Through three minutes after the Big Bang, the Universe played its first symphony. It was the big bang nucleosynthesis — the formation of light nuclei, where each element sounded its own note, and the tempo was set by the expansion of space. The standard rhythm, predicted by the Friedmann equations, kept protons and neutrons apart, preventing them from sticking together too quickly. In this score, hydrogen — the fragile isotope of hydrogen with one neutron, deuterium — was the most sensitive violin: the slightest quickening of the tempo threatened to drown out its voice. George Gamow in the 1940s first transcribed the notes of this cosmic symphony, and Georges Lemaître even earlier derived the rhythm of expansion — Hubble's law in its modern sense.

Today astronomers hear this music in two independent recordings. Cosmic microwave background (the microwave background reaching us from the recombination epoch) sets the baryon density — the overall weight of the orchestra — with an accuracy of half a percent. And observations of primordial deuterium in ancient clouds give its abundance with a one percent uncertainty. But the two scores don’t match: the discrepancy reaches 3,1σ. The tension is exacerbated by early dark energy (EDE) models, proposed to resolve the famous Hubble constant problem. Adam Riess, who discovered the acceleration of the Universe, first noticed this discord: if we add a bit more mysterious dark energy to the young Universe, more baryons are required. But then too little deuterium survives — it gets burned up by the extra heat.

The idea came from an unexpected place: what if the acceleration was not constant, but a brief impulse? Thus, 'very early dark energy' (vEDE) appears in the symphony of nucleosynthesis — a light scalar field that momentarily speeds up the tempo and then vanishes. The key moment is the temperature 3,4×$10^8$ K, when helium-4 has already sounded its note, while deuterium is still burning up. Numerical simulations with the PRIMAT code showed: a 9% acceleration ($δ_D$ = 0,087) during this narrow interval preserves the observed deuterium abundance (about 2,5×10⁻⁵) and barely changes helium. According to the calculations, the deuterium abundance depends much more strongly on the baryon density than helium: D/H ∝ $ω_b^{-1.65}$. That’s why such a sensitive indicator responds even to a faint whisper of extra energy. The relative acceleration $δ_D$ = $H_{tot}$($T_D$)/$H_{std}$($T_D$) − 1 shows how much the expansion rate exceeded the standard one at the moment $T_D$. Using Bayesian inference, scientists compared models. The standard scenario gave a discrepancy of 3,1σ. A constant increase in radiation did not improve the picture. But vEDE reduced the statistical tension to 0,7σ — almost ideal agreement, and the logarithm of the critical redshift is bounded from above: log₁₀ $z_c$ < 8,3. The Akaike information criterion also strongly favored the intermittent acceleration.

The result shows that the symphony of the first minutes was more complex than it seemed. Very early dark energy doesn’t just inject energy, but does so at a precisely determined beat. This fine-tuning hints at the existence of a whole spectrum of axion-like fields — an 'axiverse' pervading the history of the Universe from inflation to the present-day acceleration. Like hidden instruments in an orchestra, these fields could have played out in different epochs, and vEDE is just one detected voice. Ahead lies the analysis of full data sets: cosmic microwave background, large-scale structure, supernovae, and light element abundances. The cosmic symphony continues — and each new chord brings us closer to deciphering the mystery.