Ultralight Scalar Dark Matter Overview

• Ultralight scalar dark matter is a class of bosonic fields with masses far below the eV scale, oscillating coherently with macroscopic de Broglie wavelengths.
• It exhibits diverse phenomenology through effective couplings and portal structures, including linear and quadratic interactions with Standard Model fields.
• Production via misalignment and environmental mass effects drives its cosmological evolution, while precision experiments and astrophysical observations set multifaceted constraints.

Ultralight scalar dark matter (ULSDM) denotes bosonic dark matter candidates with masses far below the eV scale, often described as coherently oscillating classical fields because their occupation numbers are enormous and their de Broglie wavelengths are macroscopic. In the literature, the term covers both “fuzzy” dark-matter scales around $m_\phi \sim 10^{-22}\,\mathrm{eV}$ and higher ultralight masses extending to the $\mu\mathrm{eV}$–$\mathrm{meV}$ regime in specific portal constructions. Its phenomenology is correspondingly broad: the scalar may be generated by standard misalignment, thermal or environmental mass effects, or non-minimal gravitational couplings; it may couple linearly or quadratically to Standard Model operators, universally through an effective metric, to neutrinos, or through flavor-violating portals; and it can be constrained by Big Bang Nucleosynthesis (BBN), compact objects, pulsar timing, clocks, atom interferometers, neutrino experiments, and other precision probes (Davoudiasl, 2024, Banerjee et al., 2022, Murgui et al., 2023).

1. Coherent-field description and kinematic regime

A standard starting point is a real scalar field $\phi$ with ultralight mass $m_\phi$, treated as a classical background because of its large occupation number. In halo applications the field is written as

$$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$

with $\omega \simeq m_\phi$ and $k \simeq m_\phi v$ for a nonrelativistic halo. The coherence scales are

$$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$

and for virialized dark matter $v\simeq 10^{-3}c$ the quality factor is $\mu\mathrm{eV}$0 (Brzeminski et al., 22 Jan 2026, Zhao et al., 2021).

Once oscillations begin, the scalar behaves as pressureless matter. In the bare-mass regime,

$\mu\mathrm{eV}$1

and in WKB regimes with time-dependent effective mass one instead has $\mu\mathrm{eV}$2, but still $\mu\mathrm{eV}$3 after oscillations start (Sibiryakov et al., 2020, Bouley et al., 2022).

The classical-field description is not the only useful language. For high-energy neutrino propagation through an ultralight background with $\mu\mathrm{eV}$4, ULSDM has also been modeled as a cold, nonrelativistic particle bath with $\mu\mathrm{eV}$5, because the relevant observable is scattering with absorption and regeneration rather than coherent forward scattering (Reynoso et al., 2016). A common misconception is therefore that ULSDM must always be treated in one universal way; in practice, the appropriate description depends on the observable.

2. Effective couplings and portal structures

A prominent class of models uses universal couplings that preserve the weak equivalence principle by coupling $\mu\mathrm{eV}$6 to Standard Model matter through an effective metric,

$\mu\mathrm{eV}$7

In the Einstein frame the leading interaction is $\mu\mathrm{eV}$8, so fermion and vector masses rescale as $\mu\mathrm{eV}$9, while WEP preservation also forces $\mathrm{meV}$0 and $\mathrm{meV}$1 in the Jordan frame. Physical observables are frame-invariant even though practical implementations differ (Sibiryakov et al., 2020).

A second broad EFT parameterization employs Damour–Donoghue-type linear and quadratic couplings. For quadratic interactions one writes

$\mathrm{meV}$2

with the induced fractional shifts

$\mathrm{meV}$3

The general phenomenology of linear and quadratic $\mathrm{meV}$4 couplings includes oscillations at $\mathrm{meV}$5 and $\mathrm{meV}$6, DC offsets for quadratic couplings, and source-dependent screening effects near massive bodies (Bouley et al., 2022, Banerjee et al., 2022).

ULSDM portals are not limited to diagonal “dilaton-like” deformations. Neutrino-specific couplings can modulate neutrino masses and mixings through

$\mathrm{meV}$7

while right-handed-neutrino portals use

$\mathrm{meV}$8

and make the scalar evolution sensitive to the finite-density neutrino background (Krnjaic et al., 2017, Plestid et al., 2024). Flavor-violating constructions provide another qualitatively distinct possibility,

$\mathrm{meV}$9

which induces $\phi$0 CKM modulation and $\phi$1 down-quark mass shifts (Guo et al., 18 Mar 2026). Gauge-kinetic couplings,

$\phi$2

open yet another sector of ULSDM phenomenology, including late-time magnetogenesis and oscillatory variation of $\phi$3 (Kamali et al., 1 Jan 2026).

3. Production mechanisms and cosmological evolution

The canonical production mechanism is misalignment: a light scalar displaced from its minimum begins oscillating once $\phi$4 drops below the relevant mass scale and then redshifts like cold matter. In models with quadratic couplings this simple picture is modified because the scalar acquires a finite-temperature or finite-density effective mass,

$\phi$5

and the onset of oscillations satisfies $\phi$6. For electron couplings,

$\phi$7

while for photon couplings the leading high-temperature contribution is two-loop and scales as $\phi$8 (Bouley et al., 2022).

In universally coupled models the environmental mass is governed by the Standard Model trace,

$\phi$9

During radiation domination relativistic species contribute negligibly, but around BBN the $m_\phi$0 plasma becomes nonrelativistic and generates a large $m_\phi$1. This produces three characteristic regimes: Hubble friction, a bare-mass oscillatory regime, and an induced-mass regime. For $m_\phi$2, induced-mass domination can move the onset of oscillations earlier and suppress $m_\phi$3 by BBN; for $m_\phi$4, the induced mass becomes tachyonic and can trigger exponential growth unless the initial conditions are extremely fine-tuned (Sibiryakov et al., 2020).

Several works study departures from ordinary misalignment. “Gravitational misalignment” introduces a non-minimal curvature coupling,

$m_\phi$5

so that during radiation domination trace-anomaly effects make $m_\phi$6, and $m_\phi$7 can generate a tachyonic phase with

$m_\phi$8

For the benchmark $m_\phi$9, $$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$0, $$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$1, and $$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$2, one finds $$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$3, $$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$4, and $$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$5 (Davoudiasl, 2024).

A further departure arises when ULSDM couples to right-handed neutrinos. In that case the relic neutrino density generates an asymmetric finite-density effective potential, producing a “brick-wall” regime in which the scalar energy density redshifts nontrivially and can undergo an $$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$6 drop between recombination and today. This implies that CMB neutrino-mass bounds and laboratory oscillation bounds constrain different regions of parameter space rather than the same naive misalignment amplitude (Plestid et al., 2024).

4. Big Bang Nucleosynthesis and early-universe limits

BBN provides one of the most developed cosmological probes of ULSDM because scalar backgrounds can shift the effective values of Standard Model masses and couplings during neutron freeze-out and the deuterium bottleneck. In universally coupled models, the relevant variations include

$$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$7

The helium-4 mass fraction depends mainly on the neutron-to-proton ratio at freeze-out and on neutron decay before nucleosynthesis,

$$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$8

and the observationally allowed fractional deviation is

$$\phi(t,\mathbf{x}) \simeq \phi_0 \cos(\omega t - \mathbf{k}\!\cdot\!\mathbf{x} + \theta),\qquad \phi_0=\sqrt{\frac{2\rho_{\rm DM}}{m_\phi^2}},$$9

A full kinetic treatment shows that precision helium-4 measurements set the dominant constraint, deuterium is subdominant, and the lithium effect is insufficient to solve the cosmological lithium problem (Sibiryakov et al., 2020).

The same conclusion persists in more general quadratic-coupling analyses, but with important refinements. Once the finite-temperature thermal mass and the full weak freeze-out kinetics are included, BBN constrains $\omega \simeq m_\phi$0 up to $\omega \simeq m_\phi$1, with qualitatively different scaling regimes depending on whether the field is Hubble-frozen, thermally-induced-mass dominated, or bare-mass dominated during freeze-out. The inclusion of thermal mass and kinetic freeze-out weakens the electron-coupling bound by about two orders of magnitude at $\omega \simeq m_\phi$2, strengthens the quark-coupling bound by a factor of about 2 in the same range, and yields new gluon-coupling constraints (Bouley et al., 2022).

A central technical result is that frame invariance is preserved only if freeze-out is handled beyond the instantaneous approximation. In the Einstein frame ULSDM appears as time-dependent masses and widths, whereas in the Jordan frame it appears as a modified Hubble rate,

$\omega \simeq m_\phi$3

The numerical implementation in AlterBBN v2.2 therefore proceeds through a modified “dark density”

$\omega \simeq m_\phi$4

with conservative cuts imposed once the linearized treatment begins to fail (Sibiryakov et al., 2020).

The sign of the coupling matters sharply. For $\omega \simeq m_\phi$5, BBN bounds are strongest at very low $\omega \simeq m_\phi$6, weaken for $\omega \simeq m_\phi$7 because induced-mass suppression lowers $\omega \simeq m_\phi$8 at BBN, and develop oscillatory exclusion bands at intermediate masses. For $\omega \simeq m_\phi$9, a large tachyonic region during BBN requires extreme fine-tuning of the initial field value to avoid dark-matter overproduction and is effectively excluded; outside that region, the BBN effect is negligible, though neutron-star scalarization may remain relevant (Sibiryakov et al., 2020).

5. Observational searches from clocks to pulsar timing

Laboratory and space-based frequency metrology probe ULSDM through periodic shifts of atomic transition frequencies, masses, and effective constants. ISS-based clocks are especially relevant for quadratically coupled ULSDM because atmospheric shielding suppresses the ground-level field for sufficiently strong couplings, whereas at orbital altitudes the orbit-averaged monopole coefficient approaches $k \simeq m_\phi v$0 for $k \simeq m_\phi v$1. When $k \simeq m_\phi v$2, equivalently $k \simeq m_\phi v$3, the scattered profile around Earth develops a large dipole with $k \simeq m_\phi v$4 at ISS altitude in the geometric-optics regime. In this setting, optical clocks can be comparable to MICROSCOPE for some couplings, and LEO nuclear clocks can set world-leading bounds for quark/gluon couplings (Brzeminski et al., 22 Jan 2026).

Long-baseline atom interferometers provide a complementary route. In ZAIGA, the ULSDM-induced differential phase contains contributions from changes in atomic internal energy levels, atomic masses, and the gravitational acceleration. Under future assumptions, the vertical configuration can improve over MICROSCOPE by about $k \simeq m_\phi v$5–$k \simeq m_\phi v$6 orders of magnitude for $k \simeq m_\phi v$7, while the horizontal configuration can exceed MICROSCOPE by more than three orders of magnitude for $k \simeq m_\phi v$8 and $k \simeq m_\phi v$9 in the range $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$0–$$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$1 (Zhao et al., 2021). A distinct atom-interferometric channel uses oscillations of nuclear charge radii induced by scalar or axion dark matter, converting them into differential phases through field-shift-sensitive clock transitions such as the Yb $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$2 line; projected bounds on $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$3, $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$4, and $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$5 were found to improve current levels and complement other experiments (Zhao et al., 2024).

Neutrino oscillation experiments probe ULSDM differently. Fast scalar modulation of neutrino masses and mixing angles can average into “Distorted Neutrino Oscillations” (DiNOs), with the time-averaged two-flavor expressions involving Bessel-function suppression,

$$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$6

Existing data already imply $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$7 from T2K, $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$8 from Daya Bay, and $$\lambda_{\rm dB}=\frac{2\pi}{m_\phi v},\qquad \tau_c\sim \frac{2\pi}{m_\phi v^2},$$9 from KamLAND, while projections indicate JUNO sensitivity at the $v\simeq 10^{-3}c$0 level and DUNE sensitivity of order half its energy resolution for atmospheric-sector distortions (Krnjaic et al., 2017). In the ESS$v\simeq 10^{-3}c$1SB study, the averaged spectral-distortion parameter $v\simeq 10^{-3}c$2 is constrained at $v\simeq 10^{-3}c$3 confidence to $v\simeq 10^{-3}c$4 for $v\simeq 10^{-3}c$5 and $v\simeq 10^{-3}c$6 for $v\simeq 10^{-3}c$7, with a mass window $v\simeq 10^{-3}c$8 in the averaged-distortion regime (Cordero et al., 2022).

Muonium and muonic-atom spectroscopy target scalar–muon couplings that are otherwise hard to access because stable bulk matter contains no muons. The ULSDM-induced fractional muon-mass oscillation scales as $v\simeq 10^{-3}c$9 for linear couplings and $\mu\mathrm{eV}$00 for quadratic couplings. Existing and ongoing spectroscopy can probe scalar–muon interactions up to 12 orders of magnitude more stringently than astrophysical bounds, while muonium free-fall experiments can improve complementary fifth-force limits by up to 5 orders of magnitude (Stadnik, 2022).

At nanohertz frequencies, pulsar timing arrays search for the oscillating gravitational potential generated by ULSDM pressure oscillations at frequency $\mu\mathrm{eV}$01. In the NANOGrav 11-year analysis, no detection was found once solar-system ephemeris effects were modeled, and the strongest $\mu\mathrm{eV}$02 upper limit on the local dark-matter energy density was $\mu\mathrm{eV}$03 at frequency $\mu\mathrm{eV}$04, corresponding to $\mu\mathrm{eV}$05; the limit is below $\mu\mathrm{eV}$06 over the band $\mu\mathrm{eV}$07 to $\mu\mathrm{eV}$08 (Kato et al., 2019). The Parkes PTA similarly found $\mu\mathrm{eV}$09 for $\mu\mathrm{eV}$10 at $\mu\mathrm{eV}$11 confidence (Porayko et al., 2018). A related proposal considers LISA observations of quasi-monochromatic white-dwarf binaries: if some binaries reside in ULSDM clumps with density $\mu\mathrm{eV}$12 times the local density, the resulting frequency modulation could be detectable for $\mu\mathrm{eV}$13–$\mu\mathrm{eV}$14 (Wang et al., 2023).

6. Specialized scenarios, compact objects, and theoretical issues

Flavor-violating portals extend ULSDM phenomenology into time-domain flavor physics. In off-diagonal down-quark models, the coherent field produces $\mu\mathrm{eV}$15 CKM modulation and $\mu\mathrm{eV}$16 quark-mass shifts. The resulting constraints are complementary across scales: near $\mu\mathrm{eV}$17, a NANOGrav reanalysis yields $\mu\mathrm{eV}$18, $\mu\mathrm{eV}$19, and $\mu\mathrm{eV}$20; at $\mu\mathrm{eV}$21, Rb/Quartz clock data imply $\mu\mathrm{eV}$22, $\mu\mathrm{eV}$23, and $\mu\mathrm{eV}$24; and in the quantum regime invisible meson decays require $\mu\mathrm{eV}$25, $\mu\mathrm{eV}$26, and $\mu\mathrm{eV}$27 (Guo et al., 18 Mar 2026).

Compact objects are unusually sensitive to environmental or curvature-induced instabilities. In universally coupled quadratic models with negative coupling, there exists a band where cosmology is stable but neutron stars may spontaneously scalarize when

$\mu\mathrm{eV}$28

with $\mu\mathrm{eV}$29 and $\mu\mathrm{eV}$30 (Sibiryakov et al., 2020). In the purely gravitational misalignment model,

$\mu\mathrm{eV}$31

gives the conservative neutron-star bound

$\mu\mathrm{eV}$32

for $\mu\mathrm{eV}$33, $\mu\mathrm{eV}$34, and $\mu\mathrm{eV}$35 fixed by the dark-matter abundance (Davoudiasl, 2024). These results sharpen a common misconception that ULSDM without direct Standard Model couplings is necessarily inaccessible to astrophysical tests.

Quadratic models also raise distinct theoretical issues. Near massive bodies, sufficiently large $\mu\mathrm{eV}$36 produce screening, so terrestrial sensitivities to quadratic couplings can be suppressed even while space-based observables remain sensitive. At the same time, generic quadratic couplings induce sizable radiative corrections to $\mu\mathrm{eV}$37, motivating symmetry-protected constructions based on $\mu\mathrm{eV}$38, shift symmetry, clockwork, or relaxed-relaxion dynamics (Banerjee et al., 2022). A particularly explicit UV-linked construction is the Coleman–Weinberg model with three right-handed neutrinos and a soft Higgs portal $\mu\mathrm{eV}$39, which prefers

$\mu\mathrm{eV}$40

and predicts GeV-scale right-handed neutrinos when the scalar relic abundance is set by dynamically induced misalignment (Murgui et al., 2023).

Gauge-kinetic couplings add a late-time electromagnetic consequence. For

$\mu\mathrm{eV}$41

the post-recombination oscillation $\mu\mathrm{eV}$42 periodically makes the canonical photon frequency imaginary for

$\mu\mathrm{eV}$43

producing tachyonic amplification of electromagnetic modes and potentially Mpc-scale magnetic fields. The same mechanism induces $\mu\mathrm{eV}$44, and a conservative bound $\mu\mathrm{eV}$45 from $\mu\mathrm{eV}$46-variation still leaves viable parameter space for late-time magnetogenesis (Kamali et al., 1 Jan 2026).

Taken together, these specialized scenarios show that ULSDM is not a single model but a class of coherent-boson dark sectors whose phenomenology depends sensitively on portal structure, environmental masses, finite-density effects, and the distinction between vacuum, thermal, and dense-matter backgrounds. This suggests that progress will continue to come from combining cosmological calculations, compact-object theory, and precision time-domain experiments rather than from any one observable alone.