What the Study Found
- A model drawing dark energy from the strong nuclear force’s vacuum, fit to supernova, galaxy and microwave background data, matched rival models.
- With all data combined, a Bayesian test that penalizes extra parameters gave the model a log Bayes factor of 2.0, against 1.0 for a popular rival.
- On early-universe or late-universe data alone, the strong force model could not be told apart from the standard cosmological constant.
- A strong force model of dark energy puts its turning point near redshift 0.67, earlier than the 0.3 to 0.4 most rival descriptions favor.
Dark energy makes up about 70% of the universe’s energy, and nobody knows what it is. Most proposed explanations call for new fields or particles that no experiment has ever detected. Physicists at the University of Sheffield and the University of British Columbia have now tested a route that needs none. In their model, dark energy comes from the vacuum of the strong nuclear force, and it fit current measurements of cosmic expansion as well as its leading rivals. The work, published in The Astrophysical Journal, rests on one theoretical step that is still a conjecture.
The puzzle has stood since supernova surveys showed the expansion speeding up, a discovery honored with the 2011 Nobel Prize in Physics. The simplest account is a cosmological constant, a fixed energy belonging to empty space itself. It works, but particle theory makes a mess of it: quantum estimates of that vacuum energy overshoot the observed value by as much as 122 orders of magnitude. The new work, a computational modelling study led by PhD researcher Dong Ha Lee at Sheffield, asks whether a better answer was already sitting inside the Standard Model of particle physics.
A Vacuum That Responds to Expansion
Quantum chromodynamics (QCD) is the theory of the strong force that binds quarks inside protons and neutrons, the same force ScholarPeer has followed in the hunt for particles made only of its own carriers and in the shapes it gives atomic nuclei. Its vacuum is not blank. It can settle into several distinct configurations that have the same energy, and quantum tunneling between them shifts the vacuum’s energy.
The paper offers an analogy. It compares the vacuum to a quantum particle in a row of identical dips: each dip has the same energy, but tunneling between them shifts the energy of the particle’s resting state. The dips are not places in space, so the tunneling does not make bubbles; it changes one vacuum state for the whole universe. The proposal is that this tunneling contribution depends on the large-scale geometry of spacetime, so it differs between a static universe and an expanding one, by an amount proportional to the expansion rate.
A rough calculation from the strong force’s own energy scale lands on a dark energy density and expansion rate of the same order of magnitude as those observed today. “Our theoretical framework suggests no new particles may be required at all,” says Ariel Zhitnitsky at the University of British Columbia, whose ideas the model is built on. The authors argue the effect is unique to the strong force: the weak force gets its mass from the Higgs mechanism, which exponentially suppresses the same kind of tunneling.
How the Strong Force Model Fared Against Survey Data
The team tested the model on microwave background data from the Planck satellite and the Atacama and South Pole telescopes, galaxy maps from the Dark Energy Spectroscopic Instrument (DESI), and two supernova catalogs: Pantheon+ and a recalibrated set from the Dark Energy Survey (DES) called DES-Dovekie. DESI’s own analysis had already found that an evolving dark energy fit better than a constant, at 3.1 standard deviations when combined with the microwave background. The question was whether a model with a physical origin could do the same.
On raw fit it did, roughly matching the popular two-parameter recipe for evolving dark energy. The harder test was Bayesian evidence, which penalizes every extra adjustable number, and that penalty usually swallows the gain: one Cambridge reanalysis, not yet peer reviewed, found that with the DES-Dovekie supernovae the Bayesian evidence for evolving dark energy vanishes. Here, on early or late data alone, the strong force model was indistinguishable from the standard cosmology. With everything combined, it scored a log Bayes factor of about 2.0, against about 1.0 for the evolving recipe, which the authors place in the moderate range. Its results also held steady when one supernova catalog was swapped for the other.
The model leaves fingerprints of its own. It places dark energy’s turning point, when its behavior crosses from one regime to another, near redshift 0.67, a measure of how far back in time; most rival descriptions put it around 0.3 to 0.4. It also reproduces an observed weakness in the very largest-scale ripples of the microwave background more closely than the evolving recipe does. “Testing models based on this idea against cosmological data is an essential step,” says Lee.
The Unproven Step and the Two Extra Parameters
The university’s release quotes Ludovic Van Waerbeke saying the observed scale emerges “without the need for any additional fitting parameters.” That holds for the energy scale. The cosmological fit is another matter. The effect had to be suppressed in the early universe, and since the authors cannot compute that suppression exactly, they imposed it with a switch function tuned by two parameters. The full model has eight, the same as the evolving recipe and two more than the standard model’s six.
The deeper limitation is theoretical. Computing the strong force vacuum in an expanding universe is currently infeasible, so the model rests on a conjecture that an exact result for a simpler geometry carries over. It is supported by a simplified version of the theory and by lattice simulations of a different quantity, the rate of particle production. Only one switch function was tried, and the Bayesian scores were estimated from sampling chains rather than the more thorough nested sampling the paper itself describes. Zhitnitsky and Van Waerbeke, who proposed the model, are among its testers.
If the idea holds, dark energy would stop being a placeholder for unknown physics and become a side effect of a force the Standard Model already describes. The authors suggest a laboratory check: the same tunneling should add a tiny extra pressure in the Casimir effect, the attraction between closely spaced plates in a vacuum. That correction has not been measured, and the team plans to test the model against data from next-generation space telescopes.
Reference
Lee, D. H., van de Bruck, C., Di Valentino, E., Van Waerbeke, L., & Zhitnitsky, A. (2026). Evolving Dark Energy Is Vacuum Energy After All. The Astrophysical Journal, 1012(2), 172. https://doi.org/10.3847/1538-4357/aeb9cc
- Study type: Computational modelling study comparing a theoretical dark energy model with cosmological data; peer reviewed, published in The Astrophysical Journal.
- Sample size: Not applicable; no participants or samples. The model was fit to public microwave background, galaxy survey and supernova datasets.
- Model: Dark energy as a correction to the strong nuclear force (QCD) vacuum energy proportional to the expansion rate, switched on at late times by a two-parameter function; eight parameters in total.
- Inputs and assumptions: Planck, Atacama Cosmology Telescope and SPT-3G microwave background data; DESI DR2 galaxy distances; Pantheon+ and DES-Dovekie supernovae. Assumes a conjectured extension of an exact vacuum calculation to an expanding universe.
- Comparison: Fit and Bayesian model selection against the standard cosmological constant model and the popular two-parameter evolving dark energy recipe.
- Time horizon: Expansion history from the early universe to today, with the model approaching a steady accelerating state in the far future.
- Funding / conflicts of interest: EPSRC studentship, STFC grant, Royal Society Dorothy Hodgkin Fellowship and NSERC. Conflicts of interest: not reported.
- Data availability: The authors’ modified cosmology code and their CosmoCAT analysis package are posted on GitHub.
- Preregistration: Not applicable.
- Main limitation: The key theoretical step is a conjecture, and early-time suppression is imposed by a fitted switch function rather than computed. Not author-stated: two co-authors proposed the model being tested.
FAQ
How could a force that works inside protons affect the whole universe?
A force that works inside protons could reach cosmic scales, the authors argue, because the effect comes from the structure of its vacuum rather than from particles traveling through space. Tunneling between equivalent vacuum configurations changes a single vacuum state shared by the whole universe, and the model proposes that this shift depends on how fast space is expanding.
Does this mean the dark energy mystery is solved?
The dark energy mystery is not solved by this work. The model fits current data well and was mildly to moderately preferred when all datasets were combined, but its central theoretical step is a conjecture and its early-time behavior is set by two fitted parameters. Independent tests are still needed.
What is the difference between a cosmological constant and evolving dark energy?
A cosmological constant is an energy of empty space that never changes, while evolving dark energy changes over cosmic history. Recent galaxy surveys from DESI hinted at evolution, but statistical tests that penalize extra parameters have often still favored the constant.
Could the idea be tested in a laboratory?
The idea could in principle be tested in a laboratory, according to the paper. The same tunneling physics should add a small extra contribution to the Casimir effect, the attraction between closely spaced plates in a vacuum, but that correction has not yet been measured.
Cite This Page
