Curvaton Scenario in Inflationary Cosmology

• The curvaton scenario is an alternative mechanism where a subdominant light scalar field converts its isocurvature fluctuations into adiabatic curvature perturbations after inflation.
• It predicts significant local-type non-Gaussianity and a varying tensor-to-scalar ratio, distinguishing it from the standard single-field inflation model.
• Various model realizations, including mixed inflaton–curvaton and MSSM embeddings, highlight its practical relevance under current observational constraints.

Searching arXiv for recent and foundational papers on the curvaton scenario to ground the article in published work. The curvaton scenario is an alternative to the simplest single-field inflationary mechanism for generating the primordial curvature perturbation. In this framework, inflation is still driven by an inflaton, but the dominant contribution to the observed perturbation $\zeta$ is generated by a second light scalar field that is energetically subdominant during inflation and converts its initially isocurvature fluctuations into adiabatic curvature perturbations only after inflation, typically when it oscillates and decays. In the literature this idea appears in standard post-inflationary form, in mixed inflaton–curvaton models, in “inflating curvaton” realizations, and even in matter-bounce analogues where an entropy field plays the curvaton role (Mazumdar et al., 2011, Byrnes et al., 2014, Cai et al., 2011).

1. Basic mechanism

In the standard inflationary picture, a single scalar field both drives quasi-exponential expansion and sources the primordial perturbations. The curvaton scenario separates these roles. During inflation, the inflaton dominates the energy density, while the curvaton $\sigma$ is present but subdominant. A successful realization requires the curvaton to be light during inflation,

$m_\sigma^2 \ll H_{\rm inf}^2,$

so that it acquires nearly scale-invariant fluctuations with amplitude

$\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$

Because the curvaton does not dominate the total energy density during inflation, these fluctuations are initially isocurvature rather than adiabatic (Mazumdar et al., 2011).

After inflation, the inflaton decays into radiation. The curvaton remains displaced from its minimum and begins to oscillate when $H \sim m_\sigma$. For a quadratic potential, the oscillating curvaton behaves like non-relativistic matter, $\rho_\sigma \propto a^{-3}$, while the background radiation redshifts as $\rho_r \propto a^{-4}$. Consequently,

$\frac{\rho_\sigma}{\rho_r} \propto a,$

so the curvaton energy fraction grows with time. When the curvaton eventually decays, spatial fluctuations in its local energy density are transferred to the total energy density, converting the original isocurvature perturbations into the adiabatic curvature perturbation $\zeta$ (Mazumdar et al., 2011).

This mechanism differs from single-field inflation in two structurally important ways. First, the amplitude of $\zeta$ is not fixed solely by inflaton dynamics. Second, local-type non-Gaussianity can be appreciable if the curvaton is still subdominant at decay, whereas the simplest single-field slow-roll models generically predict very small local non-Gaussianity.

2. Transfer to curvature perturbations and non-Gaussian structure

For a quadratic curvaton potential, the curvaton energy density during oscillations scales as $\sigma$0, so a field fluctuation produces

$\sigma$1

If $\sigma$2 denotes the curvaton energy fraction at decay, one obtains schematically

$\sigma$3

and, more precisely in the standard quadratic case with negligible inflaton perturbations,

$\sigma$4

The same setup gives the familiar local-type non-Gaussianity parameter

$\sigma$5

which reduces to

$\sigma$6

for $\sigma$7. Small curvaton energy fraction at decay therefore implies large local non-Gaussianity (Mazumdar et al., 2011).

A broader $\sigma$8 treatment shows that the quadratic, matter-like result is only one limit of a more general dependence on both the curvaton potential and the curvaton equation of state at decay. In the generalized formulation of the non-linearity parameter,

$\sigma$9

the parameter $m_\sigma^2 \ll H_{\rm inf}^2,$0 encodes the potential dependence and $m_\sigma^2 \ll H_{\rm inf}^2,$1 the effective curvaton equation of state. For a quadratic curvaton, $m_\sigma^2 \ll H_{\rm inf}^2,$2, which reproduces the standard result in the matter-like limit $m_\sigma^2 \ll H_{\rm inf}^2,$3. In that case the observational constraint on local non-Gaussianity requires roughly $m_\sigma^2 \ll H_{\rm inf}^2,$4, while for $m_\sigma^2 \ll H_{\rm inf}^2,$5 one finds approximately $m_\sigma^2 \ll H_{\rm inf}^2,$6. By contrast, the $m_\sigma^2 \ll H_{\rm inf}^2,$7 limit corresponding to a secondary curvaton-driven inflation generically drives $m_\sigma^2 \ll H_{\rm inf}^2,$8 large and is strongly constrained (Liu et al., 2020).

The usual sudden-decay transfer parameter $m_\sigma^2 \ll H_{\rm inf}^2,$9 is useful but not unique. A more robust quantity for gradual decay is the global transfer parameter

$\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$0

defined in terms of the final radiation sourced by curvaton and inflaton decay. In the sudden-decay approximation $\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$1, but $\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$2 remains well-defined when the decay is extended in time and when thermal effects are important (Kitajima et al., 2014).

3. Realizations and model-building frameworks

The simplest concrete realization consists of two non-interacting massive scalar fields with

$\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$3

In this setup the inflaton drives the first inflationary phase, the curvaton is a light spectator during that phase, and the final curvature perturbation can be inflaton-dominated, curvaton-dominated, or mixed. The same model also contains an “inflating curvaton” regime in which the curvaton itself drives a second period of inflation. The mixed regime is controlled by the inflaton fraction of the scalar power, conveniently parameterized by $\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$4, where $\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$5 is the inflaton mass that would reproduce the observed amplitude in single-field quadratic inflation (Byrnes et al., 2014).

A particularly explicit particle-physics embedding identifies both the inflaton and the curvaton with MSSM flat directions. In that construction the curvaton is the $\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$6 flat direction and the inflaton is an orthogonal $\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$7 flat direction. Because both fields belong to the visible MSSM sector, their couplings are fixed by known gauge and Yukawa interactions, their decay products thermalize efficiently, and the model predicts

$\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$8

with the magnitude determined solely by weak-scale physics and Standard Model Yukawa couplings. Gauge-mediated curvaton decays yield $\delta \sigma \simeq \frac{H_{\rm inf}}{2\pi}.$9, while Yukawa-mediated decays can give $H \sim m_\sigma$0 (Mazumdar et al., 2011).

The post-inflationary evolution of spectator fields can also be altered by Hubble-induced masses generated by the kinetic energy of an oscillating inflaton. In that case the effective interaction

$H \sim m_\sigma$1

induces

$H \sim m_\sigma$2

The resulting power-law evolution of the spectator field between the end of inflation and the onset of curvaton oscillation modifies the curvaton amplitude by a factor

$H \sim m_\sigma$3

and can substantially enhance the curvaton energy density at decay. For negative $H \sim m_\sigma$4, this effect relaxes the usual upper bound on the curvaton reheating temperature by a factor $H \sim m_\sigma$5 and makes curvaton domination easier to achieve (Fujita et al., 2016).

Other variants alter not the potential minimum but the origin of the perturbation transfer. In the hybrid curvaton, the source of $H \sim m_\sigma$6 is an inhomogeneous phase transition: a rolling field $H \sim m_\sigma$7 modulates the onset of oscillations of a waterfall field $H \sim m_\sigma$8. In this case the relevant perturbation is not the fluctuation of the oscillating field itself, but the modulation of the onset time $H \sim m_\sigma$9, with

$\rho_\sigma \propto a^{-3}$0

This mechanism supports both oscillating and inflating curvaton regimes within the same hybrid potential (Dimopoulos et al., 2012).

A supergravity realization, sometimes called the supercurvaton, embeds the curvaton in the simplest model of chaotic inflation in supergravity. In that setting the non-Gaussianity parameter naturally lies in the observationally interesting range from $\rho_\sigma \propto a^{-3}$1 to $\rho_\sigma \propto a^{-3}$2, and the spatial regions where $\rho_\sigma \propto a^{-3}$3 is especially large form a “curvaton web” resembling a net of thick domain walls, strings, or global monopoles (Demozzi et al., 2010).

4. Decay physics, thermal effects, and isocurvature

Realistic curvaton phenomenology depends sensitively on how the curvaton decays and how efficiently its decay products thermalize. A temperature-dependent decay rate,

$\rho_\sigma \propto a^{-3}$4

modifies the detailed evolution of the coupled curvaton–radiation system. Numerical analysis shows that thermal dependence can shift the final curvature perturbation amplitude by up to $\rho_\sigma \propto a^{-3}$5 for intermediate transfer efficiency, while the effect on $\rho_\sigma \propto a^{-3}$6 is very small. A naive sudden-decay treatment including a large modulation term $\rho_\sigma \propto a^{-3}$7 can significantly overestimate the impact of thermal effects; the simpler approximation

$\rho_\sigma \propto a^{-3}$8

with the global transfer parameter $\rho_\sigma \propto a^{-3}$9 is typically more accurate (Kitajima et al., 2014).

A distinct variant allows the inflaton itself to decay partly into curvaton particles during reheating. In the quadratic inflaton+curvaton model with branching ratio $\rho_r \propto a^{-4}$0, these decay curvatons are initially relativistic and later become non-relativistic, thereby modifying the background composition. The main result is that this changes only the background history and the maximum achievable curvaton fraction at decay; once the observed amplitude is matched, there are no additional distinct signatures in $\rho_r \propto a^{-4}$1, $\rho_r \propto a^{-4}$2, or $\rho_r \propto a^{-4}$3. Quantitatively, branching ratios of order $\rho_r \propto a^{-4}$4 are typically incompatible with the simplest curvaton scenario unless $\rho_r \propto a^{-4}$5, while $\rho_r \propto a^{-4}$6 has negligible impact on the allowed parameter space (Byrnes et al., 2016).

Isocurvature perturbations are not generic in all curvaton models, but they are generically possible. The detailed outcome depends on whether baryon number, lepton number, and CDM are produced before curvaton decay, by curvaton decay, or after curvaton decay. A comprehensive CMB analysis of all 27 such decay histories found that 18 scenarios remain consistent with current data. Some scenarios require the curvaton fraction at decay $\rho_r \propto a^{-4}$7 to be extremely close to unity, while others allow compensated isocurvature perturbations in which baryon and CDM isocurvature nearly cancel in the total matter sector. In the scenario $\rho_r \propto a^{-4}$8, the preferred value is $\rho_r \propto a^{-4}$9, and in $\frac{\rho_\sigma}{\rho_r} \propto a,$0 it is $\frac{\rho_\sigma}{\rho_r} \propto a,$1 (Smith et al., 2015).

Visible-sector embeddings can sharply simplify the isocurvature problem. In the MSSM curvaton model, both inflaton and curvaton decay into MSSM degrees of freedom, and the final thermalization temperature after curvaton decay is estimated as

$\frac{\rho_\sigma}{\rho_r} \propto a,$2

for Yukawa couplings $\frac{\rho_\sigma}{\rho_r} \propto a,$3. Because this is well above the electroweak scale and BBN, there is ample time to establish a single thermal bath, so residual isocurvature is effectively absent unless the LSP is a gravitino or axino (Mazumdar et al., 2011).

5. Observational constraints and phenomenological status

Within the simplest quadratic two-field model, the spectral index is

$\frac{\rho_\sigma}{\rho_r} \propto a,$4

and the tensor-to-scalar ratio is

$\frac{\rho_\sigma}{\rho_r} \propto a,$5

These expressions show that a pure curvaton limit suppresses tensors and tends to produce a slightly blue or nearly scale-invariant scalar tilt if the curvaton is very light, whereas an admixture of inflaton power restores the red tilt and larger tensor signal characteristic of quadratic inflation. A comprehensive parameter-space analysis concluded that the fully curvaton-dominated regime is in some tension with observational data, while an admixture of inflaton-generated perturbations improves the fit (Byrnes et al., 2014).

The same paper emphasized that a large tensor signal requires a significant inflaton contribution to the scalar spectrum. In the BICEP2-era interpretation adopted there, the claimed tensor amplitude ruled out the usual curvaton scenario in which inflaton perturbations are negligible, though not the admixture regime where both inflaton and curvaton contribute to the spectrum. The inflating curvaton regime, by contrast, mimics the predictions of Nflation and yields

$\frac{\rho_\sigma}{\rho_r} \propto a,$6

the same tensor-to-scalar ratio as single-field quadratic inflation with the same total number of e-folds (Byrnes et al., 2014).

Applications to specific inflationary potentials further illustrate the curvaton’s role as a phenomenological repair mechanism. In chaotic inflation, the curvaton can suppress a tensor-to-scalar ratio that would otherwise be too large, and a negative curvaton mass-squared can account for the observed red tilt. In the generalized exponential runaway model, where the inflaton does not oscillate and standard reheating fails, curvaton decay provides an efficient reheating channel. In that framework the reheating temperature is constrained for both the dominating and sub-dominating curvaton cases and remains compatible with nucleosynthesis (Sharma et al., 2019).

Non-Gaussianity remains the most characteristic signature. In the quadratic curvaton with generalized $\frac{\rho_\sigma}{\rho_r} \propto a,$7 treatment, most parameter space satisfies current observational constraints, but the $\frac{\rho_\sigma}{\rho_r} \propto a,$8 secondary-inflation limit is strongly disfavored. For the pseudo-Nambu–Goldstone curvaton, the second short inflationary process is ruled out in light of observations, whereas matter-dominated and radiation-dominated decay regimes retain substantial viable parameter space (Liu et al., 2020).

6. Extensions beyond standard post-inflationary conversion

The curvaton idea is not restricted to the standard radiation-dominated post-inflationary background. In the matter-bounce curvaton scenario, a light entropy field $\frac{\rho_\sigma}{\rho_r} \propto a,$9 in a matter-dominated contracting universe acquires a scale-invariant fluctuation spectrum from vacuum initial conditions. These isocurvature fluctuations are converted into a scale-invariant curvature spectrum during a non-singular bounce, with additional amplification near the bounce. The mechanism can both enhance scalar perturbations and suppress the tensor-to-scalar ratio, yielding

$\zeta$0

and it predicts local-type non-Gaussianity of order a few, typically negative in the examples studied (Cai et al., 2011).

Small-scale curvaton dynamics also connect the scenario to primordial black holes and induced gravitational waves. In the non-minimal curvaton model with kinetic term

$\zeta$1

a sharp feature in the field-space metric $\zeta$2 can generate a narrow peak in the curvaton fluctuation spectrum on small scales. The fully nonlinear mapping from the Gaussian curvaton contrast to the curvature perturbation must then be used to compute PBH formation. In the region where the resulting PBHs do not overclose the universe, the curvature non-Gaussianity is still well approximated by a local quadratic form with

$\zeta$3

so induced gravitational waves can be computed using the standard local ansatz. For PBHs constituting all dark matter in the asteroid-mass window, the asymptotic peak gravitational-wave amplitude in the small-$\zeta$4 limit is $\zeta$5, with a peak frequency around $\zeta$6; in the $\zeta$7 limit, the PDF becomes similar to that of ultra-slow-roll inflation (Pi et al., 2021).

These extensions make clear that the curvaton scenario is not merely a two-field add-on to inflation. It is a general mechanism for sourcing primordial curvature perturbations from a spectator degree of freedom whose late-time dynamics, decay channel, equation of state, and microphysical embedding all remain observationally consequential. Across its many realizations, the recurring control parameters are the curvaton fluctuation amplitude during the primordial phase, the transfer efficiency at decay, and the extent to which decay or background dynamics generate local non-Gaussianity or residual isocurvature.