Cosmology is living through an era of precision: the cosmic microwave background (CMB), governed by Planck's law, tells us the baryon density with an error under 0,5%. At the same time, astronomers measure primordial deuterium in ancient gas clouds to within one percent. These two numbers should match within the framework of big bang theory and nucleosynthesis (BBN), refined since the 1940s by the work of George Gamow and his colleagues. But lately a rift has appeared between them. According to Hubble's law, the speed at which galaxies recede is proportional to their distance, and measuring the Hubble constant by different methods has yielded the famous discrepancy. The tension grows if we include early dark energy (EDE) — a hypothetical component invoked to solve the Hubble constant problem and which requires a slightly larger amount of baryons. Adam Riess, who discovered the accelerating universe, was the first to notice this puzzle. Back in the 1920s, Georges Lemaître laid the foundations of the expanding universe theory, and now his equations serve to test exotic hypotheses. The idea of a non-standard expansion during the nucleosynthesis era was born from this contradiction.
To test the hypothesis, the authors used the PRIMAT code — a modern implementation of BBN theory, rewritten in Python. Three scenarios were compared: standard ΛCDM, a model with a constant radiation excess (parameter ΔN_eff), and a model with very early dark energy (vEDE). vEDE describes a scalar field that froze until a critical redshift, then rapidly decayed, temporarily boosting the expansion rate. The key parameter is the relative speedup $δ_D$ = ΔH/H at a temperature $T_D$ = 3,4×$10^8$ K (≈0,03 MeV): at this point deuterium was still burning out, while helium-4 had almost fully formed. This choice isolates the impact on deuterium without affecting helium. Abundances were calculated along with Bayesian inference using MCMC and likelihood profiling, accounting for experimental and theoretical errors.
In standard ΛCDM, the baryon density from BBN ($ω_b$ ≈ 0,02188) disagrees with the CMB estimate in the EDE model ($ω_b$ = 0,02272) at the 3,1σ level. Adding a constant radiation component ΔN_eff does not help: it raises the helium abundance too much, and the tension remains at 3,1σ. vEDE behaves differently: the speedup by $δ_D$ = 0,087 (8,7%) during the deuterium burnout era keeps D/H at its observed level ($10^5$ D/H = 2,527), while helium changes by only a fraction of a percent. A joint analysis of BBN and CMB yields $δ_D$ = 0,087±0,036, with an upper bound on the logarithm of the critical redshift: log_10 $z_c$ < 8,3 (95% confidence). The statistical tension drops to $Q_{DMAP}$ = 0,7σ, i.e., it vanishes entirely. The Akaike information criterion strongly favors vEDE over ΛCDM (ΔAIC ≈ −5,3), whereas the ΔN_eff model is disfavored.
The result implies that nucleosynthesis data are sensitive not only to the amount of extra energy, but also to when exactly that energy was injected into the expansion. This contrasts with the usual excess of neutrino-like particles and demands a very specific scenario: a light scalar field with a mass tuned to dynamics at temperatures around 30 keV. Such a component fits naturally into the idea of the "axiverse" — a spectrum of axion-like fields threading cosmological history from inflation to the current acceleration. vEDE thus becomes another window into this dark sector, alongside classical dark energy and possibly inflation.
The vEDE hypothesis paves the way for reconciling BBN with CMB in extended models. Testing it requires a joint analysis of multiple datasets: CMB, large-scale structure, supernovae, and light-element abundances. Future work will incorporate vEDE into multicomponent fits, such as EDE + vEDE, and assess the impact of the vEDE transition on the small-scale matter power spectrum. Improved nuclear data for deuterium reactions could either strengthen or weaken the discovered tension, refining the field's parameters.
The finding affects all of observational cosmology: from the cosmic microwave background to large-scale structure and primordial nucleosynthesis.
A full joint fit of EDE and vEDE models to data from Planck, ACT, SPT, DESI, and SH0ES is planned, along with refinements of nuclear reaction rates in the lab.
The article touches on several unsolved mysteries at once: [open_problem:проблема постоянной Хаббла], [open_problem:природа тёмной энергии], [open_problem:проблема лития], and [open_problem:расхождение барионной плотности] between BBN and CMB. It shows how microphysics at the scale of 0,03 MeV can manifest in the global dance of galaxies.
This section is addressed to the author of the paper and their colleagues. Neighbouring works were found by meaning-based search across our archive, not by word overlap. We do not review: we show the field around the work.
What the work does
The authors examined the hypothesis of very early dark energy (vEDE), which temporarily accelerated the expansion of the Universe during the epoch of nucleosynthesis. Using the PRIMAT code and Bayesian analysis, they showed that an acceleration of 8.7% at a temperature of about 0.03 MeV reconciles the measured deuterium abundance with cosmic microwave background data, without affecting helium-4.
Where its strength is
The fine-tuning of vEDE to a specific temperature range—where deuterium is still sensitive to the expansion rate but helium is not—allows the tension between BBN and CMB to be resolved in isolation, without disrupting predictions for other elements.
Where it could go next
Worth trying
Why this may matter
If vEDE or similar mechanisms are confirmed, they would point to new physics in the early Universe, possibly related to phase transitions or non-trivial dynamics of scalar fields, and could help close the chain of consistency from the first minutes to the present-day acceleration.
Nearby works in our archive
The archive is crowded around this work: 4 reviews nearby, closest at 0.70. This ground is well walked — and there is plenty to compare against.
In the time it takes you to read this paragraph, the universe could have produced its entire present-day stock of deuterium — had the expansion rate in the first minutes been just 9% higher than standard for only a fraction of a second.