Born–Oppenheimer Berry Phase

The Born–Oppenheimer approximation separates slow nuclear motion from fast electronic motion. In its geometric form, the nuclear coordinates play the role of parameters, and the electronic eigenstates form a bundle over nuclear configuration space. Transporting the nuclei around a loop can then give a Berry phase.

This page owns the geometric mechanism. Born–Oppenheimer Approximation as Scale Separation owns the mass expansion, exact channel equations, potential-energy surfaces, and nonadiabatic validity diagnostics. Nonadiabatic Coupling owns off-diagonal molecular transfer, surface hopping, vibronic models, and the distinction between passage through a coupling region and adiabatic transport around it. Conical Intersections owns the local degeneracy geometry, seam dimension, branching plane, MECI interpretation, and molecular computation. The key idea here is:

$$\text{slow nuclear loop} \quad \longrightarrow \quad \text{electronic eigenstate holonomy} \quad \longrightarrow \quad \text{molecular Berry phase}.$$

Electronic Problem at Fixed Nuclear Geometry

Let $R$ denote all nuclear coordinates and $r$ all electronic coordinates. After center-of-mass separation and ordinary nonrelativistic approximations, the molecular Hamiltonian is schematically

$$H_{\mathrm{mol}} = T_N + H_e(r;R),$$

where $T_N$ is the nuclear kinetic energy and $H_e(r;R)$ is the electronic Hamiltonian with the nuclei held fixed.

At each nuclear geometry, solve the electronic eigenvalue problem

$$H_e(R)\lvert\phi_a(R)\rangle = E_a(R)\lvert\phi_a(R)\rangle.$$

Here $\lvert\phi_a(R)\rangle$ is a state in the electronic Hilbert space, and $R$ is a parameter. The electronic energy $E_a(R)$ becomes a potential-energy surface for nuclear motion.

Born–Oppenheimer Product and Its Limitation

The simplest one-surface Born–Oppenheimer ansatz is

$$\Psi(r,R) = \chi_a(R)\phi_a(r;R).$$

Here $\chi_a(R)$ is the nuclear wavefunction on electronic surface $a$. This looks like a product, but $T_N$ differentiates with respect to $R$. Therefore it also differentiates the electronic state $\phi_a(r;R)$. Those derivatives produce two kinds of terms:

• off-diagonal derivative couplings that connect different electronic surfaces;
• diagonal geometric terms that act like a gauge connection on the nuclear wavefunction.

The first kind controls nonadiabatic transitions. The second kind is the Born–Oppenheimer Berry connection.

Molecular Berry Connection

For one isolated electronic state, define

$$\mathbf A_a(R) = i \langle\phi_a(R)|\nabla_R\phi_a(R)\rangle_e.$$

The subscript $e$ means that the inner product is over electronic coordinates only. This is exactly the Berry connection, with nuclear geometry $R$ serving as parameter space.

Under an electronic phase convention change

$$\lvert\phi_a(R)\rangle \mapsto e^{i\lambda(R)} \lvert\phi_a(R)\rangle,$$

the connection transforms as

$$\mathbf A_a \mapsto \mathbf A_a-\nabla_R\lambda.$$

The total molecular wavefunction must remain unchanged, so the nuclear amplitude transforms oppositely:

$$\chi_a(R) \mapsto e^{-i\lambda(R)} \chi_a(R).$$

This is why the Berry connection enters the effective nuclear equation as a gauge potential.

Effective Nuclear Momentum

In the one-surface adiabatic approximation, the nuclear kinetic energy becomes a covariant kinetic energy. Suppressing tensor and mass-index details, the ordinary nuclear momentum is replaced by

$$-i\hbar\nabla_R \quad \longmapsto \quad -i\hbar\nabla_R-\hbar\mathbf A_a(R).$$

Thus the effective nuclear Hamiltonian has the schematic form

$$H_{\mathrm{nuc}}^{(a)} = \sum_\alpha \frac{ \left( -i\hbar\nabla_{R_\alpha} -\hbar\mathbf A_{a,\alpha} \right)^2 }{2M_\alpha} + E_a(R) + \Phi_a(R).$$

The term $\Phi_a(R)$ is the diagonal Born–Huang scalar correction. One common expression is

$$\Phi_a = \sum_\alpha \frac{\hbar^2}{2M_\alpha} \left( \langle \partial_\alpha\phi_a| \partial_\alpha\phi_a\rangle_e - \left| \langle\phi_a|\partial_\alpha\phi_a\rangle_e \right|^2 \right),$$

where $\partial_\alpha$ differentiates with respect to $R_\alpha$. The vector connection $\mathbf A_a$ is the part responsible for Berry phase around nuclear loops.

Molecular Geometric Phase

For a closed nuclear path $C$ on one adiabatic surface,

$$\gamma_a[C] = \oint_C \mathbf A_a(R)\cdot dR.$$

The phase factor $e^{i\gamma_a[C]}$ is gauge invariant modulo $2\pi$. Physically it can affect nuclear interference, rovibrational level patterns, tunneling paths, and selection rules when the molecular configuration space has nontrivial loops around degeneracies.

The phase does not require that the electronic state be complex everywhere. In many molecular problems without magnetic fields or spin–orbit coupling, electronic wavefunctions can be chosen real locally. Then $\mathbf A_a=0$ in that patch. A Berry phase can still appear globally if the electronic eigenvector changes sign after transport around a loop.

Conical Intersections

A conical intersection is a degeneracy between electronic potential-energy surfaces as a function of nuclear geometry. Near a simple two-state conical intersection, the electronic Hamiltonian can often be modeled locally by a two-level form such as

$$H_{\mathrm{eff}}(X,Y) = X\sigma_z+Y\sigma_x.$$

The degeneracy is at $X=Y=0$. A loop around the origin keeps the electronic levels nondegenerate along the path but encloses the degeneracy. In the two-level geometry, that loop corresponds to a great-circle loop around the Bloch sphere, enclosing solid angle $2\pi$.

The Berry phase for one adiabatic electronic state is therefore

$$\gamma = \pi \quad \operatorname{mod}2\pi,$$

up to sign convention. Equivalently, in a real local gauge the electronic eigenvector returns with a minus sign after one circuit around the conical intersection.

The total molecular wavefunction must be single-valued. If the electronic factor changes sign, the nuclear factor must acquire a compensating sign or boundary condition. This is the molecular geometric phase.

Relation to Aharonov–Bohm Physics

The nuclear Berry connection near a conical intersection acts like a gauge flux in nuclear configuration space. In the simplest conical-intersection case, the phase around the intersection is $\pi$, like an effective half-flux quantum for the nuclear motion.

This is why molecular Berry phases are sometimes compared to the Aharonov–Bohm Effect. The comparison is structural:

• the nuclei move in a space with an excluded or singular region;
• a loop around that region carries a phase;
• locally the connection may be removable in patches;
• globally the holonomy affects interference and boundary conditions.

The physical origin is different. The molecular phase comes from electronic eigenstate transport over nuclear-coordinate space, not from an external electromagnetic vector potential.

Off-Diagonal Nonadiabatic Couplings

The full expansion

$$\Psi(r,R) = \sum_a \chi_a(R)\phi_a(r;R)$$

contains derivative couplings

$$\mathbf d_{ab}(R) = \langle\phi_a(R)|\nabla_R\phi_b(R)\rangle_e.$$

For $a\ne b$, these couplings drive transitions between electronic surfaces. For $a=b$, the imaginary diagonal part is the Berry connection:

$$\mathbf A_a = i\mathbf d_{aa}.$$

Near a conical intersection the electronic gap closes, so off-diagonal nonadiabatic effects can become large. This is not a contradiction. The Berry phase is an adiabatic holonomy for loops that avoid the degeneracy; actual molecular dynamics near the degeneracy may require multi-surface nonadiabatic treatment.

Real Gauges and Sign Changes

When the electronic Hamiltonian is real and the electronic level is nondegenerate on a simply connected patch, one can often choose $\phi_a(R)$ real. Then

$$\langle\phi_a|\nabla_R\phi_a\rangle_e=0,$$

so the local Berry connection vanishes.

Around a loop enclosing a conical intersection, however, a real eigenvector can return as

$$\phi_a(R(T)) = -\phi_a(R(0)).$$

This sign change is a Berry phase $\pi$. It cannot be removed by one globally continuous single-valued real gauge on the punctured parameter region. The local connection is zero in patches; the global patching carries the phase.

Boundaries of This Page

This page gives the geometric mechanism behind molecular Berry phases. It does not attempt to replace:

• molecular electronic-structure theory;
• the full Born–Oppenheimer approximation and its validity tests;
• nonadiabatic molecular dynamics;
• conical-intersection chemistry and photochemistry;
• rovibrational spectroscopy with geometric phase corrections.

Those topics need their own canonical pages. Here, the canonical content is the bridge from Berry connection to nuclear-coordinate dynamics.

Common Mistakes

• Treating the Born–Oppenheimer product ansatz as if nuclear derivatives never act on electronic states.
• Forgetting that the electronic phase convention forces an opposite gauge transformation on the nuclear amplitude.
• Assuming a locally real electronic wavefunction means the molecular Berry phase must vanish globally.
• Applying a one-surface adiabatic Berry phase through the conical intersection itself, where the electronic gap closes.
• Confusing nonadiabatic transitions with the diagonal Berry connection; both arise from derivative couplings, but they play different roles.
• Saying “conical intersection gives Berry phase” without specifying the loop and the electronic surface.
• Treating molecular geometric phase as a small correction in all settings; in loops around conical intersections it can impose a sign-changing boundary condition.

Cross-Links

• Adiabatic Theorem Reminder
• Berry Phase
• Berry Connection
• Berry Curvature
• Holonomy
• Parallel Transport
• Berry Phase for Spin-1/2
• Dirac Monopole Preview
• Aharonov–Bohm Effect
• Molecular Rotation Applications
• Adiabatic Theorem
• Adiabatic Approximation as a Method
• Nonadiabatic Coupling
• Conical Intersections
• Berry Connection as a Mathematical Object
• U(1) Bundles and Quantum Phase
• Chern Numbers

References

• M. Born and R. Oppenheimer, “Zur Quantentheorie der Molekeln,” Annalen der Physik 389, 457-484, 1927.
• H. C. Longuet-Higgins, U. Öpik, M. H. L. Pryce, and R. A. Sack, “Studies of the Jahn–Teller effect II. The dynamical problem,” Proceedings of the Royal Society A 244, 1-16, 1958.
• C. A. Mead and D. G. Truhlar, “On the determination of Born–Oppenheimer nuclear motion wave functions including complications due to conical intersections and identical nuclei,” Journal of Chemical Physics 70, 2284-2296, 1979.
• M. V. Berry, “Quantal phase factors accompanying adiabatic changes,” Proceedings of the Royal Society A 392, 45-57, 1984.
• C. A. Mead, “The geometric phase in molecular systems,” Reviews of Modern Physics 64, 51-85, 1992.
• D. R. Yarkony, “Diabolical conical intersections,” Reviews of Modern Physics 68, 985-1013, 1996.

Exercises

1. Show the gauge covariance of the effective nuclear derivative.

Let $\phi_a\mapsto e^{i\lambda}\phi_a$, $\chi_a\mapsto e^{-i\lambda}\chi_a$, and $\mathbf A_a\mapsto\mathbf A_a-\nabla\lambda$. Show that

$$D\chi_a = \left( \nabla_R-i\mathbf A_a \right)\chi_a$$

transforms as $D\chi_a\mapsto e^{-i\lambda}D\chi_a$.

Solution

Compute

$$\begin{aligned} D'\chi_a' &= \left( \nabla_R-i(\mathbf A_a-\nabla\lambda) \right) \left( e^{-i\lambda}\chi_a \right)\\ &= e^{-i\lambda} \left( \nabla_R\chi_a -i(\nabla\lambda)\chi_a -i\mathbf A_a\chi_a +i(\nabla\lambda)\chi_a \right)\\ &= e^{-i\lambda} \left( \nabla_R-i\mathbf A_a \right)\chi_a. \end{aligned}$$

Thus the covariant derivative transforms with the same phase as the nuclear amplitude.

2. Why can a real electronic eigenvector still produce a Berry phase $\pi$ around a conical intersection?

Solution

On a local patch, a real normalized eigenvector has

$$\langle\phi|\nabla_R\phi\rangle=0,$$

so the local Berry connection vanishes. Around a loop enclosing a conical intersection, however, the real eigenvector can return as $-\phi$ rather than $\phi$. That sign change is a global patching effect. It corresponds to a phase $\pi$ even though the local connection can be zero in patches.

3. Use the spin-$1/2$ solid-angle formula to estimate the Berry phase of the local two-level conical-intersection model.

The loop around $H_{\mathrm{eff}}(X,Y)=X\sigma_z+Y\sigma_x$ maps to a great circle enclosing solid angle $2\pi$. What is the Berry phase modulo $2\pi$?

Solution

For a two-level system, one common convention gives

$$\gamma = -\frac{\Omega}{2}.$$

With $\Omega=2\pi$,

$$\gamma=-\pi.$$

Modulo $2\pi$, this is equivalent to $\pi$. The sign depends on orientation and eigenstate convention; the physically important phase factor is $e^{i\gamma}=-1$.

4. Distinguish diagonal and off-diagonal derivative couplings.

What role is played by $\mathbf d_{aa}$, and what role is played by $\mathbf d_{ab}$ for $a\ne b$?

Solution

The diagonal coupling satisfies

$$\mathbf A_a=i\mathbf d_{aa},$$

and acts as a Berry connection on the nuclear wavefunction in the adiabatic one-surface approximation.

The off-diagonal couplings $\mathbf d_{ab}$ with $a\ne b$ connect different electronic surfaces. They drive nonadiabatic transitions and become especially important near small gaps or conical intersections.

5. Explain why the total molecular wavefunction remains single-valued even if the electronic factor changes sign around a loop.

Solution

The product form is

$$\Psi(r,R)=\chi_a(R)\phi_a(r;R).$$

If the electronic factor changes sign after a loop,

$$\phi_a\mapsto-\phi_a,$$

then the nuclear factor must also change sign,

$$\chi_a\mapsto-\chi_a,$$

so that

$$\Psi\mapsto(-\chi_a)(-\phi_a)=\Psi.$$

The Berry phase is therefore encoded as a boundary condition or sign structure in the nuclear wavefunction.