192 | Nicole Yunger Halpern on Quantum Steampunk Thermodynamics
What this covers
Nicole Yunger Halpern joins Sean Carroll to map the current state of quantum thermodynamics—a field that fuses quantum mechanics, information theory, and classical thermodynamics to ask what happens when genuinely quantum resources reshape how energy is extracted, stored, and dissipated. The conversation follows the historical threads connecting these domains: how Maxwell's demon and Szilard's engine revealed that information and entropy are thermodynamic currencies, and how Landauer's principle closes the apparent loopholes by pricing the cost of erasing information. From there, Yunger Halpern traces how quantum features—entanglement, contextuality, and boson statistics—can outperform classical approaches in concrete thermodynamic tasks, while also explaining why quantum effects are so fragile in warm, biological systems that monogamy of entanglement spreads quantum correlations too thin to be useful.
A central tension runs through the discussion: the field still lacks a settled definition of quantum work and heat, and Yunger Halpern argues against the physics community's instinctive unification drive in physics, proposing instead that different definitions may serve different experimental contexts. She also describes how fluctuation relations—exact mathematical statements that govern systems far from equilibrium—have proven their worth in small-scale experiments stretching individual DNA strands. The conversation touches on institutional geography: why European and Canadian physicists led the field's early, abstract phase, and how recent bridges to experimental US disciplines like atomic-molecular-optical physics are shifting momentum. Throughout, the discussion treats thermodynamics as an operational story about agents with incomplete information trying to accomplish work, a framing that makes entropy as information not metaphorical but fundamental.
Yunger Halpern argues that quantum thermodynamics—the marriage of quantum mechanics, information theory, and thermodynamics ('Quantum Steampunk')—is a maturing field where genuinely quantum resources like entanglement, contextuality, and spin statistics can reshape thermodynamic tasks, even though the field has lacked a unifying definition of quantum work and heat and is only now reaching experimental viability.
- Information and thermodynamics are deeply linked, as shown by Maxwell's demon, Szilard's engine, and Landauer's erasure cost
- Quantum features such as boson statistics and contextuality can outperform classical resources in thermodynamic tasks
- Quantum effects are fragile in warm, wet, large systems because entanglement spreads thin (monogamy), making macroscopic quantum thermodynamics hard
Information and entropy are intrinsically linked because agents accomplish thermodynamic goals only insofar as they possess information about the systems they operate.
- Information and entropy enter thermodynamics because thermodynamics is an operational theory about agents who run engines, cool systems, or charge batteries to accomplish goals, and they can accomplish those goals only insofar as they have information about the system (e.g., where a gas is compressed)—so a gas's entropy and the agent's information are intrinsically linked.
“thermodynamics is an operational theory. It is about agents who run engines, who want to cool systems down or to charge batteries, so they want to accomplish goals, and they can accomplish goals only in so far as they have the information”
- Maxwell's demon appears to violate the second law by sorting fast and slow molecules to create a temperature difference and extract perpetual work, but the resolution (Bennett, building on Szilard and Landauer) is that the demon accumulates information in its memory and must erase it to truly reset the system; since Landauer showed that erasing information costs thermodynamic work, the erasure cost equals the work gained, netting zero and forbidding perpetual motion.
“the demon needs to erase this information from his memory, and Landauer showed that erasing information costs thermodynamic work, so the work that the demon gains by running the engine cycle is lost in the erasure, and so the demon nets zero work after all and can't run a perpetual motion machine.”
- Szilard showed that if you have information plus a source of heat, you can convert random uncoordinated heat into useful work: with one bit of information about which half of a partitioned box a particle is on, you can hook up a weight so that as the particle pushes the partition across the box it lifts the weight, drawing heat from the bath; by the end you have lost the information (the particle could be anywhere) but gained lifted weight—so the fuel is heat directed by information.
“Szilard showed that if you have information and you also have a source of heat, then you can use that information to turn this random uncoordinated heat into useful work.”
- Erasing information increases entropy and costs thermodynamic work (Landauer's bound sets a minimum entropy created per erased bit), which is why irreversible computations heat up our laptops; reversible computation is possible but would require idealized systems running impractically slowly, so real-world computers are far more wasteful than Landauer's limit.
“We could do our computation reversibly, but we would need an idealised system, it would run very, very, very, very, very slowly. So it won't be very practical for doing your taxes”
Fluctuation relations are exact equalities governing far-from-equilibrium systems and prove especially useful for small systems where random fluctuations dominate behavior.
- Fluctuation relations are especially useful for small systems because in large systems behavior follows the average and fluctuations away from it lose significance, whereas in small systems like strands of DNA or RNA the fluctuations are important—which is why such relations have been tested by stretching single DNA strands with optical traps, where different trials require different amounts of work due to random water-molecule kicks.
“Fluctuation relations are especially useful when the system is small. If a system is large, then it will very often, behave as it does, on average. And fluctuations away from the average have less and less significance.”
- Fluctuation relations are stronger versions of the second law: whereas the second law is an inequality that only says entropy increases or stays constant without specifying how much, fluctuation relations are exact equalities that govern systems even far from equilibrium and let you predict the statistics of work needed across repeated trials—remarkable because far-from-equilibrium systems are otherwise chaotic and hard to describe with general results.
“Fluctuation relations are... Can be thought of as stronger versions of the second law... The second law of thermodynamics is an inequality... Fluctuation relations are equalities. And they govern systems even very far from equilibrium”
Quantum thermodynamics blends Victorian steam-engine thermodynamics with futuristic quantum information and computation, and bridges to experimentally developed fields are reviving US interest.
- Quantum thermodynamics blossomed mainly in Europe, Canada, and Singapore over the past decade because its early character was abstract, mathematical, and philosophical, fitting Europe's philosophical tradition, whereas the US's culture of efficiency and applications led it to lag, though recent bridges to experimentally developed US fields (AMO, condensed matter, high energy) have revived US interest.
“Europe has this millennia old wonderful tradition of philosophy, and the US has had, let's say, at least a stereotype of efficiency and productivity and applications. So the nature of the quantum science done in the US has been very different”
- Quantum thermodynamics is the bringing together of quantum physics, information theory, and thermodynamics; it is partly Victorian (rooted in 19th-century thermodynamics motivated by steam engines) and partly futuristic (quantum information and quantum computers), giving it an aesthetic shared with the steampunk genre.
“So quantum thermodynamics, this bringing together of these three fields is partially Victorian, but quantum information is part of the science and technology of the future.”
Quantum thermodynamics tools like resource theories apply across disciplines and reveal that quantum statistics like boson bunching can beat classical work-extraction results.
- A most exciting frontier for Quantum Steampunk is taking the mathematical and conceptual tools of quantum thermodynamics—such as resource theories—around to other disciplines (chemistry, condensed matter, AMO, high-energy and black-hole physics) both to help solve their problems and to draw inspiration for new quantum-thermodynamics questions, as exemplified by using a resource theory to derive a bound on a chemist's molecular switch.
“taking the mathematical and conceptual tools of this quantum theory of thermodynamics, taking it around to different disciplines and saying, "How can we help with our toolkits?”
- In a multi-particle version of Szilard's information engine, the average work extractable depends on particle statistics: fermions (in a certain setup) yield zero work, classical particles yield more, and bosons yield even more—because bosons tend to pile into the same quantum state and exert greater pressure—so the spin statistics of bosons can help beat the best classical work-extraction result.
“the amount of work that you could get from fermions would be in a certain setup would be zero, the amount you could get with classical particles would be greater, and the amount that you could get with bosons would be even greater.”
Work transfers coordinated useful energy through changes in extensive parameters; heat transfers random molecular jiggling; Carnot's efficiency bound constrains all engines operating between two heat baths.
- Work and heat are the two forms of energy transferred between bodies: work is directly useful, coordinated energy that can be harnessed (e.g., to push a rock or charge a battery), defined rigorously as energy transferred by changing extensive thermodynamic parameters like volume, while heat is the random, uncoordinated jiggling of molecules—everything left over after the work is accounted for.
“work is energy that's being transferred that's directly useful... it's, in a sense, coordinated, whereas heat is random energy. So it's uncoordinated”
- Carnot calculated the best possible efficiency of any engine operating between just two heat baths at two temperatures (the Carnot bound); when developing new engine cycles, including quantum ones, physicists check that an engine satisfying Carnot's assumptions does not break this bound, because breaking it would violate the second law of thermodynamics.
“Carnot calculated the best possible efficiency achievable by this engine or by any engine that's in contact with just two different heat baths of two different temperatures.”
- A refrigerator is Carnot's engine cycle run backward: instead of getting work out by moving heat from hot to cold, you put work into the cycle to move heat from the cold bath to the hot bath, which is why cooling never happens spontaneously without energy input.
“If we want to run Carnot's engine backward, then instead of getting work out of the cycle, we have to put work into the cycle.”
Entanglement creates non-classical correlations where system information exceeds any part's information, and contextuality proves phenomena are genuinely quantum rather than classically mimicked.
- Entanglement is a strong relationship between quantum particles such that the whole is greater than the sum of its parts: there is information in a system of entangled particles that is not in either particle alone nor in the sum of the two considered separately, but resides in the whole—and this entanglement automatically endows each subsystem with entropy (measured by the von Neumann entropy).
“There is some information in a system of entangled particles that is not in just one particle, it's not in just the other particle, and it's not in the sum of the two particles separately, but it's sort of between the particles that's in the whole.”
- While many quantum features (like discretized energy) can be mimicked classically, contextuality is a very high bar that, if met, proves a phenomenon is truly non-classical; in quantum theory the context of a measurement (e.g., what other observable you choose to measure) can affect measurement statistics in ways classical physics would deem irrelevant, and contextuality underlies what makes entanglement genuinely non-classical and provides some quantum-computing speed-ups.
“there's a very high bar that you can set such that if you prove that some phenomenon meets this bar, then it is truly non-classical, that bar is contextuality.”
Theorists' proposed experiments often prove feasible through experimentalist collaboration, as shown by realizations of Landauer and near-feasibility of Szilard's engine.
- Theorists have a reputation for proposing experiments that are impossible for the next 20 years, but Yunger Halpern has been pleasantly surprised at how often experimentalists respond that they can actually do a proposed experiment and offer to collaborate—evidenced by recent realizations of the Landauer ratio experiment on multiple platforms and the near-feasibility of Szilard's engine.
“theorists have a reputation for proposing experiments that are impossible for the next 20 years, but I've been very pleasantly surprised at how often experimentalists have said, "Huh. Oh, yeah, I could do that. Sure, let's collaborate."”
- Because measuring a quantum system disturbs it, defining quantum heat and quantum work is genuinely problematic and contested—many competing definitions exist—and Yunger Halpern argues, contrary to physics' stereotypical drive toward unification, that there may be no single definition of quantum work and heat 'to rule them all,' nor does there need to be, since different definitions and measurement strategies are useful in different situations.
“in quantum thermodynamics, there are these many different definitions of work and heat that seem to be useful depending on how you poke the system... there might not be any one definition of quantum work and heat to rule them all, but maybe there doesn't need to be”
Early thermodynamicists debating particles versus macroscopic properties foreshadowed quantum theory; atomism won and presaged the quantum revolution.
- During the Industrial Revolution some thermodynamicists argued large-scale changes could be described purely by macroscopic properties (energy, temperature, pressure) and dismissed talk of invisible particles as unscientific, but the thermodynamicists who insisted on individual particles ended up being right, and that legacy of atomism foreshadowed quantum theory.
“they sort of beat down on the thermodynamicists who wanted to think about individual particles, but fortunately, those other thermodynamicists did have their say and they ended up being right, that there are individual particles out there”
- Statistical mechanics describes a large material as many particles whose individual states could in principle be computed from known theories, but since that is impractical and unnecessary, we use probability distributions—assigning particles probabilities of being in various positions and momenta—to recover the material's large-scale average behaviors without knowing each particle's exact state.
“we can also, not knowing exactly what the particles are doing, use probabilities to say the particles have some probability of being here with these momenta and some probability of being there with those momenta. We don't know, but we don't actually need to know in order to recover the large-scale behaviors”
One can bend around Carnot's bound by accessing quantum resources outside Carnot's assumptions, and Szilard showed heat plus information converts random molecular motion to useful work.
- Claude Shannon, founder of information theory, was unsure what to call his uncertainty-quantifying function until John von Neumann advised him to call it entropy, both because it already had that name in statistical mechanics and because no one really understands entropy, so in a debate Shannon would always have the advantage.
“von Neumann said, "This thing you should call it entropy because it's already been called that in statistical mechanics. It already exists. And also, no one really understands entropy, so in a debate, you will always have the advantage."”
- To her knowledge the Carnot bound has never been broken by any engine, classical or quantum; however, one can 'bend around' it by proposing an engine that accesses a slightly different resource—such as a particular quantum resource—so that it no longer falls under Carnot's assumptions, giving the appearance of breaking the bound while actually evading it.
“you could propose an engine that on the face of it might seem like it should obey the Carnot bound, but actually accesses some slightly different resource, like a certain quantum resource that really doesn't enable the engine to fall under the assumptions of Carnot, and so you can, "break the Carnot bound," but actually, you're just bending around it.”
Biological systems are too warm, watery, and large to sustain usable entanglement for quantum tasks due to monogamy of entanglement across environmental particles.
- Intrinsically quantum phenomena like entanglement are not ubiquitous in biology because biological systems are warm, watery, and large, so quantum effects die off quickly; by the principle of monogamy of entanglement, a particle in such an environment shares only a tiny bit of entanglement with each of many stray particles rather than concentrated entanglement with one, so effectively there is no usable entanglement to perform a quantum task.
“biological systems are warm, they're watery, they tend to be large. And in these conditions, quantum phenomena such as entanglement tend to die off quickly.”
Practical quantum thermodynamic devices should exploit naturally existing conditions providing necessary quantum control rather than imposing artificial requirements on new systems.
- Because quantum systems are very difficult to control, a promising path to practical quantum thermodynamic technology is to find 'quantum analogues of Southern California'—existing settings (such as certain labs) that already happen to provide the specific conditions and quantum control a device needs, so a quantum thermodynamic device dropped into the right place could yield benefits using resources already present.
“I like to look for what I think of as quantum analogues of Southern California. Southern California is... It just happens to be very well-suited to solar panels”
Physicists model DNA as a flexible spring for applications rather than as a carrier of genetic information.
- To physicists, DNA is interesting not for the genetic information encoded in its base pairs but as a very flexible molecule that behaves like a good spring for tiny applications—reflecting the broader physicist tendency to model most objects as springs.
“To physicists, most things are springs.”