Cavity-mediated localization and collective electron correlation phases Preprint: APS/123-QED

Dominik Sidler, Michael Ruggenthaler, Angel Rubio, UnMySt Collaboration · 2026

Read Cavity-mediated localization and collective electron correlation phases Preprint: APS/123-QED by Dominik Sidler, Michael Ruggenthaler, Angel Rubio, UnMySt Collaboration (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-01551.txt.

Introduction

Dominik Sidler, Francesco Greco, Michael Ruggenthaler, and Angel Rubio predict new collective phases for molecules coupled strongly to light inside an optical cavity. Cavity coupling adds correlation channels that span the whole ensemble, on top of electron correlations that are already hard to describe, which makes the joint problem look untractable.

The team show it maps onto spin glass theory, in fact onto a classic model that can be solved exactly. Two phases appear beyond the normal uncorrelated regime: a paracorrelated phase and a spin-glass correlation phase. Both come from an entropy-driven shuffle that moves molecular states into shared, cavity-dressed states with many equal members. The collective state can stay insulating or, if the entropic filling falls right, turn metallic. The result gives cavity chemistry a rare case where strong collective correlation can be reasoned about exactly.

Catalog record

CREATORDominik Sidler, Michael Ruggenthaler, Angel Rubio, UnMySt Collaboration
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.01551

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Cavity-mediated localization and collective electron correlation phases Preprint: APS/123-QED Dominik Sidler, Michael Ruggenthaler, Angel Rubio, UnMySt Collaboration Abstract Collective strong coupling of molecular ensembles to optical cavities opens a route to modifying matter through genuinely collective electronic correlations. Yet even in the absence of a cavity, Coulomb correlations are notoriously difficult to describe, and cavity coupling adds transverse correlation channels extending over the entire molecular ensemble. Here we show that this seemingly intractable problem admits a controlled description by mapping the collective intermolecular electronic correlations to the analytically solvable spherical Sherrington-Kirkpatrick model. The resulting theory predicts two collective correlation phases, a paracorrelated phase and a spin-glass correlation phase, beyond the conventional uncorrelated molecular regime. These phases reveal an entropy-driven localization-delocalization mechanism that transfers molecular electronic states into collectively correlated cavity-dressed states. Our work establishes cavity-mediated electron correlations as a microscopic mechanism for emergent phases in strongly coupled molecular ensembles. Over the past decades, seminal experiments have demonstrated that collective strong coupling of matter in optical cavities can be utilized to alter and design material and molecular properties. 5 ; 18 ; 22 ; 6 ; 4 ; 17 ; 2 ; 12 ; 23 ; 13 ; 25 Despite considerable theoretical advances, the relevant underlying physical mechanisms remain largely unknown. One of fundamental question is how collective strong coupling can be sufficiently localized such that they can induce significant chemical (local) changes as observed in experiments. In the following work, we focus on the collective strong coupling of molecular ensembles. Cavity-modified electron correlations appear to be critical for many of the observed changes. However, accurately treating Coulomb correlations is already a notoriously difficult problem to solve outside of a cavity. Inside a cavity, transversal collective electron correlation effects spanning the molecular ensemble seem completely inaccessible. However, thanks to a recently proposed mapping of the collective transversal electron correlations to the analytically solvable spherical Sherrington-Kirkpatrick model of a spin glass, we can subsequently identify and investigate different phases of collective electron correlations. Notably, the design of spin glass conditions in optical cavities by structuring the photonic environment has been extensively discussed in Refs. 7 ; 8 ; 10 ; 9 ; 14 . In our setup, however, the glassy picture emerges from the complex (random) molecular structure and the Pauli principle of the electrons. 21 To investigate collective transverse electron correlations, we extend recently proposed theoretical framework from collective configuration interaction singles 24 ; 11 (CIS) to a multireference picture based on reduced density matrix 3 (RDM) theory. Within the cavity-Born-Oppenheimer partitioning, the electronic Hamiltonian is given as, 17 \displaystyle\hat{H}^{e} \displaystyle=\sum_{i=1}^{2N}\bigg\{\frac{\hat{\vec{p}}_{i}^{2}}{2m}+V_{\rm Coul}(\hat{\vec{r}}_{i},\bm{R})-\vec{\lambda}\cdot e\hat{\vec{r}}_{i}\bigg(\vec{\lambda}\cdot\sum_{j=1}^{N_{\rm nuk}}Z_{j}e\vec{R}_{j}-\omega_{\beta}q_{\beta}\bigg)\bigg\} \displaystyle+\underbrace{\frac{1}{2}\sum_{i=1}^{2N}\sum_{j=1,j\neq i}^{2N}\frac{e^{2}}{4\pi\epsilon_{0}|\hat{\vec{r}}_{i}-\hat{\vec{r}}_{j}|}}_{\hat{W}_{\parallel}}+\underbrace{\frac{\big(\vec{\lambda}\cdot\sum_{i=1}^{2N}e\hat{\vec{r}}_{i}\big)^{2}}{2}}_{\hat{W}_{\perp}}. (1) The first line contains solely one-electron terms, with kinetic energy, Coulomb attraction of the nuclei V_{\rm Coul} and the linear coupling to a single effective cavity mode q_{\beta} of frequency \omega_{\beta} , linear polarization \vec{\lambda}/|\lambda| and light-matter coupling parameter \lambda=|\vec{\lambda}| . The number of electrons is 2N and the positive nuclei enter with \vec{R}_{j} and charge Z_{j}e . The second line, contains longitudinal (Coulomb) two-electron operators \hat{W}_{\parallel} and the transversal dipole self-energy \hat{W}_{\perp} . We start by expressing the total energy of the ensemble of interacting molecules, i.e., we consider the wave function of all the molecules in the cavity, in terms of the one-body reduced density matrix (1-RDM) \gamma and the two-body reduced density matrix (2-RDM) \Gamma . Using the natural orbitals \phi_{p} we find the full energy by minimizing \displaystyle E \displaystyle= \displaystyle\sum_{pq}h_{pq}\gamma_{pq}+\frac{1}{2}\sum_{pqrs}\bra{pq}rs\rangle\Gamma_{pqrs}, (2) \displaystyle= \displaystyle\underbrace{\sum_{pq}h_{pq}\gamma_{pq}+\frac{1}{2}\sum_{pq}n_{p}n_{q}(2J_{pq}-K_{pq})}_{=E_{\rm EHF}} \displaystyle+\underbrace{\frac{1}{2}\sum_{pqrs}\bra{pq}rs\rangle\Lambda_{pqrs}}_{=E_{\rm corr}} (3) with \gamma_{pq}=Tr[\hat{\Gamma}a_{q}^{\dagger}a_{p}] , 2-RDM \Gamma_{pqrs}=Tr[\hat{\Gamma}a_{q}^{\dagger}a^{\dagger}_{p}a_{s}a_{r}] and density-matrix operator \hat{\Gamma}=\sum_{k}w_{k}\ket{\Psi_k}\bra{\Psi_k} with normalized ensemble weights \sum_{k}w_{k}=1 . If the ground state of the ensemble is non-degenerate, we would find a pure state, i.e., w_{0}=1 with a correlated ground-state wave function \Psi_{0} . Here we have used that we can re-express the full 2-RDM in terms of a Hartree-Fock-like term and corrections \Gamma_{pqrs}=\gamma_{pr}\gamma_{qs}-\gamma_{ps}\gamma_{qr}+\Lambda_{pqrs} , and the diagonal of the 1-RDM is given as \gamma=\sum_{p}n_{p}\ket{\phi_p}\bra{\phi_p} for all the natural orbitals, 15 with fractional occupations 0\leq n_{p}\leq 2 that sum up to the total umber of electrons \sum_{p}n_{p}=2N . We want to investigate next how specifically cavity-induced long-range correlations influence the electronic structure of the ensemble. For this we make the ansatz, based on the fact that a large ensemble of molecules will have a very dense spectrum and potentially large degeneracies, that E_{\rm EHF} can be approximated as \hat{\Gamma}_{\rm EHF}=\sum_{k}w_{k}\ket{\Phi_k}\bra{\Phi_k} with \Phi_{k} being Slater determinants of the natural orbitals of the 1-RDM. That is, the fractional occupations are due to large degeneracies and we approximate this by weighted Hartree-Fock wave functions. We denote the corresponding reference energy by \displaystyle E_{0}=Tr\left[\hat{\Gamma}_{\rm EHF}\hat{H}^{e}\right], (4) where we assume that each \Phi_{k} essentially contributes the same energy (on the scale of the molecular ensemble) such that we have also many energetically degenerate orbitals. Practically this means that we solve the Hartree-Fock equation for the many-molecule wave function multiple times, which due to the non-linearity of the theory will lead to different self-consistent solutions. Each self-consistent solution is then taken into account such that we have an ensemble of many-molecules Slater determinants, which will lead to fractionally occupied natural orbitals. Now, for each \Phi_{k} the simplest form of cavity-induced long-range correlations can be derived from a configuration-interaction-singles (CIS) ansatz 21 , such that |\Phi^{\mathrm{CIS}}_{k}\rangle=s_{0}|\Phi_{k}\rangle+\sum_{m\in N_{\mathrm{occ}}}\!\sum_{p\in N_{\mathrm{virt}}}s_{m}^{p}\,|\Phi_{m}^{p}\rangle, (5) where |\Phi_{m}^{p}\rangle=a_{p}^{\dagger}a_{m}|\Phi_{k}\rangle denotes a singly excited determinant and orthogonality imposes the normalization condition \langle\Phi^{\mathrm{CIS}}_{k}|\Phi^{\mathrm{CIS}}_{k}\rangle=s_{0}^{2}+\sum_{m,p}(s_{m}^{p})^{2}=1 . Since we assume equally-weighted contributing determinants, the s_{0} corresponds to the the so-far uncorrelated (but fractionally occupied) \hat{\Gamma}_{\rm EHF} and we have many different single-excitations that explore the low-energy density of states of the molecular ensemble. We then propose to use the CIS correlation functional for long-range transversal electron correlations \hat{W}_{\perp} as discussed in Ref. 21 . In this regards we are primarily interested in electron correlations that emerge collectively from the set of degenerate (valence) orbital energies \epsilon_{i}=\epsilon_{j},\;i,j\in N_{\mathrm{occ}} spread across our large molecular ensemble (within the quasi-dilute gas approximation, see Ref. 21 ). Furthermore, we assume not all of these degenerate natural orbitals are occupied in our ensemble reference states, i.e., there are l -virtual orbitals with \epsilon_{i}=\epsilon_{l},\;i\in N_{\mathrm{occ}},\;l\in N_{\mathrm{virt}} . When restricting the different CIS constructions to those N_{\rm deg}={N_{\rm occ}}{N_{\rm virt}} degenerate orbitals only, the resulting correlation energy minimization problem has the following form 21 E_{\rm CIS}\sim E_{0}\underbrace{-\sum_{i<j}^{N_{\rm deg}}J_{ij}\,\sigma_{i}\sigma_{j}}_{E_{\rm CIS,corr}},\qquad\sum_{i=1}^{N_{\rm deg}}\sigma_{i}^{2}=1-s_{0}^{2}=\epsilon, (6) where \sigma_{k} are normalized CIS amplitudes emergent from s_{m}^{p}\mapsto\sigma_{i} , playing the role of continuous spin variables. The random transversal interactions J_{ij} , emerge from two-electron integrals of the dipole-self energy, i.e., from \hat{W}_{\perp} , in combination with random molecular orientation (rotation) with respect to the linear cavity-polarization 21 . In principle we would need to construct all possible CIS states for the respective Hartree-Fock references and consider their effect. Specifically, since there will be many CIS configurations close to the reference states of the molecular ensemble, their entropic contributions can become decisive. In contrast, in the following we use the fact that Eq. ( 6 ) has known solutions, provided we also know \displaystyle\epsilon=(1-s_{0}^{2}),\quad\epsilon\in[0,1]. (7) This, we will determine in a next step. In the following, the random-interaction are assumed to belong to the equivalence class of zero mean with \displaystyle{\rm Var}(J_{ij})=(\epsilon\sigma)^{2}N_{\rm deg} (8) (see follow-up publication for detailed discussion of random-interaction,

Read the complete text