Stranger than a constant. If w is below -1, dark energy strengthens forever and rips the universe apart.
Phantom Energy
Dark energy with an equation of state below -1: its density grows as the universe expands, ending in a Big Rip that tears apart every bound structure.
Placeholder for a 3D visualisation of Dark Energy Candidates. The interactive scene will land in Phase 3. Roughly 69% of the universe's energy density drives accelerating expansion. Plain ΛCDM models it as a cosmological constant with equation of state w fixed at -1. This family asks what dark energy is as a physical substance, not how to parameterize it. The leading physical candidate is quintessence, a dynamical scalar field with w greater than -1 that evolves over cosmic time; phantom energy is its mirror image with w less than -1. DESI's 2024 hint of evolving dark energy (w0 greater than -1, wa less than 0) is exactly the signature a physical field would leave, which is why these models are an active frontier rather than a settled question.
§1 · The claim, in one sentence
Phantom energy is dark energy with equation of state w less than -1. Its density grows as space expands, rather than staying constant or diluting, driving a runaway acceleration that ends in a Big Rip.
§2 · Why it might be true
Three possibilities sit side by side. A cosmological constant has w exactly -1, its density never changing. Quintessence has w greater than -1, its density slowly diluting. Phantom energy is the third option, w less than -1, where the density actually grows as the universe expands. This is the strangest case, and the reason this entry exists opposite quintessence: the two bracket the cosmological constant from above and below.
A growing dark energy density means acceleration that intensifies without limit. Caldwell, Kamionkowski, and Weinberg (2003) traced where it leads, a Big Rip. The expansion rate diverges in finite time, and in the final moments it overwhelms gravity, then electromagnetism, then the nuclear forces, unbinding galaxy clusters, then galaxies, then solar systems, then atoms and nuclei.
Phantom energy is not a comfortable theory; it is studied because the data keep flirting with the w less than -1 side. Fits to supernovae, BAO, and the CMB often place the best-fit w slightly below -1, and DESI's evolving-dark-energy posteriors cross the phantom divide in the past. Whether that crossing is physical or a sign of unmodeled systematics is the open question.
The family stance
Dark energy may not be a constant. If its equation of state differs from -1, or changes with time, a cosmological constant cannot be the whole story and a dynamical component, most naturally a scalar field, is required. The case that the universe accelerates is overwhelming; the case for any one physical model of what drives it is not.
§2.5 · Evidence
- Best-fit values of w from several data combinations have repeatedly landed slightly below -1, and DESI 2024 posteriors cross into the phantom regime in the past (the wCDM and w0waCDM entries cover the parameterized fits).
- Phantom behavior is straightforward to write down as an effective equation of state, which makes it a natural benchmark against which a true cosmological constant is tested.
§3 · What you'd need to test it
- An equation of state w less than -1, with the dark energy density increasing over cosmic time rather than staying constant or diluting.
- A future Big Rip: the scale factor diverges in finite time, unbinding structures from clusters down to atoms in a fixed sequence.
- A measurably different expansion history from quintessence: at the same value of w today, phantom energy was less dense in the past, quintessence denser.
- A crossing of the phantom divide (w = -1) visible in the reconstructed w(z) from combined supernova, BAO, and CMB data.
§4 · Where it breaks
- A canonical scalar field cannot reach w less than -1; phantom energy requires a non-canonical kinetic term, a ghost field with negative kinetic energy, which is unstable as a quantum field theory because the vacuum decays.
- Phantom energy violates the dominant energy condition (rho + p less than 0), which most physicists regard as a serious warning rather than a feature.
- The Big Rip is avoided if the apparent w less than -1 is instead a modified-gravity effect or a supernova-calibration systematic; no measurement yet establishes a genuine phantom-divide crossing.
Go deeper
For a fluid with constant equation of state w, the density evolves as rho proportional to a^(-3(1+w)). For w less than -1 the exponent is positive, so rho grows as the scale factor a grows, which is the defining phantom feature. Caldwell (2002) coined the term in 'A Phantom Menace?' and flagged the energy-condition violation; Caldwell, Kamionkowski, and Weinberg (2003) computed the Big Rip timescale, finding bound systems torn apart in reverse order of their binding energy as the Hubble rate diverges.
Realizing w less than -1 in a field theory is the hard part. A scalar with a wrong-sign (negative) kinetic term gives w less than -1 but is a ghost: its energy is unbounded below and the vacuum is unstable to runaway pair production. Viable phantom-like behavior therefore tends to come from non-minimal kinetic terms, coupled multi-field models, or modified gravity that mimics w less than -1 without a fundamental ghost (the Geometric Modified Gravity family treats this geometric route, the same acceleration data explained without a dark-energy fluid), rather than from a fundamental phantom scalar.
Variants in this family
▸§5 · Who built it, and when(3 sources, 3 established)
- EstablishedCaldwell (2002). A phantom menace? Cosmological consequences of a dark energy component with super-negative equation of state. Phys. Lett. B 545, 23
- EstablishedCaldwell, Kamionkowski & Weinberg (2003). Phantom energy and cosmic doomsday. Phys. Rev. Lett. 91, 071301
- EstablishedCaldwell, Dave & Steinhardt (1998). Cosmological imprint of an energy component with general equation of state. Phys. Rev. Lett. 80, 1582
Up next
Spotted an error? Have a source to add?
Prefer email?
You can also send a prefilled email with the variant URL already filled in.