Materials science fundamentally has a math & computability problem. Rigorous materials simulations at a nanometre scale seem now feasible - so you could model physical properties of small groups of atoms - eg. Modeling molecular reactions at a sub-atomic level. Bulk property simulations at a centi and up scale also work and so you can do first principles based design and use tricks like finite element methods to deal with the underlying stochasticity to a degree. Materials properties however largely arise from features between nano and micro scale - grain boundaries, dislocations etc. These are computationally infeasible today - there are neither engineering solutions, not math tricks that make this tractable and so materials science and engineering becomes a grind of experimentation and metrology.
This is interesting in that it seems to be making the grind more efficient. I think the true breakthrough will likely be proper scale quantum computing to make the first principles design feasible
Lab to product scale-up is a well known hard problem in materials (and chemistry), lots of public and private investment has been aimed at accelerating this for decades
I think the distinction between discovering a material and processing / scaling it up is a bit artificial. A lot of people think of a new material as just the crystal structure or something, but really all the defects and complex multiscale structure is just as much part of what defines a material, and controlling all that is why materials development is hard, and why you need so many different complementary measurements to understand what’s going on
I was a bit underwhelmed by this writeup because it’s a bit generic. I didn’t really see any specific new ideas on how to accelerate this process, or to differentiate from the main stream of materials discovery research which has been pretty dominantly AI forward for at least five years now
EDIT: I checked out some of their case studies and they’re pretty interesting and exploring some new materials characterizations territory. They’d be more impactful if they were more than just text IMO but much more concrete and less generic than the linked post
> THE MATERIALS OF THE FUTURE ALREADY EXIST IN THE LAB
Do they? There’s plenty of stuff in our labs, most of them are completely useless, some that were bought to be useless become fashionable again, and we get new and exciting ones every day. There are a lot of issues in going from concept to useable material, and "scaling up" is only one of them.
> FRONTIER INTELLIGENCE WILL BRING THEM TO THE WORLD.
It will probably help, but I doubt it will do it by itself.
> Put simply, materials innovation has a scale-up problem, not a discovery problem.
I just don’t think that’s true. It’s also a scale-up problem, but discovery itself is not solved.
The problem spaces keep getting larger (composites! nanostructures! High-entropy!). High-throughput thermodynamic and electronic structure calculations, automated characterisation and testing, and things like that are being developed because we just don’t know what materials could exist and what could be their properties. The problem is that while there is room for AI there, particularly in automation, even cutting edge models are very dodgy to extrapolate materials properties outside their training sets, which are utterly negligible compared to the size of the search space.
> The bottleneck has never been a shortage of promising candidate materials. It is the decades of trial and error it takes to manufacture even one of them reliably.
It’s worse than that. The first sentence is true (ideas are cheap), but the main bottleneck is to try to figure out the properties of the damn thing and whether some of them are deal breakers or not. The vast majority of materials we come up with never see any application, not because we don’t have processes at the right scale, but because they just have terrible properties.
All the properties have to be right and the price has to be right too.
Mainstream plastics, like all the ones that have their own recycling symbols, are made from monomers that cost about 50 cents a pound. There are thousands and thousands of polymers you've never heard of, some of which are very high performance, which are many times more expensive.
I think of the story that Silicon is not that good of a semiconductor as semiconductors go, but boy do people know how to make things out of it.
Think of how many man-hours over the past ~35 years have been spent in trying to make graphene and carbon nanotubes into useful products. It's easily somewhere in the millions. And really all we have to show for it are resins with slightly improved thermal conductivity and a rogues' gallery of grifters and failed startups.
Discovery's easy, scaling in the real world is hard, and scaling something that makes good commercial sense is very hard.
Very promising but I think it’s more important for “cheaper” technologies. For cutting edge 2nm logic where Angstrom level uniformity is required, the tool vendors like AMAT, KLA, Onto have invested in metrology and data synchronization. For cheaper technologies like III-V compound semiconductors where the tools are smaller and less sophisticated, this could be very beneficial.
Materials science fundamentally has a math & computability problem. Rigorous materials simulations at a nanometre scale seem now feasible - so you could model physical properties of small groups of atoms - eg. Modeling molecular reactions at a sub-atomic level. Bulk property simulations at a centi and up scale also work and so you can do first principles based design and use tricks like finite element methods to deal with the underlying stochasticity to a degree. Materials properties however largely arise from features between nano and micro scale - grain boundaries, dislocations etc. These are computationally infeasible today - there are neither engineering solutions, not math tricks that make this tractable and so materials science and engineering becomes a grind of experimentation and metrology.
This is interesting in that it seems to be making the grind more efficient. I think the true breakthrough will likely be proper scale quantum computing to make the first principles design feasible
Lab to product scale-up is a well known hard problem in materials (and chemistry), lots of public and private investment has been aimed at accelerating this for decades
I think the distinction between discovering a material and processing / scaling it up is a bit artificial. A lot of people think of a new material as just the crystal structure or something, but really all the defects and complex multiscale structure is just as much part of what defines a material, and controlling all that is why materials development is hard, and why you need so many different complementary measurements to understand what’s going on
I was a bit underwhelmed by this writeup because it’s a bit generic. I didn’t really see any specific new ideas on how to accelerate this process, or to differentiate from the main stream of materials discovery research which has been pretty dominantly AI forward for at least five years now
EDIT: I checked out some of their case studies and they’re pretty interesting and exploring some new materials characterizations territory. They’d be more impactful if they were more than just text IMO but much more concrete and less generic than the linked post
> THE MATERIALS OF THE FUTURE ALREADY EXIST IN THE LAB
Do they? There’s plenty of stuff in our labs, most of them are completely useless, some that were bought to be useless become fashionable again, and we get new and exciting ones every day. There are a lot of issues in going from concept to useable material, and "scaling up" is only one of them.
> FRONTIER INTELLIGENCE WILL BRING THEM TO THE WORLD.
It will probably help, but I doubt it will do it by itself.
> Put simply, materials innovation has a scale-up problem, not a discovery problem.
I just don’t think that’s true. It’s also a scale-up problem, but discovery itself is not solved.
The problem spaces keep getting larger (composites! nanostructures! High-entropy!). High-throughput thermodynamic and electronic structure calculations, automated characterisation and testing, and things like that are being developed because we just don’t know what materials could exist and what could be their properties. The problem is that while there is room for AI there, particularly in automation, even cutting edge models are very dodgy to extrapolate materials properties outside their training sets, which are utterly negligible compared to the size of the search space.
> The bottleneck has never been a shortage of promising candidate materials. It is the decades of trial and error it takes to manufacture even one of them reliably.
It’s worse than that. The first sentence is true (ideas are cheap), but the main bottleneck is to try to figure out the properties of the damn thing and whether some of them are deal breakers or not. The vast majority of materials we come up with never see any application, not because we don’t have processes at the right scale, but because they just have terrible properties.
All the properties have to be right and the price has to be right too.
Mainstream plastics, like all the ones that have their own recycling symbols, are made from monomers that cost about 50 cents a pound. There are thousands and thousands of polymers you've never heard of, some of which are very high performance, which are many times more expensive.
I think of the story that Silicon is not that good of a semiconductor as semiconductors go, but boy do people know how to make things out of it.
Yeah.
Think of how many man-hours over the past ~35 years have been spent in trying to make graphene and carbon nanotubes into useful products. It's easily somewhere in the millions. And really all we have to show for it are resins with slightly improved thermal conductivity and a rogues' gallery of grifters and failed startups.
Discovery's easy, scaling in the real world is hard, and scaling something that makes good commercial sense is very hard.
Even if the properties are good, there isn't an easy way to get material data downstream into design and manufacturing software.
Very promising but I think it’s more important for “cheaper” technologies. For cutting edge 2nm logic where Angstrom level uniformity is required, the tool vendors like AMAT, KLA, Onto have invested in metrology and data synchronization. For cheaper technologies like III-V compound semiconductors where the tools are smaller and less sophisticated, this could be very beneficial.
In software, AI seems to have inverted the equation.
Building is cheap. Distribution, differentiation and discovering unmet demand are becoming the expensive parts.
Starlite what?
also, append `?tune` for fun times!