like trying to calculate the traffic flow of an entire country just to understand how one car moves.
• If you try to simulate just the tiny "lightbulb" (the defect) by cutting it out of the diamond, you run into a different problem. The atoms at the edge of your cut piece are "dangling." They are missing neighbors, so they act weirdly and give you the wrong answer. It's like trying to study a fish by cutting it out of the water; it doesn't behave the same way in a bowl as it does in the ocean.

The Solution: The "Smart Bubble" (Capped-DFET)

The author of this paper, John Mark P. Martirez, developed a clever new method called capped-DFET (capped Density Functional Embedding Theory).

Here is how it works, using an analogy:

Imagine you want to study a specific conversation happening in a crowded, noisy stadium.

1. The Old Way (Full Simulation): You try to record every single person in the stadium. The data is too massive, and your computer crashes.
2. The "Cut-Out" Way: You cut a small square of the stadium floor out and put it in a quiet room. But now, the people at the edge of your square are confused because they can't see the rest of the crowd. They act strangely.
3. The "Smart Bubble" Way (This Paper's Method): You take a small group of people (the defect and its immediate neighbors) and put them in a room. But instead of just leaving the edges open, you put up a smart, invisible force field around them.
   • This force field mimics the pressure, noise, and atmosphere of the rest of the stadium.
   • It tells the people in the room, "Hey, the rest of the stadium is still there, acting normally, so you don't need to worry about the edges."
   • Crucially, this force field is "capped." If a person in your group reaches out to grab a neighbor who isn't there, a "placeholder" hand (a capping atom) grabs them instead, so they don't feel lonely or confused.

What Did They Discover?

Using this "Smart Bubble" method, the author simulated the diamond defect with incredible accuracy. Here are the key takeaways:

• It's Accurate: The method predicted the energy levels of the defect (how much energy it takes to make it glow) with an error of less than 0.1 electron-volts. That's like hitting a bullseye on a dartboard from across the room.
• It's Efficient: You don't need a massive diamond to get a good answer. You only need a tiny cluster of about 40 atoms (the defect plus its immediate neighbors). The "Smart Bubble" does the rest of the work.
• It Solves the "Charged" Problem: This defect has an extra electric charge. In physics simulations, charged objects usually repel each other in weird ways if you don't have a huge container to hold them. This method cleverly avoids those messy long-range electrical interactions, meaning the size of the "container" (the supercell) doesn't mess up the results.
• It Handles "Spin": The defect has a property called "spin" (like a tiny magnet). It can be in different states (singlet or triplet). The method correctly figured out the complex quantum dance between these states, which older, simpler methods failed to do.

Why Does This Matter?

This paper is a "how-to" guide for the future of quantum computing.

Before this, scientists had to choose between accuracy (simulating the whole diamond, which was too hard) and simplicity (simulating just the defect, which was inaccurate).

This new method gives us the best of both worlds. It allows scientists to:

1. Design new types of quantum bits without needing a supercomputer the size of a building.
2. Predict how new materials will behave before they even build them in a lab.
3. Speed up the development of quantum sensors, secure communication, and quantum computers.

In short: The author built a "virtual microscope" that lets us zoom in on a tiny diamond defect and see exactly how it works, without needing to simulate the entire diamond. It's a faster, cheaper, and more accurate way to build the quantum computers of tomorrow.

Technical Summary

1. Problem Statement

Diamond defects, particularly the negatively charged nitrogen-vacancy center ($NV^-$), are leading candidates for solid-state qubits due to their long quantum coherence times and optical addressability. However, accurately modeling their electronic and optical properties is computationally challenging for several reasons:

• Multiconfigurational Nature: The defect states involve strong static and dynamic electron correlation, particularly in spin-singlet states, which single-determinant methods like standard Density Functional Theory (DFT) fail to describe accurately.
• System Size vs. Accuracy Trade-off: Accurate wavefunction methods (e.g., CASSCF/NEVPT2) are too expensive for large periodic supercells required to isolate defects and minimize spurious interactions between periodic images.
• Charged Defect Artifacts: Periodic Boundary Condition (PBC) calculations for charged defects suffer from slowly converging Coulomb interactions (monopole-monopole) that scale as $1/L$, requiring extremely large supercells for convergence.
• Limitations of Existing Embedding: Many quantum embedding methods still rely on large supercells or complex orbital localization schemes that do not fully eliminate the dependence on system size or the computational cost of large clusters.

2. Methodology

The authors propose and validate a Capped Density Functional Embedding Theory (capped-DFET) combined with Multireference Second-Order Perturbation Theory (NEVPT2).

• System Partitioning: The system is divided into a small "cluster" containing the defect and its nearest neighbors, and an "environment" representing the bulk crystal.
• Capping Scheme: To handle the broken covalent bonds at the cluster boundaries, the method uses "capping" atoms (F, O, B) to satisfy valency. Crucially, an auxiliary fragment composed of these capping atoms is introduced to project out their densities. This ensures the embedding potential remains local and avoids the need for orbital localization or projector orbitals.
• Embedding Potential ($V_{emb}$):
  • The environment is treated with DFT (using the HSE06 hybrid functional for accuracy).
  • An optimized effective potential ($V_{emb}$) is derived variationally to ensure the sum of the cluster and environment densities reproduces the full system density.
  • This potential captures the dielectric response of the environment without explicitly including the slowly converging long-range Coulomb interactions of the charged defect.
• Wavefunction Theory:
  • The embedded cluster is treated with State-Averaged CASSCF (SA-CASSCF) to capture static correlation (degeneracies).
  • NEVPT2 is applied to capture dynamic correlation.
  • The active space (AS) was systematically tested, with (6e, 8o) (6 electrons in 8 orbitals) identified as the optimal balance between accuracy and cost.
• Validation: The method was tested on the $NV^-$ center in diamond using supercells of varying sizes (64, 128, and 160 atoms) and cluster sizes (15 to 36 carbon atoms).

3. Key Contributions

• First Application to Condensed Matter Defects: This is the first successful application of capped-DFET to characterize localized defects in a solid-state environment.
• Decoupling from Supercell Size: The study demonstrates that the embedding potential, once optimized, effectively mimics the infinite bulk limit. Consequently, the calculated excitation energies are independent of the supercell size used to generate the potential, even for relatively small supercells (63 atoms).
• Elimination of Coulomb Artifacts: By performing the high-level wavefunction calculation on a non-periodic, embedded cluster, the method completely avoids the $1/L$ convergence issues associated with charged defects in PBC calculations.
• Superiority over Bare Clusters: The authors show that bare clusters (without embedding) require prohibitively large sizes to converge to the same accuracy as the embedded small clusters, highlighting the efficiency of the embedding approach.

4. Key Results

• Accuracy: The method reproduces experimental vertical excitation energies (VEE) for both triplet and singlet states of the $NV^-$ center with errors < 0.1 eV.
  • Triplet ($^3A_2 \to ^3E$): Calculated VEE $\approx$ 2.25 eV (Exp: 2.18 eV).
  • Singlet ($^1E \to ^1A_1$): Calculated VEE $\approx$ 1.28 eV (Exp: 1.26 eV).
  • Intersystem Crossing (ISC): The energy difference between the excited singlet and triplet states was also accurately predicted.
• Active Space Sensitivity: Including the nitrogen lone pair orbital in the active space (moving from 4e/6o to 6e/8o) was critical for accuracy, reducing errors by up to 0.07 eV.
• Cluster Size Independence:
  • Embedded Clusters: VEEs remained stable (variation < 0.05 eV) when the cluster size increased from 15 to 21 carbon atoms, regardless of whether the embedding potential came from a 64, 128, or 160-atom supercell.
  • Bare Clusters: Bare clusters showed significant size dependence and failed to converge to the experimental values even at larger sizes (36 C atoms) without embedding.
• Comparison to Other Methods: The capped-DFET/NEVPT2 approach outperformed or matched more computationally expensive methods (GW+BSE, CI-cRPA, DMET) that required supercells of 250–500+ atoms to achieve similar accuracy.

5. Significance

This work establishes capped-DFET as a robust, high-accuracy, and computationally efficient framework for simulating quantum defects in solids.

• Scalability: It enables the use of medium-sized supercells to predict properties of defects in the dilute limit, bypassing the need for massive supercells that are currently prohibitive for high-level wavefunction theories.
• Reliability: The method is free from the slowly converging Coulomb errors that plague charged defect simulations in periodic systems.
• Future Impact: This approach paves the way for the ab initio design and characterization of new defect-based qubits and quantum sensors in diamond and other wide-bandgap materials, moving beyond the limitations of DFT and single-reference theories.