Now, I'm gonna turn on 2000 volts.

What?

And this is the first step in creating snowflakes in the lab.

This is totally wild.

What?

Crazy, huh?

The tips of those needles are like a hundred nanometers in diameter.

That is so wild.

Dr. Ken Libbrecht is the snowflake guy.

I was the snowflake consultant for the movie, "Frozen."

It's okay to conjure snowflakes out of your fingertips, but they have to be real snowflakes, or people aren't buying it.

(Ken and Derek laughing)

The US Post Office made snowflake stamps using my pictures.

It's not the kind of thing you normally think of when you start doing physics, that you'd be on a postage stamp.

You've written the book on snowflakes, literally.

So I had like two successful books in a row, so we just kept making books, until finally (laughing) they sold zero copies, and then we stopped.

(dramatic orchestral music)

So, are you kind of like a snowflake artist?

I call it a designer snowflake.

Because yes, I am designing this on the fly.

I don't have a computer that does all this for me.

I just do it by hand, so every one's a little different.

Ken knows so much about snowflakes, he can design and construct them to his own specifications.

So what's happening now is it's growing and doing its thing at some...

It's at -13 Celsius now, but I want to make some branches.

I'll just turn this down to -15.

And then I'm gonna increase the humidity a little bit, the super saturation, and you'll start to see branches come out there.

See you changing those conditions, just caused the plate to kind of stop and become really--

I changed the growth conditions to prefer branches.

This little thing right there, that little nub, that's the only thing that touches the sapphire substrate.

The rest of this is all growing above.

This increases the air flow.

Those are droplets forming, and now I'm really kicking it into gear.

(mystical orchestral music)

Now, what I'm gonna do is I'm gonna turn that humidity down to zero, so the droplets are starting to recede, and this will stop growing and kind of start to facet a little bit.

Let's say I want branches again, now I'm gonna really hammer on it.

So you're giving it a lot of moisture.

A lot of moisture now, but you'll see side branches.

You really start to feel you understand what's going on, when you can say, "Now I'm gonna do this," and then it happens.

It's fun. I can predict the future.

(laughing)

I like to think they're better than nature, and the reason is that the facets of just sharp.

All the edges on these things are just sharp and crisp, whereas in the sky they have to fall, and by the time they fall, and you pick them up, and you put them under a microscope, they've started to evaporate a little.

Boy, these are just, (snaps) bang, just crisp.

The first close-up photograph of a snowflake in the wild was taken in 1885 by American meteorologist, Wilson A. Bentley.

It was Bentley who originated the idea that no two snowflakes are alike, and he would know.

Over the course of his life, he took more than 5,000 photos of snowflakes, a selection of which appear in his book, "Snow Crystals," which is still in print today.

But most snowflakes don't look like the ones Bentley photographed, because he selected only those in pristine condition with uncommon beauty and symmetry.

I mean, when you're looking for snowflakes, I'll take a big piece of cardboard, and you just glance at it.

(exhaling) Crap, nothing, brush them aside, more.

No, each brush is a thousand snowflakes.

They're hard to find.

They're one in a million, I mean, literally.

We are all so used to seeing pictures like this, that we are blind to the mysteries of the snowflake.

Like, why do they all have six fold, radial symmetry?

Why are they so intricate, and yet so different from each other?

How do opposite arms of a snowflake mirror each other so perfectly?

I mean, how does one side of the snowflake know what the other side is doing?

And why are snowflakes flat?

They're usually millimeters in diameter, but micro meters thick.

The edges of a plate can be as narrow as razorblades, but the mystery goes even deeper.

Everyone pictures snowflakes like this, but the truth is they take all sorts of different forms, like this, a hollow column.

That is a snowflake?

That is a snowflake.

There are also needles, cups and bullets.

This is like my favorite kind of snowflake.

It is a capped column.

It started out growing as a column, but then the temperature changed, and then you've got plates growing on either end.

(bright orchestral music)

It's just a cacophony of different shapes.

All of these appear spontaneously.

There's no DNA or any kind of a blueprint for what's going on.

It's just water vapor freezing into ice, and all this happens.

So you've identified 35 different types of snowflakes.

Yes, there's no really one way to define a type of snowflake.

The first chart of snowflakes is like 41, I think, and then it got bigger - 60 or 70, and the latest one by some Japanese physicists, I think had 108 different types of snowflakes, and I found 108 was too many.

(Ken laughs)

(snow blowing)

How does simple ice create so many distinct forms?

(curious orchestral music)

All snowflakes form in much the same way.

Water evaporates into water vapor, individual molecules bouncing around in the atmosphere, and as this vapor rises, it cools and becomes super saturated, meaning there are more water molecules in the air than there would be in equilibrium at this temperature.

Water molecules condense onto dust particles to form tiny droplets.

And although the temperature may be below freezing, the droplets don't immediately freeze, but at some point, one droplet will freeze.

Inside, the water molecules lock into place, forming a hexagonal crystal.

This structure results from the peculiarities of water molecules.

Oxygen atoms attract electrons more than hydrogen.

And since the molecule has a bent shape, it's polar with oxygen being slightly negative, and the hydrogens, slightly positive.

Since unlike charges attract, hydrogen from one molecule will sit next to an oxygen from another molecule, forming a so-called hydrogen bond, and this is what creates the hexagonal molecular lattice.

But how does this microscopic lattice grow into a hexagonal crystal that we can see?

So you start with a chunk of ice, and these little guys are meant to be water molecules.

And what happens is there are these flat surfaces, which are the facet surfaces, and at a molecular scale, they're very smooth and flat.

And so when a molecule hits, a water vapor molecule hits that smooth and flat surface, it tends to bounce off, whereas here, it's rough.

There are a lot of dangling molecular bonds over here.

That's a rough surface.

And so when these molecules hit, they tend to stick.

It's a statistical thing, of course, but the probabilities are high that they stick here and low that they stick here.

So if you take any shape, and you just let it grow for a little while, the rough areas fill in, and the flat areas don't grow very fast, and you end up with a faceted shape.

And that's how we get from the quantum mechanics that governs a water molecule, to a hexagonal prism of ice.

This prism has two basal facets and six prism facets, which is important.

If the basal facets grow fast, you get a column.

If the prism facets grow faster, you get a flat snowflake.

Once there is a seed crystal, nearby water droplets evaporate and deposit water molecules onto the growing snowflake.

Since the corners of the hexagonal prism stick out farther into humid air, they grow faster, and now they extend even farther, so they grow even faster in a positive feedback loop.

This gives rise to six radial branches.

At the corners of these branches, additional branches can form for the same reason.

Around a hundred thousand droplets are required to make a single snowflake, and the process usually takes 30 to 45 minutes.

In the 1930s, Ukichiro Nakaya was systematically studying snowflakes at the University of Hokkaido in Japan.

He discovered that the different types of snowflakes don't all occur under the same conditions.

Instead, two factors, the temperature and level of super saturation determine what type of snowflake grows.

His findings are summarized in the Nakaya Diagram, but it's not a simple pattern.

Around -2 Celsius, you get plates.

At -5 Celsius, columns and needles form.

At -15 Celsius, it's plates again, and then below -20, you get columns and plates.

The Nakaya Diagram allows us to understand a rough history of any snowflake.

Does each snowflake in essence reveal its history through its shape?

Yeah, absolutely, to some degree.

You can definitely look at a snowflake and say, "Yeah, I know what conditions "that crystal grew under, more or less."

Your typical weather patterns, fronts, cold front, that produces a lot of capped columns, because as the cloud moves up, it starts to get colder and initially start to freeze at around -6, -10.

That makes columns, and as it gets colder, then it makes branches and plates, and so you get capped columns.

This also explains why snowflakes are so intricate.

The temperature and humidity at each moment of growth determines the structures formed in that moment.

The symmetry you see is not because one side somehow knows what the other side is doing, but because both sides of a single snowflake grow in the exact same conditions.

When the crystal changes its position, the temperature will change, say, and all six branches will see the same temperature change, and so they'll all respond the same way.

Different snowflakes, on the other hand, each take a unique path, and therefore they experience a unique set of conditions, which is why no two snowflakes are alike.

But in the lab, you can carefully control the conditions, so theoretically it should be possible to create almost identical snowflakes, and indeed, Ken has.

What's in here?

Poke your flashlight in there, and you'll see.

Ah ha, I'm imagining these are seed crystals of a sort?

Those are a little sparkly snow crystals.

Yeah.

You know this is another little chamber.

It's just a cold plate, and there's a little sapphire disc in there, and then I'm going to push this thing, my sapphire, all the way in here, and the crystals will waft onto to it and hopefully stay there.

The idea popped in, it's like, oh, if I grow two next to one another, they'll be kind of identical, and I call them identical twin snowflakes, 'cause they're like identical twin people.

They're not exactly the same, but clearly more alike than you would ever expect.

(gentle orchestral music)

Is it really true that no two snowflakes are alike?

You know, that's just a silly question.

(Ken laughs)

It's silly because no two trees are like, no two grains of sand are like, no two anything are alike.

Anything that has any complexity is different from everything else, because once you introduce complexity, then there's just an uncountable number of ways to make it.

If a pair of twins snowflakes are growing too close together, they end up competing for moisture between them, stunting both of their growths.

The Nakaya Diagram allows us to understand a lot about snowflake formation.

Ken has used his experiments to build his own version of the chart.

But what it doesn't explain is why do ice crystals form this way in the first place?

I mean, why do we get plates and then columns and then plates and columns again?

This has been a mystery, essentially since Nakaya introduced his diagram back in the 1930s, but Ken believes he now has an answer.

Anytime you have a crystal, the reason why you get these smooth flat facets is because it's not easy to grow more crystal on top.

There are so-called nucleation barriers.

What you need is a critical density of additional molecules of the substance before they can come together to form a little island that is stable enough to grow and add another layer onto the crystal.

When you're first forming a snowflake, you're always gonna start with a hexagonal prism with its two basal facets and six prism facets around the side.

And the nucleation barrier for the basal facets is different than that of the prism facets.

If the nucleation barrier is lower for the prism facets, then they grow faster, and you get plate like structures.

If the nucleation barrier is lower for the basal facets, then they grow faster, and you end up with column-like structures.

Now, the nucleation barriers of ice are known as a function of temperature, and this explains why around -2, the prism facets grow faster, and you get plates, because their nucleation barrier is lower.

You can also see why below -20 or so, well then you get columns, because the basal facet nucleation barrier is lower at those temperatures.

But what doesn't make sense is why we should get columns at around -5 Celsius and then plates again at -15.

So what is happening?

Well Ken's hypothesis is that these nucleation barriers are valid only for large flat facets, but if you had really narrow facets, well, the nucleation barriers would be different.

So Ken proposes that narrow basal facets have a dip in their nucleation barrier around -4 Celsius, and narrow prism facets have a dip at -15, so his hypothesis is that the graph should look like this.

This then is consistent with all the different forms of snowflakes that grow at different temperatures.

But what accounts for these dips?

Well, let's say we have a narrow prism facet, so we're growing a plate snowflake.

Water molecules that hit the basal facets are unlikely to reach the critical density required to overcome the nucleation barrier, so that surface grows only slowly.

But on either side of this narrow prism facet, water molecules can stick on the rough edges, and to minimize surface energy, the ideal shape of this face would be semicircular.

But if only the top few layers of water molecules are mobile to try to lower the surface energy, many of them diffuse onto the prism facet and in the process, they exceed the critical density required to overcome the nucleation barrier, and so they can grow the crystal on the prism facet.

So due to this narrow edge, the nucleation barrier is effectively lower than it would be for a large prism facet.

A similar effect happens for the basal facets, just at a different temperature.

And Ken has done experiments to investigate whether these effects are observed in the lab.

And so I did a series of experiments using that apparatus and man, it's just like, boom, just like that.

(laughing) Whoa!

When you make a model, and you sort of find it's supposed to do something, and it sorta does, it's just like, this might be right!

So far, the results agree nicely with the hypothesis.

So after 85 years, maybe we now understand the molecular physics of ice well enough to finally explain why snowflakes grow into such a diverse collection of columnar and plate-like forms.

And I've done a lot of my career in astronomy and astrophysics.

Nobody ever asks you what it's good for, I mean, never.

Not even once did anyone say, "What are those black holes gonna be used for?" No,

(Ken and Derek laughing)

Saturn's rings, "Why do you care about Saturn's rings? "What's the motivation for studying Saturn," nobody asks that.

Every time I give a talk, people are like, "What are you doing? What on earth is this for?"

I'll tell you the real reason, the real reason that I got into this.

You look at a snowflake and you kind of go, "Um, actually, (laughs) we don't have any idea "how that works."

Well, that doesn't work.

We have to know how that works, dammit!

Well, I want to be the guy that figures out how snowflakes work.

That's always been a driver.

You know, as a scientist, you want to figure something out.

(bright electronic music)

Hey, this video was sponsored by Brilliant, the interactive learning platform that lets you tackle concepts in math, science and computer science.

You know, YouTube videos are great for finding out about new areas of interest, but if you really want to master a topic, you have to try problems for yourself, and that's what Brilliant allows you to do.

They've recently revamped a lot of their courses to make them even more interactive, like their course on logic.

Here's a puzzle where you have to sort robots.

The questions increase in difficulty as you go, but they've always got helpful hints, so you don't get stuck.

Now, if you enjoyed this video, I'd recommend you check out their course on beautiful geometry.

And for viewers of this channel, Brilliant are offering 20% off a yearly subscription to the first 200 people to sign up.

Just go to brilliant.org/veritasium.

Plus, this offer is valid for subscriptions that you gift to others.

So, if you know someone who would benefit by exploring STEM concepts in a fun, interactive way, a Brilliant subscription makes the perfect gift.

So, I want to thank Brilliant for sponsoring this video, and I want to thank you for watching.