Excitonic Gap Phase

• Excitonic gap phase is an interaction-driven state where Coulomb attraction binds electrons and holes into excitons that condense to open a many-body excitation gap.
• Spectroscopic and transport studies, including STM, ARPES, and Raman, consistently reveal characteristic gap signatures in systems like Ta₂NiSe₅ and InAs/GaSb bilayers.
• External controls such as doping, strain, pressure, and optical excitation tunably modify the gap, underscoring the sensitivity of electron–hole correlations and lattice effects.

The excitonic gap phase is an interaction-driven gapped state in which Coulomb attraction between conduction-band electrons and valence-band holes produces bound excitons and, under appropriate conditions, a condensate or coherent electron–hole hybridization that opens a many-body excitation gap. In narrow-gap semiconductors and semimetals, the canonical instability occurs when the exciton binding energy exceeds the bare single-particle gap or band overlap, so that the insulating gap is not a simple band-structure gap but a correlation-generated one (He et al., 2020, Yu et al., 2017). Across current literature, the term encompasses equilibrium excitonic insulators, field- or pressure-tuned condensates, and photo- or doping-assisted excitonic gap states in systems ranging from Ta$_2$NiSe$_5$ and InAs/GaSb bilayers to quantum-limit graphite, strained graphene, bilayer graphene, quantum-confined Sb nanoflakes, doped SnSe$_2$, and La$_3$Cd$_2$As$_6$ (Volkov et al., 2020, Zhu et al., 2015, Sharma et al., 2017, Mo et al., 11 Jul 2025, Kengle et al., 10 Jun 2025).

1. Excitonic instability and phase-space structure

In its standard form, the excitonic instability is formulated for a narrow-gap semiconductor with $E_G>0$ or a semimetal with band overlap $|E_G|$. Thermally excited electrons and holes bind into excitons of binding energy $E_B$, and if $E_B>E_G$, excitons condense spontaneously below a transition temperature, producing an excitonic insulator with a many-body excitation gap $_5$0 around the chemical potential (He et al., 2020). Early mean-field treatments summarized for zero-gap InAs/GaSb bilayers express the criterion in the compact form $_5$1, emphasizing that the ordered state can open a spectral gap even when $_5$2 or slightly negative (Yu et al., 2017).

A recurring organizing principle is the dependence on the underlying band gap. In Ta$_5$3NiSe$_5$4, a useful phase-diagram picture places $_5$5 and $_5$6 on a dome as a function of bare $_5$7, peaking near $_5$8, while on the semimetallic side free-carrier screening rapidly suppresses $_5$9 (He et al., 2020). By contrast, the extended three-dimensional Falicov–Kimball model separates the temperature $_2$0 for exciton-pair formation from a lower temperature $_2$1 associated with phase coherence, yielding a broad intermediate regime with a charge gap but no long-range phase order when $_2$2 (Apinyan et al., 2012). This separation is one concrete expression of the BCS–BEC crossover.

The crossover language is central to Ta$_2$3NiSe$_2$4. Pump–probe analysis found a largely temperature-independent gap up to approximately $_2$5–$_2$6 K, together with an additional temperature-dependent component emerging above that range, which was interpreted as placing the material in the middle of the theoretical BEC–BCS crossover (Werdehausen et al., 2018). Raman work similarly identified strong departures from mean-field behavior, a large gap-to-$_2$7 ratio, and an exciton coherence length comparable to the inter-exciton spacing, again situating the system in a strongly correlated crossover regime rather than a weak-coupling limit (Volkov et al., 2020).

2. Mean-field description of the excitonic gap

The minimal two-band description introduces conduction- and valence-band fermions coupled by an attractive electron–hole interaction. In the formulation summarized for Ta$_2$8NiSe$_2$9, a standard Hamiltonian is

$_3$0

with the excitonic order parameter

$_3$1

Mean-field factorization yields hybridized quasiparticle branches

$_3$2

and a self-consistent gap equation of the form

$_3$3

This structure makes the excitonic gap explicitly a many-body hybridization between conduction and valence sectors (He et al., 2020).

In Ta$_3$4NiSe$_3$5, the low-temperature gap is often written as

$_3$6

or equivalently $_3$7, where $_3$8 denotes the bare semiconductor gap and $_3$9 the excitonic order parameter (Mor et al., 2016). This decomposition is useful because time-resolved ARPES can track transient shifts of the valence-band top and conduction-band bottom and thereby infer changes in $_2$0. In the pump–probe interpretation, flattening of the valence-band top near $_2$1 and displacement of the conduction-band minimum are direct spectroscopic signatures of the correlation-driven term $_2$2 (Mor et al., 2016).

Related mean-field structures appear in several other platforms. In Sb nanoflakes, an indirect-gap two-band model with $_2$3 leads to Bogoliubov quasiparticle branches and a self-consistent gap equation involving $_2$4 (Li et al., 2024). In graphene and strained graphene, the same logic appears as an excitonic mass gap generated by Coulomb interactions; the self-energy acquires a $_2$5 component, and the quasiparticle dispersion becomes $_2$6 (Sharma et al., 2017). In quantum-limit graphite, the experimentally relevant activation gap follows

$_2$7

which explicitly combines the single-particle band gap $_2$8 with the excitonic correlation gap $_2$9 (Zhu et al., 2015).

3. Spectroscopic signatures, collective modes, and deviations from simple mean field

The excitonic gap phase is distinguished experimentally not only by the existence of a gap but also by how that gap appears in spectroscopy and by the collective dynamics that accompany it. In Ta$_6$0NiSe$_6$1, low-temperature STM at $_6$2 K with large tip–sample distance reveals a fully open, temperature-independent gap of approximately $_6$3 eV, whereas slightly below $_6$4 only a V-shaped pseudogap with finite zero-bias conductance is observed because of thermal broadening (He et al., 2020). Pump–probe reflectivity independently extracted $_6$5–$_6$6 meV from amplitudes and $_6$7 meV from relaxation times, with amplitudes and lifetimes essentially constant below $_6$8 K (Werdehausen et al., 2018).

Polarization-resolved Raman spectroscopy provides a complementary view of collective excitonic degrees of freedom. In Ta$_6$9NiSe$E_G>0$0, the $E_G>0$1 response contains a relaxational electronic continuum whose characteristic energy softens linearly according to $E_G>0$2, with $E_G>0$3 K, identifying a critical electronic mode distinct from optical phonons (Volkov et al., 2020). The same study found that coupling to noncritical optical phonons raises the transition temperature to $E_G>0$4 K, and including exciton–strain coupling further raises it to the observed $E_G>0$5 K. Below $E_G>0$6, a gap feature at approximately $E_G>0$7 meV appears in the $E_G>0$8 channel with a square-root divergence at onset, while the corresponding $E_G>0$9 intensity is nearly zero at the gap edge, matching coherence-factor selection rules for excitonic hybridization (Volkov et al., 2020).

Nonequilibrium studies expose additional structure beyond equilibrium mean field. Time- and angle-resolved photoemission on Ta$|E_G|$0NiSe$|E_G|$1 found that a $|E_G|$2 eV pump produces transient gap narrowing below a critical fluence $|E_G|$3 mJ/cm$|E_G|$4, but above $|E_G|$5 the flat valence-band peak at $|E_G|$6 reverses and shifts to larger binding energy after a delay of approximately $|E_G|$7 fs, yielding a transient gap enhancement by approximately $|E_G|$8–$|E_G|$9 meV (Mor et al., 2016). The accompanying Hartree–Fock analysis attributes this to a nonthermal carrier distribution together with Hartree shifts that transiently increase the self-consistent $E_B$0. A separate theory of photo-induced enhancement showed that interband phonon coupling makes the phase mode massive and allows long-lived amplitude oscillations at frequencies near both the excitonic gap and the phonon frequency, with transient increases of $E_B$1 and the minimum gap by order $E_B$2 (Murakami et al., 2017).

4. Representative material realizations

The excitonic gap phase has been reported in equilibrium, driven, and quasi-equilibrium settings with distinct microscopic emphases.

System	Reported gap behavior	Distinguishing observation
Ta$E_B$3NiSe$E_B$4 (He et al., 2020)	Fully open gap of $E_B$5 eV below $E_B$6 K	Reversible STM-tip-induced collapse to a zero-gap state
Ta$E_B$7NiSe$E_B$8 (Volkov et al., 2020)	Gap feature at $E_B$9 meV in Raman	Soft electronic $E_B>E_G$0 mode and coherence-factor contrast
InAs(10 nm)/GaSb(5 nm) (Yu et al., 2017)	Activated gap $E_B>E_G$1 meV at charge neutrality	Narrow $E_B>E_G$2 k$E_B>E_G$3 resistivity peak stable up to $E_B>E_G$4 T
Quantum-limit graphite (Zhu et al., 2015)	Field-tuned excitonic phase across a band-gap opening at $E_B>E_G$5 T	Asymmetric phase boundary around $E_B>E_G$6
Sb(110) nanoflakes (Li et al., 2024)	STS gap $E_B>E_G$7 meV centered at $E_B>E_G$8	$E_B>E_G$9 charge order without detectable periodic lattice distortion
Doped SnSe$_5$00/SnSe$_5$01 (Mo et al., 11 Jul 2025)	Anisotropic conduction-band gap, with $_5$02 meV and $_5$03 meV	Quasi-steady dark excitons detected by ARPES replica bands
La$_5$04Cd$_5$05As$_5$06 (Kengle et al., 10 Jun 2025)	Excitonic gap $_5$07 meV below $_5$08 K	Highly insulating state with no accompanying structural transition

These examples span several limiting cases. In zero-gap InAs/GaSb bilayers, the large resistivity peak at the charge neutrality point, its activated temperature dependence above $_5$09 K, and its stability against quantizing magnetic fields were argued to reflect an interaction-driven gap rather than a single-particle gap (Yu et al., 2017). In quantum-limit graphite, the excitonic phase evolves continuously from weak-coupling momentum-space pairing below $_5$10 to strong-coupling real-space pairing above $_5$11, with the maximum $_5$12 coincident with the band-gap opening (Zhu et al., 2015). In Sb nanoflakes, the defining observation is a charge density wave without periodic lattice distortion, which the authors treat as the hallmark of an excitonic rather than Peierls phase (Li et al., 2024). In La$_5$13Cd$_5$14As$_5$15, the absence of a structural transition and the failure of DFT+$_5$16SOC, DFT+U, Peierls-type distortions, or Cd-vacancy arrangements to open a gap are used to support an excitonic interpretation (Kengle et al., 10 Jun 2025).

5. External control: doping, light, strain, pressure, and nonequilibrium populations

One of the defining properties of the excitonic gap phase is its sensitivity to perturbations that modify screening, carrier balance, or the underlying band alignment. The clearest local control experiment is the STM-tip study of Ta$_5$17NiSe$_5$18. There, the tip–sample distance $_5$19 controls the local electric field through a work-function difference $_5$20 eV, with $_5$21 eV and $_5$22 eV. As $_5$23 is reduced from $_5$24 nm to $_5$25 nm, three regimes appear: for $_5$26 nm the $_5$27 eV gap is stable, for $_5$28 nm the gap rapidly shrinks, and for $_5$29 nm a V-shaped zero-gap local density of states emerges (He et al., 2020). The collapse is fully reversible, and the extracted gap plotted against surface electron density shows that a density of order $_5$30 electrons per unit cell suffices to destroy the gap. The proposed mechanism is tip-induced electrostatic charge accumulation at the topmost Se layer, which screens the Coulomb attraction and unbalances electron–hole populations.

Optical excitation provides a second control axis. In Ta$_5$31NiSe$_5$32, fluence below $_5$33 mJ/cm$_5$34 transiently narrows the gap via free-carrier screening, whereas above $_5$35 the nonthermal distribution and Hartree shifts transiently enhance the order parameter before relaxation on an approximately $_5$36 ps timescale (Mor et al., 2016). A time-dependent self-consistent $_5$37 treatment of a generic excitonic insulator identified two dynamical transition points under photoexcitation: a nonthermal trapping point at $_5$38 and a thermal critical point at $_5$39, with impact ionization identified as the main mechanism for gap melting (Golež et al., 2016). In pumped Dirac materials, population inversion described by separate electron and hole chemical potentials reduces the critical coupling for transient excitonic instability and can produce gaps of several to tens of meV, with formation times $_5$40 and observability controlled by $_5$41 (Triola et al., 2017).

Band anisotropy and lattice tuning provide further routes. In uniaxially strained graphene, the critical coupling for an excitonic mass gap decreases monotonically as the velocity-anisotropy parameter $_5$42 is reduced below $_5$43, indicating that strain supports excitonic gap generation (Sharma et al., 2017). In band-gap-tuned Ta$_5$44Ni(Se,S)$_5$45, ARPES locates the normal-state Lifshitz crossing near $_5$46 sulfur substitution, yet the broken-symmetry phase is continuously suppressed from the semimetal side to the semiconductor side rather than peaking at the Lifshitz point (Chen et al., 2023). Under pressure, Ta$_5$47NiSe$_5$48 first enters a semimetal at $_5$49 GPa and then develops a low-temperature partial gap of $_5$50–$_5$51 eV that is suppressed to zero at $_5$52 GPa, where superconductivity with maximum $_5$53 K emerges (Matsubayashi et al., 2021). In doped SnSe$_5$54, ARPES instead reveals a quasi-steady excitonic gap phase generated by photo-created holes and doped electrons forming momentum-indirect dark excitons; the gap magnitude scales with electron doping while the exciton binding energy remains essentially constant (Mo et al., 11 Jul 2025).

6. Interpretation, controversies, and conceptual boundaries

A central issue in the field is whether a measured gap is dominantly excitonic, lattice-hybridization-driven, or an intertwined state with both components. Ta$_5$55NiSe$_5$56 is the most developed example of this ambiguity. Several studies emphasize evidence for a many-body excitonic gap: the reversible tip-induced collapse of the $_5$57 eV gap under minute carrier doping, the nearly temperature-independent low-temperature gap, the quadrupolar Raman mode that softens as an electronic critical fluctuation, and ultrafast enhancement of the gap through transient increase of $_5$58 (He et al., 2020, Werdehausen et al., 2018, Volkov et al., 2020, Mor et al., 2016). These observations are difficult to reconcile with a rigid single-particle semiconductor gap, since in a conventional semiconductor doping would shift the chemical potential rather than close the gap (He et al., 2020).

At the same time, strong evidence exists that electron–phonon coupling is not a minor correction in Ta$_5$59Ni(Se,S)$_5$60. The band-gap-tuned ARPES/XRD study found that the broken-symmetry phase is most enhanced on the semimetal side and is monotonically suppressed toward the semiconductor side, contradicting the classic dome expected from a purely Coulomb-driven excitonic instability centered at the Lifshitz point (Chen et al., 2023). First-principles and model analyses there identify strong interband electron–phonon coupling, with $_5$61–$_5$62 meV and $_5$63 meV, as crucial to the observed symmetry breaking. The high-pressure study strengthens this view by arguing that the pressure-induced partial gap in the semimetallic phase originates primarily from symmetry-allowed Ta–Ni hybridization associated with monoclinic distortion, rather than from an excitonic instability surviving strong screening (Matsubayashi et al., 2021).

This tension has sharpened the distinction between systems with clear structural entanglement and those with more purely electronic phenomenology. Sb nanoflakes are presented as a case where a $_5$64 charge density wave is directly observed without detectable periodic lattice distortion, and La$_5$65Cd$_5$66As$_5$67 is presented as a bulk narrow-gap semiconductor with a highly insulating state below $_5$68 K and no accompanying structural transition (Li et al., 2024, Kengle et al., 10 Jun 2025). A plausible implication is that the phrase “excitonic gap phase” now covers a spectrum of situations: a near-ideal electronically driven excitonic insulator, an exciton–phonon cooperative instability, and quasi-steady excitonic gap states sustained by nonequilibrium or doped carrier distributions.

Across these variants, the unifying element is that the gap is tied to electron–hole correlations rather than solely to a preexisting one-electron band structure. That commonality underlies proposals for nanoscale phase switching, “excitonic transistor” functionality, ultrafast gap control, and excitonic circuits based on tiny charge modulations or light-induced carrier distributions (He et al., 2020, Mor et al., 2016, Mo et al., 11 Jul 2025).