Knots smaller than human hair make materials unusually tough | Hacker News Reader

Knots smaller than human hair make materials unusually tough

caltech.edu

133 pointsgmays47 comments

> The knots were not tied but rather manufactured in a knotted state by using advanced high-resolution 3D lithography capable of producing structures in the nanoscale.

This would be difficult to scale. And it's possible, even likely, that it can't be applied to existing strong fibers, which are drawn at very low diameters. The paper does its best to bury the lede, but the polymer they were using was acrylic-based, and likely very weak in comparison with, e.g., Dyneema or aramid fibers.

There are quite a lot of papers which examine the mechanical properties of nano-lattice and nano-architected materials, made with lithographic techniques, but I think that the commercial viability of these materials is effectively zero. And, in many cases, as here, it's not clear that they'd be superior to the standard high-strength materials that are already so ubiquitous.

I don't see a reason why the lithography method is necessary here. One could design a manufacturing process which ties the knots mechanically, just like woven fabrics.

Caltech probably already has the lithography equipment and expertise on hand, so they did it that way. They aren't trying to do their research in a manufacturing-friendly way, just the way that is easiest for them.

Possibly lithography was used to avoid stress concentrations and kinking, which can potentially be troublesome in mechanically-tied knots. (Presumably, especially at that scale.)

With respect to knotted and woven fabrics, there's quite a lot of interesting research in "3D-woven" fabrics for high-impact applications like body armor. 3D fabrics are pinned by fibers running through the weave top-to-bottom (Z axis) and are effectively macroscopically knotted. They're also commercially available:

https://compositeenvisions.com/product/fiberglass-e-glass-3d...

Interesting stuff, but still very niche and not always clearly superior to plain woven composite materials.

IIRC with 3D weaves, the ratio of strands to empty space (that would get filled with epoxy in typical rigid part applications) is significantly lower.

Why would lithography not scale? We've got plenty of mass production lithography with microchips.

Economically mass producing microchips is only possible because they are 100s of square millimeters in size and very valuable per square mm. Any bigger than that and error rates in the fab process start to destroy the yield.

These fibers are used in much bigger applications like body armor or mechanical composite parts where the surface area is on the order of square meters, not millimeters.

Modern panel fabs use fine masking over like 10sq. meters with precision on the same order of this (~dozens of micrometers), and people have successfully shown lithography over similar panel sizes by separate exposures.

The feature size is a LOT bigger than microchips though:

"Each knot is around 70 micrometers in height and width, and each fiber has a radius of around 1.7 micrometers"

That is, cheaper lithography methods could presumably be used.

Older processes are cheaper, but 5nm node costs something like $25,000 per square foot (with around 0.1 defects per square cm, which may result in waste) and doesn't do pieces larger than about 1 foot. That's pretty bad as far as textiles go.

I'm shocked daily by HN comment depth, what a breadth of experience people have. Thanks for the insight.

This place is all culture. People are smart and varied everywhere. But here we try to enforce a culture is that assigns low status to low effort trash. It doesn't always work. But that's the expectation, I don't think anything else could work. Comments should be for the community not to spread low information barnacles on every piece of original thought.

This place also suffers from the Dunning-Kruger-Effekt.

I guess you could argue, that most places suffer from that, as everyone has a boundary, at which they don't understand things. But the effect itself, is probably more relevant in an organization, as it was used to describe management.

There's a fine line between a Cunning Drueger and Dunning Kruger.

I want to offer a tempered view. The comment does seem deep and it probably is. But I work in physical prototyping - new products, new technologies. I do the "design and manufacture from real materials in the real world" part. I own a few hundred thousand dollars in advanced machinery and make parts for FAANGs and tiny research startups. I'm a named inventor on something like 16 patents. And yet daily, I read confident, well worded answers here -- about my domain -- which reflect views completely divorced from reality, or in living in some strange parallel reality.

Personally, I try to recognize this as an opportunity for me to be humble (as I never know who may read my comments from a higher place of skill and domain knowledge), and also to take what is posted here with a grain of salt.

Anybody who's worked with cables or ropes knows they spontaneously form knots quite easily. Would it be possible to utilize this property in manufacturing?

Shake a bunch of dry fibers before applying the matrix? Are they too stiff?

You're more-or-less describing nonwoven fabrics, which are used in a huge variety of applications. There's different fabrication methods, but they are typically not as strong as standard weaves since the patterning can't really be controlled.

https://en.wikipedia.org/wiki/Nonwoven_fabric

I believe nano-scale polymer materials composed of randomly formed knots are normally just called "plastic".

Could one shake that plastic while the chains are mobile to make these knots to make the material tougher?

That seems likely, though it does still seem like their results might fall into the "things behave differently at micro-scale and macro-scale" category. Normally we'd expect the (macro) knots we're familiar with to weaken -not strengthen - the cord or rope they're tied in (how much varies quite a bit with which knot we're talking about, but I don't know of any that don't have at least some negative impact). The little video almost implies the effect is similar to stretch, that the tangled version can't tighten up (with friction presumably absorbing some energy) like the knotted one does.

Interesting behaviour, usually any normal rope knot makes the rope weaker at the point of maximum curvature/choking etc. Would be enlightening to get a comparison why the different behaviours, just a scale/friction non linearities?

It looks like the loop of the knot permits slack which in turn causes stress on the fiber to be taken up by straining neighbouring weaves, rather than causing the stressed fiber to immediately strain to its failure point.

They claim the structure is tougher (absorbs more energy) not stronger (maximum force before breaks). As far as I see they say nothing about if it is stronger or weaker than straight lines. Likely because it is weaker due to the curvatures leading to stress peaks in the material before the ultimate force of two straight filaments was reached. They should have presented the force-displacement diagram too which is essential data (maybe the original article has it). I'd be curious seeing the force-displacement diagram of the two illustrated experiment compared to the diagram of two straight fibres tested.

Also I guess the toughness is just a relative matter to the dimensions of the structure but since uses much more material than two straigh fibres it is less tough by weight (due to the decreased strength). If the same amount of material was connecting the top and bottom with straight lines then that would lead to the toughest situation of all (absorbing the most energy). Again, guessing.

In the video the woven material may have tighter threads with stronger friction or more uneven friction distribution leading to reaching the yield limit of the filaments quicker. The woven with some 'lubrication' should have had similar properties assuming the same amount (length or weight) of material included. I'd also be curious then about the reproducibility of the results on the same kind of structure. Like if making the same knotted pattern would lead to the same results or slight deviation of geometry was affecting the end results significantly.

Not sure "knot" is the best wording. This looks like knitting. It is well known that knitted material is very strong and very flexible (much more elastic than weaving). And that all of knitting/weaving/braiding already enable to improve properties from initial material. Not so sure how the scale is relevant. I guess the contribution from the paper is that this is in the context of 3d printing.

Coincidentally I was looking into simulating rope and knots for a side project, and I don’t know how to really research this other than googling, does someone interested in these kinds of things know? I’d like to simulate friction and tensile strenghs, and other mechanical deformations (pulls, obviously).

You could probably look at protein folding libraries like pymol. Although not exactly the same, it’s generally close enough to ropes and it has enough knobs where you should be able to run the simulations you want

Hmmm. PyMol is molecular viewer, which has a python scripting interface to various libraries ... but is not a folding library I would say.

There is some research into the topology of DNA (especially circular DNA, like plasmids which is a knot) that considers things like writhe and twist. Not sure it would adapt too well to macroscopic systems though.

There is also the excellent KnotPlot for actually drawing the knots. I bought a licence once on a whim, but rarely use it :)

Wouldn't really be helpful- first, pymol is a visualizer, not a folder. Also, proteins have properties very, very different from typical knots.

That does raise an interesting question- can proteins spontaneously fold into knots? The answer is yes although I was discouraged when I originally asked this question in grad school.

Maybe you can look into (extended) position based dynamics https://positionbaseddynamics.readthedocs.io/en/latest/about... for ropes.

If knots are what's interest you more, then maybe you can have a look at the self-collision avoidance of "Repulsive Curves" http://www.cs.cmu.edu/~kmcrane/Projects/RepulsiveCurves/

In general, the interesting properties of real ropes are not well-modeled by simulation. Dirt, abrasion, water, stress concentrations, and other real factors will dramatically impact the results you get out in the real world. That's why these things are experimentally verified and ropes used in safety-critical applications (are supposed to) have huge safety margins, regular monitoring, and ideally regular replacement. Use the manufacturer numbers for real ropes.

But if you just want to simulate ropes, there's a few models out there from academics going "let's try and model this difficult system more accurately". Take a look at stuff like imc-der [0] and ridgerunner [1].

[0] https://github.com/QuantuMope/imc-der

[1] https://jasoncantarella.com/wordpress/software/ridgerunner/

I've been curious about this too- there seems to be an underexplored area of physics around designing new knots with useful properties ("easy to untie under load") and it's not addressed by knot theory at al.

I wonder if it is possible to drive multiple carbon nanotubes through C60 buckyballs to create a fabric which has tensile strength of C-C binding.

Maybe the pressure the knots hold is enough to maintain rows of sections of the material in close enough range for the casimir effect to apply, making making some part of it flexible (the material) and some part of it super resistant (the quantum fluctuation zone) which ends up giving this property.

Way too big for quantum stuff, it is just plain friction.

Most probably, but throwing geckos at knots is more fun.

It won't pass the IRB if you describe it like that.

The adhesion of the gecko’s foot is caused by van der waals forces, not the Casimir effect, just fyi

Thank you for correcting me.

The title calling this a "material" is disingenuous. One would assume that they somehow tested a "bulk" material coupon even if just a few mm^2. But the image in the article shows them testing a single knot.

In the paper, shows a test of eight knots (two sets of four in series)

So if you tie a knot on a rope, and pull on both ends to breaking point. Is the claim that the rope would not break at the knot. Not sure how true that is.

The rope would be a series of knots. Kind of like comparing the tensile strength of a scarf to synthetic wool strings (since regular wool is always knotted anyway.)

No, that is not the claim.

I know from fishing that if you have a spontaneous tie/knot on your 0,2mm line than it is dead but 0.4mm is almost always able to be unknot.

Is anyone able to reach this site? the server seems to be down

Works fine for me, but here https://archive.is/mwRR1

https://news.ycombinator.com/item?id=35094074

Reply on news.ycombinator.com