A series of quantum physics experiments and theoretical papers now suggest that heat can flow spontaneously from a cold object to a hot one, provided the right quantum conditions are in place. This finding challenges one of the oldest rules in physics, the Clausius statement of the second law of thermodynamics, which holds that heat never moves from cold to hot without external work. The results do not break thermodynamics outright, but they expose a gap in how the classical rules account for quantum correlations and coherence, forcing physicists to reconsider the boundaries of a law once thought to be absolute.
Heat Flowing the Wrong Way in a Quantum Lab
The classical Clausius statement, attributed to R. Clausius, strictly applies only when two systems are initially uncorrelated. When quantum correlations or coherence link the systems before they interact, the usual direction of heat flow can reverse. A peer-reviewed experiment published in a quantum thermodynamics study demonstrated exactly this using a nuclear magnetic resonance (NMR) setup with two spin qubits. Researchers prepared the qubits so that one was effectively “hot” and the other “cold,” then allowed them to exchange energy. When the qubits started out uncorrelated, heat flowed normally from hot to cold until both reached the same temperature. But when quantum correlations were engineered into the initial state, energy moved spontaneously from the colder qubit to the hotter one. The team quantified this reversal using mutual information and quantum discord, showing that the information encoded in the correlations acted as a thermodynamic resource.
That NMR result, whose original preprint first appeared on arXiv, established a clear experimental precedent: quantum correlations can make heat run backwards without any external energy input. A more recent preprint extends this line of work by demonstrating that internal quantum coherence alone, not just initial correlations between a system and its environment, can reverse heat flow in a multipartite spin system. This distinction matters because it widens the set of conditions under which the reversal can occur. The earlier NMR experiment relied on carefully prepared entanglement between the two qubits before they interacted. The newer work, according to its authors, shows that coherence within the system’s own energy basis is sufficient, even when the system starts out uncorrelated with its surroundings, hinting that reversed heat flow could arise in more natural or less fine-tuned quantum settings.
Coherence as a Thermodynamic Fuel
What makes these results more than a laboratory curiosity is the growing theoretical framework that explains precisely when and why heat reversal happens. A peer-reviewed theory paper in Physical Review Research showed that heat-flow reversal can occur without reversing the arrow of time, and that internal coherences and correlations, rather than system-bath initial correlations, can drive the effect. This is a critical clarification for anyone tempted to read the headline as “time runs backwards.” It does not. The arrow of time remains intact; what changes is the direction in which energy redistributes among quantum subsystems. A separate theory preprint developed formal mathematical bounds showing that sufficiently large coherence is both necessary and sufficient to reverse heat flow between two qubits, giving the field a quantitative threshold rather than just a qualitative observation.
The practical implication is striking. In the quantum domain, correlations and coherence behave like consumable fuel that can be converted into directed heat flow. A 2022 analysis argued that such non-classical resources can, in principle, be spent to move heat from cold to hot without the external work normally required to run a refrigerator. In this view, entanglement and coherence sit alongside energy and entropy as core thermodynamic quantities, with their own conservation-like constraints. Once the correlations are used to power a reversed heat flow, they are degraded or lost, and cannot be freely reused without investing work to restore them. This picture preserves the generalized second law, because the total entropy of the universe, including the information stored in correlations, still increases, but it means the familiar Clausius version of the law, written for uncorrelated classical systems, is incomplete.
Open Systems, Work Extraction, and Quantum Heat Machines
The story becomes richer when the reversed heat flow is considered within open quantum systems, where a small quantum device interacts with a larger thermal environment. A peer-reviewed paper in Physics Letters A examined anomalous heat flow in such systems when initial correlations are present, and found that the reversal can enable work extraction and hybrid heat-machine behavior. That means a quantum device could, under the right conditions, simultaneously act as a heat engine and a refrigerator, extracting useful work from the “wrong-way” heat flow while also pumping heat from a colder reservoir to a hotter one. This is not perpetual motion; the correlations are a finite resource that gets depleted as the device runs. But it suggests new design principles for nanoscale thermal machines that have no classical analog and that harness information-like quantities as operational fuel.
Related theoretical work has started to map out how such quantum thermal machines might be optimized and controlled. A recent preprint on quantum thermodynamic protocols investigates how coherence and correlations evolve under repeated interactions, and how these dynamics affect both heat flow and work output. By treating the environment as a sequence of small quantum units rather than a featureless bath, the authors show that memory effects and structured correlations can significantly modify the efficiency and operating regimes of quantum engines. In scenarios where heat flows from cold to hot, these models clarify how much useful work can be harvested before the very resource that enables the anomaly (coherence or entanglement) has been exhausted, offering a blueprint for future experiments that aim to turn reversed heat flow into a practical technology rather than a one-off demonstration.
Scaling Up and Redefining the Second Law
The challenge, as with much of quantum thermodynamics, is scaling these effects beyond a handful of qubits. Experiments testing multi-particle quantum correlations under strict conditions have confirmed that interest in scaling these systems is driving the experimental frontier, but violations of classical expectations become harder to observe as systems grow, according to research published in a study of multi-particle correlations. Maintaining coherence and correlations in larger systems remains one of the hardest problems in quantum science, and until that barrier is addressed, backwards heat flow will stay confined to carefully controlled laboratory setups. Decoherence from environmental noise tends to wash out the delicate quantum features that drive the anomalous heat currents, restoring the familiar classical direction of heat flow at macroscopic scales.
Even so, the conceptual impact is already reshaping how physicists phrase the second law. Instead of a blanket statement that “heat never flows from cold to hot,” the more accurate formulation distinguishes between uncorrelated and correlated initial states, and between classical and quantum regimes. In uncorrelated classical systems, the Clausius statement remains an extremely reliable rule of thumb. In correlated quantum systems, however, heat can flow in either direction, constrained by generalized entropy balances that include information and coherence. This shift does not license perpetual motion or time travel, but it does reveal that the second law is less a single inviolable rule and more a family of statements whose precise form depends on what kinds of physical resources (energy, entropy, correlations, and coherence) are allowed to come into play.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
