SCIENCE AT THE SHINE DOME canberra 30 April – 2 May 2003
Symposium: Nanoscience – where physics, chemistry and biology collide
Friday, 2 May 2004
Dr Angela Belcher
John Chipman Associate Professor of Materials Science and Engineering
and Bioengineering,
Massachusetts Institute of Technology
Angela Belcher is John Chipman Associate Professor of Materials Science and Engineering and Bioengineering at Massachusetts Institute of Technology, a position she took up in 2002. She is a pioneer in the area of using biological systems to make novel electronic and magnetic materials and to pattern materials on the nanometre-length scale. In 1997 she received a PhD in inorganic chemistry from the University of California, Santa Barbara. In 1999 she joined the Department of Chemistry and Biochemistry at the University of Texas at Austin, where her work focused on understanding and exploiting natural biological processes for manufacturing materials. Some of her recent awards include the 2002 World Technology Award, the 2001 Harvard University Wilson Prize in Chemistry and the 2000 Presidential Early Career Award for Science and Engineering.
Viruses put to work to make high-tech materials
We like to understand how nature makes materials, and when I say 'nature' I mean biological nature. We like to apply the ideas that nature has already used to make materials, to work on materials that we like to say nature hasn't had the opportunity to work with yet.
One of the things we work on is rapidly evolving organisms, to live with and work with electronic and magnetic materials, and to try to incorporate them into growing technologically important structures.
Click on image for a larger version
We like to look at how nature makes materials. An example is an abalone shell. Abalone is a biocomposite material made up of 98 per cent inorganic material in the form of calcium carbonate and 2 per cent of organic material in the form of protein. Yet it is 3000 times tougher than its geological counterpart. It is made in the ocean, at ocean temperatures, with ocean pressures, using non-toxic materials, and the reason that it is such a great material is that it is constructed on the nanoscale. The abalone has, over millions of years, evolved the ability to control crystal structure and make the metastable crystal structure of calcium carbonate, all using protein control.
So the kinds of things that nature does that we like to do are basically self assembly, molecular scale recognition, nanoscale regularity and self correction. One of our goals is to have biological molecules that can actually grow and assemble electronic materials in one beaker at the benchtop and assemble them on the nanoscale and also act like enzymes and self correct. So as they start to grow the materials, if there is a mistake they would be able to correct themselves.
Click on image for a larger version
There are many examples of biological materials. Another example is Coccolithophora, a calcium carbonate. It is a unicellular algae made out of calcite, the most stable form of calcium carbonate – again abalone. And if you look at a cross-section of abalone you will notice that it is made out of plates, which are the metastable crystal structure. They are very uniform in thickness, depending on the species, but it is in the area of hundreds of nanometres in one dimension and tens of microns in the other, and they actually naturally stack on top of each other to give a brick-wall-like structure. In between each individual inorganic tablet there are proteins.
Another example is diatoms, which are made out of silica (SiO2). They make perfect nanospheres of nanospheres, and they are all made very quickly – again using non-toxic materials. Another example is a magnetotatic bacteria which makes small, perfect, single domain magnetite (Fe3O4) particles that are used in navigation. This is the area that we like to look to for inspiration for making other types of materials.
Abalone secrete proteins containing very highly negatively charged amino acids with carboxylic acid groups. The distance between the carboxylic acid groups of the protein are thought to atomically match the calcium in the calcium carbonate. You can think of the proteins as being able to chelate calcium out of a solution, but chelating it in a very specific crystal structure so they actually start and nucleate materials. But as a materials scientist, something that is equally as interesting in terms of nucleating material would be to have another organic material that could bind. It could act as a site-specific poison to force crystals to grow in a different way. Basically, this is crystal engineering, using naturally evolved proteins.
In nature the elements used in bio-molecular materials synthesis are primarily calcium, in the form of calcium carbonate or calcium phosphate – like bones – a little bit of barium in the form of barium sulfate, iron in the form of Fe3O4, and silicon in the form of silic carb. Nature doesn't really work with the whole periodic table. In my lab we want to force organisms to work with other elements, and force them to grow materials that we are interested in. So we work in the part of the periodic table with semiconductors like gallium arsenide and silicon, and zinc sulfide and cad sulfide. We also work with magnetic materials, used for magnetic storage.
So what we are looking at is designing biologically-based building blocks for nucleating and patterning electronic and magnetic materials on nano-length scales, using tools from nature. Nature has already evolved the ability to work with materials like calcium carbonate. How do you force them to work with materials like gallium arsenide or indium phosphide or gallium nitride?
Our original idea was to do a peptide combinatorial approach, where a peptide is just a short protein. We decided to start with a billion different proteins and sift through them to look for ones that would be specific for materials that we were interested in. So we were looking at selecting peptides that have a high affinity for a particular semiconductor material. But we not only want to have peptides that have a recognition for a particular semiconductor material; we want them to have a recognition for one crystallographic face versus another – very, very high specificity. And then we want to molecularly imprint them and use that protein to grow a material at the benchtop.
Click on image for a larger version
This was based on an idea, called phage display, that we borrowed from the drug discovery industry. I will tell you a little bit about it.
How do you force interactions with materials that nature has not evolved interactions with? We decided to use a virus. This virus is called a bacteriophage, because its host is a bacteria. The particular virus we used is non-infectious to humans, non-infectious to mammals. The way a materials scientist looks at a virus is that it is 880 nanometres in one dimension and 6.6 nanometres in another. It is made mostly of a couple of different types of proteins. There is a major protein coat, where there are 2700 copies – it is self-assembled – and inside the protein coat there is a single strain of DNA. Also on one end of the virus there are five copies of a minor coat protein called pIII. Random DNA inserts into Gene 3 code for random peptide inserts on one end of the virus. What that allows you to do is have a billion different viruses that are all genetically similar, except that they differ from each other based on a small peptide on each end.
We can have about a billion different viruses in a 1-microlitre sample, and we force them to interact with a semiconductor wafer. We tried to use single crystals when possible, because we were looking for viruses that would interact with one side of a crystal and not the other, very, very specifically. So we throw a billion possibilities at a semiconductor wafer or whatever kind of material we are interested in. Most of them won't have any chemical affinity for it so we wash them away. Any that do have a chemical affinity for it, we remove from the surface by lowering the pH and disrupting their interaction, or by ion competition. Then we are left with a collection of viruses, based on a peptide amine, that has some affinity for this material.
Viruses cannot replicate themselves, so we have to infect them into a bacteria. We infect them into a bacteria, and the bacteria makes a million copies of the viruses that recognised the surface. At this point we call it a clone. We then actually take the population that had an affinity for the surface, and we re-expose it to a fresh semiconductor material and ask it to interact again but under more chemically stringent conditions. So what we are doing is saying, 'Okay, there is one collection that bound, but now can it bind?' Most of the viruses won't bind, and we wash them away. Those that do have an affinity we collect, put into bacteria and make millions of copies. We go through this process about seven times. We think of it as a Darwinian kind of process: we are looking for the ones that survive under the conditions that we are interested in.
We then infect the viruses into a bacteria, plate them out on a Petri dish, and each single infection event represents one virus that recognised this material. We use a toothpick, to pull out the virus and sequence the DNA, and from the DNA sequence we know something about the protein that interacted with that material. So in my group we can evolve interaction for materials in about one week.
So far we have looked at semiconductors, magnetic materials; we are talking to industry and working on materials they are interested in; and we are trying to do 'green' chemistry for the synthesis of these materials.
The initial idea that I was thinking about in the late '90s was: Can you find a virus that will recognise gallium arsenide? As you go down or up a column of a periodic table and you start substituting different atoms into a lattice, when will the specificity fall off? Can you find one that will recognise gallium arsenide but not aluminum gallium arsenide or not aluminum arsenide, or iron oxide Fe3O4 but not iron oxide Fe2O3? We were able to find both of these in my group in the first year.
We are getting interested in viruses as materials themselves, and there are other genes that you can actually manipulate in the virus and express different proteins. This would be a handle for growing and organising different materials, and so we have been manipulating many genes for different functions of growing materials.
One of the first materials we worked with was gallium arsenide, and we selected viruses for the crystallographic orientation of gallium arsenide that had both gallium and arsenic on the surface.
We then took a gallium arsenide wafer, and using traditional etching we etched into it 1-micron lines with 4-micron spaces and we filled in the spaces with SiO2. We then took viruses we had selected for gallium arsenide and we fluorescently labelled them, and then took a chip and dipped it into a virus solution. You could see that the viruses only bound to the gallium arsenide, not to the SiO2. This was our first proof of concept, that you could actually select a virus to be specific for a totally non-biological material like gallium arsenide. We were very excited about this, and we published it in Nature in 2000, but it didn't get to the idea of being able to grow a material. I wanted to grow and assemble a material.
So we looked to two column VI semiconductor materials, like zinc sulfide and cad sulfide – this work was really spearheaded by one of my graduate students. We went through and we selected for viruses that could bind to hexagonal versus cubic structures of the sulfides.
We sequenced protein after the third, fourth and fifth round of selection. By the fifth round we actually get to one predominant sequence, a clone that we call A7, which has the best affinity for zinc sulfide.
We then took that virus and we basically dumped zinc and sulfur in solution. We selected viruses that could grow, by molecular recognition, zinc sulfide nanoparticles at their tips, and then used them as templates to grow semiconductor quantum dots.
Well, we were pretty excited about that, but we wanted to push it a little further. This is in collaboration with some colleagues of mine at the University of Texas. We wanted to take advantage of the beautiful structure of the virus itself. We had focused on the pIII minor coat protein at the tip of the virus in doing genetic modifications to grow materials. But on the virus itself, along the length, there are 2700 copies of a major coat protein. One of the things that is so interesting about these is that not only do they self assemble – the length of coat protein is actually genetically controlled, so the virus is monodispersed, which is great for a materials scientist – but it is actually crystalline, so the proteins are all crystallographically related to each other.
We thought, 'Wouldn't it be interesting if you could take the DNA sequence for the pIII protein and actually clone it and express it on the major coat of a virus, and use that as a template to grow semiconductor wires?' So we were able to do that. We actually took that DNA sequence, cloned it into the major coat of the virus, and expressed it as a protein fusion along the major coat of the virus. We then dumped zinc and sulfide, and what we were able to do was grow virus-based semiconductor wires.
We made scanning transmission electron micrographs of these virus-based semiconductor wires. The wires were visible and not optically transparent to the electron beam because they had inorganic materials on them. But to further prove the presence of inorganic materials, we actually did elemental mapping, which showed that the zinc and the sulfur are actually mapped predominantly to the virus itself.
Then we did electron diffraction to look at the crystal structure along the coat of the virus. What we found was actually very, very interesting. Even though this virus had nucleated and grown semiconductor particles of very small sizes, about 3.9 nanometres, the particles themselves were all crystallographically related to each other. Actually, they act as a single crystal. The reason that they did that was that they had a most perfect template to start with. The virus template was perfectly organised; we expressed a protein and locked it into the coat of the virus and used it as a template to grow materials.
Click on image for a larger version
We also used selection to grow cad sulfide semiconductors. This is a cad sulfide based semiconductor virus wire, made at room temperature.
We produced a hybrid virus, where we expressed the ability to grow two different materials simultaneously on the coat of the virus. This virus can simultaneously grow zinc sulfide and cad sulfide.
One thing that was interesting, that actually broke one of our rules of only using room temperature and pressure, was that we found that if we actually heated these virus-based wires to about 500°, all the biological components of it burned off. What you were left with was perfect single crystal wires of the material. We could do this with semiconductor materials and magnetic materials, and they have the same aspect ratio as the virus did. So by manipulating the DNA in the virus you can manipulate the length of the virus, and by manipulating the length of the virus you can manipulate the length of a semiconductor wire. It is all genetically controlled.
Now what we are doing is selecting viruses that can bind to different electrodes on both ends, having different gene-based control of being able to grow materials between two electrodes. Then you can actually grow this virus-based wire and burn off the biological material, and you are left with a single crystal wire between two electrodes. We can do that for many different kinds of materials.
My student Dan Solis went and worked with IBM; he was working on cobalt platinum-based materials for magnetic storage. We made cobalt platinum-based virus wires and when we burned these off we actually had single crystals. This showed that we had been able to nucleate and specifically grow these materials – all at room temperature. These are usually materials that are grown at high temperature, at 300° and then eventually at 600°.
So we showed that we could grow materials, we showed that we could assemble them in highly oriented materials. But one of the things with nanoscience is you need to be able link it up to the rest of the world. So what we decided to do was see if we could extend this to basically supramolecular assembly. What my student Lee said was, 'Okay, let's select for the ability to do something, like growing a nanoparticle. But let's take advantage of the shape of a virus, which is about 1 micron by 6 nanometres, and see if we can use viruses to build liquid crystals.' And he was successful in doing this.
Click on image for a larger version
Using viruses we can grow a smectic phase liquid crystal, where they are all aligned; we can grow cholesteric liquid crystals, where they have a helical pitch to them; or we can grow nematic liquid crystals, where they are laterally offset. We can grow them for centimetres in length, and fill any spaces with whatever we want. We can control the cholesteric pitch by varying the concentration, and then separate the nanoparticles at different lengths.
And we can cast viruses as solids on a surface. I usually travel with samples, but I was afraid to bring genetically engineered viruses into Australia. But you can make viruses as a solid, containing 99 per cent virus and 1 per cent quantum dots. The solid is optically transparent, and if I shined a laser through it you could actually see the diffraction pattern on the wall, it is so highly organised. The viruses can pack closely together to form a crystalline material.
Click on image for a larger version
We were able to get viruses to pick up gold particles, just one gold particle per virus, and self assemble into a gold-based film. And the gold-based film is red, because of the size of the particles. We just got on the cover of Advanced Materials this week, for being able to show that you can put anything – organics, inorganics, biologicals – onto the tip of a virus and then self assemble it into a multidimensional structure.
We can put in enzymes, or phycoerythrin, self organise them into a material and form really nice crystalline material that you can pick up and move around with forceps.
Question: What are the applications of such technologies in society and in the future?
AB: We are looking at a lot of different applications. My first love is self assembling electronics, and that is the furthest out. We are looking at the ability to make growing, connecting wires, specific places on chips. We are looking, in our liquid crystal displays, at whether one day your laptop will be able to be made out of virus liquid crystal displays. We have a lot of applications in controlling crystal structure and particle size. It could be anywhere from making nanoparticles for computing to making diagnostics, we are looking at drug delivery, we are looking at imaging contrasting agents for detection, we are looking at handheld detectors for the field, for the Department of Defense.
We found by accident that our viruses are really, really stable when stored as solid films, and we are looking at them as a way of storing enzymes as a solid. They are stable for up to a year without any loss in their infectability. We are also looking at them as a way of storing vaccines that could be transported to Third World countries. We are looking at them as separation for environmental aspects.
We have too many possibilities and not enough people right now.
Question: Can you see applications based on using spherical icosahedral viruses?
AB: We definitely can. We haven't really extended to that yet. One of the reasons was that we did not want to mix kinds of viruses that we had in the lab. The other one was that we fell in love with the shape of these viruses, in terms of their ability to self assemble.
There are definitely possibilities, and other groups have looked at putting gold particles on different surfaces, different faces of icosahedrals and using those as connectors, so it is a possibility. Right now we are going to stick with the M13 for a while.
Question: Last week, at the Materials Research Society meeting in San Francisco, Mark Alper gave a talk on the possibilities that arose out of this type of work, and one thing he said – I guess with a whole bunch of funding agency people sitting in the audience – was not to hold your breath for applications. I was very surprised that he said at the start, 'There are no commercial applications of these types of activities. This is really showing some things that are really exciting, but it may take some time before they actually make it into commercial products.' Could you comment on that?
AB: I think that he was being conservative, which is what we always try to do. I think that some of the applications for the electronics are further out; I wouldn't say that in the electronics we are going to have anything in the next 10 years. I am not sure how long he was wanting to hold his breath or not hold his breath.
I have my own company that works on developing this technology for long-term applications, but there are already other very successful companies, for example ones developing the idea of the zinc oxide for sunscreens. There are also companies that are making semiconductor quantum dots for cell labelling and things like that, and possible diagnostics. I think that is much more in the short term.
In the short term we are looking at more applications in the biotechnology end than we are in the electronics end. I think it will come, but it is going to be a while because it is possibly a disruptive technology – you have to start thinking about how you are going to integrate wet things with traditional semiconductor clean-rooms and things like that. So it is going to take a while, but I am confident that new technologies will come from this.
Question: There is a similar scheme for selecting RNA aptamers to different types of surfaces. Have you or anybody else in the nanoscience field considered using them for organising particles and things like that?
AB: I considered it when I first started thinking about using proteins. I thought about RNA and DNA as well. I think that there are going to be groups who will start looking at this. I have not seen anything in the literature, but I have talked to people. One is Andy Ellington – I don't know if you know him – at the University of Texas. He does a lot of aptamer work. There are also people at Caltech who will possibly be looking at this. It is definitely an interesting possibility.
The reason I chose proteins was that at the time when I originally started working with this it was a really high-risk thing to work on, and I thought that we would have a better opportunity with proteins and amino acids because that is the way it already evolved naturally. But I don't see a reason why aptamers should not be used. It is going to be different, and it will be interesting to see what comes out of it.
Question: Can you tell us how you settled on using sevenmers – peptides with seven amino acids?
AB: We have looked at very many different sizes. For my PhD I worked on how abalone grow shells. Those proteins that control calcium carbonate are actually quite a bit bigger, and that is one of the things that kept me awake a lot as a graduate student, in terms of: if you really have control at the lattice level, why do they need to be so big?
So we were going to look at what the smallest size possible was. The biggest we have looked is about 15, and so we were trying to approach the size of a unit cell. The smallest we looked at is seven. I didn't really talk about what we found, that the most effective is seven amino acids that are constrained within a disulfide bond. You can't always get selectivity for that, but if you can get them to work they are the most reliable and the most reproducible.
So we have looked at the minimum sizes. Actually, with seven linear you don't get the kind of selectivity that you get with 12 or that you get with the nine in the disulfide loop, but I wanted to see how small we could push it.
