Field-Driven Excitonic Condensate

• Field-driven excitonic condensates are ordered states characterized by coherent interband mixing that is tuned via magnetic, electric, or optical fields.
• Magnetic fields (orbital or Zeeman) and electric fields modulate the condensate’s phase and internal structure, enabling transitions between distinct quantum phases.
• These condensates are explored in various systems, from lattice models to TMD bilayers, offering insights into topological states and potential device applications.

Field-driven excitonic condensates are excitonic ordered states whose condensate amplitude, phase, internal flavor structure, topology, or spatial profile is controlled by a magnetic, electric, or optical field. In the works considered here, the relevant order parameter is usually an interband coherence such as the local excitonic average $\Delta=\langle d_i^\dagger f_i\rangle$ or the momentum-resolved quantity $$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$, but the same language is also used for field-induced condensation of spin excitons and for multicomponent spin-valley condensates. The field may act through orbital motion and Hofstadter quantization, Zeeman splitting, Stark trapping, resonant optical pumping, or a coherent excitonic polarization that becomes an internal bosonic drive (Pradhan et al., 2015, Pareek et al., 2024, Sotnikov et al., 2016, Yu et al., 12 Nov 2025, Qi et al., 16 Mar 2026).

1. Definitions and order parameters

In lattice excitonic-insulator models, the basic condensate variable is an interorbital coherence. In the extended spinless Falicov–Kimball setting, the central quantity is the local excitonic order parameter

$\Delta=\langle d_i^\dagger f_i\rangle,$

where $d$ denotes an itinerant orbital and $f$ a localized or weakly dispersive orbital. A nonzero $\Delta$ represents coherent mixing between the conduction $d$ band and the localized $f$ level, and in the spinless, parity-different context discussed there it corresponds to an electronic ferroelectric state if the orbitals have opposite parity (Pradhan et al., 2015).

In dense photoexcited semiconductors, the order parameter is naturally momentum dependent. For monolayer WS$_2$, the coherent exciton population produces

$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle \neq 0,$$

which acts as an off-diagonal self-energy coupling valence and conduction bands. In that formulation the condensate is BCS-like rather than a dilute bosonic gas, and the resulting quasiparticles are Bogoliubov bands with a Mexican-hat dispersion in the valence sector (Pareek et al., 2024).

In spin-triplet excitonic magnets the order parameter is vectorial and complex. For the two-orbital Hubbard problem studied with DMFT, the triplet condensate is

$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$0

so the condensate carries both amplitude and internal orientation in spin space. The distinction between unitary and non-unitary condensates, and between SDW-, SCDW-, and FMEC-type states, is encoded in the relative geometry of $$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$1 and $$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$2 (Geffroy et al., 2018).

The term also extends beyond charge excitons in the narrow sense. In CeCoIn$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$3, the field-driven condensate is a zero-energy condensation of a spin-1 collective mode into static spin-density-wave order at finite wave vector, while in MoSe$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$4/hBN/WSe$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$5 the condensate is a multicomponent fluid of interlayer excitons with four spin-valley flavors (Michal et al., 2011, Qi et al., 16 Mar 2026).

2. Routes by which fields control the condensate

Orbital magnetic fields act through Peierls phases in the hopping and therefore restructure the underlying kinetic spectrum before any mean-field instability is formed. In the two-dimensional extended spinless Falicov–Kimball model this produces a Falicov–Kimball analogue of the Hofstadter problem for rational flux $$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$6, with magnetic unit cells enlarged by $$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$7 and subband splitting inherited from the Hofstadter butterfly. The excitonic condensate is then formed on a flux-dependent background of bandwidths and gaps, so its magnitude and real-space structure become strongly commensurability dependent (Pradhan et al., 2015).

Zeeman fields act differently. In LaCoO$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$8, the field lowers one branch of a dispersive intermediate-spin exciton band until the gap closes and a condensate of mobile IS excitations forms. In CeCoIn$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle$$9, the field softens a finite-$\Delta=\langle d_i^\dagger f_i\rangle,$0 spin resonance until the static transverse susceptibility diverges and the spin exciton condenses into SDW order within the superconducting phase. In spin-triplet electron–hole systems with Rashba SOC, Zeeman splitting selects spin-up triplet excitonic pairing and can stabilize a topological Chern phase. In MoSe$\Delta=\langle d_i^\dagger f_i\rangle,$1/hBN/WSe$\Delta=\langle d_i^\dagger f_i\rangle,$2, modest magnetic fields do not merely polarize a preexisting condensate but switch its flavor composition from an intravalley two-component condensate to an intervalley two-component condensate and finally to a fully polarized single-component state (Sotnikov et al., 2016, Michal et al., 2011, Phan, 9 Dec 2025, Qi et al., 16 Mar 2026).

Electric fields appear in two distinct ways. They can define the condensate energetically, as in double quantum wells or bilayers where layer separation and band alignment tune the interlayer pairing kernel, and they can define it spatially, as in nanoscale Stark traps where a vertical electric-field landscape confines excitons and sets the finite-size scale of the condensate. In the latter case the trap is not ancillary: removing the field-defined confinement removes the extended coherence signatures (Hakioğlu et al., 2012, Özgün et al., 2015, Yu et al., 12 Nov 2025).

Optical fields can act either externally or effectively internally. In driven excitonic insulators, periodic laser fields excite amplitude and phase modes of the condensate. In dense WS$\Delta=\langle d_i^\dagger f_i\rangle,$3, by contrast, the coherent excitonic polarization itself acts as a monochromatic bosonic field, formally analogous to a Floquet drive with amplitude proportional to $\Delta=\langle d_i^\dagger f_i\rangle,$4, and it is this internal field that reshapes the quasiparticle bands (Davari et al., 2023, Pareek et al., 2024).

3. Orbital magnetic fields, Zeeman fields, and lattice realizations

The orbital-field response is most explicit in the extended spinless Falicov–Kimball model on a $\Delta=\langle d_i^\dagger f_i\rangle,$5 square lattice. There the excitonic average is generally suppressed by increasing magnetic flux, but the suppression is not monotonic. For $\Delta=\langle d_i^\dagger f_i\rangle,$6 and small $\Delta=\langle d_i^\dagger f_i\rangle,$7, the calculations show a prominent peak in $\Delta=\langle d_i^\dagger f_i\rangle,$8 at $\Delta=\langle d_i^\dagger f_i\rangle,$9, a dip at $d$0, and an extremum at $d$1 where $d$2 is smallest; the behavior is symmetric around $d$3. At small $d$4 and small $d$5, however, there are regions where the field slightly enhances $d$6, specifically at $d$7 and very small $d$8. The same study shows that the interband Coulomb interaction exponentially enhances $d$9, nonzero $f$0-electron hopping suppresses it, commensurate flux keeps $f$1 uniform, and incommensurate flux generates a one-dimensional striped modulation of $f$2 (Pradhan et al., 2015).

A later Hartree–Fock study of the magnetic-field effect in the extended Falicov–Kimball model reaches a closely related but more topological high-field picture. There the excitonic order parameter again shows a nonmonotonic field dependence, but sufficiently high fields suppress the condensate and produce either a fully orbital-polarized phase or a disordered insulating state with partial occupation of the two orbitals and nonzero Chern numbers. In the excitonic supersolid, excitonic and orbital orders coexist at zero field, yet the orbital order remains robust under magnetic fields while excitonic condensation is suppressed (Ohta et al., 17 Jan 2025).

Zeeman-driven condensation in correlated oxides and heavy fermions follows a different logic. In LaCoO$f$3, the decisive diagnostic is the temperature dependence of the critical field: $f$4 is compatible with field-induced exciton condensation of mobile intermediate-spin excitations, whereas the competing spin-state-ordering scenario predicts the opposite low-temperature slope. In CeCoIn$f$5, the field drives the transverse spin exciton resonance continuously down to zero energy at an incommensurate $f$6, so the ordered SDW phase inside the superconducting state is interpreted as the zero-energy condensation of spin excitons (Sotnikov et al., 2016, Michal et al., 2011).

4. Internal structure, topology, and multicomponent condensates

Field control is not restricted to the magnitude of the order parameter; it also selects its internal symmetry. In a two-dimensional interacting electron–hole system with Rashba SOC and external magnetic field, unrestricted Hartree–Fock yields a spin-up triplet excitonic condensate with a nonzero Chern number. The order parameters are spin selective,

$f$7

and the topological phase appears only in a finite window of magnetic field and interaction, away from topologically trivial singlet and spin-down triplet regions. Dynamical excitonic susceptibility further shows strong spin-polarized triplet fluctuations that soften toward condensation in the spin-up channel (Phan, 9 Dec 2025).

In the two-orbital Hubbard model close to half filling, DMFT reveals a different kind of field-controlled internal structure. The triplet order parameter $f$8 generally tends to lay perpendicular to the applied field, although the $f$9-wave cross-hopping case shows exceptions to strict orthogonality. For solutions with $\Delta$0-odd spin textures, the work emphasizes that Bloch’s theorem forbids spontaneous net spin current in the absence of spin–orbit coupling, yet the local-self-energy nature of DMFT fails to respect this constraint and produces finite equilibrium spin currents in the approximation (Geffroy et al., 2018).

A still richer field response occurs in MoSe$\Delta$1/hBN/WSe$\Delta$2 electron–hole bilayers, which host four spin-valley exciton flavors. Magneto-optical spectroscopy identifies three condensate phases. At zero magnetic field, the ground state is a coherent superposition of two simultaneously condensed intravalley flavors. Under a weak magnetic field, this state switches through a first-order quantum phase transition to a two-component intervalley condensate, and at higher fields it becomes a fully polarized single-component condensate. The two-component condensates persist up to approximately $\Delta$3 K, and the phase boundaries are detected directly in the spin susceptibility of constituent electrons and holes (Qi et al., 16 Mar 2026).

5. Optical driving, internal bosonic fields, and condensate dynamics

The most explicit realization of an internally driven condensate is the dense-exciton regime of monolayer WS$\Delta$4. There the coherent excitonic polarization creates an off-diagonal self-energy $\Delta$5, so the excitons themselves act as a classical monochromatic field coupling valence and conduction bands. Time- and angle-resolved photoemission then shows the valence band evolving from a conventional paraboloid into a Mexican-hat Bogoliubov dispersion, with the onset of strong reshaping around $\Delta$6 and a clear Mexican-hat at $\Delta$7. The inferred hybridization scale is $\Delta$8 meV at $\Delta$9, whereas an optical Floquet drive of comparable pulse duration gives only $d$0 meV (Pareek et al., 2024).

Driven condensate dynamics in more conventional excitonic-insulator models separates amplitude and phase responses. In the periodically driven spinless two-band lattice model, the condensate amplitude oscillates in time during irradiation and the oscillations survive after the light is switched off. For a condensate originating from purely electronic correlations, the phase changes linearly with time. With electron–phonon coupling, that linear behavior is replaced by harmonically oscillating phase dynamics, which the analysis identifies as a manifestation of gapped phase modes due to relative band charge symmetry breaking. Bicircular and circular drives, both of which break time-reversal symmetry, excite the phase mode more strongly than linear polarization, and the primarily electronic and primarily lattice regimes show distinct phase-mode frequencies (Davari et al., 2023).

A related one-dimensional topological $d$1–$d$2 model reaches a similar conclusion from a different starting point. There the excitonic insulator exists only on the band-inverted side of the noninteracting problem, the BCS/BEC crossover appears in the low-$d$3–$d$4 hybridization regime, and a pump pulse reveals oscillations of exciton states that depend strongly on the laser frequency. The corresponding optical conductivity contains both single-particle features and a low-energy many-body peak associated with collective excitonic dynamics (Khatibi et al., 2020).

Optically driven condensates need not be predominantly photonic. In a ZnO microwire, pulsed excitation realizes a one-dimensional polariton condensate with approximately $d$5 photonic and $d$6 excitonic fraction. The first-order coherence extends over distances as large as $d$7 $d$8m, and a driven-dissipative mean-field model attributes the finite correlation length primarily to shallow disorder under non-equilibrium conditions (Trichet et al., 2013).

6. Electric-field landscapes, mechanical consequences, and unresolved issues

An electric-field landscape can itself define the condensate. In monolayer WSe$d$9, nanoscale spacing-graded Stark traps generate a center-to-edge field difference of about $f$0 mV/nm and a trap depth $f$1 meV, far above $f$2 at $f$3 K. With off-axis non-resonant injection, this field-defined trap accumulates dark excitons into a quasi-equilibrium condensate showing a sharp degeneracy threshold, first-order spatial coherence extending to about $f$4, algebraic decay of $f$5 with exponent $f$6, and identical condensation signatures in over $f$7 independent samples. A useful contrast is provided by proposed equilibrium heterobilayers with doubly-indirect overlapping bands and broken type-III gap, which were introduced precisely to avoid the applied-voltage doping that makes other bilayers essentially non-equilibrium (Yu et al., 12 Nov 2025, Gupta et al., 2019).

The condensate can also feed back mechanically on its host structure. In double quantum wells, the condensation energy depends on layer separation $f$8, leading to an attractive exciton-condensate force

$f$9

estimated to be of order $_2$0 N for a sample area $_2$1 and $_2$2. In TMDC-inspired two-layer models with coupled CDW and EC orders, the same logic yields an EC pressure of order $_2$3 Pa, large enough to be discussed as a possible driver of the structural lattice deformation observed in $_2$4-TiSe$_2$5 below $_2$6 K (Hakioğlu et al., 2012, Özgün et al., 2015).

Several recurrent misconceptions are corrected by the literature. A field-driven excitonic condensate is not always a charge-exciton condensate in a semimetal; in CeCoIn$_2$7 it is the zero-energy condensation of a spin exciton into SDW order (Michal et al., 2011). A magnetic field does not invariably suppress the condensate; orbital fields usually do, but weak-coupling windows such as $_2$8 in the Falicov–Kimball model can slightly enhance $_2$9 (Pradhan et al., 2015). Nor does “field-driven” necessarily mean that the field alone creates the condensate from an otherwise featureless state; in WS$$\Delta_{\mathbf{k}}^{X}=\langle c^\dagger_{c\mathbf{k}}c_{v\mathbf{k}}\rangle \neq 0,$$0 the coherent excitonic population is itself the driving field once resonant pumping has established it (Pareek et al., 2024).

The main limitations are also consistent across platforms. Static Hartree–Fock and related mean-field treatments can overestimate the stability of ordered phases, especially in two dimensions; the spinless lattice models omit Zeeman, spin–orbit, and triplet sectors; on-site excitons can behave differently from extended bilayer excitons; and local-self-energy DMFT can violate the Bloch theorem for equilibrium spin current (Pradhan et al., 2015, Geffroy et al., 2018). Experimental realizations add further constraints: transient lifetimes in optically driven states, proximity to the Mott transition at high density, and incomplete access to topological invariants or collective-mode spectra remain open problems rather than settled features (Pareek et al., 2024, Qi et al., 16 Mar 2026).