Frank Wilczek had a thought back in 2012. Ordinary crystals (salt, ice, diamond etc.) get their structure from atoms locked into a pattern that repeats across space. What, Wilczek wondered, if you ran the same trick in time? You would theoretically get a system that ticked forever, by itself, with no battery, no winding, no energy added. Wilczek already had a Nobel prize so people listened, even though the idea sounded suspiciously like perpetual motion, a thing physics has spent several centuries firmly disallowing. 

But quantum mechanics, it turns out, leaves a loophole: perpetual motion is allowed in the quantum realm, but only if nobody is looking. Physicists confirmed time crystals existed in 2016, at which point everyone agreed they were pretty useless. Looking at one, by definition, killed it.

However, a team of researchers in Finland have just done the looking, and somehow the magic survived. They built their time crystal in a bath of superfluid helium-3 cooled to near absolute zero, then attached it to a tiny mechanical oscillator. Attaching the oscillator should, in theory, have collapsed the whole thing, but it didn’t…

The crystal kept ticking for around 10⁸ cycles, which is several minutes by a wristwatch and a comfortable eternity by quantum standards. The team could measure and even tune it from the outside without killing it. The result, published in Nature Communications, drags Wilczek's curiosity firmly into the realm of working hardware.

🧐 What's in it for me? Leaving the existential questions about infinity aside, short term practical applications include a stubborn bottleneck in Quantum computing. Qubits are notoriously forgetful, decaying in microseconds and taking the calculation with them. Time crystals on the other hand are stubborn enough to last minutes. If they can be slotted into actual quantum hardware as memory without popping out of existence, the timeline for genuinely useful quantum computing shortens considerably.

💵 Out of the Lab: Quantum memory is the unglamorous but strategic layer of the quantum stack. A working physical substrate that holds state for minutes rather than microseconds rewrites who's competitive, and possibly who's solvent.

What if you could diagnose depression with a blood test? It turns out this would be unbelievably helpful, especially given current diagnosis relies on what people are willing to tell their doctor, which is a less than ideal foundation for a branch of medicine concerned, in large part, with people who would rather not say. 

Worse, depression doesn't show up the same way in everyone. Some get the textbook physical symptoms (fatigue, appetite changes, restlessness etc). Others mainly get the emotional and cognitive ones, like hopelessness or anhedonia, the technical name for losing interest in things you used to enjoy.

Now though, a team of researchers has just published data measuring biological aging in 440 women using two epigenetic clocks. One looked across multiple cell types; the other was specific to monocytes, the white blood cells that get caught up in HIV infection and tend to spike during depression. The general clock told them very little. The monocyte clock, however, lined up cleanly with the emotional and cognitive symptoms of depression. The cells your immune system uses to fight infection are also seemingly keeping a written record of how sad you feel. 

🧐 What's in it for me? If this holds up, depression diagnosis finally becomes objective. That matters most for people whose physical symptoms are routinely chalked up to a chronic illness rather than a mental health condition. A biomarker that can distinguish emotional from somatic depression also opens the door to matching specific drugs to specific subtypes.

💵 Out of the Lab: Mental health diagnostics has been one of medicine's stubbornly subjective backwaters, kept afloat largely by self-report and educated guesswork. An objective biomarker, even a partial one, opens a real market.

Computer memory has been on a collision course with physics for several decades, and physics, predictably, has been winning. Memory cells work by controlling whether a tiny current can flow through a material. The smaller you make the cell, the more that current starts leaking out through the cracks between the material, which become a more prominent feature than the material itself. Engineers spend their lives trying to plug this leakage, and conventional wisdom holds that smaller is better only up to a point.

Enter Yutaka Majima and his team, who've just produced a ferroelectric memory cell measuring 25 nanometres across, roughly one three-thousandth the thickness of a human hair, that works better than larger versions, even though it shouldn't…

Below a certain size, smaller actually starts working again. The cracks between the material (in this case, crystal grains) stop being a feature of the device, and the cell starts behaving better, not worse. The leakage problem wasn't solved so much as evicted which is more significant than it sounds: ferroelectric tunnel junction memory was first proposed in 1971, and for over half a century the leakage barrier has undone every attempt to scale it down. Until this one.

🧐 What's in it for me? Smartwatches that last months on a single charge. AI accelerators that draw a fraction of current power. Sensor networks that don't need batteries replaced every few months. Memory is, in energy terms, one of the more demanding parts of any modern device, and a ferroelectric cell this small could reorder the energy economics of basically all consumer electronics. 

💵 Out of the Lab: Hafnium oxide is already in everyone's chip stack, which means the upgrade path is unusually clear. The race is to whoever yields this manufacturing process reliably at scale.

A question oft asked by six-year-olds and senior natural historians, and not many people in between. A team at Nagasaki University, ignoring this convention, recently put 50 species into plastic arenas and mapped the results onto a crab family tree.

It turns out the answer to why the sideways shuffle is down to a single ancestral crab, around 200 million years ago, who took a well timed sidestep at the tail end of a mass extinction. 

To set the scene, thanks to some over eager volcanoes, roughly half the planet had just died, but out of the ash and rubble, the crab found itself in possession of a surprisingly useful trick. The ability to sprint full-tilt in either direction made it much harder to catch than its more predictable cousins. The catch being that sideways walking makes nearly everything else considerably more awkward. The crab, undeterred, kept on

What's strange is that nothing else has ever copied it. The crab body shape (technically carcinisation) has independently evolved at least five separate times, in hermit crabs, in lobsters, and in things barely related to true crabs at all. Evolution, given enough time, apparently cannot stop accidentally inventing crab. The walk, however, is a one-off. Bodies converge. Behaviour, apparently, does not.

Eight thousand species are coasting on a single piece of footwork from the wreckage of deep history, and have yet to be persuaded otherwise.

Until next time.

Like what you're reading? Share toast with a friend.