Phantom-Divide Crossing in Dark Energy

• Phantom-divide crossing is the transition of dark energy’s equation-of-state parameter through w=-1, delineating quintessence-like (w>-1) and phantom-like (w<-1) regimes.
• The phenomenon arises in models with modified gravity, multi-component interactions, or dissipation, often tied to changes in the cosmic expansion rate and underlying dark-sector composition.
• Observational analyses using DESI, CMB, and supernova data suggest evolving dark energy with crossing behavior, yet statistical and parametrization challenges complicate its definitive detection.

Searching arXiv for recent and foundational papers on phantom-divide crossing to ground the article in published work. Phantom-divide crossing denotes the evolution of a dark-energy equation-of-state parameter through the value $w=-1$, the boundary between quintessence-like behavior ($w>-1$) and phantom-like behavior ($w<-1$). In contemporary cosmology the term is used in two closely related senses: a crossing of a fundamental dark-energy component, and a crossing of an effective dark-energy sector reconstructed from the background expansion. The distinction is central. In viable $f(R)$ gravity, for example, future crossings arise generically from the approach to a stable de Sitter state (Bamba et al., 2010), whereas recent DESI-motivated analyses emphasize that low-redshift crossing can be a genuine dynamical signal, an effective artifact of a composite dark sector, or even a consequence of statistical fluctuations and parametrization choice (Keeley et al., 18 Jun 2025).

1. Definition and kinematic characterization

The basic quantity is

$w \equiv \frac{p}{\rho},$

with $p$ the pressure and $\rho$ the energy density of the component identified as dark energy. The value $w=-1$ corresponds to a cosmological constant. A phantom-divide crossing occurs when the evolving equation of state passes through this boundary, so that $1+w$ changes sign.

At the level of the homogeneous background, the useful kinematic relation is

$w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$

When nonrelativistic matter becomes negligible, one has $w>-1$0, so the sign of $w>-1$1 determines the regime: $w>-1$2 corresponds to quintessence-like evolution, and $w>-1$3 to phantom-like evolution. In this sense, a crossing of $w>-1$4 is equivalent to a sign change of $w>-1$5 whenever dark energy dominates strongly. In numerical studies the crossing is often displayed by plotting $w>-1$6; the condition $w>-1$7 marks the event itself (Bamba et al., 2010).

A second distinction concerns effective versus fundamental crossing. In multi-component or interacting models, the quantity reconstructed from distances and expansion rates can be

$w>-1$8

even when no individual component ever crosses $w>-1$9. This suggests that observationally inferred phantom-divide crossing need not imply a single phantom field (Gómez-Valent et al., 1 Aug 2025).

2. Why crossing is theoretically nontrivial

For a single minimally coupled scalar field with standard kinetic term, smooth crossing of $w<-1$0 is highly constrained or impossible. Canonical quintessence obeys $w<-1$1, while a fundamental phantom scalar with $w<-1$2 typically requires a negative kinetic term and is associated with ghosts and vacuum instabilities. This is why phantom-divide crossing has long been treated as a marker of noncanonical dynamics, multiple degrees of freedom, or modified gravity rather than ordinary single-field quintessence (Gómez-Valent et al., 1 Aug 2025).

This theoretical difficulty is sharpened by recent data analyses. DESI BAO, Planck CMB, and supernova compilations favor evolving dark energy, and CPL fits with $w<-1$3 and $w<-1$4 imply a transition from quintessence-like behavior today to phantom-like behavior at higher redshift. Yet a Monte Carlo analysis based on 1,000 mock datasets generated from a non-crossing Padé-$w<-1$5 quintessence model found that in $w<-1$6 of realizations CPL with phantom crossing not only fits better than the non-crossing model, but exceeds the real-data improvement $w<-1$7. This establishes that evolving dark energy is a robust signal, but that the precise claim of crossing is not yet definitive (Keeley et al., 18 Jun 2025).

A common misconception is therefore that any preferred $w<-1$8 interval demands a ghostly phantom field. The current literature instead treats crossing as a model-selection problem involving background degeneracies, perturbative stability, and the distinction between physical and effective dark-energy sectors.

3. Effective and composite crossings in interacting or multi-component sectors

One major line of work interprets phantom-divide crossing as an emergent property of a composite or interacting dark sector. In the two-field “standard quintessence + negative quintessence” construction, the dark-energy sector is the sum of a canonical scalar and an exotic component with negative energy density and positive pressure, both decreasing in absolute value and tending to zero in the future. Neither component individually crosses $w<-1$9, but their sum generates an effective $f(R)$0 that moves from phantom-like to quintessence-like and produces a peak in $f(R)$1 around $f(R)$2. Relative to $f(R)$3CDM, this model improves the fit by $f(R)$4 and is preferred at $f(R)$5, comparable to the CPL level of exclusion (Gómez-Valent et al., 1 Aug 2025).

A second mechanism uses dissipation. In a canonical scalar-field model with energy transfer into dark matter, the dark-energy field remains nonphantom, but the inferred effective dark-energy equation of state can cross $f(R)$6. In the dissipative scenario proposed to explain DESI, the interaction is governed by

$f(R)$7

and even weak dissipation is sufficient to reproduce an evolving $f(R)$8 and recent phantom-divide crossing without invoking any pathological phantom-like dynamics for the quintessence field (Chanda et al., 3 Jun 2026).

A third realization is an interacting dark-sector field theory in which fermionic dark matter is Yukawa-coupled to a Born–Infeld tachyonic scalar. The scalar equation of state remains nonphantom and bounded, while the effective dark-energy equation of state undergoes a recent double crossing due to the sign-changing interaction term. This construction was confronted with DESI DR2 BAO, Planck 2018 distance priors, and several supernova compilations, and the reconstructed background allows a recent double crossing of the phantom divide (Abdalla et al., 19 May 2026).

Closely related is exponentially coupled quintessence with an exponential potential and an exponential coupling to CDM. When $f(R)$9 eV is fixed, the data favor a nonzero coupling $w \equiv \frac{p}{\rho},$0 at more than $w \equiv \frac{p}{\rho},$1, and specifically the $w \equiv \frac{p}{\rho},$2 branch, where the energy transfer changes sign and the effective equation of state crosses $w \equiv \frac{p}{\rho},$3. When the effective neutrino mass parameter $w \equiv \frac{p}{\rho},$4 is allowed to take negative values, that extra freedom weakens the preference for the coupling and makes the $w \equiv \frac{p}{\rho},$5 and $w \equiv \frac{p}{\rho},$6 branches nearly indistinguishable in $w \equiv \frac{p}{\rho},$7, so a noncrossing explanation becomes viable as well (Wang et al., 20 Jun 2026).

These constructions collectively suggest that a reconstructed crossing can reflect dark-sector composition, interaction, or misidentification of a multi-fluid system as a single fluid.

4. Modified-gravity realizations

Modified gravity offers a geometric route to phantom-divide crossing in which the effective dark-energy sector is built from departures from Einstein gravity rather than from a fundamental phantom field.

Before listing specific theories, two general features recur. First, the crossing is often tied to the sign change of $w \equiv \frac{p}{\rho},$8, hence to an oscillatory or nonmonotonic approach to a late-time attractor. Second, the resulting $w \equiv \frac{p}{\rho},$9 is effective: it arises from the gravitational field equations rewritten in fluid form.

Framework	Dynamical origin of crossing	Characteristic behavior
Phantom Crossing DGP	$p$0 in the DGP correction term	Single smooth crossing at $p$1, with best-fit $p$2 and $p$3
Viable $p$4 gravity	Scalaron oscillations around a stable de Sitter point	Multiple future crossings, oscillatory $p$5, $p$6, and horizon entropy
$p$7 gravity	Nonlinear torsion functions	Model A gives double crossing; Model B allows direction-controlled single crossing
$p$8 gravity	Sign change of $p$9 in combined models	Combined theory crosses $\rho$0; future crossings are generic in feasible models

In DGP cosmology, the original self-accelerating model and the Dvali–Turner extension with constant $\rho$1 do not cross $\rho$2: $\rho$3 for $\rho$4, $\rho$5 for $\rho$6, and $\rho$7 for $\rho$8. The “Phantom Crossing DGP” model instead promotes the exponent to $\rho$9, so that the sign of $w=-1$0 changes during cosmic evolution and the effective equation of state crosses the phantom divide smoothly. A joint SNIa+CMB+BAO fit yields $w=-1$1, $w=-1$2, and $w=-1$3, better than both the original DGP and the Dvali–Turner model (Hirano et al., 2010).

In viable $w=-1$4 gravity, the archetypal result is that future phantom-divide crossing is generic in the Hu–Sawicki, Starobinsky, Tsujikawa, and exponential models. These theories admit a stable late-time de Sitter attractor; the scalaron oscillates as it relaxes to that attractor; and $w=-1$5, $w=-1$6, and $w=-1$7 oscillate accordingly. Because

$w=-1$8

each sign change of $w=-1$9 induces a crossing of $1+w$0. The same oscillatory dynamics causes the cosmological horizon entropy

$1+w$1

to oscillate in time. Present-day values are $1+w$2 to $1+w$3, while the first future crossings occur around $1+w$4 to $1+w$5 (Bamba et al., 2010).

In $1+w$6 gravity, explicit torsion-based models can realize both single and double crossing. Model A,

$1+w$7

crosses only if $1+w$8 for $1+w$9, and once it crosses, it must cross twice. Its best fit is $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$0, $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$1, with $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$2. Model B,

$w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$3

crosses when $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$4 and $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$5 have the same sign; its best fit is $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$6, $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$7, $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$8, with $w_{\rm eff}=-1-\frac{2}{3}\frac{\dot H}{H^2}.$9 (Wu et al., 2010).

In $w>-1$00 gravity, the exponential and logarithmic models remain on one side of the divide, but a combined exponential–logarithmic theory produces genuine crossing. The paper’s reconstruction from cosmographic data leads to $w>-1$01, and the authors conclude that future crossings of the phantom-dividing line are a generic feature of feasible $w>-1$02 gravity models (Arora et al., 2022).

5. Scalar-tensor, vector-tensor, and exact-solution mechanisms

Beyond fluid reconstructions and modified curvature/torsion/non-metricity theories, phantom-divide crossing appears in a wide range of scalar-tensor, vector-tensor, and exact-solution constructions. In a tachyon model with general non-minimal kinetic couplings $w>-1$03 and $w>-1$04, the crossing condition is formulated in terms of $w>-1$05 with $w>-1$06, and the authors show that the crossing can occur even if $w>-1$07 and $w>-1$08, i.e. even when the potential is asymptotically at its minimum. For the parameter sets displayed, the sound speed satisfies $w>-1$09, indicating classical stability during the crossing (Banijamali et al., 2012).

A different stability question arises in Horndeski and generalized Proca theories. In shift-symmetric Horndeski and generalized Proca models with luminal gravitational-wave speed and no direct couplings to dark matter, the paper argues that a low-redshift transition from $w>-1$10 to $w>-1$11 is generically difficult without ghosts, strong coupling, or Laplacian instabilities. Breaking the shift symmetry in Horndeski by adding a potential changes this conclusion: an explicit scalar-tensor model with a potential, a Galileon self-interaction, and a higher-order derivative term realizes the desired crossing while avoiding ghosts and Laplacian instabilities (Tsujikawa, 24 Aug 2025).

Generalized Proca theory supplemented by a canonical scalar provides a vector-tensor realization of the same idea. The vector sector alone can sustain $w>-1$12, but the addition of a canonical scalar field with a potential allows a low-redshift crossing from $w>-1$13 to $w>-1$14. The theory admits a parameter region in which all scalar, vector, and tensor degrees of freedom are free from ghost and Laplacian instabilities, and the quasi-static observables $w>-1$15 and $w>-1$16 can remain close to 1 because the longitudinal scalar sound speed $w>-1$17 is influenced by the transverse vector mode (Tsujikawa, 29 Jan 2026).

Exact-solution approaches reach similar conclusions from a different angle. Chiellini-integrable scalar cosmologies derived from a canonical scalar minimally coupled to Einstein gravity with an extended Higgs-like potential produce exact background solutions in which the effective dark-energy equation of state crosses the phantom divide smoothly while the scalar remains canonical. The reconstruction gives $w>-1$18 and a reduced tension relative to $w>-1$19CDM (Chakrabarti et al., 14 Jan 2026). In another exact construction, a spatially flat FRW universe containing a stiff fluid, a cosmological constant, and a classical Dirac field yields an effective dark-energy component

$w>-1$20

with

$w>-1$21

so the phantom divide is crossed precisely at

$w>-1$22

Here the crossing is driven by the negative-energy part of the Dirac sector rather than by a phantom scalar (Cataldo et al., 2010).

These models reinforce a broad conclusion: canonical or otherwise healthy underlying fields can generate effective phantom-divide crossing once derivative couplings, additional vector modes, exact nonlinear structures, or multifluid decompositions are present.

6. Observational status, ambiguities, and cosmological implications

The observational status of phantom-divide crossing is suggestive rather than settled. Planck CMB, DESI BAO, and supernova samples repeatedly point toward evolving dark energy; CPL fits often prefer $w>-1$23 and $w>-1$24, implying a recent crossing; and composite or interacting models can reproduce the same background behavior with comparable statistical quality. Yet the interpretation remains model-dependent. A crossing can be genuine, apparent, or simply overfitted by a flexible parametrization (Keeley et al., 18 Jun 2025).

A recent data-driven program within VCDM makes this point sharply. VCDM is a minimally modified gravity theory with only the two tensor degrees of freedom of GR, but it allows both background and linear perturbations to evolve consistently across the phantom divide. Bayesian spline reconstruction of $w>-1$25 from CMB, BAO, and type-Ia supernova data finds a preference for smooth, monotonic phantom-crossing trajectories, while Bayesian evidence disfavors increasingly complex spline models. Exhaustive symbolic regression then identifies a simple one-parameter form,

$w>-1$26

which naturally crosses the phantom divide for $w>-1$27, suppresses early dark energy, and predicts a transient accelerating and phantom phase without a future big-rip singularity. Bayesian comparison yields mild-to-moderate support relative to standard two-parameter alternatives and stronger evidence relative to $w>-1$28CDM (Borghetto et al., 16 Jun 2026).

Several implications follow. First, crossing does not by itself imply a Big Rip. In viable $w>-1$29 gravity the universe instead asymptotes to a stable de Sitter state after damped oscillations of $w>-1$30, $w>-1$31, and horizon entropy (Bamba et al., 2010). Second, the crossing can be future, recent, single, multiple, or double depending on the dynamical mechanism. Third, perturbations remain decisive: many background-level constructions that cross $w>-1$32 are excluded once ghosts, Laplacian instabilities, growth-rate constraints, or ISW–galaxy correlations are imposed.

The contemporary literature therefore treats phantom-divide crossing not as a single phenomenon but as a diagnostic of dark-sector structure. It can signal oscillatory modified gravity, dark-sector interactions, effective multi-fluid composition, or a genuinely new low-complexity background law. Current data favor dynamical dark energy strongly enough that crossing remains a live and technically rich possibility, but they do not yet force a unique physical interpretation.