🔬 materials science

Exchange-Correlation Functionals in 2D Materials: Applications, Challenges, and Limitations

This review examines the critical role, challenges, and limitations of exchange-correlation functionals within density functional theory for accurately predicting the properties of two-dimensional materials, while evaluating advanced computational approaches and emerging strategies like machine learning to overcome current deficiencies in modeling quantum confinement and van der Waals interactions.

Original authors: Ahsan Javed, Mahvish Shaheen, Muhammad Shahbaz, M. Sufyan Ramzan, Rafi Ullah, Wei Jiang

Published 2026-07-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ahsan Javed, Mahvish Shaheen, Muhammad Shahbaz, M. Sufyan Ramzan, Rafi Ullah, Wei Jiang

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 are trying to build a house, but instead of bricks and wood, you are building with atoms. In the world of two-dimensional (2D) materials, these "houses" are incredibly thin—often just one atom thick. Because they are so thin, they behave very differently than the thick, chunky materials we are used to. They are like a sheet of paper compared to a brick wall; the paper bends, ripples, and reacts to the wind in ways the brick wall never could.

To predict how these atomic houses will behave, scientists use a powerful computer simulation tool called Density Functional Theory (DFT). Think of DFT as a sophisticated architectural blueprint generator. However, the blueprint isn't perfect; it relies on a specific set of rules called Exchange-Correlation (XC) Functionals.

You can think of an XC functional as the "rulebook" the computer uses to decide how the tiny electrons inside the material interact with each other. If the rulebook is wrong, the blueprint will be wrong, and the house might collapse or not work as intended.

This paper is a review of which "rulebooks" work best for these ultra-thin 2D materials and why some fail. Here is the breakdown:

1. The Problem: The "Flatland" Effect

In normal, thick materials (bulk), electrons are surrounded by neighbors on all sides, which helps them screen out (block) each other's electric charges. But in 2D materials, the electrons are stuck on a flat plane. They have nowhere to hide.

The Analogy: Imagine a crowded party in a large ballroom (bulk material). People can move around easily, and no one feels too much pressure from a single neighbor. Now, imagine that same party squeezed onto a narrow, flat stage (2D material). Everyone is right up against each other, and the interactions become intense and chaotic.
The Consequence: Standard rulebooks (like LDA or PBE) assume the "ballroom" scenario. When applied to the "flat stage," they fail. They underestimate the energy gaps (how hard it is to turn the material on), miss the weak glue holding layers together, and get the magnetic properties wrong.

2. The Ladder of Rulebooks (Jacob's Ladder)

The authors describe a hierarchy of rulebooks, climbing up from simple to complex, like a ladder:

Bottom Rungs (LDA/GGA): These are the simple, old-school rulebooks. They are fast to use but often get the details wrong for 2D materials. They might tell you a material is a conductor when it's actually an insulator.
Middle Rungs (Meta-GGA like SCAN): These are smarter rulebooks. They look at not just where the electrons are, but how fast they are moving. The paper finds that SCAN is excellent for predicting the physical size and shape of these 2D materials, often matching real-world experiments perfectly.
Top Rungs (Hybrids and GW/BSE): These are the "super-accurate" rulebooks. They mix in exact math to fix the errors of the simpler ones.
- Hybrids (like HSE06): These are great for figuring out the "band gap" (the switch that turns electricity on or off). However, the paper warns they aren't perfect; sometimes they accidentally create a gap in materials that should have none (like graphene), like drawing a wall where there should be an open door.
- GW + BSE: This is the "gold standard" for light and magnetism. It accounts for the fact that when an electron jumps, it leaves a "hole" behind, and they attract each other like a magnet. In 2D materials, this attraction is huge. Only these complex methods can accurately predict how the material will glow or absorb light.

3. The "One Size Does Not Fit All" Rule

A major takeaway from the paper is that there is no single "magic bullet" rulebook.

If you want to know the size of the material, use SCAN.
If you want to know the electrical switch (band gap), HSE06 is usually good, but be careful with special materials like graphene.
If you want to know how it glows (optical properties), you must use the expensive GW+BSE method.
If you want to know about magnetism, SCAN is surprisingly good, but you have to be careful with spin effects.

Trying to use one rulebook for everything is like trying to use a hammer to fix a watch, a car, and a house. You might get lucky, but you'll likely break something.

4. The New Helper: Machine Learning

Because the most accurate rulebooks (GW+BSE) take a massive amount of computer time—like trying to simulate a hurricane in a bathtub—scientists are turning to Machine Learning (ML).

The Analogy: Imagine you have a master chef who can cook a perfect meal but takes 10 hours. You train a robot (ML) to watch the chef. The robot learns the patterns and can cook a meal in seconds that tastes 90% as good.
The Catch: The robot is only as good as the chef it watched. If the chef (the reference data) made mistakes, the robot will too. The paper suggests using ML to speed up calculations, but we need to make sure the robot is trained on high-quality data, not just the "fast but inaccurate" rulebooks.

Summary

This paper is a guide for scientists navigating the tricky world of 2D materials. It tells us that because these materials are so thin and unique, we cannot rely on the old, simple rules. We need to choose our "rulebook" carefully based on what we are trying to measure. Sometimes we need a fast, simple tool; other times, we need a slow, expensive, high-precision tool. And while Artificial Intelligence is coming to the rescue to speed things up, it still needs to be guided by the most accurate physics we have.

Technical Summary: Exchange-Correlation Functionals in 2D Materials: Applications, Challenges, and Limitations

Problem Statement
The rapid expansion of two-dimensional (2D) materials research has revealed that their physical properties—ranging from structural stability to optoelectronic response—differ fundamentally from their bulk counterparts due to quantum confinement, anisotropic dielectric screening, and enhanced many-body interactions. While Density Functional Theory (DFT) is the standard computational tool for material discovery, conventional exchange-correlation (XC) functionals (such as LDA and GGA) often fail to accurately describe these unique 2D phenomena. Specifically, standard functionals suffer from self-interaction errors, a lack of derivative discontinuity, and an inability to capture long-range van der Waals (vdW) interactions and strong electron-hole correlations. These deficiencies lead to significant discrepancies between theoretical predictions and experimental measurements, particularly in band gaps, exciton binding energies, and interlayer binding forces. The paper addresses the critical need to evaluate, compare, and select appropriate XC functionals and advanced many-body methods to achieve "chemical accuracy" in modeling 2D systems.

Methodology
This work is a comprehensive topical review that systematically evaluates the performance of various XC functionals across different classes of 2D materials. The methodology involves:

Theoretical Framework Analysis: A detailed examination of the hierarchy of XC approximations, ranging from Local Density Approximation (LDA) and Generalized Gradient Approximation (GGA) to meta-GGA, hybrid functionals (e.g., HSE06), and many-body perturbation theory (GW and Bethe-Salpeter Equation, BSE).
Benchmarking Against Experiment: The authors compare theoretical predictions from various functionals against experimental data for key properties, including lattice constants, band gaps (direct vs. indirect), valence band splitting (VBS), charge density waves (CDW), magnetic transition temperatures, and thermal conductivity.
Case Studies: Specific material families are analyzed, including Transition Metal Dichalcogenides (TMDCs), hexagonal Boron Nitride (hBN), MXenes, and magnetic 2D materials (e.g., CrI3, FePS3).
Machine Learning Integration: The review assesses the emerging role of Machine Learning (ML) in correcting functional errors, generating transferable models, and accelerating high-throughput screening, while highlighting data requirements and transferability challenges.

Key Contributions and Results

Structural Properties: The review demonstrates that semi-local functionals (LDA/GGA) often yield inaccurate lattice constants and interlayer distances. Meta-GGA functionals, particularly SCAN and its regularized version r2SCAN, are identified as superior for predicting lattice constants and bond lengths across diverse 2D families, often matching experimental values within 1%. For vdW-dominated interlayer spacing, the combination of SCAN with non-local corrections (e.g., SCAN+rVV10) provides the most reliable results.
Electronic Properties:
- Band Gaps: Semi-local functionals consistently underestimate band gaps. Hybrid functionals like HSE06 offer significant improvements and are generally the best choice for predicting band gap magnitudes and nature (direct/indirect). However, HSE06 is noted to artificially open gaps in gapless materials like graphene and silicene.
- Quasiparticle and Excitonic Effects: For accurate quasiparticle gaps and exciton binding energies (EBE), the GW approximation combined with the Bethe-Salpeter Equation (BSE) is deemed essential. The review highlights that reduced dielectric screening in 2D materials leads to large EBEs (0.3–2.0 eV), which standard DFT cannot capture.
- Spin-Orbit Coupling (SOC): The inclusion of SOC is critical for heavy-element 2D materials (e.g., WSe2, WS2). The review notes that G0W0+SOC calculations align most closely with experimental valence band splitting values.
Magnetic and Thermal Properties:
- Magnetism: SCAN and r2SCAN functionals outperform GGA and GGA+U in predicting magnetic exchange constants and Néel/Curie temperatures for 2D magnets, largely due to reduced self-interaction errors.
- Thermal Conductivity: For thermal properties, PBE and PBEsol remain efficient choices, though the review notes that higher-order interatomic force constants (4th order) are highly sensitive to the choice of functional, potentially introducing large errors in predicted thermal conductivity if not carefully managed.
Machine Learning: The paper identifies ML as a powerful tool for accelerating material discovery. ML models trained on high-fidelity data (GW, BSE, or experimental) can predict properties like EBEs with high accuracy (e.g., Random Forest models achieving $R^2 = 0.84$ for EBEs). However, the review emphasizes that transferability remains a bottleneck, and ML models must be constrained by physical laws (e.g., density scaling, Lieb-Oxford bounds) to avoid unphysical extrapolations.

Significance and Claims
The authors claim that the central finding of this review is the non-universality of XC functionals in 2D materials. No single functional performs optimally across all properties; for instance, a functional that accurately predicts lattice constants may fail to capture excitonic spectra or magnetic transition temperatures.

Consequently, the paper proposes a property-dependent functional selection framework:

Use SCAN/r2SCAN (with vdW corrections) for structural and vdW-dominated properties.
Use HSE06 for general band gap trends, but resort to G0W0 for quantitative quasiparticle gaps.
Employ GW+BSE for optical properties and excitonic effects where strong electron-hole attraction is present.
Utilize SCAN/r2SCAN for magnetic exchange and transition temperatures.
Apply PBE/PBEsol for thermal transport, while acknowledging the sensitivity of higher-order force constants.

The review concludes that the future of predictive modeling for 2D materials lies not in a single "universal" functional, but in a hierarchical approach that balances physical fidelity with computational cost. It advocates for the development of dielectric-dependent and range-separated functionals, the construction of high-fidelity benchmark datasets, and the integration of uncertainty-aware machine learning to bridge the gap between low-cost screening and high-accuracy many-body methods.

1. The Problem: The "Flatland" Effect

2. The Ladder of Rulebooks (Jacob's Ladder)

3. The "One Size Does Not Fit All" Rule

4. The New Helper: Machine Learning

Summary

More like this