Hacker News Reader

Comment by tyingq | Hacker News Reader

I read the headline and suspected it was overstated, but the content of the article...for once, seems to match the headline.

That is, the long-accepted “solution-diffusion” theory that the water dissolved and diffused across the membrane is wrong. That instead, the membrane has spontaneous tunnels created through the pressure/heat/friction. And the tunnels are ~5 angstroms in diameter, allowing clusters of ~1.5 angstrom water molecules through.

That's fairly exciting, as material designers can now target and optimize for that behavior.

I was hoping for more detail on how it keeps salt from following the same path. Individual NaCl bonded clusters are larger than water molecules, but searching around suggests they aren't always much larger.

View on HN

Had a skim, this seems relevant [0]

> The Na+ hydration shell, which is formed by three water molecules, is quite stable while the K+ ion can only perturb the water structure in its immediate neighborhood so that it is unable to attract the water molecules. As a result the diameter of the hydrated Na+ is about 0.5 nm while the K+ diameter is equal to its ionic diameter.

Basically in water Na+ and Cl- are surrounded by water molecules attracted by electrostatic forces. This will add bulk to the naked ion and they are reasonably stable/robust

You can get an intro into solvation shells here[1]

[0] https://www.sciencedirect.com/science/article/pii/0005272896....

[1] https://en.m.wikipedia.org/wiki/Solvation_shell

NaCl isn't bonded in water, it's ionized and dissolved. Each ion (Na+ or Cl-) is surrounded by a large number of water molecules due to the electric field of the ion.

That's not completely helping me visualize why it doesn't pass through the membrane. Are you saying that each water molecule surrounded (Na or Cl) ion stays intact as a group, and is too collectively too large to pass?

The water molecules aren't chemically bonded to the ions, but they are "bonded" by intermolecular forces. Although weaker than a chemical (intramolecular) bond, the intermolecular forces are still strong enough to "bond" water molecules to the ion. So either only water molecules not "bonded" to ions can pass through these channels, or the pressure differential across the channel can free the water molecules from the ions.

It's not quite a pressure differential, but a chemical potential differential (but thinking of it as a 'pressure' gives a useful mental model).

It’s force per area so it’s definitely a pressure, the difference is just that the driving mechanism is chemical potential instead of a difference in number of particles.

Tough a system put into contact with a resevoir of a constant concentration should try to increase in volume until the concentrations on both sides are equal.

It's probably the 'constant concentration' part that is confounding the two. It is connecting number of particles with volume, whereas the canonical ensemble has them as separate terms.

i always wondered why membranes. can we not drill 0.2nm holes in silicon for instance?

If you had a thin sheet of silicon filled with 0.2nm holes, wouldn't that be a membrane?

Not if it’s thick and/or rigid, right?

What would be the advantage? Presumably it would have to be very thin, so rigidity seems like it would just mean fragility. Are you seeing it differently?

Just speculating!

Yes, that would be a membrane. There a number of membranes with larger pores that are fabricated like above, but out of polycarbonate (track etch membranes).

Some people have also experimented with single and double layers of graphene with a few very small pores. Because they are so thin, you don't need as much porosity for high transport rates.

You still have fouling issues, though.

What you're picturing has a name: https://en.wikipedia.org/wiki/Fritted_glass. It's used in a https://en.wikipedia.org/wiki/B%C3%BCchner_funnel instead of filter paper, to efficiently separate liquids from solids.

Also, if you take some fritted glass and wrap it around a solid core, such that the holes don't go anywhere — i.e. they're just pits that water molecules (but not anything larger) can enter, and then get stuck there by surface tension — then you've get https://en.wikipedia.org/wiki/Molecular_sieve s.

> If you had a thin sheet of silicon filled with 0.2nm holes

That's literally individual silicon atoms removed

"The nearest neighbour distance is 0.235 nm" https://www.princeton.edu/~maelabs/mae324/glos324/silicon.ht...

Silicon lattice spacing is ~0.5 nm - you're talking about a single atom

> can we not drill 0.2nm holes in silicon for instance?

I think that this is interesting question, and should not be waved away with "just drill some 0.2nm holes, use a real small drill bit"

0.2nm is tiny, given that it's on a scale that impacts how H2O molecules behave.

It's also smaller than our ability to make the circuits in computer chips, which are currently > 1 mn (1).

Use a laser? Light with a comparable wavelength is x-rays. As used in the chip manufacturing (2)

This is about size of the spacing between individual atoms (3)

So it seems like a material with 0.2nm holes "drilled" into it would be a major engineering feat at least.

1) https://research.ibm.com/blog/1nm-chips-vtfet-ruthenium

2) https://en.wikipedia.org/wiki/X-ray_lithography

3) https://en.wikipedia.org/wiki/Atomic_spacing

I’m not sure we can do that. Even if we could, drilling millions of holes is a lot of work.

Also, for membranes, it’s not only the size and shape of the ‘holes’ that count. Many have specific chemical properties that do not solely filter on particle size.

That would cost 10000000x as much to manufacture. Think making a net versus hollowing a solid cube of steel into a net.

no shade but this is peak "can't we just" hacker news energy

What's wrong with "can't we just" questions?

Lots of curious laypersons here, and it's often a useful way to learn. Usually it's pretty easy for an expert to answer such questions.

nothing wrong with it that's why i put no shade

It's only wrong if you push the follow-through onto others. That's lazy. If you're really curious, spend 5 minutes on it.

In this case, if you ask "can we just do this" you should follow through with "how big is that actually, what's of comparable size?" We can all google this, or learn how to. It's a useful way to learn. You won't learn how to find things out, by always just asking "can't we just" questions and walking away.

In this case, soon you'd find out that 0.2nm is on the same scale as the spacing between individual atoms in a silicon crystal.

Literal "drill" bits for removing 1 atom at a time are obviously out of the question.

It is 10x smaller than the 2nm, the scale of the best x-ray lithography computer chip manufacturing processes.

I’d say that peak “can’t we just” comes with an assumption that we could just, which is quite different from this comment which comes off to me as hoping to learn why we can’t just.

I often want to honestly ask "why we can't just do X", but I feel like I end up coming off as asking "can't we just do X" no matter how hard I try to phrase it otherwise. I feel like this is a place where English grammar just doesn't give you any good options.

I’ll often ask “what are the problems with doing X?”.

It doesn’t imply that there should be a solution but you still want to learn why it might not work. Often you can continue the conversation enough to either exclude it or think some more about the problems that are described.

i don't think you quite grasp how small that is

Yeah, the smallest drill bits regularly commercially available are on the order of 1mm. That’s 6 orders of magnitude.

Individual silicon atoms in a crystal are around 0.2nm apart.

https://www.princeton.edu/~maelabs/mae324/glos324/silicon.ht...

So no, you don't get drill bits that are 1-2 atoms wide.

Also it would take a long time to "drill" a useful number of holes if you do it one by one.

Use X-ray lithography maybe? The wavelength can be short enough and it might be more feasible to do in bulk.

You can "drill" 5nm holes in aluminum oxide[1]. Which is pretty sweet, but calling that process "drilling" is a gross misnomer. The finest drills[2] out there are measured in hundreds of nm, and still take several seconds to get through relatively thin material. To drill the billions of holes you'd need for a reverse osmosis membrane, you'd die before it was finished.

[1] https://en.m.wikipedia.org/wiki/Anodic_aluminium_oxide

[2] https://www.tpc.toray/en/technology/hole/hol_005.html

Reply on news.ycombinator.com