Engineered Conical Intersections

• Engineered conical intersections are deliberately created degeneracies in potential-energy surfaces or energy-momentum spectra that control nonadiabatic transitions and geometric phase effects.
• They are implemented across platforms—from molecular photochemistry and cavity QED to circuit QED and trapped ions—using reduced Hamiltonians and convex optimization to restore accurate conical topology.
• Key design variables such as laser intensity, spectral chirp, cavity coupling, and lattice symmetry enable tunable control over wave-packet branching, quantum yield, and transport phenomena.

Searching arXiv for recent and relevant papers on engineered conical intersections and related control platforms. arXiv search query: "engineered conical intersections control photochemistry cavity Rydberg ions" Engineered conical intersections are deliberately created, restored, or manipulated degeneracies between adiabatic states in a reduced coordinate space, with the aim of controlling nonadiabatic transfer, branching, coherence, geometric phase, and transport. In molecular photochemistry they appear in nuclear configuration space as crossings of electronic potential-energy surfaces; in periodic photonic and matter-wave media they appear as conical points in the energy-momentum spectrum governed by an effective Dirac–Weyl Hamiltonian. Recent work treats engineering in several distinct senses: enforcing the correct conical topology in electronic-structure theory, imposing light-induced or cavity-modified intersections, constructing synthetic intersections in circuit QED and trapped-ion platforms, and designing lattice symmetries that host Dirac- and Dirac-like cones (Rossi et al., 22 Apr 2026, Li et al., 2023, Wang et al., 2022, Leykam et al., 2016).

1. Topological structure and minimal models

A conical intersection in molecular physics is a point, or seam, in nuclear configuration space where two electronic states become degenerate and exchange character. Its local topology is defined by the branching space spanned by the gradient-difference vector $g$ and derivative-coupling vector $h$, and the characteristic local energy dependence is the double-cone form. In the two-state region discussed for CVX-DFT, the restored adiabatic states $\mathcal E_0(R)$ and $\mathcal E_1(R)$ cross conically along a one-dimensional seam, with the proper branching-plane topology; in the benchmark systems, the branching-plane displacements faithfully reproduce the linear dependence $E_1-E_0 \propto \sqrt{x^2+y^2}$ required for a proper CI (Rossi et al., 22 Apr 2026).

The standard reduced Hamiltonian used across several molecular control studies is a two-state, two-mode vibronic model with a tuning coordinate $Q_t$ and a coupling coordinate $Q_c$. In the diabatic representation, the diagonal terms depend on $Q_t$ and the off-diagonal term depends on $Q_c$, so that the degeneracy condition is simultaneously $V_{11}=V_{22}$ and $h$0. Zhang et al. use exactly this structure in a model vibronic system driven by a chirped Gaussian pulse, and Ye et al. use the same two-state/two-mode form in a cavity-coupled setting; both place the CI at $h$1, $h$2 in the reduced model (Zhang et al., 28 Dec 2025, Ye et al., 12 Nov 2025).

In engineered periodic media, the analogous local description is a $h$3 expansion near a degeneracy point in the Brillouin zone. The universal effective Hamiltonian is

$h$4

where $h$5 is the cone speed and $h$6 are generators of the spin-$h$7 irreducible representation of $h$8. This formulation covers half-integer pseudospin Dirac cones and integer-pseudospin higher-order cones, including the $h$9 case with a flat band at $\mathcal E_0(R)$0 (Leykam et al., 2016).

A common misconception is that engineered conical intersections are confined to molecular chemistry. The cited literature instead places them in at least four settings: molecular potential-energy surfaces, field-dressed atom-ion collisions, hybrid qubit-oscillator simulators, and two-dimensional periodic lattices for light and matter waves (Li et al., 2023, Wang et al., 2022, Leykam et al., 2016).

2. Restoring conical topology in electronic-structure theory

A central engineering problem is not only to locate a CI, but to preserve its topology within an approximate electronic-structure method. Conventional Kohn–Sham DFT treats the ground state via a separate variational self-consistent field problem and excited states via linear-response TDDFT or the Tamm–Dancoff approximation. Near a CI, this decoupling breaks the coupled two-state topology. The explicit failures listed for conventional DFT/TDDFT are orbital Hessian non-convexity, unphysical energy gaps, and broken branching space, leading to bifurcated ground-state surfaces, discontinuities of the $\mathcal E_0(R)$1 gap, spurious intersection lines, and even plateaus of negative gaps (Rossi et al., 22 Apr 2026).

Convex DFT (CVX-DFT) is introduced to remedy this by enforcing strict convexity of the orbital optimization problem in the subspace responsible for the CI. Within the TDA, the total energy is defined as

$\mathcal E_0(R)$2

and orbital variations are parameterized by occupied–virtual rotations. The key engineering step is to diagonalize the orbital Hessian $\mathcal E_0(R)$3, identify the problematic eigenvector $\mathcal E_0(R)$4 whose eigenvalue approaches zero or becomes negative near the CI, and project that direction out of both the gradient and the orbital updates:

$\mathcal E_0(R)$5

By construction, the reduced Hessian has all eigenvalues $\mathcal E_0(R)$6, so the projected problem has a unique minimum and continuous dependence on nuclear displacements (Rossi et al., 22 Apr 2026).

The final electronic states are recovered by diagonalizing a Hamiltonian $\mathcal E_0(R)$7 in the space spanned by the optimized Kohn–Sham determinant, the excluded mode $\mathcal E_0(R)$8, and the remaining single excitations. This CIS-like eigenproblem yields two adiabatic states that cross conically along a one-dimensional seam. Algorithmically, the workflow begins from a conventional KS-DFT solution, performs standard SCF iterations, forms and diagonalizes the TDA orbital Hessian, projects out the lowest mode, iterates the convex projected SCF problem to threshold, constructs the final $\mathcal E_0(R)$9, and then computes gradients and derivative couplings as in standard response theory. The dominant cost remains the underlying SCF and Hessian builds, with Hessian construction scaling as $\mathcal E_1(R)$0 per iteration and the overall method retaining the favorable scaling of KS-DFT/TDA (Rossi et al., 22 Apr 2026).

The benchmarks are explicitly given for protonated formaldimine, azobenzene, and the retinal model PSB3. In those cases CVX-DFT restores the correct double-cone topology in quantitative agreement with SA-CASSCF and XMS-CASPT2: a closed “Mexican-hat” seam for formaldimine, two distinct conical points for azobenzene rather than two continuous crossing lines with unphysical negative gaps, and smooth continuous $\mathcal E_1(R)$1 surfaces for PSB3 where conventional TDA fails to converge near the CI (Rossi et al., 22 Apr 2026).

3. Optical, cavity, and laser-field engineering

One route to engineered conical intersections is to impose them with external radiation fields. In the ultracold K–Ca$\mathcal E_1(R)$2 atom-ion system, Li et al. construct a field-dressed $\mathcal E_1(R)$3 Hamiltonian in the dressed-state or Floquet picture. The light–matter coupling is

$\mathcal E_1(R)$4

and a true degeneracy occurs when the dressed diabatic energies are resonant and the coupling vanishes at $\mathcal E_1(R)$5. The CI condition is therefore

$\mathcal E_1(R)$6

This produces a light-induced conical intersection whose radial position is set by the laser frequency and whose avoided-crossing geometry away from the seam is controlled by the intensity through $\mathcal E_1(R)$7 (Li et al., 2023).

The dynamical consequences in that system are irregular interference effects in the charge-exchange rate coefficients as functions of laser frequency, arising because two LICIs are present. The difference between the full calculation and a comparison model in which the CIs are “removed” can be as large as $\mathcal E_1(R)$8 within the relevant laser-frequency window. The same study states that rate enhancements can reach $\mathcal E_1(R)$9–$E_1-E_0 \propto \sqrt{x^2+y^2}$0 over the field-free value $E_1-E_0 \propto \sqrt{x^2+y^2}$1, while the operating regime remains at laser intensity $E_1-E_0 \propto \sqrt{x^2+y^2}$2 and ultracold temperatures below $E_1-E_0 \propto \sqrt{x^2+y^2}$3 (Li et al., 2023).

A second route is coherent control near an existing CI with shaped pulses rather than CI creation per se. Zhang et al. excite a model vibronic system with a linearly chirped Gaussian pulse,

$E_1-E_0 \propto \sqrt{x^2+y^2}$4

and track wave-packet populations and coherence dynamics through the degeneracy region. The long-time branching ratio is sampled at $E_1-E_0 \propto \sqrt{x^2+y^2}$5–$E_1-E_0 \propto \sqrt{x^2+y^2}$6, with $E_1-E_0 \propto \sqrt{x^2+y^2}$7 for $E_1-E_0 \propto \sqrt{x^2+y^2}$8 and $E_1-E_0 \propto \sqrt{x^2+y^2}$9. Negative chirp with $Q_t$0 increases the yield to $Q_t$1, while positive chirp gives $Q_t$2. The paper attributes the enhancement for red-first pulses to pulse lengthening and lower kinetic energy at the CI, which preserves vibrational phase, whereas blue-first pulses desynchronize vibrational components and produce rapid dephasing. The same study emphasizes that under one-photon weak fields the effect is typically $Q_t$3–$Q_t$4, and that stronger dephasing suppresses the chirp dependence to below $Q_t$5 when $Q_t$6 (Zhang et al., 28 Dec 2025).

Cavity QED supplies a third optical control modality. Ye et al. couple the molecular coupling mode $Q_t$7 to a single cavity mode through

$Q_t$8

Their central result is spectroscopic rather than purely dynamical: a nuclear wave packet that splits on the lower adiabatic surface and encircles the CI along two trajectories accumulates a geometric phase difference of $Q_t$9, so the total transition amplitude is proportional to $Q_c$0 and the excited-state-absorption contribution vanishes for perfectly symmetric splitting. Turning on the cavity coupling breaks the equality $Q_c$1 and revives the ESA cross peak in the 2DES signal. In the reported simulations, increasing the coupling parameter from $Q_c$2 to $Q_c$3 steadily restores the ESA peak, while high-$Q_c$4 cavities preserve the coherent photon–vibrational coupling long enough for the wave packet to encircle the CI (Ye et al., 12 Nov 2025).

4. Synthetic quantum simulators and laboratory analogues

Hybrid quantum hardware provides platforms in which conical intersections are created directly in a programmable Hamiltonian. In the circuit-QED processor of the study “Observation of wave-packet branching through an engineered conical intersection,” the effective Hamiltonian is

$Q_c$5

where the two bosonic modes act as tuning and coupling coordinates. The corresponding adiabatic surfaces are

$Q_c$6

so the CI sits at $Q_c$7, $Q_c$8 and the splitting is locally linear in the two coordinates. The experiment reports direct time-domain tracking of both the reactive wave packet and the electronic qubit, and identifies dephasing of the electronic qubit as the mechanism that drives wave-packet branching along the reactive coordinate (Wang et al., 2022).

The same platform engineers dissipation through the lossy “Bob” resonator, giving an effective continuous measurement of $Q_c$9 at rate

$Q_t$0

for $Q_t$1. This is interpreted as a measurement-induced dephasing term $Q_t$2 in the adiabatic limit of Bob. Representative operating values are $Q_t$3, $Q_t$4, $Q_t$5, and $Q_t$6. At the CI point, $Q_t$7 collapses almost immediately, whereas for displaced initial conditions the wave packet first oscillates and then shows an enhanced kink in the decay upon return to $Q_t$8 (Wang et al., 2022).

Trapped Rydberg ions provide a second synthetic realization. In the two-ion formulation, the field-free spinor Hamiltonian in relative coordinates is

$Q_t$9

with adiabatic surfaces

$Q_c$0

The CI is placed at

$Q_c$1

and in practice $Q_c$2 can be engineered so that the intersection lies at the origin and the remaining linear dependences $Q_c$3, $Q_c$4 yield a textbook conical topology in the $Q_c$5 plane. Adjustable knobs explicitly listed are the trap parameters, static field, Rydberg polarizabilities, inter-ion spacing, and microwave detuning (Belfakir et al., 14 Sep 2025).

Belfakir et al. then use quantum optimal control with a time-dependent electric field to drive the motional wave packet between prescribed spatial configurations on microsecond timescales. In the CI-enabled case, the packet passes through the CI region twice at early times, undergoes significant nonadiabatic transfer, and is routed directly to the target with final fidelity $Q_c$6. In the Born–Oppenheimer limit, the same control machinery reaches the target only through symmetric, multi-cycle oscillations. The comparison is explicitly presented as a difference in underlying trajectory rather than endpoint reachability (Belfakir et al., 14 Sep 2025).

Earlier trapped-Rydberg-ion work emphasizes a related effect: when the CI is perfectly symmetric, the wave packet splits and acquires a geometric phase difference $Q_c$7 along two paths, producing destructive interference that inhibits nuclear motion. If the CI is shifted off symmetry, interference becomes partial and large oscillations in diabatic populations are restored. These dynamics unfold on microsecond timescales and can be measured by direct spectroscopic monitoring of Rydberg populations (Gambetta et al., 2020).

Ultracold atomic aggregates and mobile Rydberg networks extend the same logic to many-body dipole-coupled systems. Wüster, Eisfeld, and Rost show that dipole–dipole coupled ultracold atoms generically possess CIs and propose a circular trimer in a ring trap as a minimal example (Wüster et al., 2010). Leonhardt, Wüster, and Rost then exploit a planar trimer embedded in a T-shaped Rydberg aggregate to create a CI at the equilateral configuration and use it as a switch for exciton pulses. Near the symmetric point one finds $Q_c$8, whereas asymmetric geometry suppresses the hop and steers the pulse along a selected branch (Leonhardt et al., 2013).

5. Lattice, photonic, and matter-wave realizations

In two-dimensional periodic lattices, engineered conical intersections are usually discussed as Dirac or Dirac-like cones. The local Hamiltonian $Q_c$9 supports different qualitative classes depending on pseudospin. For $V_{11}=V_{22}$0, the eigenvalues are $V_{11}=V_{22}$1, the Berry phase is $V_{11}=V_{22}$2 modulo $V_{11}=V_{22}$3, and the two-dimensional density of states vanishes linearly. For $V_{11}=V_{22}$4, the eigenvalues are $V_{11}=V_{22}$5 for $V_{11}=V_{22}$6, so two linearly dispersing bands coexist with a flat band at $V_{11}=V_{22}$7, and the Berry phase is $V_{11}=V_{22}$8 modulo $V_{11}=V_{22}$9 (Leykam et al., 2016).

The design principles separate into symmetry-protected and accidental degeneracy routes. Honeycomb geometry with two equivalent sublattices and $h$00 symmetry guarantees $h$01 Dirac cones at the $h$02 and $h$03 points. By contrast, accidental degeneracy can be induced at $h$04 or $h$05 by tuning rod radius, lattice spacing, refractive-index contrast, or mode content so that two or more modes become degenerate. The review explicitly lists zero-index metamaterials with an $h$06 conical point at $h$07, plasmonic nanoparticle arrays, phononic crystals, Lieb lattices, laser-written waveguide arrays, patterned photonic crystals, and cold-atom lattices with laser-assisted tunneling between internal states (Leykam et al., 2016).

These momentum-space intersections are engineered for their transport and wave-propagation consequences. The cited applications include zero-index metamaterials, single-mode large-area photonic-crystal slab lasers, orbital angular momentum generation through conical diffraction, quantum simulation of relativistic phenomena such as Dirac and Klein tunnelling, pseudo-diffusive transport, and flat-band localization with enhanced nonlinearities (Leykam et al., 2016).

This broadens the meaning of engineered conical intersections beyond chemistry. In molecular and hybrid quantum systems, the engineering objective is typically control of nonadiabatic population transfer and branching; in periodic media, it is control of singular dispersion relations and the pseudospin-dependent wave dynamics that follow from them. The shared mathematical element is the deliberate shaping of a two-parameter degeneracy with linear splitting away from the degeneracy point (Leykam et al., 2016).

6. Design variables, observables, and emerging applications

Across platforms, engineered conical intersections are characterized by a relatively small set of control variables. In molecular electronic structure, the decisive variable is the problematic orbital-Hessian mode removed from SCF optimization and then reintroduced in the final-state Hamiltonian (Rossi et al., 22 Apr 2026). In ultrafast coherent control, the control knobs are spectral phase, chirp, and pulse duration, which alter vibrational coherence and kinetic energy at the CI (Zhang et al., 28 Dec 2025). In LICIs, the governing parameters are the laser frequency, intensity, and polarization geometry (Li et al., 2023). In cavity-enhanced spectroscopy they are the cavity resonance frequency, light–matter coupling strength, and quality factor (Ye et al., 12 Nov 2025). In circuit QED and trapped-ion simulators the knobs are drive amplitudes, detunings, sideband strengths, decay rates, static fields, trap frequencies, and exchange couplings (Wang et al., 2022, Belfakir et al., 14 Sep 2025).

A concise cross-platform summary is useful because the engineering objective differs by platform.

Platform	Primary control variables	Reported consequence
CVX-DFT	Hessian projection of $h$08	Restored double-cone topology
Chirped-pulse vibronic control	$h$09, $h$10	Modified branching and quantum yield
LICI in atom-ion collisions	$h$11, $h$12, polarization angle	Irregular interference in charge-exchange rates
Cavity-enhanced 2DES	$h$13, $h$14, $h$15	GP-induced ESA cancellation or revival
Circuit QED / trapped ions	Drives, detunings, decay, trap parameters	Wave-packet branching or directed motion
Periodic lattices	Symmetry, lattice geometry, accidental tuning	Dirac or Dirac-like cones

The principal observables are equally platform-specific. Molecular simulations monitor energy gaps, branching-plane topology, wave-packet populations, coherences, and quantum yields (Rossi et al., 22 Apr 2026, Zhang et al., 28 Dec 2025). The cavity study focuses on the third-order 2DES signal and, specifically, an ESA amplitude zero as a spectroscopic manifestation of geometric phase (Ye et al., 12 Nov 2025). Circuit QED directly measures qubit coherence, conditional Wigner functions, and branching ratios (Wang et al., 2022). Trapped-ion realizations read out diabatic or adiabatic populations and track the motional wave packet in real time (Gambetta et al., 2020, Belfakir et al., 14 Sep 2025). Periodic media emphasize density of states, conical diffraction, transport scaling, and band topology (Leykam et al., 2016).

Several recurring points qualify the scope of control. First, engineered intersections do not automatically guarantee useful dynamics; the literature repeatedly emphasizes that dissipation, dephasing, and multi-mode coupling can reduce or wash out the effect. Zhang et al. state that rapid dephasing and multi-mode couplings in real molecules dilute weak-field coherent control and can reduce chirp-induced modulation below $h$16 (Zhang et al., 28 Dec 2025). Second, approximate electronic-structure methods may destroy the very topology one intends to exploit, which is why the convex reformulation in CVX-DFT is presented as a prerequisite for reliable non-adiabatic simulation beyond conventional TDDFT (Rossi et al., 22 Apr 2026). Third, geometric phase is not merely a formal property: in trapped Rydberg ions and cavity-enhanced 2DES it appears as a directly observable interference effect, either freezing motion through destructive interference or cancelling an ESA contribution (Gambetta et al., 2020, Ye et al., 12 Nov 2025).

A plausible implication of these converging results is that “engineering” now spans three levels at once: Hamiltonian design, topology restoration, and dynamical steering. The cited work does not collapse these into a single methodology, but it does establish a common research program in which conical intersections are treated as controllable resources rather than only unavoidable features of excited-state dynamics (Rossi et al., 22 Apr 2026, Belfakir et al., 14 Sep 2025, Leykam et al., 2016).