SCIENCE AT THE SHINE DOME canberra 5 - 7 may 2004
Symposium: A celebration of Australian science
Friday, 7 May 2004
Professor Marcela Bilek
Head, Applied and Plasma Physics, School of Physics, University of Sydney
Marcela Bilek was appointed Professor of Applied
Physics at the University of Sydney in 2000 and awarded an Australian
Research Council Federation Fellowship in 2003. She holds a PhD in engineering
from the University of Cambridge in the UK, a BSc in Physics from the
University of Sydney and an MBA from the Rochester Institute of Technology
in the USA. Prior to her present appointment she held a Visiting Professorship
at the Technische Universitat Hamburg-Harburg, Germany, and a Research
Fellowship at Emmanuel College, University of Cambridge. She also worked
as a visiting research scientist at the Lawrence Berkeley Laboratory,
University of California. Aside from her academic experience, Marcela
has spent time working in industry as a research scientist at Comalco
Research Centre, Melbourne, and at the IBM Asia Pacific Group Headquarters
in Tokyo. Her research focus is plasma processing for materials synthesis
and surface modification. She has published more than fifty refereed journal
articles. She received the 2002 Malcolm McIntosh Prize for Physical Scientist
of the Year.
Energetic ions and fine structure in the synthesis of new materials
A few of the slides will overlap with what I talked about yesterday, but I am going to expand on that now. I introduced you yesterday to energetic ions, but today we are going to take on also the fine structure in materials. That is another important area, and one where, basically, looking at what nature does tells us that we can use this kind of fine structure at the nanoscale to introduce even more interesting properties into materials – make them stronger, for example. I will walk you through some of those examples from nature and also from our own work.
I will explain why we are so interested in surfaces and why we are so interested in modifying them, and why we use ions. Why are ions such useful tools to do this, and how do we make them?
I will look at some of the new surfaces that we have made, and some of the ways that we use to control their properties. Then we will move on and look at what nature has told us about structure at the nanoscale, and some of the methods that we have been using to try and emulate that. And then a bit about the future and where these methods might go – some of the future applications.
The action happens at the surface
(Click on image for a larger version)
So why are surfaces important? For any object or device that we end up using – for example, here is an artificial hip joint – you will choose a material primarily based on its bulk properties. In this case, weight is very important; so is strength of the material. And in almost all applications, cost is very important as well: it depends on the value added by that application as to how high you can go in the cost area. So that sets the bulk material that we choose to underpin this device.
However, the performance of the device is very, very strongly influenced – in many cases, dominated – by the properties of the surface, and very often the optimal choice for the bulk properties of the device don't give us the best surface for the performance that we seek. This is where surface modification comes in, and that can be done in a number of ways. One is to produce new materials which we coat onto the surface – make them stick on – and produce an entirely new surface. Another way is to actually implant ions into the existing surface and change its properties that way. We are working in both those areas.
Plasma and ions
(Click on image for a larger version)
The tools that we are using, and which we have found to be extremely useful nanoscale tools, if you like, for these surface treatments, are ions, essentially. They are the important part. We make those by generating a plasma. Plasma is the fourth state of matter. If you heat a solid you get a liquid; a liquid goes to a gas with more energy; if you put even more energy into the material from the gaseous phase, you start to tear some of the electrons from the atoms and you end up with ions – these are positively charged cores missing the electrons – and electrons swimming around in a soup, basically. And that's what a plasma is.
So we generate these. We generate them from gases and also from solids. From solids we generate them by putting in either an optical or an electronic pulse, a very high-energy pulse, that ablates some material from that solid surface.
Given that we can generate them from gases and solids like that, we have really good control over the composition of the plasma that we make. That is important because one of the things we want to do to our surfaces is to control the chemistry that is on the surface, and that is done by controlling the composition.
It turns out, however, that energy is really crucial also to the structures that we get. The microstructure is basically controlled by the energy, the stress in the material is also dependent on that energy. As I said, we can control the composition, and the energy it turns out we can also control quite easily and precisely in these systems, due to the fact that the ions are positively charged, we can apply electric fields to give them whatever types of energy distributions we want over time.
The importance of energy
(Click on image for a larger version)
This is to show you the importance of energy in producing a microstructure. The composition in all these cases is the same. Let's imagine that this is maybe carbon or some other material. At low energies the ions come down and they don't have a lot of energy so they just drift and they sit on top of whatever structure they first hit. You can imagine that as they start to grow, you end up with little voids appearing, and since these ions are coming in from all directions like this, the probability of their actually going down one of these voids is very low compared with the probability of their sticking on the top. So you end up growing these sorts of columns with gaps in between. This is in general not a very good surface coating. You can imagine that its corrosion properties are not very good if it is full of voids. So this is not ideal.
For the next generation of materials, people discovered that using moderate energy – say, 10-100 eV – produced a much denser material that now didn't have any voids. The way that happened is that this sort of energy allows the ions to burrow in one or two monolayers and implant just under the surface. Unfortunately this produces a fairly stressed material. You are shoving atoms in where there are already lots of atoms, and the whole thing would actually like to expand a bit but it can't because it is held on this substrate. So you actually see that coatings like this will bend a substrate. If you have a thin silicon substrate, the force that this applies to the substrate will produce a bend in it. And these things have a tendency to peel off – so also not the best design.
If we integrate this sort of growth energy with very high energies we get something that has a much lower stress. And another neat feature that comes out of this is that as we first start growing it we manage to implant some of these ions below the surface, so we also produce a mixed interface here, which is much, much stronger. So there are a lot of benefits of integrating these medium energy depositions with very high energies.
Simulations of surface coatings
(Click on image for a larger version)
[Includes moving simulation – not visible on static slide]
In a lot of our work we actually tackle these problems on two fronts. One is in the computer; we simulate some of these growth processes in the computer. I have got one here, where there is a diamond surface and we are putting down some carbon films. This material is growing in the moderate energy regime, so it is about 25 eV, I believe, for this one. I will just show you that [as a moving simulation].
That simulation is done using an empirical code which was developed from a quantum mechanically based code. Because this is quite a large system, it would be a little bit too computationally expensive to do with the ab initio code. You can see that we can simulate the growth process quite well, and in fact the properties of the material as we vary the energy emulate quite closely what we see experimentally.
We also have very high-accuracy codes that examine the quantum mechanical nature of the electrons, and actually treat the valence electrons – that is, the ones involved in chemical bonding – as quantum mechanical objects, so as wave functions. The atomic cores here are treated as a pseudo-potential. The whole thing runs dynamically, and in the end of the processes we cool the object – so we simulate the ion impact as a heating of the structure and we cool it – over certain time periods corresponding to the size of the impact. We can then find out details of the chemical bonding that occurs. Here you can see these black dots representing silicon cores, silicon atoms, and here we have carbon. The colour you see surrounding it is actually a density map of the electronic density of the valence electrons, so you can see here a silicon-silicon bond and here a silicon-carbon bond, and in fact you even see the fact that the carbon attracts the electrons much more than the silicon in this bond, and therefore the electron density around the carbon atom is higher.
There is a lot of work going on in that sort of area, and also in looking at the trade-offs between accuracy and the scales of time and length that we can model, in trying to make compromises and even integrate different types of models to treat different parts of the system.
Back in the lab
(Click on image for a larger version)
We also have quite a large experimental program, which looks at actually producing these materials in the laboratory, and looking at their microstructures and their properties to determine what we have done and comparing that back with the computational data.
Here is one of our latest pieces of equipment that we built. It cost quite a bit of money; we recycled quite a lot of equipment from the University of Sydney. It essentially is a multisource cathodic arc plasma – this is one that generates plasmas from solid targets, and we can put in a number of these targets at once. You will see that we use this to make multilayers of different compositions.
We have also put quite a lot of effort into having some in situ measurements, and we have just finished putting on and developing a very accurate spectroscopic ellipsometry measurement, which basically optically looks at the surface of the material as it grows in the vacuum system. (This is all done in a big vacuum system.) And it can feed back to us, down to about a 1 Ångstrom of material, what we have put onto that surface, by looking at the optical properties, and the trick here is to look at the phase as well as the intensity. That's how ellipsometry works. We are now working on an in situ stress measurement as well, so that we can watch the development of the stress in our materials and the relief process, in situ, as it is grown.
Here you can see some really beefy power supplies. In order to make those plasmas from the solid materials, we have to put a very solid pulse in there. These run up to about 5 kilo-amps in each pulse.
A new coating for blood contacting surfaces
(Click on image for a larger version)
This coating was basically only carbon; there is no other material in this. So this is a single-source product. However, we varied the energy. In fact, the 200 layers here that you can see are not even so much the product of varying the energy but the product of switching off the system for 30 seconds. So we ran it for 30 seconds, depositing as well as pulsing – that gives a nice stress-relieved layer – and then we switched it off for 30 seconds, and then kept repeating the process. It was repeated about 200 times, hence the 200 layers. So there is some interesting surface relaxation that goes on in that stop, that break that we give the system. Actually, the layer that we have deposited starts to relax back from the surface.
We studied that quite closely and in fact have determined that that is what is going on. You can see, this is what gave us the clue. There are two layers here that are thicker than the other dark layers. This one corresponds in the laboratory book of the student that was doing it to a quick toilet break of about 15 minutes, and here he left the system for 45 minutes, suffering some sort of exhaustion and having to go and get a tea break. So that was really interesting. In fact, the microstructure, when we look more closely at that, is different in those layers.
With and without high energy ions
(Click on image for a larger version)
Looking at the performance of these films: that energetic ion pulsing which relieves the stress actually does make a big difference, and we can see that in the resistance to cracking of these materials. This one [on right of slide] is much, much stronger; there was the same deposition technique, except this one is with the 1 per cent of highly energetic ions incorporated.
Performance enhancements by multilayering
(Click on image for a larger version)
Once we discovered that we had actually made a multilayer system with this switching off and switching on of the system – which was originally just designed to keep the temperature low – we started varying the thickness of those layers and seeing what effect it had on the properties of the material. It turns out that there is quite a strong effect.
This particular one was our best-performing film, and it has layers of about 24 nanometres, stacked one on top of the other. This [2nd] dot underestimates the performance, in fact. This is a time to failure on a pin-on-disc test. The material is coated onto a disc, and a titanium pin is sitting on top of it as the disc spins underneath. We occasionally pick up this pin and drop it, to give an impact. So it is a really rigorous test. This particular film did not fail in the time that we tested it, so that point is actually not the failure but just the time when we stopped testing. So you can see that there is a link to the period in the toughness of these materials.
Stabilization of metastable phases
(Click on image for a larger version)
Also, we work in collaboration with a group in Germany, and this work [on slide] was done in Germany by our German colleagues. They were looking at carbon nitride, which is a similar system to carbon in that it has a graphitic form, this hexagonal boron nitride, and a cubic form which is equivalent to diamond. In their system, where they were putting this down, they used a similar technique with the high-energy ion pulsing to change the levels of stress. Here they plot the stress as they deposit the material, and they can actually control quite nicely which phase they get. When they are using the lower-energy ions, the lower stress appears and they get the hexagonal phase. As they switch to the high-energy ions, they get this transition region, which contains a mixture of both phases, and then it switches into the very hard cubic phase. So they also have been able to make multilayers of hard-soft materials, and the performance is also enhanced.
Time variations in ion energy
(Click on image for a larger version)
We took this a step further again in the graphite system. This time, instead of simply the relaxation process to make our layers, we actually tried to change our energies to create the layers. The reason they are different widths is that we actually did that on purpose; they are different times of deposition for the different widths.
Essentially, we ended up producing hard layers – these are the dark ones in the TEM picture – which are high density. They are basically diamond-bonded; they are an amorphous diamond form. And that was done simply by depositing this material at 150 volts bias, so that's about 150 eV energy. When we switched over then to higher energy, 750 eV, we started to grow these lighter layers. They are light because the density is lower, and their bonding is sp2; it is similar to graphite.
The other thing that is quite interesting in the microstructure of these layers is that although the dense ones are amorphous, these ones have produced crystallites which you can see here, columnar-type crystals, and if we look at the electron diffraction picture we can see that there is a preferred orientation in these crystals as well. In fact, these two spots correspond to the fact that the graphite planes are aligned this way: they are aligned perpendicular to the direction of the growth face.
So again we have just the use of energy to produce an interesting multilayer structure like that, with one constituent in this experiment.
Structure at the nanoscale
(Click on image for a larger version)
Our interpretation as to why these things probably do show those toughness enhancements that we saw in the carbon system is that you don't have to look too far, you can look at what nature has done in the abalone shell. Basically, the abalone shell is a whole lot of little plates of calcium carbonate, which is quite a brittle, chalky material, and in between is a protein called chitin – very small layers of that, which you actually can't see because those layers are about 100 times smaller than these. That produces an extremely tough material, something that as you all know is very hard to break, from two materials which on their own do not perform to that sort of level.
Photonic crystals in butterflies
(Click on image for a larger version)
Nature can also teach us about other types of structure that can produce interesting effects. This is an example in a butterfly, where this is all done from a protein. This is chitin again, but this time it has grown in a sort of a photonic crystal structure – a three-dimensional structure here, and when light penetrates through this structure some wavelengths go through and others are strongly reflected. That gives this bright colour – in this case it is a bright green colour – on the butterfly, because the green light is specifically reflected by this design of microstructure.
Time variations in condensing species and surface mobility
(Click on image for a larger version)
We then tried to extend this project, and started looking at materials of different compositions from layer to layer. We picked this one, a titanium nitride-titanium [TiN-Ti] stack, because we thought it would emulate the neat thing that nature has done in the abalone shell: a very hard material and a softer, squishier material.
It turns out that we were stumped a little bit; this often happens when you try and perform nature's tricks. Although we did make this multilayer, and it looks lovely here in the electron micrographs, we found that it wasn't very hard. It didn't perform better than titanium nitride on its own. And so we asked why.
The importance of interfaces
(Click on image for a larger version)
When we studied it further we discovered, looking at the electron diffraction pictures and then a dark field image – in fact, if you look very carefully you might be able to see that the layers are about this thin so there are a lot of layers through each of these columns – that what was happening here was that epitaxy was occurring at our interfaces, and we were growing one huge column. Okay, it was Ti-TiN stacked, but there were still interfaces between the columns and that is where the material failed. So obviously the trick is to not have that sort of growth structure. We have to break that epitaxy.
The area that we are now looking at is how do we try and synthesise these nanostructures. One of the paths that have been most looked at so far is nanomachining; that is the sort of thing that you see in etching. The idea there is that we use our traditional construction techniques – chisel and hammer, break things up – or in fact use building blocks to put them together, painfully. Nature is a little cleverer than that, though. In nature, most of the structures that I have shown you now, in fact all of them, are self-assembled. It just happens automatically, and it is basically some sort of a clever energy minimisation process under some system constraints.
Nanomachining – top down
(Click on image for a larger version)
This is what has been achieved so far with the nanomachining style. One thing to notice about all these images is that the smallest features are about 0.25 microns, so not that small compared with some of the multilayer stacks I have been showing you and the sorts of things that nature can do. Getting smaller with this method is just very, very difficult, so we won't go much smaller with this sort of nanomachining.
Atomic level construction with UHV atomic force microscope
(Click on image for a larger version)
Another technique is to use an atomic force microscope tip to pick up atom by atom and move them around. We have all seen nice pictures of what that can achieve. But again that is very, very painful and very expensive, so for a lot of materials it is not the way to go either.
Nature's way: Self-assembly
(Click on image for a larger version)
So in our next lot of work we are looking at self-assembly, and trying to use constraints on systems and the properties of the systems themselves, to allow them to create forms. For example, in the atoms and ions game, if the bonding energy between this material is much less than this material that it is sitting on, it will tend to form these islands, or globules. If, however, the energies are closer together, you will get much more expanded blobs, and in fact if the energies are very, very close, you will get a continuous coating. So there are some neat things we can do in this realm. We can also use temperature to change those processes on the surface.
Biology, of course, has some special sites that recognise each other, and we have also got a project looking at making surfaces that will tether biomolecules onto them so that we can use that specific natural interaction as well.
Question: I am interested in the pit stop phenomena that you saw when someone stopped and went to lunch. What occurs to me – I don't know what vacuum you use in your system – is that surfaces don't stay clean for long if there is any oxygen around. I am just wondering whether those films are pure carbon, or whether there are mixtures in there.
There is a little bit of oxygen in the system, so you are right. However, basically when we saw this phenomenon we had the contamination theory and the relaxation theory, and these two were vying for what was the correct view. We did look very carefully at the composition of these layers and we did some other experiments where we just had one surface and then we looked at what happened to that, but the composition itself – although there is an increased layer of oxygen – does not explain the sorts of depths that we see and the depth/time relationship. And also, looking at the microstructure under the electron microscope, although the material is amorphous – both the dark and the light layers – at the interfaces to those layers there is actually an alignment of graphite-like planes. So something is going on in terms of relaxation. We don't fully understand it, and there is still work going on to try and determine what the actual process is, but it is time-dependent and it decays with time. So it is rapid initially and then it gets slower. It can't just be contamination, because there simply isn't enough and there is no way for that contamination to penetrate the material.
Question: A couple of years ago I went to the technology park at Cambridge, where there were some people trying to sell a new process for making coatings of the type you defined: a really hard coating. But they were doing it in solution, so they essentially had a little plasma bubble. The processes you have described for large-scale industrial purposes – I am thinking of Colorbond and other things that we were looking at – are not very easy to do in a high vacuum. Are you familiar with any processes where you can do these coatings in liquids of this type?
I guess the most common process they use is electroplating – that is done in a solution and is a very common industrial process. There is another process that I am aware of that uses a plasma in a liquid environment, but what that creates is nanotubes. Generally it is a carbon arc which creates nanotubes. I haven't actually seen a process that uses a plasma in a liquid to create a layered coating, like a Colorbond. I can see that there is probably a problem with that, because your plasma is likely to be localised – I can't see how you would make a plasma in a liquid over an extremely large area. So it is going to run into the same expense problems as you face with doing it in a pure plasma.
The vacuums that we use don't have to be that high – 10-4 is reasonable – and people do use plasmas to do coating of textiles and so on, so there are atmospheric plasma regimes. But I can't see that a liquid-based technique would have advantages over that, if it is localised.
Question (cont'd): It wasn't localised. That was the interesting bit. There was light all over the surface, little bubbles, almost like soda chemistry, only different.
Question: I am interested in where we are in real life with these, as distinct from in a laboratory. For example, in your first slide you showed a hip replacement prosthesis, and then we have seen some wonderful laboratory experiments. Are we at the stage where you can implement some of these laboratory discoveries to coat hip joints, knee joints or that sort of thing so that they are useful, or is there still a period of development that needs to take place before we are at that level?
It varies from application to application. Obviously, in the medical field there is a lot of approval that has got to go on – a lot of clinical trials and so on. That multilayer carbon coating that I showed is in fact at the clinical trial stage. That is being used to coat blood-contacting surfaces, and there are actually heart pumps in people at the moment that incorporate that surface. Surfaces for the hip joints are at the stage of looking at how they perform, still in laboratory tests but with cell cultures, so actually looking at the adhesion of cells onto those surfaces, and improvements both by chemical flavouring of the surfaces and by incorporating different types of layers.
Question: You have concentrated a good deal in your presentation on getting good adhesion and hitting them with the ions to get that good adhesion. My experience – and I used to have a couple of groups working in an area very similar to yours – is that very often, although this is a very desirable characteristic, it is not the primary aim of going for a thin film. It might be optical properties in the thin film you are looking for, it might be thermal conductivity of the film, it might be electrical properties of the film and so on; I could go on and on.
More often than not, we found we had to compromise on adhesion, because if you just belt it in as best you can and just go all out for adhesion, you can't control these other, what you might call the primary, objectives. I wonder if you have any comments about that.
To some extent that is true, but I think it is also dependent on the way you try and control these things. For example, the optical properties and mechanical properties and so on you can control to a large extent by selecting your components either in your nanolayer or even in your single layer, just by the pure chemistry. The ion energy allows you to get that adhesion. So with a particular film with the same optical properties, for example, we can get poor adhesion quite easily and we can also get very good adhesion by doing some ion mixing, at least at that initial interface if the rest of it doesn't have a lot of stress. So you can actually use the ions in combination with the composition and what you know about the materials to engineer for both.
Although I didn't focus on other properties – optical and mechanical or electronic – in this talk, of course they are crucial to the application and they have to be satisfied first before you consider adhesion. But it is no good having wonderful properties there, particularly in something as sensitive as the biomedical area, and not having good adhesion. The coating simply won't perform. So you have got to focus on both. And it is possible, in my experience, in most cases to get both of those sorted out.
