Paul Mulvaney is a Reader in Physical Chemistry at the University of Melbourne, where he also received his BSc and PhD degrees. He spent four years working as a research scientist at the Hahn-Meitner-Institute in Berlin and has been a visiting researcher at the Australian National University, the Max-Planck Institute of Colloids and Surfaces and the Argonne National Laboratories in Chicago. In 1998, he took leave to work at a start-up company in California, pioneering the bioapplications of luminescent quantum dots. His current interests involve the fundamental properties and applications of nanocrystals in optical and electronic devices. He is the recipient of the David Syme and Grimwade Prizes and was a Humboldt Research Fellow in Germany in 2000. He has published over 90 scientific papers and his work has accumulated more than 2400 citations. He is a member of the editorial board of Advanced Functional Materials and the Royal Society journal Physical Chemistry Chemical Physics.

Nanocrystals – controlling the shape and colour of the future
by Dr Paul Mulvaney

This is a fairly simple presentation about the beauty and elegance of nanocrystals, and why materials scientists are so fascinated by what happens when you shrink things down in size. I have called it 'Controlling the Shape and Colours of the Future' because I hope, in 20 minutes or so, to convince you that within the next 10 or 20 years nanocrystals are going to be incorporated into all sorts of things in our lives.

What are nanocrystals? I am going to talk about crystals which are between 1 and 20 nanometres in diameter. That corresponds to particles that contain from 100 up to 100,000 atoms or molecular units. I am primarily interested in inorganic crystals, but you could work equally well with nanocrystals made from organic materials.

The reason we have chosen that length scale is that it is where we start to see the merging of molecular properties and crystal properties. It is a transition region, and I want to try and show you something about that transition region.

Let's start off by taking sodium chloride – which is table salt, for the non-chemists – and imagine what happens when we shrink it down. Take a large piece of table salt and cut it in half, nothing much happens. You have two pieces of table salt with the same properties. Keep cutting it down and nothing much seems to happen; the crystal basically maintains the same properties.

Well, take that down to its logical limit and imagine a crystal which contains only about 10 or 12 sodium and chloride ions. If you look at the lattice in the [Figure , only the central chloride ion, which is surrounded by six sodium atoms, actually believes that it is inside a crystal any more. The rest of the ions, 95 per cent of them, are on the surface and they are still half-bonded to the surrounding solution. So this crystal cannot possibly behave like a piece of table salt any more, because there is only one ion in there that actually thinks it is in a crystal. The rest of them are in a halfway world between the solution and the crystal. This chloride here on the surface is bonded to three or four sodiums, but the rest of the time it is bonded to the solution around it – or the air, or whatever medium it is in. So this crystal cannot be a bulk crystal. It is going to have to have some unusual properties.

Click on image for a larger version of figure 1

For every material, the same question is going to arise. When I shrink it down, what is going to happen to its properties, over what size regime is it going to change, and how is it going to change? So there is this transition region.

Let's start off with one very well-known example – gold. Nuggets of gold: chop them down, chop them down. How long do I have to chop them down before the gold will not be gold? And what would I see if I could keep chopping up a piece of gold? What is going to happen? Will it be a metal right down to the point where it is atoms, or is something else going to happen?

Obviously, the chopping down route is going to take quite a long time, so the way to build a gold nanocrystal would be to build it from the atom upwards. There are two ways to do that: a physical approach, or a chemical approach.

The physicist's approach can be illustrated by the first nanocrystal that was ever actually made, hand-built, by Eigler (figure 2). It was made by joining together caesium and iodine atoms, about eight of them. He did this at 4 kelvin, in a vacuum. It took him several months to build one single nanocrystal. So this isn't a very practical way to go, although this image is one of the ones that inspired the whole concept of nanotechnology. In fact, it led to a rash of incredibly silly statements such as we're "going to build consumer goods made from cheap, raw atoms".

Click on image for a larger version of figure 2

There is a much simpler way. We are simply going to use chemical forces. Chemistry works best between one and 10 nanometres. Joining atoms together is what chemists do. Reduce a gold salt in solution with the right reagent – sodium citrate – and you get this beautiful ruby red solution of gold nanocrystals (figure 3). It takes 10 minutes and can be done in your kitchen.

Click on image for a larger version of figure 3

All of a sudden you have got a completely different material to play with. These crystals are dispersed as molecules, so you no longer do the sorts of experiments you would normally do with a gold surface. You now treat this stuff as a chemical. You dissolve it, you play with it, it has got the same chemical resilience as gold but it has completely different colours. So you now have a way to now use gold in all sorts of optical devices and other things, based on this new colour.

How do we know that they are real gold particles? If you do very good high-resolution electron microscopy you can show that each of the particles is made up of a single crystal. The white dots you can see in figure 4 are the individual gold atoms lined up in the classic cubic array that bulk gold has. You can actually see the facets, the little bit of flattening of the surfaces – this particle is actually a cuboctahedron – and the space between the gold atoms is 2.0 Ångstroms, exactly as in the bulk material.

Click on image for a larger version of figure 4

For some reason, when we shrink the material down, it changes colour. There has been a lot of work spent on explaining this, and all sorts of models, but qualitatively you can understand some of it this way:

In a metal, electrons are free to move around. And because they are free to move around, it turns out they can absorb most forms of light. So most metals are very opaque: they absorb the light energy and the electrons move around a little bit faster. Once electrons are trapped in a box, in a container like this one, although they are ostensibly free they keep bouncing off the surfaces. As they rattle backwards and forwards, they actually find that there are only certain frequencies or wavelengths they like to absorb. So the electrons, because they are forced to bounce around inside the confined walls of the particle, reject certain forms of light. For gold in particular, red and infrared light are no longer absorbed by the metal.

So it is the presence of a surface which causes this colour change and the electrons see this surface most strongly as they are shrunk down in size. They see the walls closing in on them.

I would like to claim that we were the first to work with colloidal gold, but honesty forbids. It was Michael Faraday, back in 1850, who was the first person to say that this red colour is due to the finely divided state of gold. People had played with gold particles and made reddy-purple solutions before then; but he was the first one to have said, 'I think it is due to metallic gold.' And the sample shown in figure 5 is 150 years old and is still sitting in the Royal Institution in London.

Click on image for a larger version of figure 5

But even Faraday cannot claim to have been the first to have done nanocrystal science. I think for the purposes of this symposium it is worth trying to be historically accurate. The vase in figure 6 dates back to the 4th century AD. That is 1500 years ago, or in my way of thinking, 500 rounds of ARC funding ago! Romans playing around with glass had put gold salts into the glass and they had found that when they heated that glass up – the sand and the sodium oxide – they could get a sort of red colour coming out. They only managed to do it a few times; despite a lot of searches, there are really only three or four examples in the whole world of these red-coloured glasses from Roman times.

Click on image for a larger version of figure 6

They had phenomenal understanding of things like Fresnel coefficients. The glass is red in transmitted light, but as you turn it round, the light source around it changes to green. It is an incredibly spectacular piece of glass, and a very complex optical material.

Simply, the vase shows a guy, King Lycurgus, being strangled by grapevines. I have decided that that was probably the punishment for being drunk and disorderly while in the nude.

So we have got gold particles. They are grown as spheres, and they have different colours when they are very small. That opens up a lot of very, very simple physical questions. What happens if I now decide I want to use those spheres to make something? For example, suppose I want to make a film of those nice red particles, and I would like to have a red-gold film rather than a gold-gold film. Can I just put those particles down on a film and have a nice red film?

First of all you would say, 'Well, why not?' But then you would start to think, 'Hang on, if I bring all those red particles together, aren't they going to reassemble into a piece of bulk gold?' There is a really big problem, once you have made nanomaterials, in how you are going to use them, because they are only beautiful while they are small and discrete. As soon as you start to reassemble them, they are going to interact with each other. They are going to want to form a bulk material again. So of course, if I stick all those small red particles together, they are going to form a piece of bulk gold again, so there are going to be some real questions about how we go from 'nano back to macro'.

We got rather interested in the question of how many gold particles could you put down on a film before they started to talk to each other, before they started to think they were actually a piece of metallic gold again.

To answer that question, we took a whole lot of these nice red gold particles and we coated them with silica, each one, ..one by one…, …millions and trillions of them. The silica shell thicknesses are different (figure 7). What this means now is that when we pack these particles onto a film they will come close to each other but they won't actually be touching.

Click on image for a larger version of figure 7

Where the particles are a long way apart, you might think, 'Well, that's going to give me a nice red film.' But what about something which is 40 or 50 per cent gold mixed with glass. Is that still going to be red? Surely we have got a lot of metal here. Surely this material is now starting to think, 'I should be gold.' Or is it still going to show the ruby-red properties of the single particles?

The answer, is in fact, that as you bring gold particles down onto a film and bring them together, bit by bit the colours change (figure 8). Where we have very thick silica shells, the films are red. As we pack the particles closer and closer together, they start to talk to each other and the colours vary. We can get an incredible range of colours. This is all from the same red ruby glass particles – red ruby colloids in glass. We can make films of tunable optical properties, from red to crimson to purple, turquoise, cyan or blue, just by controlling how the red particles talk to each other.

Click on image for a larger version of figure 8

What is happening, qualitatively again, is that each particle has electrons in it which are fluctuating, bouncing around in the particle. When two particles come close to each other, if the electrons in one particle are moving continuously from side to side, the electrons in the nearby particle want to move in phase. So if the electrons go to the left in one particle, the electrons in the next one will try to move to the left as well. They can't talk very well when they are a long way apart, but as they are brought closer and closer together they start to oscillate together. This alters which colours they like to absorb. So these colours come about when all the particles communicate and create one large optical effect.

So if someone ever asks you, 'When is gold not gold?', the answer is fairly obvious. It is, 'When it's red.'

You have seen now how we can build nanocrystals as spheres. For over 2000 years, the one thing the Romans and Michael Faraday and no-one else managed to do was to make something other than a sphere. It has got the lowest surface energy. Most of the atoms want to be inside, they don't want to be on the surface. Spheres are the lowest surface to volume structure. Most crystals try to grow as spheres or something like it.

About three or four years ago the first papers started to come out suggesting that we might be able to have some sort of shape control. That is vital for nanoscience and nanotechnology. It is all very well to just grow particles, but if you want to grow a structure – a machine, electric circuits, bone – you have got to make nanocrystals grow in the direction you want. You have got to drive them.

At the moment this sort of research is largely serendipitous, but about three years ago it started to become possible to steer gold formation, and to actually start to form rods. This is the first stage. You can call them rods or, if you want to be more practical, you might think of these as starting points for wires – actually being able to grow a wire on a surface. The other interesting thing to point out about these rods is that each one, because it has got a different length, has a different colour. So the shortest rods are like spheres, they are red; the larger ones go purple and then go blue, which is one of the reasons I was interested in them. The shape controls the colour, the colour tells you about the shape.

Again if you zoom in on these with a very good electron microscope, they are truly single crystals of gold. The atoms are spaced just as in the normal case. It is just that for some reason they prefer to grow in one direction – and I am not going to go into why that is.

The reason I am telling you this is that we think in the last couple of months we may have slowly got to the point where we might be able to do something quite dramatic. What I would like to be able to do is to change direction. I would like to take a growing rod and push it off to one side, to say, go left. The idea of doing that in a homogeneous solution sounds quite strange, but what we have found is that under certain conditions we can start to get growth occurring at an angle. We have only just reached the starting point, but this will at least give you an idea about where things are going and what people are trying to do.

What we end up with is what we call dumbbells. We have got a rod, and then while it is sitting in solution what we have started to do is to direct the growth onto the ends. We think the reason it goes onto the ends is electrostatic. I don't want to go into it in detail, but it is enough to say that we have preferentially started to grow off the ends of the rods. If we can continue that, then we should start to be able to actually direct the shape of gold crystals. If we can understand why that is, we are well on the way to trying to make quite complex nanostructures.

When you get these sorts of structures, you can ask another very simple question, because one issue that I think will come up in the next few years is nanomechanics. Take an incredibly small piece of metal, it has a lot of atoms on the surface. Is it going to have the same strength as a bulk material? That is a very difficult question to answer. How do you pick up a nanorod and bend it to see how strong it is?

It turns out that if you hit them with a laser, their colours fluctuate. We think they are bending and stretching. So John Sader in the applied maths department has done a quick calculation – when I say 'quick', I think it took a few months to do the simulation. He has now shown that when a laser impacts on a rod-like material, it is going to start to stretch and contract as you see in the movie.

One of the few concerns that we have at the moment is that, in fact, if we start to build circuitry or any sort of object on this scale, thermal energy alone is going to be quite disruptive on this length scale, and the material may start to stretch, bend and break very easily.

By comparing the experiments and the theory, we actually believe at the moment that the Young's modulus – that is the stiffness of this material – is about 20 per cent less than bulk gold. It is not as strong as the bulk material. That's still not too bad.

I have shown you that gold is a material which, when we shrink it down, changes colour. It even changes colour when we change its shape. This is completely counter-intuitive to our normal knowledge about life around us. Things don't change shape and then change colour.

That is primarily with metals, and virtually all metals do the same sorts of things. Silver becomes yellow when you shrink it down, sodium and potassium become purple and red – all sorts of colours turn up when you shrink metals down.

But other materials will also change their properties. One of the most spectacular nanocrystals, one which is being researched widely, is cadmium selenide. Cadmium selenide is a semiconductor; it is normally a brown-black colour and it can be grown as single crystals for optoelectronics. But I am interested in what happens when you take it down to say, 3 or 4 nanometres in size. Again I wonder, 'If I keep cutting this piece of material in half, what happens to its properties?'

I would like to try to explain what actually happens in these materials. In most materials, any crystal, any material you look at, absorbs certain wavelengths of light and rejects others. It is controlled basically by the energy levels where the electrons are, and an excited state which is empty. So there are empty levels and filled levels (figure 9).

Click on image for a larger version of figure 9

In cadmium selenide, the bulk material, the gap corresponds to a fairly small energy, and it basically means it can absorb red light, it can absorb yellow light, green light, it absorbs all colours of light. And it emits red light. If I shrink the material down enough, what should happen is that this energy level will split and that the material should be able to emit different colours. Again why the 10 to 20 nanometre regime? Because theory predicts that only under 20 nanometres will this change in energy structure start to occur. Above 20 nanometres most materials have their bulk properties.

So if you can make the nucleation go fast enough to make cadmium selenide small enough, and you can stop them growing, and if you can stop them sticking together, and if you can make them with no defects, and if you can make them all the same size, then...

...if you have got a student like Craig Bullen, patient enough to spend three years doing it, you can actually start to make materials which show some pretty dramatic effects.

What you end up with is a material which can emit any colour you like across the visible part of the spectrum. Start with cadmium selenide around 7 nanometres in size, which is close to bulk – it still emits red colours and it is a brown looking material – as you shrink the crystals down in this very critical region between one and 20 nanometres the energy levels within the crystal are diverging, the electrons in their excited states have to make bigger jumps down to the ground state, and they emit photons ranging from blue down to red.

Click on image for a larger version of figure 10

To show you that this not a graphic but a real thing, I have brought some samples along. [Shows four sample fluorophores in vials.] These materials have something like 80 per cent quantum yield. That is higher than pretty well everyone else in the world has managed to achieve. It comes about from understanding about surface states.

These materials are more photo-stable than any conventional dye molecule. They are more robust than laser dyes, and you can make them any colour you like across the visible part of the spectrum. All you have to be able to do is to control the crystal size. So cadmium selenide is another extremely exciting nanocrystal.

The opportunities for this sort of controlled fluorescence are obviously very wide, ranging from tunable lasers and tunable LEDs – these will probably come out this year – to tunable LEDs based on nanocrystals. Biolabelled materials are also appearing on the market.

I hope that I have been able to show you that down at this nanometre length scale, all sorts of materials are changing properties. All I have simply tried to show you today is that they become size- and shape-dependent. The whole nanoworld looks quite different. If we can understand these materials and exploit these quantum size effects, we are going to see a host of new devices and processes.

I'd like to thank a host of colleagues, students and of course the Academy for its sponsorship of the Nanoscience symposium and the ARC for funding our research.

Questions/discussion

Question: In the crystal growth area, I guess you are technically growing your crystals with faceted faces. Is there a transition to more dendritic type growth, or is that further up in size?

PM: If you grow the material at low temperature, you can get more amorphous structures. You won't get dendritic structures, but you can get amorphous structures with different shapes. We deliberately aim to get as crystalline a material as possible, which requires us basically to do the nucleation at about 350°. That is because we want the particles to crystallise. We don't want those other shapes, because we only understand the energy levels in spherical crystals. So it is probably possible, but so far we have focused on crystals.

Question: I am curious as to the effect of the size distribution of your crystallites, especially for the non-spherical shaped crystallites. How does that influence your optical properties, and electronic properties as well?

PM: That is an extremely good question. At the moment, the way we are doing this is to cheat, that is, we use chemistry to make trillions and trillions of these things. But there is a size distribution, so all the optical effects that you see are in fact averages or ensemble effects. So the sharpness of the colour depends on the sharpness of the size distribution. To get yellow different from green, all the particles have to have within 5 per cent the same size.

We now do experiments on single nanocrystals. We take one crystal out of these solutions and we are doing the spectrum on that single crystal, because that is a very big problem, having to average the optical effects out.

Question: Could you possibly manipulate these properties by manipulating the size distribution – let's say it had monomodal characteristics – to manipulate the various optical and electronic properties as well?

PM: When we make the crystals, quite often when we first make them they are not as nicely distributed as the final electron micrographs show. We purify them and take fractions out, of different sizes, to try and get the properties that we are looking at as sharp as possible.

Question: Is it possible to use these laser sorting arrangements to get over that problem? Presumably you could interrogate the particles one at a time and kick them into boxes.

PM: There are two ways that people go. One is to actually spread them out on a glass slide and do the spectroscopy on single crystals. That is the ultimate way to go. The other way is that you can tweak the laser across the absorption spectrum until you are selecting out only certain particles. But you have always got the problem that you are looking at a subpopulation of a sample. You get narrower and better resolution, but ultimately in nanocrystals applications, single nanocrystals will be the functional device.

Question: You said that the size of these nanocrystals influences the colour that they show. Does the size also influence electronic properties of the crystals?

PM: It does. Effectively, in a small crystal the electrons, when they are excited or you push them – are hitting the surface. So, effectively, their ability to conduct is limited by their size. The models which predict the colour changes show that the resistance effectively varies inversely with the radius. As the particle gets smaller and smaller, the electrons can't move as far. So, effectively, the conductivity of the particle is decreasing with radius. For metal particles, their conductivity is dropping; they are less metallic as you shrink them down.