Thermodynamics didn't Predate the Steam Engine
Why the next big breakthroughs will come from working to understand and improve today’s frontier technologies.
When you think about major breakthroughs in our understanding of the world, especially in physics, odds are that the first image that comes to mind is a lone theorist thinking really hard for months or years, then emerging with a fully-formed idea that changes the world. Think Newton sitting under his apple tree, Einstein in the Swiss patent office, or Wiles locked away in his attic. While these stories are (roughly) true, they’re the exception rather than the norm. Looking back through history, most of the biggest breakthroughs came instead through trying to understand and solve real-world engineering problems. My aim with this post is to convince you that, just as the steam engine pushed us to discover thermodynamics, today’s frontier technologies will push us to discover the next generation of nonequilibrium thermodynamics principles.
The story of the steam engine
One of my favourite examples is the development of the theory of thermodynamics, which stands in stark contrast to those above. For those not familiar, thermodynamics is the physics of how (free) energy is transformed from one form to another; its most important tenets are the first and second laws, namely that energy is conserved, and that the entropy of a closed system always increases. Because it deals with energy and entropy, rather than system-specific details, thermodynamics as a theory is uniquely powerful. As Einstein once said, thermodynamics is “the only physical theory of universal content which I am convinced will never be overthrown, within the framework of the applicability of its basic concepts”.
The development of thermodynamics was intimately coupled to the invention of the steam engine, a device which transformed energy from a large temperature difference into useful mechanical work. But unlike, for example, how the theory of gravity predated building rockets, the steam engine came before thermodynamics! The first industrial steam engine, used to pump water out of mines, was invented by Thomas Newcomen in 1712. By contrast, the first and second laws of thermodynamics weren’t formulated until the mid-1800s. The result most relevant to steam engines, Sadi Carnot’s proof that the efficiency of a heat engine is limited by the temperature difference between its heat source and cooling reservoir, came about as he tried to better understand the working of heat engines so that the French could catch up to more advanced British technology.
In summary, our best theory to explain how steam engines actually work was developed a hundred years after the invention of the first steam engine in an effort to learn how to build better ones.
Nonequilibrium thermodynamics: the quest to understand biology
The basic framework of Thermodynamics initiated by Carnot, and later expanded upon by Boltzmann, Maxwell, Szilard, and many many others, deals with physical systems at or near equilibrium. This is great for studying gases and liquids and thus building technology like refrigerators and internal combustion engines, and even not bad for studying less obviously equilibrium structures like hurricanes, stars, or even the universe itself. Not so great, though, for studying biology: every living organism is constantly expending energy to maintain nonequilibrium structure, and thus stave off the slow decay to equilibrium (i.e., death). In the quest to better understand biology, and ultimately to engineer biology-like machines, today one of the frontiers of physics is the study of nonequilibrium thermodynamics.
We’ve made huge progress in this quest over the last hundred years or so. From Onsager’s discovery of reciprocal relations in the 1930s that explained counterintuitive nonequilibrium phenomena like the thermoelectric effect, to the realization that information is physical and on the same thermodynamic footing as energy, to more modern developments like fluctuation theorems (1990s - 2000s) and thermodynamic uncertainty relations (2010s), we’re learning more and more about the how the laws of physics apply to nonequilibrium systems. But despite these advances, we’re still far from truly understanding all of the constraints and properties of energy and information in living organisms, and even farther from designing machines of our own that operate anywhere near the capabilities of what evolution has cooked up.
Case study: molecular machines
Modern nonequilibrium thermodynamics, and in particular stochastic thermodynamics, has been inspired in large part by the goal of understanding molecular machines: nanoscale protein structures that are responsible for many of the functions required to keep cells alive. Biology has evolved specialized molecular machines for interconverting between different types of (free) energy such as light, chemical, electrical, mechanical, and even information. One of the biggest motivations for studying these biological nanomachines is to learn design principles that will guide the engineering of synthetic nanomachines. We can use cutting edge tools from stochastic thermodynamics to derive physical limits on performance metrics like efficiency, power output, precision, or combinations of the above. As part of my PhD I showed that biological molecular machines are remarkably close to these limits. Evolution has gotten these machines to within spitting distance of what’s possible within the constraints of physics and chemistry.
But these advances in deriving fundamental physical limits have yet to lead to any engineering breakthroughs in the design, manufacturing, and control of synthetic nanomachines. Today the best synthetic molecular machines we can build are orders of magnitude away from these limits. One of the most advanced autonomous nanomachines, a 26-atom chemically-fueled motor invented by David Leigh and colleagues (I can’t overstate how impressive it is that this was built and actually works, if interested you can read more here), converts chemical fuel into directed mechanical rotation at an efficiency of about 0.000001%. Compare this to the ~70% efficiency of ATP Synthase (the machine that synthesizes ATP in your mitochondria), and you’ll see we have a long way to go to compete with evolution on the nanoengineering front.
What this ultimately tells us is that the engineering limitations holding back the development of nanotechnology are not the same as the known fundamental limits that have been derived thus far. Designing and building better molecular machines is going to take new discoveries about the physics of nanoscale machines.
Nonequilibrium thermodynamics is everywhere along today’s technology frontier
Beyond nanotechnology, today there are tons of cutting edge startups along with nonprofit and industry labs working at the technology frontier on problems where nonequilibrium thermodynamics plays a huge role. Just to give a few examples I’ve been thinking about lately:
The field of thermodynamic computing (home to startups like Normal Computing, Extropic, and Unconventional AI, as well as a whole ARIA program) is centered around the idea of leveraging noisy fluctuations to more efficiently process information. Harvesting fluctuations to do useful work, something I think is likely central to biological computing, is fundamentally a nonequilibrium thermodynamics problem.
The field of cryopreservation (home to startups like Until Labs and Tomorrow Bio) is centered around the idea of controlling rapid cooling and rewarming in a manner that avoids undesired phase transitions (in particular ice formation), which requires the precise control of far-from-equilibrium processes.
Advanced materials manufacturing processes that take inspiration from nonequilibrium biological processes are limited by a lack of theories and models for nonequilibrium assembly under complex environmental conditions. This limitation is at the heart of a new ARIA program, and being tackled startups like 3DBioFibR and nonprofits like the Impossible Fibers Lab at the Astera Institute.
The common theme across all these examples is that understanding and controlling nonequilibrium systems is a critical step along the path to developing game-changing new technologies. While there are incredibly smart people working on each of these problems, to the best of my knowledge no one is simultaneously working on all of these. I strongly suspect that in the search for general principles, there’s huge untapped potential in cross-pollinating between these diverse fields to connect the dots across the frontier.
My bet, which I’ve hopefully convinced you of here, is that the next major developments in nonequilibrium thermodynamics are going to come from working on these and other real-world problems at the frontier of technology. Just as Sadi Carnot needed the steam engine to uncover the second law, I believe the next big breakthroughs in nonequilibrium thermodynamics will come from trying to understand and improve today’s frontier technologies.
Concluding remarks
Physics has a rich history of “industrial theorists”, people like Carnot, Shannon, and Landauer, who uncovered fundamental physical laws by wrestling with the real-world engineering challenges at the frontier of their time. In recent years though, academic incentives have led to increased separation between theorists and engineers at the frontiers of their respective fields. I believe it’s mission critical that we work hard to reverse this trend.
Lately I’ve been trying to put my money where my mouth is. I’m a physicist, and have thus far spent most of my career working on developing nonequilibrium thermodynamics tools to better understand biological systems like nonequilibrium self-assembly and free energy transduction by molecular machines. Over the last year or so I’ve made it my mission to bring existing tools to bear (or develop new ones when needed) to solve applied problems in fields from cryo electron microscopy, to information-limited animal navigation, to metabolic optimization in mitochondria, to modeling advanced biomanufacturing processes. I’m actively in search of more interesting problems to sink my teeth into. If you’re building a modern-day “steam engine” and running up against problems in the arena of nonequilibrium thermodynamics, I’d love to hear from you!