Three Quantum Machines See the Same Thing When a Gluon String Snaps: It 'Gasifies' Before Breaking
> [Physics · Quantum Simulation · Strong Interactions] On September 23-24, 2026, three independent quantum computing platforms reported the same class of simulation results on consecutive days: the real-time dynamics of gluon string breaking. Lawrence Berkeley National Laboratory (LBNL) used 104 active qubits on IBM's 156-qubit Heron processor, Duke Quantum Center used a 13-qubit ion chain, and QuEra used a neutral-atom array. All three experiments observed a phenomenon previously glimpsed only in approximate classical models: the gluon string develops finite-temperature 'gasification' internally before it breaks.
The One-Sentence Conclusion
Gluon string breaking is the core dynamics behind proton formation (hadronization) and lies in the non-perturbative forbidden zone of quantum chromodynamics (QCD). Classical computers cannot simulate it in real time due to the sign problem; lattice QCD can only compute static properties. Three platforms simultaneously producing real-time simulations means quantum computing has, for the first time, become a usable tool for studying QCD dynamics. The 'gasification' observed mid-string by LBNL is a prediction classical methods had only seen in highly constrained approximations.
Putting the Scale in Perspective
First, an intuition for general readers. Atoms consist of protons, neutrons, and electrons. Protons and neutrons are not fundamental particles—they are made of quarks, which interact via gluons carrying the strong force. When you try to pull two quarks apart, nature does not allow them to truly separate. The gluon field stretches into a 'string' whose energy grows with distance; when the energy is large enough, the string breaks, and a new quark-antiquark pair pops into existence at the break point. This entire process is the physical mechanism by which the strong interaction keeps quarks permanently confined inside particles.
Why this is a hard problem in physics:
- The strong interaction is 'strongly coupled' in the low-energy regime and cannot be treated with perturbation theory
- Classical Monte Carlo simulations are limited by the sign problem—they can only compute static properties, not real-time dynamics
- Pulling two quarks apart to watch how the string breaks is a fundamental physical process requiring real-time evolution
What LBNL Saw on IBM Heron
The LBNL experiment was led by postdoctoral researcher Anthony Ciavarella (also affiliated with the University of Washington Department of Physics), using 104 active qubits of IBM's 156-qubit Heron processor (model ibm_torino). Heron's heavy-hexagonal lattice architecture gives each qubit only 2-3 neighbors, avoiding the crosstalk problems of denser connectivity; compared with the previous 127-qubit Eagle generation, two-qubit gate error rates improved 3-5x, making deep circuits feasible.
Ciavarella's key technique was SC-ADOPT-VQE (scalable circuit concurrent variational quantum eigensolver). The idea: classically optimize the vacuum-preparation circuit on small 10-12 qubit systems, observe how circuit parameters scale with system size, then extrapolate that scaling relation to construct circuits for larger systems. This avoids the cost of running full vacuum preparation on large hardware.
The experimental model is a simplified SU(2) lattice gauge theory (1+1 dimensional, heavy-quark limit). Physically it is a simplified version of QCD, but it retains string breaking as the core dynamical feature.
What the 'Gasification' Phenomenon Is
String breaking itself had already been predicted in the SU(2) simplified model, and all three platform experiments reproduced it. The genuinely new discovery is the intermediate phenomenon Ciavarella observed during string breaking:
> "In the middle of the gluon string, it starts to look like it gasifies—and this happens at some finite temperature point before the string actually breaks."
The string carries potential energy that accumulates as the quarks are pulled apart. When the energy is large enough, the string should break and produce new quarks. But Ciavarella found that before the string actually breaks, localized thermal excitations already appear in the string's middle section, looking like a gas at finite temperature. This 'gasification' agrees with theoretical predictions about gluon string behavior near the confinement-deconfinement transition, but had never been directly observed in a real-time quantum simulation.
Why this matters:
First, physics validation. If 'gasification' is reproduced across different simplified models, it is more likely a genuine feature of real QCD rather than an artifact of one particular approximation.
Second, cosmological implications. 'Gasification' means the gluon string undergoes local thermalization before breaking—directly connected to the quark-gluon plasma state in the first microseconds after the Big Bang. Seeing the embryo of this thermodynamic collective behavior in a laboratory real-time simulation opens a new window onto the matter of the early universe.
Third, methodological implications. 'Gasification' is hard for classical Monte Carlo to capture because it occurs in the transition region of string breaking—local in spacetime and requiring real-time evolution. Quantum computing has a structural advantage here.
Parallel Validation from Duke and QuEra
The Duke Quantum Center experiment was led by Professor Christopher Monroe (Gilhuly Family Presidential Distinguished Professor) and published in Nature Physics on September 23. Thirteen calcium ions were controlled in a one-dimensional chain by precisely tuned laser beams, encoding the real-time dynamics of string breaking. The paper came from a joint team spanning Duke, the University of Maryland, Oxford, Caltech, Cornell, and KU Leuven.
QuEra independently reported similar results the same day using a neutral-atom array. All three experiments use SU(2) simplified models in the heavy-quark limit, 1+1 dimensions. Three different hardware platforms, three independent groups—and yet the theoretical approximations all converge on the same phenomenon.
This kind of 'three-independent-experiments converging' is the gold standard of physics validation. It minimizes the risk of statistical fluctuations in any single experiment, making 'gasification' a serious physical phenomenon rather than a fluke.
Who to Believe
The LBNL experiment comes from an IBM-LBNL joint press release and the original paper (based on Ciavarella's PhD work at the University of Washington); the peer-reviewed version has been published. The Duke experiment was published in Nature Physics (September 23). QuEra's result was reported by Qubit Report and The Quantum Insider. All three groups' simplified models are SU(2), heavy-quark, 1+1 dimensional; 'gasification' has so far been observed only in simplified models. Confirming it as a feature of real QCD will require simulations closer to physical QCD (higher dimensions, more quark flavors, SU(3) rather than SU(2)).
Open Questions
First, the distance from simplified to full. All three current experiments use SU(2), 1+1 dimensions, and the heavy-quark limit—several steps away from real SU(3), 3+1-dimensional, light-quark QCD. Each step needs new circuit designs, new error mitigation, and possibly more qubits.
Second, whether 'gasification' is universal across simplified models. As Ciavarella himself said: "If it is reproduced in multiple simplified models, it is more likely a feature of real QCD." More models must be tested; the three current experiments use approximations that are too similar.
Third, the scaling limits of quantum hardware. IBM Heron's 156 qubits are among the largest superconducting processors available, but still only sufficient for simplified models. How many qubits a full QCD simulation requires—and when it will be possible—has no clear timetable.
Fourth, extending spacetime dimensions. 1+1 dimensions are a necessary starting simplification, but real string dynamics live in 3+1 dimensions. Increasing dimensionality raises qubit requirements dramatically—the most fundamental physical bottleneck on this path.
---
Sources: IBM Newsroom (University of Tokyo Heron deployment, May 2025); LBNL-IBM joint press release (September 24); ORNL official release on SC-ADAPT-VQE (September 24); Duke Pratt School press release (September 23); Nature Physics paper (September 23); TechTimes coverage (September 24); The Qubit Report (September 24); Scalable Circuits paper arXiv:2411.05915; Schwinger model 112-qubit paper (Ciavarella, Farrell, Illa, Savage 2024).