On the breakdown of the Born-Oppenheimer approximation in LiH and LiD

This study demonstrates that incorporating quantum nuclear effects into ab-initio density functional calculations significantly improves the agreement between theoretical and experimental electron densities in crystalline LiH and LiD, revealing a breakdown of the strict Born-Oppenheimer approximation and notable temperature dependence in these light-element solids.

The Gist

Here is an explanation of the paper, translated from complex physics jargon into a story you can visualize.

The Big Idea: The "Heavy" and the "Light" Dance

Imagine a dance floor where two types of dancers are moving: Electrons (tiny, hyperactive fireflies) and Nuclei (heavy, slow-moving elephants).

For nearly a century, physicists have used a rule called the Born-Oppenheimer Approximation to understand how these dancers interact. The rule is simple: "Treat the elephants as if they are statues."

Because elephants are so heavy compared to fireflies, the old rule says: Let's freeze the elephants in place, calculate exactly where the fireflies are dancing around them, and then move the elephants a tiny bit and do it again. It's a brilliant shortcut that makes the math possible. For most materials, this works perfectly.

But this paper asks: What if the elephants aren't statues? What if they are actually wobbling, shaking, and dancing too?

The Problem: The "Wobbling" Nuclei

The author, Ville Harkonen, looked at a crystal made of Lithium and Hydrogen (LiH) and its heavier cousin, Lithium and Deuterium (LiD). Hydrogen is the lightest element in the universe. Because it's so light, it doesn't act like a solid statue; it acts like a fuzzy cloud of probability. It jitters around its spot due to quantum mechanics.

The old "statue" rule (Strict Born-Oppenheimer) ignores this jitter. It assumes the nucleus is exactly where it's supposed to be.

The Discovery:
When Harkonen stopped treating the nuclei as statues and let them "wobble" (quantum nuclear effects), the results changed dramatically.

• The Old View: The electron density (where the fireflies are) looked like a smooth hill right next to the nucleus.
• The New View: When the nucleus wobbles, the electron density gets "smeared out." It's like taking a photo of a spinning fan with a slow shutter speed; the blades look blurry. The electron density near the nucleus dropped by huge amounts (up to 80% in some spots!) compared to the old prediction.

The Two Methods: The "Taylor Series" vs. The "Gaussian Blur"

To figure this out, the author used two different mathematical "cameras" to take a picture of the wobbling nuclei:

1. The Polynomial Approach (The "Taylor Expansion"):

   • Analogy: Imagine trying to describe a wobbly line by drawing a straight line, then adding a curve, then adding a wiggle. You are building the shape piece by piece.
   • Result: This worked well for the "fake" electrons (pseudo-electrons) used in the computer, but it got messy and produced weird "double-hump" shapes near the real nuclei. It's like trying to draw a perfect circle using only straight Lego bricks; it gets jagged at the edges.

2. The Gaussian Approach (The "Blur"):

   • Analogy: Instead of building the shape piece by piece, you just take the sharp image and apply a "motion blur" filter to it, simulating the wobble.
   • Result: This gave a much smoother, more realistic picture of how the electrons look when the nucleus is shaking.

The Temperature Twist: The "Hot Dance Floor"

The paper also looked at what happens when you heat the crystal up.

• Cold (0 Kelvin): The nuclei wobble a little (quantum jitter).
• Hot (300 Kelvin): The nuclei shake violently.

The Surprise: The author found that the heavier nuclei (Lithium and Deuterium) actually showed a stronger change in electron density when heated than the lightest one (Hydrogen).

• Why? Think of a heavy swing vs. a light swing. If you push them both, the heavy one takes longer to stop and has a different rhythm. As the temperature rises, the heavier atoms vibrate in a way that smears out the electron cloud even more significantly than expected.

The "Smoking Gun": Comparing with Reality

The author compared his new, "wobbling" calculations with old X-ray experiments done 30 years ago.

• The Old Theory (Statues): The predictions didn't match the X-ray photos very well. The math said the electrons were in one place, but the X-rays showed them elsewhere.
• The New Theory (Wobbling): When the author included the nuclear wobble, the math matched the X-ray photos much better.

This proves that the "Statue Rule" is broken for these materials. The nuclei are not statues; they are active participants in the dance.

Why Should You Care?

1. Superconductors: This is huge for materials that conduct electricity with zero resistance (superconductors), especially those involving hydrogen (hydrides). If we want to build better superconductors, we can't ignore the fact that the atoms are shaking.
2. It's Not Just Hydrogen: The paper shows that even heavier elements like Lithium (which is 7 times heavier than Hydrogen) show these effects. This means we might need to rethink how we calculate the properties of many other common materials, not just the lightest ones.
3. The Future of Computing: It tells us that to get truly accurate predictions for new materials, we have to stop treating atoms as static points and start treating them as quantum clouds.

The Bottom Line

For a long time, we thought we could freeze the heavy atoms in place to understand how electrons behave. This paper says: "Nope, the heavy atoms are dancing, and their dance moves change the electron cloud significantly."

By letting the atoms "wobble" in the math, the author finally got the numbers to match reality, solving a mystery that has puzzled scientists for decades.

Technical Summary

Here is a detailed technical summary of the paper "On the breakdown of the Born-Oppenheimer approximation in LiH and LiD" by Ville J. Harkonen.

1. Problem Statement

The Born-Oppenheimer (BO) approximation is the cornerstone of computational condensed matter physics, separating the fast-moving electrons from the slow-moving nuclei based on their mass difference. The strict BO approximation (or crude BO) treats nuclear positions as fixed parameters, implying that nuclear mass has no effect on the electronic structure.

However, experimental X-ray diffraction data from 30 years ago suggested a breakdown of this approximation in crystalline Lithium Hydride (LiH) and Lithium Deuteride (LiD). Specifically, experiments showed differences in electron density between LiH and LiD (isotopes) and deviations from theoretical predictions near nuclear equilibrium positions. While previous studies addressed quantum nuclear effects in hydrogen-rich systems, there were no ab-initio computational studies that explicitly calculated electron density beyond the strict BO approximation in solids to explain these experimental discrepancies.

2. Methodology

The author employs Density Functional Theory (DFT) using the Quantum Espresso (QE) package to compute electron densities beyond the strict BO limit. Two distinct theoretical approaches are developed to account for the quantum mechanical nature of the nuclei (nuclear zero-point motion and thermal fluctuations):

A. Theoretical Framework

The study utilizes the exact factorization of the electron-nuclear wavefunction. The total electron density $\langle n(y) \rangle$ is expressed as an integral over the nuclear density $\rho(R)$:
$$\langle n(y) \rangle = \int dR \, \rho(R) \, n_R(y)$$
where $n_R(y)$ is the conditional electron density for a fixed nuclear configuration $R$.

B. Approach 1: Harmonic (Polynomial) Approximation

• Method: The conditional density $n_R(y)$ is expanded in a Taylor series around the equilibrium nuclear positions $x$ up to the second order.
• Assumption: The system is treated within the harmonic approximation (valid for weak anharmonicity).
• Calculation: The first-order term vanishes due to symmetry. The correction is dominated by the second-order term involving the covariance of nuclear displacements $\langle \hat{u}_s \hat{u}_{s'} \rangle$.
• Formula:
  $$\langle n(y) \rangle \approx n_x(y) + \frac{1}{2} \sum_{s,s'} \frac{\partial^2 n_x(y)}{\partial x_s \partial x_{s'}} \langle \hat{u}_s \hat{u}_{s'} \rangle$$
• Implementation: Calculated using finite central differences (0.5% displacement) on supercells to obtain derivatives, combined with phonon frequencies from Density Functional Perturbation Theory (DFPT).

C. Approach 2: Gaussian (Convolution) Approximation

• Method: The strict BO electron density is fitted to a sum of Gaussian basis functions. The integral over nuclear positions is then solved analytically, effectively convolving the electronic density with the nuclear probability distribution.
• Advantage: Avoids the truncation errors of the Taylor expansion and handles the "cusps" in all-electron densities near nuclei more robustly.
• Implementation: Uses a closed-form expression involving Hermite polynomials to integrate the Gaussian nuclear density against the fitted electronic density.

D. Computational Details

• System: Crystalline LiH and LiD (Fm$\bar{3}$m structure).
• Functionals: PBE and LDA exchange-correlation functionals.
• Basis: Projector Augmented-Wave (PAW) method.
• Parameters: Plane-wave cutoffs of 100 Ry (kinetic) and 1200 Ry (charge density); $16\times16\times16$k-point grid for electronic structure;$8\times8\times8$ q-point grid for phonons.
• Temperatures: Calculations performed at 0 K and 300 K (and 93 K for structure factor comparison).

3. Key Contributions

1. First Ab-Initio Calculation: This is the first study to compute the full electron density in crystalline LiH and LiD beyond the strict BO approximation using DFT.
2. Dual Methodology: Development and comparison of two distinct approaches (Polynomial vs. Gaussian) to handle quantum nuclear effects in electron density, highlighting the limitations of Taylor expansions near nuclear cusps.
3. Resolution of Experimental Discrepancy: The study provides a theoretical basis for the long-standing experimental observation that LiH and LiD have different electron densities, attributing it to quantum nuclear effects rather than a fundamental failure of the BO approximation itself, but rather the strict version of it.
4. Temperature Dependence: Quantification of how electron density changes with temperature due to nuclear delocalization.

4. Key Results

A. Breakdown of Strict BO Approximation

• Magnitude of Effect: There is a significant reduction in electron density at the nuclear equilibrium positions when quantum nuclear effects are included.
  • Hydrogen (H): ~76% reduction at 0 K and ~82% at 300 K.
  • Lithium (Li): ~47% reduction at 0 K and ~75% at 300 K.
  • Deuterium (D): ~53% reduction at 0 K and ~62% at 300 K.
• Functional Shape Change:
  • At 0 K, the density around H/D nuclei changes from unimodal (single peak) to bimodal (double peak) due to nuclear smearing.
  • At 300 K, this bimodal shape extends to Lithium nuclei as well.
• Mechanism: The primary driver is the local position uncertainty of the nuclei (zero-point motion and thermal vibration), which "smears" the electron density. This mechanism is similar to that observed in high-pressure hydrides like YH$_6$.

B. Comparison with Experiment

• Radial Densities: The computed "full BO" densities (including quantum nuclei) show significantly better agreement with experimental X-ray diffraction data (Ref. [12]) than strict BO densities.
  • For Lithium, the full BO density matches experimental values well in the range $r \approx 1.0 - 2.2$ au.
  • For Hydrogen, the largest deviation (~20% at $r \approx 1.1$ au) remains between theory and experiment, but the strict BO approximation is far worse.
• Structure Factors:
  • At 93 K, the full BO structure factors agree with experiment within ~1.2% on average (excluding the (1,1,0) reflection).
  • The strict BO approximation shows large errors (~42% average deviation), particularly increasing with higher Miller indices.
  • The (1,1,0) reflection shows a ~18% discrepancy in full BO theory, which the author notes is an open issue but does not account for the radial density differences.

C. Isotope and Temperature Effects

• Isotope Effect: LiH and LiD exhibit different electron densities in the full BO approximation, whereas they are identical in the strict BO approximation. This confirms that the experimental observation of isotope differences is a signature of quantum nuclear effects.
• Temperature: The effect is stronger at higher temperatures. Interestingly, the temperature dependence is stronger for heavier nuclei (Li and D) than for Hydrogen. This is because heavier nuclei have lower vibrational frequencies, which are more easily excited thermally, leading to larger mean square displacements.

5. Significance and Conclusion

• Validity of Strict BO: The strict BO approximation fails significantly in solids containing light elements, even at normal pressures and room temperature. It is not just a high-pressure phenomenon.
• Beyond Hydrogen: The study demonstrates that quantum nuclear effects are significant for Lithium (7x heavier than H), suggesting that similar breakdowns of the strict BO approximation likely occur in other materials containing light elements (e.g., Carbon, Boron).
• Implications: These findings are crucial for understanding:
  • High-temperature superconductivity in hydrides.
  • Hydrogen storage materials.
  • Accurate modeling of electron density in materials science, where ignoring nuclear quantum effects leads to substantial errors in predicted properties.
• Methodological Insight: The study highlights that while polynomial expansions (Taylor series) are useful for pseudo-electron densities, they can produce non-physical results (negative densities) for all-electron densities near nuclei due to the cusp behavior. The Gaussian convolution approach is more robust for all-electron descriptions.

In summary, the paper establishes that quantum nuclear effects are non-negligible in determining the electronic structure of light-element solids, necessitating a move beyond the strict Born-Oppenheimer approximation for accurate theoretical predictions.