Scalar Chemical Potential Cosmological Collider: A CMB Lagrangian

Introduction to Inflation-Scale Spectroscopy

The Scalar Chemical Potential Framework

1. The Cosmological Collider Lagrangian

   We begin by defining the exact Lagrangian density that governs the SCP mechanism. The complete action ℒ = ℒ_inflaton + ℒ_σ + ℒ_int couples the standard slow-roll inflaton sector to a heavy complex scalar field σ. The inflaton sector is given by ℒ_inflaton = √(−g) [ M_Pl² R/2 − (∂φ)²/2 − V(φ) ], while the heavy scalar is described by ℒ_σ = √(−g) [ −|∂σ|² − M²|σ|² ]. The critical addition is the dimension-5 chemical-potential interaction, which couples the derivative of the inflaton to the U(1) current of the complex scalar, preserving the shift symmetry φ → φ + c and thus protecting the flatness of V(φ) against radiative corrections.

   ℒ_int = (1/Λ) ∂_μ φ J^μ , where J^μ = i ( σ* ∂^μσ − σ ∂^μσ* )

   On the inflating background, the inflaton acquires a time-dependent vacuum expectation value φ₀(t). The temporal derivative of this background field, φ̇₀, is related to the Hubble scale by φ̇₀^(1/2) ≈ 60H in typical single-field scenarios. This background evolution converts the interaction into an effective chemical potential ω = φ̇₀/Λ. The operator successfully harnesses the large kinetic energy of inflation, opening a phenomenological window where ω can be naturally of order M, far exceeding H.

2. Decoupling-Limit and Goldstone-π Action

   To rigorously evaluate the primordial fluctuations, we map this system onto the Effective Field Theory (EFT) of inflation (Cheung et al., arXiv:0709.0293). Operating in unitary gauge where the inflaton fluctuations are absorbed into the metric, time translations are spontaneously broken. We reintroduce the Goldstone boson π of broken time translations via the Stückelberg trick t → t + π. The quadratic action for the Goldstone mode isolates the dynamical degrees of freedom relevant for the curvature perturbation ζ ≈ −Hπ.

   S_π⁽²⁾ = ∫ d⁴x a³ M_Pl² ε H² [ π̇² − c_s⁻² (∂π)²/a² ]

   Under this transformation, the chemical-potential operator (∂_μ φ)·J^μ evaluates to ω·J⁰ on the unperturbed background. The fluctuations in the interaction sector generate linear mixings between π and the U(1) current. Specifically, the leading-order interaction in the decoupling limit is ℒ_mix = ω π̇ J⁰, dictating the tree-level exchange processes between the inflaton fluctuations and the heavy field.

3. Mode Functions in de Sitter with Chemical Potential

   The presence of the chemical potential ω radically alters the mode functions of the complex scalar σ in the de Sitter background. Solving the equation of motion in conformal time τ, we observe an explicit breaking of particle-antiparticle degeneracy. The physical dispersion relation for the positive-frequency modes undergoes a shift, directly modifying the Bogoliubov coefficients α_k and β_k responsible for particle production.

   ω_k = √( k² + a² M_eff² ) − a ω , where M_eff² = M² − ω²

   Here, M_eff represents the effective mass controlling the oscillation frequency. As demonstrated by Bodas, Kumar & Sundrum (arXiv:2010.04727), the standard Boltzmann suppression factor exp(−πM/H) is replaced by exp(−π(M_eff − ω)/H). In the parameter regime where ω ∼ M ≫ H, the argument of the exponential approaches zero. This results in 𝓞(1) particle production for the favored charge state, rendering the heavy field dynamically relevant at the time of horizon crossing without triggering perturbative breakdown.

Non-Gaussianity and the Primordial Bispectrum

1. Tree-Level Bispectrum: Single, Double, and Triple Exchange

   We compute the three-point correlation function ⟨ζζζ⟩ using the rigorous in-in (Schwinger–Keldysh) formalism. Following the full tree-level diagrammatic classification by Kumar, Lu, Reece & Welling (arXiv:2604.07434, April 2026), the bispectrum receives contributions from Single Exchange (SE), Double Exchange (DE), and Triple Exchange (TE) topologies. In the squeezed limit (k₃ ≪ k₁ ≈ k₂), the shape function B_ζ(k₁,k₂,k₃) is dominated by the interference between the heavy field's oscillations and the background expansion, producing a distinct non-analytic scaling.

   B_ζ(k₁,k₂,k₃) ≈ f_NL (1 / k₁³ k₃³) (k₃/k₁)^{1/2 ± iμ} P_s(cos θ) , where μ = √( M_eff² / H² − 9/4 )

   The parameter μ determines the frequency of the scale-invariant "clock" signal. The overall amplitude f_NL evaluates to 𝓞(0.01–10) for ω ≲ φ̇₀^(1/2), fully consistent with the theoretical bounds of the SCP framework. Intriguingly, the recent global analysis in the companion paper (arXiv:2603.15728, March 2026) reports a 1.7σ evidence for this exact oscillatory feature in the window ω − M ≈ 3H, strongly motivating a dedicated search pipeline.

2. From Bispectrum to CMB Angular Bispectrum

   To connect the primordial shape B_ζ to cosmological observables, we project it onto the CMB temperature angular bispectrum b_TTT. The line-of-sight integral incorporates both the Sachs–Wolfe effect on superhorizon scales, where the temperature anisotropy is approximately Δℓ ≈ −Φ/3, and the complex acoustic transfer functions Δ_l(k) on subhorizon scales.

   b_TTT^l₁l₂l₃ = ∫ r² dr [ α_l₁(r) β_l₂(r) γ_l₃(r) ξ_l(r) + permutations ]

   Here, the spatial radial kernels α, β, and γ are derived by factorizing the oscillatory momenta dependencies of the SCP shape. Given the highly oscillatory nature of the bispectrum for large μ, the primary sensitivity lies within the high-resolution modes of the CMB. The optimal parameter window spans ℓ_min = 2 to ℓ_max = 3000, aligning perfectly with the combined datasets of Planck PR4, ACT DR6, and the projected sensitivity bounds of the LiteBIRD satellite mission.

UV Completions and Phenomenological Comparisons

1. Modular CP-Breaking Realization

   While the Lagrangian formulation is robust within the EFT of inflation, anchoring it to a specific ultraviolet (UV) completion strengthens its theoretical validity. As explored recently by Aoki & Strumia (arXiv:2604.05548, April 2026), the chemical-potential coupling naturally emerges in modular-invariant extensions of the Standard Model. In these constructions, the inflaton τ is identified as a modulus field traversing the fundamental domain. The necessary U(1) current is constructed directly from Standard Model fermion bilinears, and the chemical potential ω is dynamically generated by the vacuum expectation value of τ as it undergoes slow-roll.

   This modular spontaneous CP breaking not only provides a concrete particle-physics origin for the operator (1/Λ) ∂_μ φ J^μ, but also guarantees that the radiative corrections generated by the massive scalar loop remain strictly subdominant to the tree-level potential, a fundamental requirement for viable inflation.

2. Contrast with Alternative Frameworks

   It is imperative to distinguish the SCP mechanism from other non-Gaussian paradigms. In the standard Quasi-Single-Field Inflation introduced by Chen & Wang (arXiv:0911.3380), a heavy scalar interacts with the inflaton via a constant mass mixing. However, without a chemical potential to shift the dispersion relation, the particle production remains strictly thermal, leading to an exponentially suppressed f_NL for M ≫ H. The SCP model explicitly bypasses this suppression.

   Furthermore, spinning-particle chemical-potential models, such as those proposed by Wang & Zhu (arXiv:2001.03879), rely on helicity-dependent couplings (e.g., φ F F̃) to amplify tensor modes or chiral fermions. While mathematically similar, these vector/tensor couplings produce distinct angular dependencies in the squeezed limit represented by higher-order Legendre polynomials P_s(cos θ) with s ≥ 1. In contrast, the scalar chemical potential yields a dominant scalar signature (s = 0), allowing precise discrimination using high-ℓ CMB temperature data.

Conclusion