Professor William (Bill) Compston is a renowned geophysicist who began his research career fingerprinting and dating rocks at the University of Western Australia before moving to the Research School of Earth Sciences at the Australian National University. He was a principal investigator dating lunar rock samples that were collected by Apollo 11, but is best known for his work developing the Sensitive High Resolution Ion Micro Probe (SHRIMP).The SHRIMP is a great achievement for Australian geology and was used to identify the world's oldest mineral, found in Western Australia. Bill is a Visiting Fellow at the Australian National University and has received many awards, including the Flinders Medal, the Mawson Medal and the Centenary Medal. He is a Fellow of the Australian Academy of Science, the Australian Academy of Technological Sciences and Engineering and the Royal Society of London.
Interviewed by Mr David Salt, 2005.
Bill, you were born in 1931 in Western Australia, a state founded on its mineral wealth, and your mother came from the WA goldfields. But I believe your connection to geology and minerals goes back even further.
Well, my mother’s and my father’s antecedents that we know of went to the Victorian goldfields, but with an interest in finding gold rather than studying it. They both arrived in the same year, 1855, and got off the ship at Portland.
Do you have any early memories of living in the west?
Oh, lots of them. As a small child I lived with my mother and father at his butcher’s shop. My schooling was a happy time for me. I always found that I was doing well at school and it didn’t require too much trouble. (I guess I didn’t muck up as much as the other children.) But during the Second World War we had to go to Toodyay, which is about 50 or 60 miles – in the old measure – from Perth, to get away from the military preparations all round Fremantle.
My father died in 1943, after which my mother managed his shop until eventually she couldn’t do so any longer and she sold it. Then we had a holiday visiting relatives in Melbourne. But we were, in effect, trapped in Melbourne, because although we got a train passage out from Perth to Melbourne we couldn’t get one back, on account of the war.
I believe you developed a love of piano playing and music during your school years.
Ah yes. I started learning the piano when I was about seven or eight, still living at my original home. Then it tailed off because I didn’t do my practice, but I took it up again as a Leaving Certificate subject at high school. That was a bit different, and practice used up a couple of hours each day – rather to the resentment, I think, of the people at school who wanted me to play in a sporting team.
You studied for a Bachelor of Science at the University of Western Australia, where you first took up the study of geology. What attracted you to this subject?
Well, I did it rather than do the fourth subject that everyone else did, which was botany or biology, and because my brother had interested me in the topic. When we were in Melbourne during the war he was also there, in the Army. Having a science degree (in geology) from West Australia, he was attached to some ordnance section to defuse hand grenades and things like that. But he took me round to the various geological ‘monuments’ in Melbourne – well-known rock outcrops, mainly fossils – and I found that very interesting.
I enjoyed geology at university. As a qualitative subject it was a matter of remembering all the names of rocks and minerals, for example, which was fairly straightforward. And I enjoyed the field trips and field camps. So I kept doing it.
Do you have fond memories of your undergraduate years, and the people you were with?
Oh yes. All memories tend to get rosy as the time goes on. I made some very good friends and connections as an undergraduate. Unfortunately, the best friends have died, which is upsetting.
Through my wife, who had been to Perth Modern School, I got in with a group of people from there. (She was in the same year as I was at the university, from first year onwards. By coincidence, one of her subjects also was geology.) This was a group of people who liked to do well, and that spurred me on to emulating them. It was very good for all of us, I think. We talked to each other about mutual problems in physics or mathematics – How do you understand this, that or the other?’ – and that’s the best way to learn.
I believe you also played classical music quite loudly.
That’s right! At lunchtimes, a couple of days a week, we would – by permission – use the big music system in Winthrop Hall to play records selected by one of the group who was especially fond of music and knowledgeable about it. We’d turn it up as loud as possible so we couldn’t hear anything else. That was our escape. And that’s what people do in their cars these days, isn’t it?
Your PhD was also at the University of Western Australia, on carbon isotope variations in rocks. What do variations in carbon isotopes tell you about rocks?
Well, plants manage to ‘fix’ carbon from the atmosphere: they absorb CO2. But they fractionate the light carbon isotopes, so the isotopic composition of carbon in plants is lighter than the average isotopic composition in the ocean and the atmosphere. This becomes a ‘fingerprint’. If inside a mineral you find a shred of something, say of graphite, which has a very light isotopic composition, it might have had a biogenic origin. So that was the interest.
This sort of study was just beginning after the Second World War, essentially led by the chemists and physicists who had been involved in or on the fringes of the Manhattan Project around Chicago. In that program they had been studying how to measure isotope ratios properly and why these were different, and as they finished and went back to the universities they spread ideas of what you can do with various sorts of isotope ratios.
One of the people who knew about all of this was Sir Mark Oliphant, who came as a founding professor of the Australian National University in its original form. He carried the message to various state universities that if we wanted to get involved with something new in departments where there was both a physics and a geological background, then it would be very good to do things like isotope dating and isotope fingerprinting. And Peter Jeffery, my supervisor, took up that stimulus.
What memories do you have of your supervisor during this time?
Well, Peter Jeffery supervised a lot of people besides me who are now in this field. Perhaps the best known are John de Laeter and Malcom McCulloch, who was a couple of years after me at the university, in the Physics Department. (Both are Fellows of this Academy.)
Peter Jeffery himself was a strange mixture of boyish enthusiasm and good physics judgment, with a practical knowledge of things as well. As a young man you tend to think that older people don’t know much, and it takes you a little while to realise that is not so. It took me about 20 years after leaving to appreciate just what he had done for me in not allowing me to give up when things got too hard, and deliberately making me do simple things of a technical nature so I would learn about them. You see, we were never taught electronics at school or at university beyond book learning, sitting down on a bench and taking lecture notes. It is quite another thing to go into the laboratory and have to fix something up, and that is one of the many things that Peter showed us.
Peter was also inspirational. He wasn’t content with anything that was second-best, and he insinuated it into us, without our knowing it, that we had always to try and do the very best, no matter how trivial the task was. He never wanted to hear that term ‘good enough’.
You gained a Fulbright Scholarship to spend a year across at the California Institute of Technology, didn’t you?
Yes. The Fulbright was a travel scholarship, and at Caltech I was offered a research fellowship because I was in that particular field. There I continued to work on fingerprinting, this time with oxygen isotopes. The carbon and the oxygen isotopes in the hydrological and the atmospheric cycle are all fractionated by the same processes but they all give a slightly different reaction to it, and it is best to have them both present at once if you can.
Why is it important to have a well-defined fingerprint of a rock, using carbon and oxygen isotope variations?
It allows us, in the case of oxygen isotopes, to deduce the temperature at which calcitic shells were formed in the past. So, if we go back 250 million years, we can tell that a certain mollusc grew at a certain temperature. There was even the thought that you could trace the annual temperature range of the water, but we now know that other factors also influence the apparent temperature of the molluscs: the water’s salinity changes and then, in turn, so does the oxygen composition of the water.
Our goal, though, was to work out palaeo-temperatures. It was known by that time, or very soon afterwards, that in the past there had been very big global temperature changes – natural changes, not man induced. (It wasn’t until later that carbon dioxide and the global warming effect were blamed.)
I should mention that before I left the United States I went across and spent three months in Washington DC at a branch of the Carnegie Institute called the Department of Terrestrial Magnetism (DTM). They no longer do terrestrial magnetism but they use modern analytical techniques including the determination of uranium and lead, and rubidium and strontium, and potassium and argon, by various sensitive and accurate methods to date rocks. I went there to learn those methods, and I was taught by a group of people such as Tom Aldrich and George Wetherill. The place was led by a former scientist, Merle Tuve, who was also involved in the Manhattan Project.
The Carnegie Institute, at that time, was like the Australian National University in its early days: funded by a block grant from the mother organisation, with no need to consult anyone else about what it would be doing. It chose to appoint scientists that it thought were good investments for their science ability, and it allowed them to do what they thought was best.
Then it was back to the University of Western Australia, this time as a lecturer in physics. But you were also given the research task of working out how to date whole rocks, using rubidium–strontium ratios. Could you tell us about this work?
Yes. In Perth I instituted the dating technology that I had learned at DTM. Peter Jeffery himself had been there earlier and so he was interested in it, and the task I was given was to develop the radioactive decay of the element rubidium, which gives off an electron to become strontium-87. That becomes a geochronometer if you can manage to measure the rubidium-87 that is left and the amount of new strontium-87 that is formed.
We were working on the local rocks – always a good start, because they are easy to access and the local geologists are happy to help. But the answers we got from the separated minerals were unexpected.
We worked on the minerals biotite and muscovite, which are micas but are rich in alkalis, including rubidium. Muscovite, in particular, is always low in common strontium, so that is a good mineral for the purpose. It is technically easy to do the analysis, but the micas were all giving the wrong answer, according to the geologists.
We weren’t sure that it was wrong, so we decided to try taking a set of 'whole-rock' samples instead. The picture was that the micas lost their radiogenic strontium – it diffused out, possibly because of later thermal events impressed upon the rocks. But how far away did it go? If it went into the nearby minerals, it should still be enclosed in the whole-rock sample. We set out, therefore, to test what an array of whole-rock samples would give. Lo and behold, they gave 2500 million years; the micas gave about 600 or 700 million years; and the geologists, who had this instinct that the rocks should be very old, were proved correct.
It was our good fortune that nobody else in the world had actually done that particular experiment. The Americans were stunned, I think, that they hadn’t done it or thought of it. In fact, the South Africans, at the Bernard Price Institute, had started working on whole rocks just a year or so before, but they had the misfortune to work on unmetamorphosed rocks where the mica ages in the whole rocks were the same. So their results didn’t command much interest, whereas ours did. And our results became a letter to Nature.
We had another letter to Nature when we showed that the strontium that was lost from the mica was in fact retained by adjacent plagioclase – you could separate the feldspar and show that it was grossly enriched in strontium-87. (It had to be somewhere, and it was in the plagioclase.) That was a great start for the laboratory in Perth and for myself as well.
You didn’t stay in Perth, however. I gather that Professor Jaeger, the head of the ANU Department of Geophysics, actually came across in the early ’60s and headhunted you for the ANU.
Well, ‘headhunted’ is the modern term; it would have been much more discreet in those days. I think it was our success in the laboratory in Perth that led Professor Jaeger to try to persuade me to come to Canberra and concentrate on research.
Were you keen to take up his invitation?
Yes, I was. ANU was regarded with a bit of awe – people hadn’t yet seen it as a big rival for funding and students. It had a reputation as a great place to go to because it was on the rise. And indeed it was, so there was not much reason for me not to go there.
I believe there were family in Canberra already, including your brother who originally got you interested in geology.
That’s right. I had two brothers and my mother living in Canberra.
Could you tell us about your work at the ANU during the 1960s?
The first thing was to get a mass spectrometer that would be suitable for the analysis of solid samples. Professor Jaeger had access to a mass spectrometer which was being used by John Richards to look at the lead isotope ratios in galenas, in ore minerals. But that, like most mass spectrometers, was a gas machine and the decision was to convert it totally to solid-source work for the rubidium–strontium studies. So that was the first job.
We bought a conversion kit from the Metropolitan Vickers Company, in Britain, which had made the mass spectrometer. (Through Oliphant’s ties with the British atomic energy establishments, he knew which of the various companies made what, and knew the people in charge. And the old boy network was very strong.)
So you had access to good technology at ANU?
Well, as good as it was anywhere at that time. The Americans would say theirs was better, of course, but this was the British equivalent.
What is the process by which you would actually make an isotope ratio?
First, whether the sample is a whole rock or a separated mineral, you have to dissolve it chemically – with clean reagents, in a clean atmospheric environment. Second, you have to add a known amount of isotope ‘tracer’. This doesn’t mean a radioactive tracer but a tracer that has a characteristic isotope ratio itself, so that you can use it to determine the amount of the rubidium-87 or the strontium-87. And the tracer has to mix fully with the sample.
Then you have to separate out the rubidium from the strontium and from everything else in the dissolved mineral, and that is done by ion exchange columns. (The chemists did a lot of chemistry and discovered this. Everything was given a huge fillip by the Second World War.) The process involves putting the sample on the top of a glass column which is packed with a brown organic resin with the capacity to absorb cations at different strengths. But you can flush the cations out progressively, like a chromatographic removal. You pour on acid of the right normality from the top and then you start the sample ions moving down, and you can separate out the rubidium and, later, the strontium. You collect these different fractions and dry them down, and you are left with them as a little bit of white powder at the bottom.
To measure the rubidium and strontium you have to pick this powder up in a drop of water – they’re all water soluble – and put it on a strip, a filament, of the metal rhenium. Rhenium is the best: it is a noble metal and it is inert. You mount the filament, by spot-welding, between two metal prongs and you dry everything down by heating it with a current. The whole assembly fits into the mass spectrometer and so you can put the filament, together with the dried-out small powder, into the machine and pump it down to create a vacuum.
When you heat the filament up again, the samples first melt on top and then start to evaporate. In a lot of black magic by way of detailed chemistry, which no-one pretends to understand, eventually the alkalis and the alkaline earths evaporate, partly as ions. And it is the electrically charged ions, evaporating in vacuum, that we can operate on. We extract them by electric fields and mass analyse them – that is, we accelerate them and send them through a magnet which bends their path according to their momentum and so separates them according to their mass.
You gained an international reputation for your ability to do this many-stepped process.
Well, we were lucky. Actually, three or four years after I went to ANU, Professor Jaeger bought us a much larger American mass spectrometer. This was funded in part by the Bureau of Mineral Resources, which had entered into an agreement with Professor Jaeger to help conduct an Australia-wide survey of the ages of various rocks that they were mapping. You see, we had very little knowledge of the absolute ages of rocks in Australia. Unless there were fossils, no-one really knew how old they were.
In 1969 your ANU laboratory, along with a handful of others around the world, was selected to undertake rubidium–strontium dating of lunar rock samples collected by Apollo 11. How did it feel to be selected for this competition?
It was very satisfying, very exciting. I should emphasise that word ‘dating’. I was a principal investigator for a small group of people who were doing the dating and chemistry of the lunar rocks and minerals by certain techniques. Other principal investigators in Canberra – Ted Ringwood and Ross Taylor – worked on other aspects of the geochemistry of minerals, by other techniques.
I understand that the various laboratories were told, ‘You are going to be given a sample of lunar rocks. You will have three months to do your analysis and then you are to come to an examination, all together, and orally present your results in 10 or 15 minutes, against all the other labs.’ A better international scientific competition I’d be hard-pressed to identify.
It was a terrific competition. I don’t think there has been one so intense since. (Maybe when the Martian samples come back there will be a similar carry-on.) It was very exciting, and it allowed us to meet all sorts of famous names involved with aspects of lunar science. We were happy with our results, but it had been a struggle. Instead of three months, we had more like two. The samples had to be released from the lunar sample handling laboratory in Houston; the package had to come out by diplomatic courier, in someone’s attaché bag; and it then had to be handed to us and received by an approved person. Also, we had to have a safe all set up and ready, where only lunar samples would be allowed, nothing else. We were not used to living like that, but that’s what we had to do. And we upgraded our laboratories because you must have a clean environment, clean reagents and a reliable mass analyser.
How did you do in this international competition?
It was like an intense examination, and I was nervous beforehand. But fortunately we found out that we had the ‘right’ answers, answers that agreed for the age of the lunar basalts with results obtained by the best US laboratory plus a couple of others. So we were confident that we were not going to make fools of ourselves, in the first place.
I used the old-fashioned petrological term ‘mesostasis’ – which the Americans had forgotten – meaning the low melting point fraction: the last fraction of a magma to crystallise, because it has a low fusion point. That is where the uranium concentrated, and the alkalis. We realised that that was where the rubidium had to be, and we were looking for rubidium-rich parts of these lunar basalts. And we were fortunate enough to find that this mesostasis adhered to ilmenite, which is a heavy mineral and doesn’t have anything much else in it. We could make a concentrate of ilmenite and know that we would have a lot of the mesostasis sticking to it, so that’s what we did.
Well then, how old were the lunar samples?
We got 3.8 billion years, 3800 million – essentially the same age as the Caltech group got, at 3700-odd million. But there were complications: there is more than one age of lunar basalt, and also a lot of the samples were of the lunar soil, which is a hotch-potch of fragments of older rocks. You would get a mixed age out of those, which added to the confusion.
Because we stayed in this program for several years we learned that some lunar rocks are, indeed, 4.4 billion years old.
What does this tell us about the Moon?
The Moon, we learned from the history of the rocks, had an initial primitive crust with an age of just over 4.4 billion years, principally made of anorthite, a calcic feldspar. (Being so rich in this feldspar, the crust rock is called anorthosite.) It floated on the top of what looks to be a lunar magma ocean, which cooled fairly rapidly because there is no enclosed atmosphere on the Moon – it is looking out at cold space.
The basalts themselves are the product of later internal heating, because the core of the Moon, or its lower parts, will contain radioactive elements. When the solar system first formed, there was a lot of short-lived radioactivity. For example, aluminium has an isotope at mass 26 that is radioactive and it decays to magnesium-26, generating heat as it does so. When this sort of thing happens inside any planet, the heat has to build up, the temperature rises and you can have a second phase of internal melting. And so the mare basalts, in the lunar ‘oceans’, came out at later times as the temperature rose.
It was found that the mare basalts range from probably 4.2 billion years down to probably about 3.2 billion. It was always known, from the astronomical observations and the crater history, that the mare basalts were younger, because they have a much lower density of craters than the so-called lunar highlands – the original crust. But how much younger? This was the question people wanted answered. The broad observation that the lunar basalts were episodic was confirmed, but we also discovered that there was a whole hierarchy in their age. (Of course, I am using ‘we’ collectively. Lots of people have contributed to this story.)
If the older rocks were 4.4 billion years old, roughly the age of the Earth, does that mean our planet has always had a moon?
Well, you are straying off into theory now. The relationship of the Moon to the Earth was, and still is, a rather famous problem. Ted Ringwood and his lunar science group – especially David Green – worked on the geochemistry of the Moon as a whole and compared it with the geochemistry of particular elements of the Earth. They certainly felt that the Moon could be formed from material that was evaporated from the Earth and recondensed around it. But in the current theory a huge, younger impacter collided with the Earth and melted it, and the Moon spun off as part of that event.
That brings us to the SHRIMP. What is this machine, and how did it get such a name?
This is a mass analyser, and its name is formed from the first letters of Sensitive High Resolution Ion Micro Probe as a sort of a pun: whereas real shrimps are small, our SHRIMP is large. Steve Clement and I realised that we had to build the machine as large as possible, in order to achieve high resolution simultaneously with high sensitivity.
Could you explain what you mean by high resolution and high sensitivity?
To do that I have to go back a bit. At the time of the first lunar science meeting in Houston, we were fed up with the labour of keeping on top of the chemical technology to get the tiny amounts of lunar minerals analysed cleanly. At that time we became aware of a new type of analysis method that used a process the physicists call ‘sputtering’. You direct a focused beam of ions to the mineral you want to analyse, and this bores in (at a slow rate, actually) to the target and emits fragments of the target – ions as well as neutral molecules – which can be mass analysed. If you have a mass spectrometer and if you can extract these charged particles electrically and send them into a mass analyser, then you can measure the isotope ratios for an in situ analysis of even a very small amount of material.
That is exactly analogous to the way the electron microprobe works, which was discovered not very long before the middle 1960s, and it featured a great deal in lunar sample analysis because it let you take a polished thin section and put your electron beam on this, that and the other spot. And we all thought, ‘Wouldn’t it be marvellous if we could do the same thing for the isotope ratios?’
The electron beam analysis can’t measure isotope ratios; it doesn’t distinguish the isotopes of one chemical element from another. But a mass spectrometer is built for that purpose. So it was decided that that would be a great thing to use.
So the process of thinking that led to the SHRIMP began in Houston?
Yes, in a way. That was where we heard that someone was working on an instrument for microanalysis which used the sputtering process. He was at that first lunar science meeting but we didn’t actually meet him then. (Incidentally, he didn’t publish his ages on the lunar samples until the next year or the year after, but he did get approximately the right answers.)
The first analyses used a small mass spectrometer to look at the ions that were sputtered, but if you are looking at lead, mass 206, for example, you get interference from other molecules which have the same nominal mass. A fragment of hafnium and silicon, a fragment of zirconium and silicon and oxygen, will all land at mass 206. You can’t afford to ignore them, you have to correct for them. The people using the small ion probe corrected for them by ‘peak stripping’: they looked at a mass where these interferences were dominant, and figured out how much of the main interference had to be under the adjacent lead peak. That technology worked quite well if you had a lot of lead in the target you were trying to analyse. But it was found fairly soon not to work well for terrestrial rocks and minerals, which were younger and did not have enough radiogenic lead to stand out above these interferences.
So we realised – as did others, no doubt, around the world – that you had to have a high mass resolution instrument that would make use of a very convenient property of atomic nuclei: the nuclei of different elements have slightly different masses. That means that a molecule of hafnium and silicon won’t have exactly the same nuclidic mass as lead-206. If you have a mass spectrometer that operates at about resolution 5000, you can separate the two, and that’s the way to go instead of peak-stripping.
Then we had the mass spectrometer companies saying, ‘Oh yes, our machine can do resolution 5000. Just buy one of our machines and attach it to your source, and the sample handling mechanism, and you’ll get what you want.’ When Steve Clement and I looked into the problem, however, we realised that the small commercial machines would not have high enough sensitivity.
To obtain high resolution you have a narrow object slit and then you focus the beam onto a narrow detector slit. But every time you narrow a slit, you are in danger of truncating the beam. If you cut off a lot of the beam, you just don’t have very many ions per second to be detected. So you really need a machine with as wide an object slit as possible, and a collector slit that is wide enough for all those ions to go through into the DC amplifier or the ion counter that you use as a detector.
Steve Clement, being a physicist, applied a technology that the particle physics people had developed called beam transport theory. It was quite clear that to avoid truncation you had to be very careful as to the size of slits, the angle of divergence and the relative locations of different parts of the mass analyser. That led to our particular design idea. We realised that you had to have as big a magnet as possible for a mass spectrometer. The normal big mass spectrometers at the time were 30 centimetres turning radius, and we made it 100 centimetres – which is probably nothing by present-day physics or engineering standards, but in the early 1970s it was the biggest we felt we could engineer.
The requirements for the magnet are that the gap has to be uniform to one in 10,000 and the iron has to be uniform magnetically. None of these are simple things, of course.
So by resolution you are talking about the capacity to identify atomic mass accurately?
Yes. And to separate them.
And sensitivity refers to the fact that you are using only a small number of atoms to make this determination?
Yes. You need high sensitivity to get precision as well. The more counts of anything that you can get per second, the higher the precision will be.
You wanted to create a machine called SHRIMP, which theoretically was possible but would be very difficult to make and had never been made before. And you yourselves were not magnet manufacturers. Yet you were given the go-ahead to try and put this contraption together. That was a pretty big gamble, I’d say.
Well, we had built a previous mass spectrometer of our own. While Steve Clement was here as a PhD student he used beam transport theory to work out the best sort of small mass spectrometer that we could have for our lunar sample analyses. And we used it.
I think the SHRIMP go-ahead was a matter of confidence in us by our Director, Professor Anton Hales. We had to fight against advice that he got from people who said we would never be able to do it, and the mass spectrometer manufacturers were each busy saying that their machine would produce the high resolution and it ought to work. We had to say, ‘Well, for this and that reason we don’t accept that. We feel the need to go ahead and do it this way.’
How many other university departments around Australia would have the capacity to embark on a gamble like this?
No geology departments, even though they are very interested in rock geochemistry and age and are the chief users of chemical analyses for minerals and rock ages. They start by getting other people, a laboratory like ours, to do their ages for them. Then they graduate to buying their own mass analysers – small mass spectrometers – and doing it themselves. But they don’t have the workshop capacity to do a job like this.
We had the great advantage of being a part of the School of Physical Sciences: it is the tradition of physicists that they have a big workshop, their business being to build experimental apparatus (which they do very well). When we separated from physical sciences, part of the arrangement was that we would take a fraction of their workshop because Professor Jaeger and Mervyn Paterson were great users of workshops. They had high pressure and rock deformation apparatus that required a building and a lot of engineering maintenance.
The Australian National University was the only place in Australia, in the early 1970s, where this building enterprise could be done. There has been a huge advance since then, however.
Building the SHRIMP was a really major project which took over five years. What were the big challenges you had to overcome?
Because we had no experience in building magnets we hadn’t realised, for example, how much a magnetic field depends on the previous cycle of current through the windings of the magnet. We knew about the problem of hysteresis, that you have to always approach the field you want from the same direction, and we always did that. But there is another phenomenon that is annoying. The shape of the effective magnetic field changes slightly with the absolute value of the field itself. The better the magnet, the less is this effect, but at first we were just not conscious of it. And so we could not understand why, although our magnet was perfectly focused at, say, mass 200, when we got up to mass 254 (uranium oxide) it would be slightly out of focus. That was perhaps the lowest morale point of the whole exercise, because we couldn’t see what we were doing wrong.
However, we didn’t give up. We discovered that if we put a mechanical bellows between the collector slit and the rest of the machine, and drove the collector slit – with all the detector on it – to and fro to preset positions, it stayed in focus. The focal point still varied, but we could not cope with it.
Then we learned that the commercial manufacturers knew about this problem. One of them, at least, fitted an electrostatic lens between the magnet and the detector, and varied its strength according to the magnet field to keep the focus. But they didn’t tell anyone, because they thought it might be deemed a defect in their magnet design. Consequently, we had been busy rediscovering the wheel. And it slowed us down, of course.
We first tested the instrument not with a sputter source but with a thermal ionisation source, a solid source. The Research School of Chemistry very kindly lent us one of their disused electron bombardment sources that we turned into a solid source, and we used thermal ions to check out the focusing. In about 1979 we realised that it was actually achieving the hoped-for performance.
But there were a whole lot of mechanical things that we had to get going too: getting the sample in and out of the vacuum reproducibly, getting the visual optics to work – you have to be able to look at the sample while it’s in the target, and we had a high-magnification reflecting microscope that was fixed in the instrument, which we had to get to focus properly. All these little things are technically soluble but each one demands a certain amount of expertise. That meant practice and experimentation if we were to get on top of everything.
I suppose that building the SHRIMP as you did, rather than buying it as five black boxes from various manufacturers, meant you had the capacity to troubleshoot the problems.
That’s right. People are scared to take apart anything that is made and sold as a working thing – whereas we, having built every single bit, put it together ourselves, had very little hesitation in stopping it, letting the air in and pulling it all apart. We had to.
Did the SHRIMP live up to its promise?
Yes. We wanted to allow a safety factor of 2 and aim to operate at 10,000 resolution with full transmission. In fact, it operated at about 5500 resolution at full transmission, which is very good and quite adequate. We seemed to lose only a very small fraction of the extracted secondary ions in the whole transmission operation. And it had very much higher sensitivity than any of the small mass spectrometers.
Our competitors, in the meantime, had got going with commercial small mass analysers, but they turned out to have far too little sensitivity. They were also restricted to using old zircons and had much less precision than we could get. So our machine was more or less automatically recognised as the correct solution. Eventually even the commercial manufacturers accepted what we were saying and built large ion microprobes.
As I understand it, the SHRIMP was built to look for zircons. You have mentioned the use of zircons found in ore bodies or rocks as time markers. What is special about zircons for this purpose?
Several things are special. First of all, uranium atoms fit easily into the zircon site in the crystal lattice. They have the same ionic radius as the zirconium, so when the zircon is crystallising, any atoms of uranium that are in the melt will slide into the growing mineral.
In contrast, lead doesn’t fit well. It has a different ionic radius and a different charge balance. So the mineral zircon strongly excludes lead. That is a very good feature, because we have to measure the amount of common lead that is in the mineral we are analysing to obtain the radiogenic lead correctly. The ion probe measures the total mass of lead-207, and the total lead-206, but each of those two isotopes starts off with a little bit of common lead-207 and a little bit of common lead-206. The less you have of the common lead the better, and that is why zircon is such a good thing.
Also, zircon is tough physically and is chemically stable, so it doesn’t dissolve during low-grade metamorphism and it stands up to being weathered out of an igneous rock and trundled down the rivers into beach sands, where it is incorporated in younger rocks. There are people now analysing the ages of zircons in sedimentary rock to get an idea of the set of rocks that were being weathered, say, 3 billion years ago when a given sandstone was deposited.
What is the importance of being able to age zircons?
Well, this is the way you discover how old rocks are. From studies of zircons by the conventional method, some zircons looked as if they were of multiple ages within a single grain. It was certainly widely recognised that the zircons within rocks called gneisses, many of which were originally sedimentary rocks but had been re-melted, had to be a mixture of old zircons and young zircons formed at the time of reheating. The traditional methods were not suitable for these – they had to use a lot of zircon and so people had to try to identify the new zircons and hand-pick them out from the old ones. This is a very tedious and generally unsuccessful process, so those zircon methods were actually measuring mixtures of ages in minerals and result in age that are neither one thing nor the other.
What we urgently wanted were single, within-grain analyses. We discovered very early on that a single zircon would be a mixture of an older core and a later mantle of younger zircon, perhaps 1000 million years later. But we couldn’t tell this in advance. Later another imaging technique, cathodoluminescence, was developed by various people – with electron bombardment you get luminescence excited – and we discovered that different parts of the zircon luminesced differently. These outlined complex growth patterns within single zircons.
So some zircon grains are actually composites of an old core and a younger skin around the edges, and the SHRIMP can pick them out, analyse bits of individual grains and tell you how old those grains are, by the ratio of uranium to lead?
Yes, that’s all true. We hadn’t realised what a great success the SHRIMP would be for zircons when we built it. And although we did apply it mainly for zircon dating, we also applied it for the study of sulfur isotope ratios in ore bodies and a range of other geological problems.
Has the SHRIMP led you to any headline-producing discoveries?
Yes, again as a result of good fortune. Derek Froude, from New Zealand, was doing a PhD with me, and the problem I gave him was to look at all the old sedimentary rocks in the Archean of Australia, get the zircons out of them and find out whether there are any older than about 3.7 billion years. (At the time, those were the oldest known igneous or sedimentary rocks in the world.) So he collected rocks, collaborating with various other geologists and geological surveys who knew the area. And at Mount Narryer, in the Murchison district of West Australia, about 100 kilometres inland from Shark Bay, he hit upon a metasedimentary rock that had plenty of zircons, one of which was about 4.1 billion years old. This was astounding. Also, it commanded world attention, which does a huge amount of good for the lab – and for everyone’s ego!
I remember the day when the minerals were analysed. There were several students running the instrument, which we had elected to run more or less on a 24-hour basis because there was so much to be done and because we were still getting it under control. You kept it running unless something went wrong and you had to stop and fix it. You certainly didn’t stop and turn it off to have dinner at night; you got someone else to run it.
When Derek saw that the computer output said 4.1 billion he couldn’t believe it and he didn’t tell anyone at first in case he had done something wrong. So he did it again. This time he got the same answer, and then he found a couple of others and felt confident that this was right. And another student, running the instrument for him for a while, also hit on one of these. This was the big excitement.
This happened in the early ’80s, just after the machine really became operational, and it was published in 1983. It had a huge impact – the public as well as scientists all round the world seem to be interested in world records, and this was the oldest mineral fragment found. So it hit the headlines all round the world.
As the oldest piece of Earth?
Yes. We even got into the New York Times when Walter Sullivan, a famous science reporter at the time, wrote an article on it.
For a few years afterwards we searched the locality where this metasediment was found, to see if igneous rocks remained which had that age. But we couldn’t find anything. If anything is ever found now, it will only be by accident because it’s not going to be different for the geologists to look at. And it doesn’t have to be in Australia, of course. It could be a sliver of old gneiss from India or Antarctica.
But another sliver of the same geological sequence was found further to the north and to the east of Mount Narryer, at Jack Hills. This was the work of a visitor we had, Bob Pidgeon, who worked on the geology of the Jack Hills region with field geologist, Simon Wilde. Isaac Asimov, the science fiction writer and science reporter, wrote about that discovery – a measure, I guess, of its global interest. The age record is now held by zircons from that site.
People are still working on the very old zircons. The oldest zircon that we know about now is about 4.4 billion years; it has crept up. There seems to be a big clump of zircons around 4.2 billion, and then a few that stretch out to 4.4. We can’t find any chemical difference between those sorts of zircons but they are still of great interest to people doing terrestrial geochemistry, because the zircons themselves enclosed other minerals within them as they grew and these other minerals give you a taste of what was in the magma.
The SHRIMP, clearly, is important to an understanding of the earliest ages of our planet. We talk about ‘the SHRIMP’ as if it is a single machine but I believe that now, some decades on, there are several SHRIMPs. Is that right?
Oh yes. The original SHRIMP was doing fine, but even though we were using it full-time there was never enough time. We realised that it would be good to have a second instrument for our own purposes. Commercial mass spectrometer manufacturers were getting into the act – the French were offering their machine, the British were offering one – and it was put to us that we should probably build a commercial prototype.
John de Laeter, who was at Curtin University in West Australia, said he thought he could get the funds to buy a SHRIMP if we would build one. So we decided to build one commercially, and that was when we liaised with ANUTECH, who set up a company called Australian Scientific Instruments for the purpose.
We started to design this second machine about nine years after SHRIMP I started to work, so we were able to put in all sorts of engineering and electronic changes. You have to realise that the electronics industry has changed hugely. Although there were transistors in the mid-’70s when our group in electronics at the Research School designed SHRIMP I, and the thing is filled up with knobs that you twist manually, the amount of computer control is very small, really just to control the magnet through its own dedicated computer. The electronics group realised that for a commercial machine they had to totally redesign the electronics, not simply to be fashionable but because you couldn’t buy the old parts any more. They had to design for the use of components that could be replaced if they failed, as well as for possible future instruments.
Also, people these days want computer control for everything, so whereas we had been considering putting all sorts of things under computer control, now we actually did it. And computers keep changing, and are much faster. We made use of all of this change in designing SHRIMP II.
So SHRIMP II is the commercial prototype. Hasn’t there been another version since that?
Well, SHRIMP II is the one that we sell. We have another model called the SHRIMP RG, meaning reverse geometry, that is not currently offered for sale. It is being used full-time but it is more complex and we still have to master certain aspects of it.
What does ‘reverse geometry’ mean?
That is an essential ingredient for a design feature to give us a lot more mass resolution at the same sensitivity. It arises from fairly recent work by a Japanese theoretician, Matsuda. I should mention that the detailed design of the mass analysers is based on his designs. In 1974 he first published the theoretical designs of families of mass analysers, one of which he built to show that it worked. They actually met all of our requirements but we faced the problem of finding the best solution to using what is called an electrostatic analyser.
This is needed because the machines we have to build are all double-focusing. The ions that are sputtered from the target have a range of starting energies up to 100 electron volts or more, and so you need an energy analyser – essentially, an electrostatic condenser that bends the beam in a particular way and makes an energy spectrum at its output.
There are advantages in using what we’ll call a spherical mass analyser but you have got to have curvature on each of these plates; they are fractions of a sphere. That is very hard to make and it was very hard to machine in 1975, whereas a cylindrical mass analyser would be more tractable in terms of machining. (Now you can get computer-controlled milling machines that will produce a cylindrical analyser.)
Anyway, that was the first constraint we had to overcome, as well as various combinations of the turning radius of the electrostatic analyser, the turning radius of the magnet, the drift length of the beam here and there – all of these ion optical components that influence the beam’s shape and divergence. Because Matsuda had discovered how to predict the optimum combinations, we decided to adopt one of his solutions. We wrote to Matsuda and with his help we went ahead.
To answer your question about the new model: he then produced a later set of solutions for higher-focusing machines that didn’t lose sensitivity, and we decided to try that. They proved to be much trickier to manage.
It is good that those weren’t the ones that he came up with in the first place.
So the SHRIMP RG is the current state of the art but as a research instrument, to figure out how it could be done better before being offered commercially. Your commercial machine, however, now exists in laboratories around the world. How many SHRIMPS are there?
We have now sold two to West Australia and we have three ourselves, counting the new model. And there will be another six overseas.
I suppose that makes the instrument one of the success stories in the commercialising of ANU technology, and also a benchmark for this type of work.
That’s right. Crazy as it may seem, ‘to SHRIMP’ is now a verb in the official geological dictionary.
Though you are retired, you still have an active interest in the SHRIMP project. I believe you even operate it from time to time.
Yes, I have operated it a number of times since retiring, though not in the past couple of years. I am now occupied with writing up the results and assessing the nature of the data, its fine structure, in terms of instrumental effects that we all need to know about to get still higher accuracy and precision. I’ll give you an example.
The SHRIMP operates as old-fashioned flame photometers used to operate: you have a standard sample and that gives a signal, then you compare the height of that signal with the signal from your sample, and the ratio of the two essentially gives the amount of the element in the sample. That is how our SHRIMP works. Essentially, as the primary standard we use a standard zircon whose age has been determined by the old-fashioned isotope-dilution method which is the only accurate way of determining the absolute value of the age.
What we have noticed quite recently is that if you plot all the ages of the standard and the sample versus time, for some samples they both go up and down slowly with time. That is an unwanted machine effect which we hadn’t noticed before. (It depends on how you put the data out; you can do it in ways that obscure this.) We had previously included the effect in the variation of the standard, and used the standard deviation as a measure of the quality of the run.
But we can do better than that, because you can allow quite easily for a time variation, even quite a complicated one, by using mathematical methods for finding the fine structure in scattered data. That is what we are now writing up.
I believe you are also taking a step back, looking at our understanding of geochronology in general – times of rocks from the earliest onward – and trying to appreciate the implications of our different dating schemes.
Dating has always been about understanding what has gone on in the geological record. The palaeontologists recognise a number of global extinction events in the past 600 million years, and the latest thing we have worked on is the extinction of the Permian faunas, at the Permo-Triassic boundary. This is a phenomenon that had been seen for a long time, and the question is exactly when it happened and how long it took.
There are lots of interesting zircon problems associated with that, because in dating these sediments you look at the ages of interbedded volcanics but the volcanics themselves can have complications. They can contain zircons inherited in the source region of the volcanics, or the volcanics may be volcanic tuffs, or ashes, with older detrital material mixed in. You have to be able to analyse single grains to get the answers.
There are standard methods of doing this and it is now possible to work on single zircons by orthodox isotope-dilution chemistry. But people using such methods hadn’t realised how complicated these ashes were. Secondly, systematic errors between laboratories seem to be showing up – they haven’t got their isotope tracers calibrated as well as they thought.
So it is an interesting time of life for this fine structure in zircon dating.
I believe that you place a very high value on unselfish cooperation in scientific research. None of your overseas colleagues working on the lunar sample dating, however, divulged their results publicly during the international competition. This confirmed a long-held belief, I think, that some regard science more as cut-throat competition than cooperation.
Yes, exactly. And it’s getting more so now because of the requirement to get external funds, which means that people are jealous about what they say they are going to do and they extol their own reputation as much as possible. It tends to make people more isolated and less communicative.
Your scientific career has been studded with achievements and breakthroughs, and yet you describe your career as having been a ‘fortunate’ one. What do you mean by that?
I think we were lucky to find the very old terrestrial zircons, for example. By contrast, a colleague of mine in Britain – Jim Long, one of the pioneers in the focusing of ion beams and applying it to geology – was just plain unlucky with his choice of samples and got sidetracked on analysing pitchblende. Pitchblende is of great interest to the ore deposits people if they want to mine uranium, but it recrystallises very easily so it loses its age very readily. The difficulty is to unscramble these age losses within the target from the errors you might be making in developing your ion microprobe analysis. But he worked with his own primary column and imaging and targeting, and with a small commercial mass analyser, and he got stuck on the problem of low sensitivity. That problem is the reason we went in this other direction. And we have been vindicated.
You knew you were taking the hard road in the creation of SHRIMP, so I suppose it is good to be vindicated. But I have the idea that you feel fortunate in the timing of your career, not just in the samples that you were looking at.
That’s correct. First there was the connection with the end of the Manhattan Project, when all of those scientists went back to universities at the end of the Second World War and spread their knowledge of what you can do with isotope ratios, how to measure them better – this was all new, and I came along just after that had started to happen. So that was fortunate.
We were very fortunate with our discovery of total-rock Rb-Sr analyses, whereas the South Africans were unlucky and didn’t use the right samples. And none of the Americans had thought of doing total-rock work. (When we did it, they saw its value and a whole lot of labs took it up.)
I won’t say we were lucky to get the right age for the lunar samples. We did well on that, and it paid off.
I think my election to this Academy was a consequence of our rubidium–strontium work in West Australia and at ANU. The SHRIMP came after that and, well, I guess we were lucky to be in a place like the ANU, with such a big workshop capacity. Otherwise, there wouldn’t have been enough confidence that we could do it and it simply would not have started.
And then, yes, we were lucky to find the old zircons.
Your scientific career has always co-existed with your family life. While you were doing your PhD you married and had a child. That would have made for a very busy PhD.
Yes. I give three points of advice to my graduate students when they first arrive: never marry if you’re doing a PhD; if you do marry, never have children; and never leave the university without actually writing and submitting your thesis. I managed to violate all of those rules! When I went overseas on the Fulbright Scholarship, I had to travel at a certain time in order to get in that year’s batch. But I hadn’t submitted my thesis, so I had to finish it at the California Institute of Technology, semi-surreptitiously, in my spare time. And there wasn’t much spare time, I can tell you.
I tell my graduate students about this and they think it is funny. They can’t see why a person would have wanted to do any of those things anyway, but they discover after three or four years that they are very happy to finish their thesis.
I believe that throughout your career, your wife and family have played an important support role.
Yes, an absolutely fundamental role. My wife was herself a scientist, educated in physics and maths as well as geology, so she could appreciate what we were up against and what we were trying to do. She was very helpful and tolerated the absences from home that were necessary. Stress can destroy marriages, of course. Well, she didn’t let it destroy our marriage, and I’m forever grateful for that.
And my wife is intensely loyal to me. If you were to interview her she would tell you very fervently some 'political' things about my career that I wouldn't want you to hear!
Could you have guessed all this when you were students together?
Oh, I didn’t think that far ahead. I’m not one of those terribly organised people who lay out the career that they are going to have. I wish I were.
Had you done that, though, you might have missed some of your opportunities.
That’s right. I think you’ve got to have the freedom to pursue a possibility. If you get stuck too rigidly on a schedule, on a program – as people are in danger of doing now – then you can miss things. The environment we had at ANU allowed us to explore channels, by ourselves and through students. You didn’t involve a student if it was deemed altogether too risky, too stupid. But ‘stupid’ things need to be explored, because they’re not all stupid.
Bill, I think that ANU, Australia and the world of geochronology are very fortunate that you have played such an important role in this field. Thank you for giving us your time today to talk about it.
© 2023 Australian Academy of Science