Excitonic Metal Phenomena

• Excitonic metal is a class of states where excitonic effects, such as finite-momentum condensation, partially gap the Fermi surface while preserving metallic conductivity.
• The phenomenon encompasses both equilibrium phases, exemplified by charge density wave formation in twisted double bilayer graphene, and transient states driven by exciton dissociation in ultrafast experiments.
• Studies reveal that excitonic correlations can coexist with conventional metallic screening, offering rich implications for non-Fermi-liquid transport and novel ultrafast dynamics.

Excitonic metal denotes a class of metallic states in which excitonic physics remains constitutive rather than merely perturbative, but the literature uses the term in several inequivalent senses. In the strictest sense, it refers to a symmetry-broken metal produced by exciton condensation that reconstructs, yet does not fully gap, the Fermi surface, as proposed for twisted double bilayer graphene near charge neutrality (Ghorai et al., 2022). In a different nonequilibrium sense, it can refer to a transient metal whose itinerant carriers are generated from an optically created exciton population, as in NiPS$_3$, where photoexcited excitons dissociate into cold carriers while long-range antiferromagnetism survives (Belvin et al., 2021). Other works use the concept more cautiously for transient or short-range excitonic correlations inside metallic environments, without establishing a thermodynamically stable excitonic phase (Cui et al., 2014, Koskelo et al., 2023). The term therefore spans equilibrium broken-symmetry metals, ultrafast conducting states of excitonic origin, and metallic systems that retain only limited excitonic phenomenology.

1. Terminological scope and classification

Across the cited literature, “excitonic metal” is not synonymous with “excitonic insulator.” An excitonic insulator is an ordered state in which electron-hole attraction opens a full gap, whereas an excitonic metal retains metallic carriers or finite density of states at the Fermi level despite substantial excitonic structure. The distinction is explicit in the twisted double bilayer graphene proposal, where finite-momentum indirect exciton condensation reconstructs the bands but leaves residual Fermi surfaces (Ghorai et al., 2022). It is equally explicit in NiPS$_3$, where the metallic response is called exciton-driven because the carriers originate from exciton dissociation rather than from exciton condensation (Belvin et al., 2021).

A second distinction separates genuine metallic phases from transient or local excitonic responses inside metals. On Ag(111), the reported object is a short-lived surface exciton that exists before metallic screening saturates, not a stable bulk excitonic phase (Cui et al., 2014). In the low-density homogeneous electron gas, the demonstrated phenomena are short-range excitonic collective modes and electron-hole localization in a metallic background, not a conventional bound-exciton condensate (Koskelo et al., 2023).

Usage in the literature	Excitonic ingredient	Metallic ingredient
Equilibrium excitonic metal	Condensation of indirect excitons at finite $\mathbf Q_i$	Residual Fermi surfaces after partial gapping
Exciton-driven transient metal	Photoexcited excitons dissociate into itinerant carriers	Drude-like THz conductivity on picosecond timescales
Excitonic response in a metal	Pre-screened or short-range electron-hole correlations	Metallic host remains metallic
Proximate or adjacent regime	Metal or semimetal near excitonic instability	No established excitonic metallic phase

This classification suggests that the phrase functions less as a single phase label than as a family resemblance term. A plausible implication is that discussions of excitonic metal are best organized by the role played by excitons: condensate, precursor, dissociation source, or short-range correlation.

2. Symmetry-broken equilibrium excitonic metals in compensated semimetals

The most explicit equilibrium realization in the supplied literature is the mean-field theory for twisted double bilayer graphene near charge neutrality (Ghorai et al., 2022). There the low-energy moiré conduction and valence bands overlap slightly because of particle-hole asymmetry and especially trigonal warping, producing a compensated semimetal with three electron pockets and three hole pockets in each valley. Since the electron and hole pockets are displaced in momentum, the relevant bound states are indirect excitons carrying finite center-of-mass momentum.

The corresponding order parameter is finite-momentum interband coherence,

$$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$

Because the condensate mixes states separated by $\mathbf Q_i$, it is simultaneously a triple-$Q$ charge density wave. The choice of equal amplitudes at the three symmetry-related $\mathbf Q_i$ preserves overall $C_3$ symmetry within a valley and avoids a net valley current. The reconstructed quasiparticle energies are

$$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$

and metallicity is diagnosed by a nonzero Fermi-level density of states,

$$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$

The decisive point is partial rather than complete gapping. At charge neutrality, the ordered state becomes fully insulating only for $_3$0. For the representative coupling $_3$1, the mean-field order parameter peaks near $_3$2, with $_3$3, so the ordered state remains metallic. The paper infers a coupling window

$_3$4

within which excitonic order forms without eliminating all Fermi surfaces. In this restricted sense, the system is neither an ordinary Fermi-liquid metal nor an excitonic insulator, but an excitonic metal.

The transport phenomenology follows from the reconstructed minibands. Using $_3$5 as a proxy for resistivity, the theory produces a double-peak structure around charge neutrality as a function of density. Near the excitonic transition, Landau-damped order-parameter fluctuations generate a quantum critical metal with

$_3$6

whereas far from charge neutrality the system returns to $_3$7. The fluctuation action is controlled by

$_3$8

so the dynamical exponent is $_3$9. The result is a non-Fermi-liquid metal tied specifically to excitonic or CDW criticality inside a compensated semimetal.

The proposal is nevertheless model-dependent. It uses a two-band reduction, approximates $\mathbf Q_i$0, neglects same-band interactions and intervalley scattering, treats the ordered state at mean-field level, and uses the density of states as a transport proxy. Those caveats do not remove the conceptual importance of the construction, but they delimit the sense in which the phase is established.

3. Exciton-driven transient metallic states

A distinct usage of excitonic metal arises in ultrafast nonequilibrium settings. In NiPS$\mathbf Q_i$1, a quasi-two-dimensional correlated van der Waals insulator, the reported state is a transient antiferromagnetic metal driven by photoexcitation of subgap spin-orbit-entangled excitons built from Zhang-Rice-derived states (Belvin et al., 2021). NiPS$\mathbf Q_i$2 has a charge-transfer gap of $\mathbf Q_i$3, while sharp excitons lie around $\mathbf Q_i$4. Pumping at $\mathbf Q_i$5 with a $\mathbf Q_i$6 fs pulse therefore excites the excitonic manifold below the gap rather than directly photodoping hot carriers across it.

The metallic response is identified in THz conductivity, not in long-lived dc transport. After pumping, $\mathbf Q_i$7 shows a low-energy Drude-like response together with a narrow first-derivative-like magnon feature. At the peak delay $\mathbf Q_i$8, the fitted Drude parameters are

$\mathbf Q_i$9

with

$$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$0

The inferred mobility is $$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$1 for the bare mass, or $$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$2 using $$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$3. The scattering rate remains essentially constant in time, which the paper interprets as evidence that the carriers are cold rather than hot photocarriers cooling through strong boson emission.

What makes the state “excitonic” is not condensation but origin. The metallic carriers are inferred to arise from exciton dissociation. The magnetic oscillation amplitude scales linearly with absorbed fluence, $$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$4, whereas the Drude amplitude scales quadratically, $$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$5. Combined with control experiments on MnPS$$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$6 and FePS$$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$7, which show no corresponding photoinduced THz signal under the same $$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$8 eV pump, this supports a mechanism in which photoexcited excitons subsequently dissociate into mobile carriers. The metal is therefore exciton-driven, but not an exciton condensate.

The simultaneous preservation of antiferromagnetism is central. Below $$\frac{V_0}{\Omega}\sum_{\mathbf k}\left\langle C_{\mathbf k,\eta}^{c\dagger}C_{\mathbf k+\mathbf Q_i,\eta}^{v}\right\rangle=\Delta,\qquad i=1,2,3.$$9, NiPS$\mathbf Q_i$0 has zigzag antiferromagnetic order, and the equilibrium zone-center magnon lies at $\mathbf Q_i$1 with linewidth $\mathbf Q_i$2 at $\mathbf Q_i$3 K. After pumping, the magnon remains sharp and is only redshifted by $\mathbf Q_i$4, with no sizeable broadening. The Drude response persists for several picoseconds and disappears by about $\mathbf Q_i$5, whereas the coherent magnon survives much longer. The paper therefore describes a transient nonequilibrium metallic state that preserves long-range antiferromagnetism, a phase inaccessible by simple thermal tuning.

Photoinduced metallization of an excitonic insulator provides a related but not identical nonequilibrium route (Ejima et al., 2022). In the one-dimensional extended Falicov-Kimball model, time-dependent photoemission develops an extra band above the Fermi energy after pumping, indicating an insulator-to-metal transition. In the special SU(2)-symmetric case $\mathbf Q_i$6, with $\mathbf Q_i$7, $\mathbf Q_i$8, $\mathbf Q_i$9, and optimal pump parameters $Q$0, $Q$1, $Q$2, $Q$3, the pairing correlator $Q$4 saturates near $Q$5 while the excitonic correlator $Q$6 is strongly suppressed. In the Ta$Q$7NiSe$Q$8-motivated case $Q$9, $\mathbf Q_i$0, $\mathbf Q_i$1, $\mathbf Q_i$2, with optimal pump $\mathbf Q_i$3, $\mathbf Q_i$4, $\mathbf Q_i$5, $\mathbf Q_i$6, the metallic state is transient and the original excitonic correlations again decrease. The result is better described as a photoinduced correlated metal descended from an excitonic insulator than as a straightforward equilibrium excitonic metal.

4. Excitonic correlations inside metallic environments

A more limited sense of excitonic metal concerns metallic systems that display excitonic correlations without forming a stable excitonic metallic phase. The clearest ultrafast example is Ag(111), where three-photon photoemission with sub-$\mathbf Q_i$7 fs pulses reveals a transient surface exciton before screening is fully established (Cui et al., 2014). The relevant resonance is

$\mathbf Q_i$8

between the occupied Shockley surface state and the first image-potential state. Because screening saturation in Ag occurs on $\mathbf Q_i$9 fs timescales, unusually slow for a metal, the experiment accesses a preasymptotic regime in which the electron-hole pair initially experiences something close to the bare Coulomb attraction.

The dominant signature is a new nondispersive feature at

$C_3$0

above $C_3$1, more than $C_3$2 times more intense than the nonresonantly excited SS and IP bands and independent of the external laser photon energy. Its momentum-space nondispersion, anomalous intensity, and polarization response pinned at $C_3$3 rather than $C_3$4 are interpreted as evidence for a transient excitonic manifold. The corresponding state is explicitly not a stable equilibrium exciton in a metal, not an excitonic condensate, and not a long-lived bulk exciton. It is a nonequilibrium, surface-localized precursor that evolves into the screened image-potential quasiparticle configuration.

The low-density homogeneous electron gas furnishes a complementary many-body example (Koskelo et al., 2023). There the question is whether excitonic physics can survive long-range metallic screening. The answer is affirmative, but only at short distance and only with vertex corrections. For $C_3$5, the static dielectric function can become negative at finite $C_3$6. Standard GW-BSE with ordinary screened interactions fails to reproduce the low-energy effect. By contrast, using test-charge–test-electron screening,

$C_3$7

the BSE yields negative static screening, negative plasmon dispersion, and an abruptly softened low-energy collective mode near $C_3$8.

At $C_3$9, the collective mode energy drops sharply slightly before $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$0, falling by more than a factor of $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$1 up to about $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$2 before damping. The mode at

$$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$3

has a correlated electron-hole amplitude localized around the hole in the plane perpendicular to $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$4, while remaining more delocalized along $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$5. The paper therefore assigns excitonic characteristics “at least in two of the three dimensions.” By contrast, the ghost poles at $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$6,

$$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$7

are judged more plasmonic than excitonic. The broader lesson is that perfect long-range metallic screening does not eliminate all excitonic behavior; it eliminates conventional long-range excitons while permitting short-range anisotropic electron-hole correlations and collective modes.

5. Proximate instabilities and interfacial hybrids

Several supplied works are highly relevant to the topic because they delimit what should, and should not, be called an excitonic metal. Ultra-thin $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$8 is a clear example of a metal or semimetal driven into an excitonic insulator rather than into an excitonic metallic state (Chen et al., 5 Mar 2025). As-grown CuS is a p-type metal with hole-like bands centered at the $$E^{\pm}_{\mathbf Q_i}(\mathbf k)=\frac{\epsilon_\mathbf{k}^c+\epsilon_{\mathbf{k}+\mathbf Q_i}^v}{6}-\mu \pm \sqrt{\frac{(\epsilon_\mathbf{k}^c-\epsilon_{\mathbf{k}+\mathbf Q_i}^v)^2}{36}+\Delta^2},$$9 point crossing $$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$0. After annealing to intermediate stoichiometry, especially around $$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$1, the system becomes less p-doped and opens a full low-temperature gap. For that sample, the fitted transition temperature is

$$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$2

with hole density

$$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$3

A $$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$4 sample has

$$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$5

The low-$$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$6 gap is about $$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$7 in ARPES, while STS gives $$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$8. STM and LEED show no superlattice modulation, no translation symmetry breaking, and no structural transition. The sequence is therefore metal or semimetal $$g(\epsilon_F)=\sum_{\pm,i,\mathbf k}\delta\!\left(E^\pm_{\mathbf Q_i}(\mathbf k)-\epsilon_F\right).$$9 excitonic insulator $_3$00 band insulator, not a demonstrated excitonic metal.

Tilted massive Dirac models for inversion-broken TMDC monolayers lead to a similar conclusion (Quintela et al., 2022). The low-energy Hamiltonian is

$_3$01

with dispersion

$_3$02

The key structural fact is that the tilt changes the indirect gap without changing the exciton spectrum in the BSE treatment. For hBN encapsulation, the exciton energies are $_3$03 and $_3$04, and the instability occurs at about $_3$05, after the system has entered a semimetallic regime with very small carrier density. For quartz, $_3$06 and the instability begins near $_3$07, within the semiconducting regime. These results make the system a candidate for an excitonic insulator and suggest proximity to semimetallic excitonic physics, but the paper does not establish a stable excitonic metal and explicitly notes the screening caveat once $_3$08.

Interfacial and hybrid structures sharpen the boundary further. In semiconductor-metal nano-hybrids, the metal modifies an otherwise ordinary semiconductor exciton through electrostatics rather than through intrinsic metallic excitonic order (Climente et al., 2011). In CdS-Au matchstick structures, a neutral metallic tip shifts the exciton by only $_3$09, whereas a charged metallic tip rapidly redshifts the exciton and makes it spatially indirect; in neutral core-shell structures, the exciton redshifts by $_3$10. Likewise, in Ag/CdS plasmonic structures, the relevant object is a Wannier Exciton Plasmon Polariton rather than an excitonic metal (Khurgin, 2018). There the very strong coupling criterion is $_3$11, the exciton radius shrinks by a factor of few, and the effective ionization energy approaches $_3$12. These systems show that nearby metals can preserve, reshape, or even strengthen excitons, but they do not constitute metallic phases made excitonic.

6. Diagnostics, misconceptions, and open problems

The supplied literature supports a layered diagnostic hierarchy. A strict excitonic metal requires more than the coexistence of excitons and metallic carriers. In the most restrictive sense, it requires a metallic state in which excitonic order or constitutive excitonic correlations reconstruct the low-energy electronic structure while leaving $_3$13, as in the finite-$_3$14 condensate scenario for twisted double bilayer graphene (Ghorai et al., 2022). A looser but still physically meaningful sense covers transient metals generated from an exciton population, such as the Drude-conducting antiferromagnetic state in NiPS$_3$15, where the free carriers are produced by exciton dissociation and the conductivity remains a THz, picosecond-scale phenomenon rather than a long-lived equilibrium dc metal (Belvin et al., 2021). Still looser usages refer only to transient or short-range excitonic correlations in otherwise metallic systems, as on Ag(111) or in the low-density homogeneous electron gas (Cui et al., 2014, Koskelo et al., 2023).

Several recurrent misconceptions follow from collapsing these categories. An excitonic metal is not, in general, an excitonic insulator with incomplete terminology. The Cu$_3$16S results, for example, support a metal-to-excitonic-insulator transition with a full low-temperature gap, not a metallic ordered excitonic phase (Chen et al., 5 Mar 2025). Nor does the observation of transient excitons in a metal imply the existence of stable equilibrium excitons under static metallic screening, a point made explicitly by the Ag(111) work (Cui et al., 2014). Conversely, the presence of a nearby metal does not automatically destroy excitons, as shown by semiconductor-metal nano-hybrids and plasmon-exciton polaritonic structures, but those phenomena remain categorically distinct from a true excitonic metallic phase (Climente et al., 2011, Khurgin, 2018).

The major open questions are correspondingly heterogeneous. For equilibrium phases, a central issue is whether partial-gapping excitonic metals survive beyond mean-field treatments and simplified interaction models. For nonequilibrium states, the key unresolved problem is whether a stable metallic phase with robust excitonic correlations can persist rather than merely appear transiently after a pump; the photoinduced EFKM results explicitly leave that question open because the realistic non-SU(2) case yields only a transient correlated metal with suppressed equilibrium excitonic correlations (Ejima et al., 2022). For metallic excitonic responses, the open problems include the universality of transient excitons across surfaces, the proper treatment of evolving screening with fully time-dependent many-body theory, and the boundary between a transient bound exciton and a broader coherent electron-hole polarization (Cui et al., 2014). For low-density metals, the unresolved issue is how broadly the short-range excitonic mode of the homogeneous electron gas carries over to real materials with lattice, orbital, and disorder effects (Koskelo et al., 2023).

Taken together, the literature defines excitonic metal less as a single settled phase than as a family of metallic regimes in which excitonic degrees of freedom remain dynamically, spectroscopically, or order-parameter relevant. The most stringent current meaning is a partially gapped, symmetry-broken metal generated by exciton condensation. Broader usages describe exciton-driven transient metals or metals with short-range excitonic correlations. The scientific value of the term lies precisely in making those distinctions explicit.