Tony Cox

Norton:

Dr. Cox, the interview is going to be taped, transcribed, and then once it is transcribed you will have a chance to look it over and make corrections, additions or whatever. It will then be archived in The Niels Bor Library at the American Center for Physics, is that okay?

Cox:

Okay.

Norton:

First off, I would like to begin by just getting some background on you, what your educational background is, and how you started doing atmospheric chemistry.

Cox:

I did a Ph.D. in Manchester in 1966 in physical chemistry. The subject of my dissertation was on thermal decomposition of chlorofluoromethanes. Subsequently it became a major part in my life in research. From there I went to National Research Counseling in Canada and do the post-doc, working on further chemistry and kinetics of pre-radical reactions.

Norton:

So you started doing atmospheric chemistry?

Cox:

No, this was not to do with atmospheric chemistry then. I had met one or two people when I was in North America who had started to think about photochemical air pollution. When I came back to the UK in 1968 to work at the Atomic Energy Research Establishment at Harwell on a project on air pollution, I started to get interested in atmospheric chemistry and in particular photochemistry.

Norton:

What year was this?

Cox:

In 1968. I started with colleagues at Harwell working on the atmospheric chemistry of sulfur dioxide, which involved work on photochemistry. Sulfur dioxide is a photo-chemically active molecule. Also, on reactions on aerosols. Since I was doing this work in the aerosol group in the UK AA, there was a natural sort of tendency to be interested in reactions on aerosols, conventional air pollution rather than radioactive substances. So I worked on sulfur dioxide oxidation, and during the course of that work we got interested in ozone.

Norton:

When did you actually get interested in the ozone? Was this in the early ’70s?

Cox:

We got interested in the ozone in the surface atmosphere not the stratosphere.

Norton:

Okay, in the troposphere.

Cox:

In the troposphere. We actually developed a fairly extensive program on research on tropospheric chemistry. During the course of this there was a development taking place, particularly in the US, on the impact of the supersonic aircraft on the stratosphere. There was a proposal by Paul Cripson [?] and Harold Johnson that these might effect the NOX emissions from these aircraft might affect the ozone layer. But we were not formally involved in that ozone layer research at that time, although we were sort of observers through the atmospheric chemistry community and what was going on. The real involvement on ozone layer research for our group in Harwell was when the CFC issue was raised by Roland and Melina [?], and they proposed the ozone depletion due to CFCs. The agency in the UK who was responsible for that was the Department of the Environment and they were funding our work at ARE Harwell, and so consequently we became the advisors to the UK government on this issue.

Norton:

When their paper came out you started doing work on chemistry in the stratosphere as well?

Cox:

That is right, but we started to think about. It took a year or two to get going. By the end of the ’80s we had fairly a tooled up, as you say, program on stratospheric ozone, which involved some modeling work and laboratory chemistry and also measurements, some of the measurements were actually taken aboard the operational Concords. That is how I became to be involved in the ozone layer research.

Norton:

Going to back to the earlier work, when you were doing the work in the tropospheric chemistry, you said you were doing work on surface reactions on aerosols. Did that involve ozone reduction in the troposphere?

Cox:

Not really, no. We were interested with the oxidation mechanisms for sulfur dioxide. Sulfur dioxide was a pretty serious concern for the UK government at that time and is still a major air pollutant. The oxidation of sulfur dioxide leads to the formation of sulfate and sulfuric acid, and we were interested in the rates and mechanisms of this transformation that was considered to be a pretty significant bit of air pollution chemistry. What we actually discovered, among other things, were the photochemical reactions that were thought to be confined to Los Angeles smog, which in fact were more widely relevant, and in fact demonstrated a connection between photochemical ozone production in the UK to the formation of sulfuric acid aerosols. But the work on heterogeneous reactions sort of faded out, because what we discovered was the gas phase reactions were a rather dominant part of this story. So we got interested more in that side.

Norton:

For the stratospheric chemistry?

Cox:

No this was for the troposphere. For the stratosphere there was no real discussion of heterogeneous chemistry as being a serious contender until much later in the mid 1980s. Initially, when the stratosphere became a focus, there was interest in a small community in the stratospheric aerosol, sulfate aerosol in the stratosphere, which had been discovered around 1960. But it was not thought to be a significant player in ozone chemistry, mainly because it was right in the lower part of the stratosphere. The conventional thinking, and still is, by and large, the ozone is not really destroyed to any extent there. Most of the chemistry occurs at the higher altitudes.

Norton:

I had read somewhere where a number of people prior to the early ’80s had proposed heterogeneous reactions for the stratosphere, but they never went anywhere.

Cox:

That is right.

Norton:

This is people just doing their yeomen-like work?

Cox:

Yes. I would say stones [?] trying to find a role for their pet theories and this sort of thing.

Norton:

Just none of them…

Cox:

None of it really made sense. It all looked to be rather unimportant sideline issues, really. Because all the calculations that you could do showed that the gas phase reactions were dominant.

Norton:

What do you see as the key thing that switched the thinking on this? Roland and Melina and the group at Harvard started working on this right before the ozone-hole was discovered, so things started moving.

Cox:

Yes. I think I can put my finger on the way thinking was. The thing is, as the gas phase chemistry became well known there was an awful lot of work, discussion, and communication. There was the evaluation of the gas phase chemistry; kinetic evaluations were starting. The gas phase chemistry was getting fairly well established by the early 1980s. There is always people thinking about what have we left out. Some of the front line people, particularly the people who were doing models of the atmosphere, and at this stage modeling was much more speculative. It was not really properly tested against observations that were by and large there weren’t the observations to test it. They had only just started to collect a lot of data on stratospheric composition in which you could test the models. So people were worried about what have we left out. I think the first questions that arose had to do with the NOX chemistry in the lower stratosphere and the coupling of the NOX and the hydrogen chemistry in the lower stratosphere and I think that Dick Zaye was one of the first persons to do calculations on this. This was this conference that we referred to this…

Norton:

The Feldafine [?]

Cox:

No the one before the Feldafine at Gurenham [?].

Norton:

It was in 1984.

Cox:

Yes, in 1984. There was a chemical manufacturers association meeting in Gurenham, and Dick Zaye [?] had presented some calculations where he pointed out that when N₂O₅ reacts with water it produces nitric acid, and then the nitric acid photolizes in the atmosphere. This actually provides a way of breaking water molecules and converting them into free radical species. He did some sums to show that if you take the estimated surface area of the aerosol in the lower stratosphere and you allocate an uptake efficiency or reactive efficiency for the heterogeneous reaction of the N₂O₅ plus water reaction that you could actually get significant amounts of NOX.

Norton:

The uptake efficiency describes the rate at which these —

Cox:

How do you represent the heterogeneous reactions, in a volume of air you have a lot of particles and you have a specific surface area. You express the reaction rate as the fraction of the collisions of the reactive molecule with that surface that actually leads to reaction. Some of the collisions do not and it bounces off again. Generally speaking, if the collision efficiencies are less than about one in ten to the four, then generally the reaction is going to be slow in the stratosphere. If the collision efficiencies get up to ten to minus two then we are starting to see significant amounts of reaction going down those pathways. That is just a broad — And at that time the thinking was that the uptake coefficients of these things was less than ten to the minus three. There was data that people had reported on these uptake coefficients.

Norton:

What explains the differences in the uptake coefficients? This is just going to be the detailed chemistry situations?

Cox:

Yes. It depends on the nature of the surface and the detailed chemical reactions that are going to take place.

Norton:

Are there any general…?

Cox:

Only now, I think, can we see in the last few years that we have seen some guidelines about how you can make judgments on reactivity. It is has only really just started to get the understanding. You can imagine that it is quite a big problem, because the first thing is people do not know what the surfaces are — what the actual material is. Quite a lot of effort has gone into understanding the microphysics of the aerosols and the ACs [?], how they are formed and what they are composed of. You have to do all that before you can really understand the chemistry.

Norton:

This conference is…

Cox:

Yes. Back in 1984 this was all really quite unknown territory.

Norton:

When was this conference again? Was this in the summer?

Cox:

I cannot remember. The CMA meetings were usually spring or autumn, but I cannot remember when. Some ideas of the potential importance of heterogeneous chemistry were actually surfacing there. That had actually raised ideas in the people who were directing this chemical manufacturer’s program. I think it was around that time that we had a research contract with the CMA. We were one of the groups that they funded research in. They asked us if we would do some experimental studies of the chlorine nitrate plus HCL reaction in our laboratories and see if we could find out some information to find out whether it was a heterogeneous reaction and try and characterize this reaction. We in fact did on a little project. At the same time there were a number of other people looking at this problem. Mario Melina had started a little program on this at JPL. The SRI group had started to look at some reactions. So there was some activity starting, but even by 1985 when we did the big ozone assessment, the so-called blue book, the really first big comprehensive ozone assessment.

Norton:

That is the summer of 1985?

Cox:

Yes, the summer of 1985. We did an assessment of the laboratory chemistry relevant to the stratosphere, and we looked rather carefully at potential roles for heterogeneous chemistry in the stratospheric ozone problem. We made some cautionary statements about the potential influence of heterogeneous chemistry on the ozone layer.

Norton:

Okay. Was this the ozone layer in general?

Cox:

This was the ozone layer in general, because we had not really started to think about dividing it up into these different packets. Thinking about the atmosphere as a three dimensional fluid wasn’t too advanced at that time, and certainly in relation to this problem. We made some cautionary statements that perhaps under certain conditions in atmosphere. For example, in the polar region conditions might be such that heterogeneous reactions could become important and thereby might affect the ozone chemistry.

Norton:

Do you have any idea what those conditions might be?

Cox:

I would have to actually go back and have a look at exactly what we said. I remember the message was that the gas phase chemistry we see is the most important overall in controlling ozone in the stratosphere where heterogeneous chemistry may become important under certain special conditions, such as after major volcanic events in polar regions. Which in fact turned out to be true, and they were important. Our message was not that you have forgotten this very important bit, but it was more that the main chemistry is reasonably well known and there may be some qualifications that one needs to add for special conditions.

Norton:

So the gas phase chemistry is pretty well understand, that if that is all that is taking place that you would expect to see this sort of behavior?

Cox:

Yes. And there was a lot of observational data in the atmosphere which had started being collected at that time, particularly on those big balloon campaigns they ran in the early 1980s, which started to give some details about chemical composition in the ozone layer. By and large it was quite a reasonably good fit between the models and the observations when one looked at what could be done at the time. That was the position, really, just before the discovery of the ozone-hole. In fact, the ozone-hole paper was actually published I think about two or three months before this meeting which was finalized the 1985 ozone report. You know, I think give the meeting due. I recognized this was something that had to be looked at over a period of time.

Norton:

Was there any talk at the meetings about the fact that at this point — this would have been in the middle of the summer.

Cox:

Yes. I would say there was actually quite a deal of skepticism about it among a good precautional of the community.

Norton:

Was it talked that you have these measurements being taken by a mersionionic [?] survey but the satellites are not showing anything? What’s going on?

Cox:

I do not think that was so much the issue, but the realization of mistakes — I think people did believe the measurements, but they did not believe the — I do not think I really realized how serious it was in terms of challenge to understanding. I didn’t really believe that it was due to CFCs.

Norton:

Most of the people at the meeting did not think it was CFCs?

Cox:

No but they did not have another alternative explanation.

Norton:

But they just did not think it was CFCs?

Cox:

I think that they could not really understand how it could be CFCs.

Norton:

Okay.

Cox:

I think that is the thing. We have this situation where heterogeneous chemistry had been suggested by Dick Zaye, and then I think the 1985 or was it early 1986 was the paper by Jerry Roland and Susan Solomon that suggested the chlorine nitrate plus HCL reaction.

Norton:

Yes. That was Solomon, Webels [?], and…

Cox:

That was early 1986. By this time it was appreciated that PFCs existed because it had shown up on the satellite. That was Pat McCormick’s discovery, the fact that you get these clouds in the stratosphere in the polar-regions that had not been appreciated before. They put that together with the idea that this chemistry might be fast on surfaces which had arisen from crude experiments, the type of experiments that Melina did and we did at Harwell and Harry had done and had published in Jafferies Ken [?]. It was rather a bad paper.

Norton:

I know which one you are talking about.

Cox:

It was just a short paper in Jafferies Ken [?], but they weren’t actually very well contrived experiments. They did not depict what was going on except to show that there was a very fast reaction in their glass vessels.

Norton:

What was not very good about it?

Cox:

It was a very crude, by the sort of standards of experiments that were really being carried out by chemical kineticists to address the problems of understanding atmospheric reactions. Which by that time it had become a fairly sophisticated methodology.

Norton:

Was there anything inherently wrong with it?

Cox:

There was nothing wrong with it, it was just very crude. It was basically what we used to call stew-pot chemistry. You put a few things in a vessel and you shine a light on it or something and you just observe the chemical changes. Everything is happening together. You then have to devise some model of all the sequential processes that are going on, and then you interpret it. There is a role for this stew-pot chemistry and that is something that we did actually visit when we had discussions in 1985 on whether we had lost any or was there any missing chemistry. But at that time this type of experiment was not generally considered the best way of approaching things.

Norton:

So there were not sufficient controls on it?

Cox:

It was difficult to diagnose transferable information. Having said that, in the end, quite a lot of things are discovered by serendipity of that kind.

Norton:

These ideas were floating around in a number of places?

Cox:

Yes, through different people. Some people rejected the whole concept of doing experiments like that and some people appreciated that you could find things out. By and large, Harry’s [?] paper was not one that really turned one on as being world shattering.

Norton:

Okay. Around 1984 before the hole was discovered, these ideas about heterogeneous chemistry were starting to come to the floor?

Cox:

Yes. It was realized by about that time that you had a lot of chlorine, the catalyst, tied up in these reservoirs. If there was a process by whereby the reservoir and molecules could be broken down and release catalytic species. Then the potential for ozone loss is much higher. It was appreciated that there was one, but nobody had really found any processes that would do this. Most of the interaction of this type was between closed shell molecules, which generally the reactions of how activation energies which temperature in the stratosphere make them very, very slow.

Norton:

It had to be something to speed them up.

Cox:

You needed something to speed them up.

Norton:

I know Roland had one of his students perform these stew-pot experiments. Then they took these reactions, the chlorine nitrate and another one, and then Webels[?] and someone else put those into one of the models they used?

Cox:

That is right, they put them into a model. They had some numbers and showed that if these reactions could go at a certain rate they got ozone depletion.

Norton:

But it was just all sort of…?

Cox:

It was speculative, but it was the right kind of speculation because it is sort of a what if. Have we left anything out? If we left this out, what are the consequences for the ozone layer?

Norton:

You knew about this stuff that Roland had been doing with this before the ozone-hole was announced?

Cox:

I did, yes.

Norton:

Did you automatically think that something like this could explain what was going on? As you say, a lot of people could not see how it could be…

Cox:

I was not convinced at that time that you could transfer that data, those laboratory observations, in a quantitative way to the stratosphere.

Norton:

What sorts of reasons were in your mind for thinking that you could not transfer them?

Cox:

I think it is because if you do a stew-pot experiment, you don’t have any confidence that you can produce a transferable number. You can’t actually get an uptake coefficient in a reliable way from a stew pot experiment. Then you are carrying it out in a laboratory vessel made of quartz or Pyrex, and what is that to do with ice particles or clouds? That is basically the basis of my skepticism.

Norton:

You said that you performed these similar experiments?

Cox:

We could measure the kinetics of chlorine nitrate lost in the presence of excess HCL in flow tubes, and we determined some kinetics and we could essentially deduce from this a reaction constant to the HCL plus chlorine nitrate reaction. What we found is that the behavior of the kinetics was incompatible with it being a gas phase reaction, so we knew that it was happening on the surface and we could make some sort of estimate of the reaction efficiency on the surface. It seemed to be pretty efficient. Again, we could not really interpret our experiments to deduce what was going on in the atmosphere.

Norton:

Did you think of trying to modify the experiment that you could get those measurements? In other words, was there just not a pressing need to do that?

Cox:

There was, but the particular experiments that we did were not predictable and not really the ideal way of studying it.

Norton:

When exactly was this experiment performed?

Cox:

It was done 1984 or 1985. Actually I wish I could find that report because that would have the date on it.

Norton:

You had done this stuff before you read about Roland’s stuff or was it around …

Cox:

No this was definitely after Feldafine. It was basically straight after Feldafine that the CMA representatives came along to us.

Norton:

What did you do with these results?

Cox:

We put them in a report and gave them to the Chemical Manufacturers Association and we reported them at their [???] meetings. At the same time other people reported experiments that gave similar conclusions that these reactions could occur rapidly on surfaces.

Norton:

Can you explain what the sort of constraints there were? I am interested in the ways in which you can constrain an interpretation of an experimental result or model results or things like that. You say you could determine that it was not a gas phase reaction. You had enough constraints in the situation of control on the experiment that you could rule that out as a possibility.

Cox:

Yes.

Norton:

What sorts of constraints did you have that allowed you to do that? Was this just general background knowledge?

Cox:

Background knowledge about the kinetic behavior of elementary gas phase reactions under certain simplicity of the kinetics that you can use to prove whether they are going on or not.

Norton:

So it was just general theoretical background indicates in experimental and…

Cox:

You measure the reaction rates and you deduce a rate constant and you deduce that based on tests of the variation of the rate reaction with the concentration of the reactants. It is pretty basic chemical kinetics.

Norton:

I guess one thing is a little unclear. Hadn’t similar experiments to this been performed before, atmospheric chemistry experiments? It seems to me that previous to all of this you would expect heterogeneous reactions to be taking place in the reaction vessels that you are doing in experiments previous to this?

Cox:

That is right. Generally speaking you try to avoid having heterogeneous reactions and lots of work went into avoiding heterogeneous reactions if you were studying gas phase reactions. I think by that time by and large the tricks of the trade had been learned, and if you set up to do gas phase studies you made sure that you basically did certain protocols to your apparatus in the way you conducted the experiments to avoid the complications.

Norton:

In your experiment you did not need to do that because you were interested in that in the first place.

Cox:

That is right.

Norton:

What other experiments did you do after this concerning the heterogeneous chemistry? Either right before [???] [???] paper came out or immediately thereafter?

Cox:

Well personally I did not do anymore work on heterogeneous chemistry in my program at Harwell. The work that we did for the Chemical Manufacturers Association at that time was really more relevant to the homogeneous gas phase chemistry. We concentrated on the mechanisms on the gas phase reactions that were involved in the ozone-hole chemistry. The papers that I published in the second part of the 1980s were all related to the work on the CLO [???] and the gas phase chemistry relative to the ozone-hole conditions. We were not equipped to do the heterogeneous chemistry there. Those people that did start to go down that road sought funding and built new apparatus, and people started actually from a fairly low level. It took a while to build up capability. I did not really start doing any heterogeneous chemical kinetics experiments until I came here in 1995.

Norton:

You said you did these other experiments on the homogeneous chemistry, the gas phase chemistry related to the ozone-hole.

Cox:

Yes.

Norton:

You said a lot of the gas phase chemistry was fairly well understood before the [???] paper came out. What sorts of additional work were you doing on the gas phase chemistry that needed to be clarified?

Cox:

Following [???] paper.

Norton:

Yes.

Cox:

It was what the mechanism was for the ozone depletion. It occurred so quickly. The atmosphere in the polar-regions after the so-called processing by the clouds was a completely different atmosphere in terms of the chlorine ozone nitrogen chemistry. The mechanisms and kinetics and even the elementary steps that were involved were not known at that time. The reactions involving CLO forming the [???] had been proposed actually by myself and other people toward the end of the ’70s, but they were actually not considered to be important at all in the atmosphere because none of the calculations…

Norton:

Didn’t Melina suggest the CLO [???] was important for the reactions of the ozone-hole depletion?

Cox:

Actually it was quite an interesting paper. Mario Melina published a paper I think in 1984 where he did some experiments in the lab where he measured some reaction products from the CLO radical reactions. Which he described incorrectly, as it happens, to the CLO [???]. He then went on to suggest that this reaction may be important in the stratosphere if you had sufficiently high CLO concentrations, which of course you did, once you got in the ozone hold conditions, which were not known at that time. The spectral interpretation was actually wrong in that paper. The product that we saw was not the CLO [???] at all. That is one of the things that we actually had some indications from earlier work that the experiments they were doing were going to give them incorrect answer. We started then to look further into this gas phase chemistry, and I published a paper with my colleague, Gary Hyman [?] I think in 1988. We published the first, what turned out to be the correct spectrum for the CLO [???].

Norton:

You sort of somewhat redid those experiments figuring out what was going on actually?

Cox:

Yes. That was my main contribution to the ozone-hole chemistry in the 1980s.

Norton:

Were the experiments that you did on that prompted by Melina’s work?

Cox:

No they weren’t. They were prompted by a long interest I had in this particular problem.

Norton:

Okay. When you read his paper originally, you said you had some indication from previous work that that experiment was not going to give him the right answer.

Cox:

Yes.

Norton:

So then how did you get around to doing these experiments eventually?

Cox:

I thought we had a better way of doing it. You could trace the work I did on the CLO [???] in the literature, but Mario Melina was a good friend of mine so I did not jump up and say, “You are wrong, you are wrong!” I started to think that does not look quite right. I was not completely convinced that he was wrong when I first read the paper, but the more I thought about it the more I thought he was. Actually at the time we were doing some work on chlorine nitrate where we were trying to unravel another little problem that Mario had turned up in some work he was doing on chlorine nitrate that subsequently we found. It was all to do with this same incorrect interpretation of experiments involving CLO and OCLO.

Norton:

It had to do with the same sort of misinterpretation as in the other paper?

Cox:

Yes. It was not a misinterpretation. It is very difficult to explain. Maybe misinterpretation possibly in the broadest sense, but there was things going on that we could not quite understand but were affecting the kinetics of CLO plus CLO reaction and CLO plus NO₂ reaction, particularly when you were doing experiments with chlorine dioxide, OCLO, in the reaction systems. What it turned out is that OCLO and CLO react together, they form not a diad but another chlorine oxide that is fairly unstable but interferes with the kinetic behavior of the system. The importance of this was not really appreciated, and it led to a number of difficulties that took a few years to actually find out what was going on.

Norton:

Can you tell me what one of the difficulties were?

Cox:

Basically you form this unstable chlorine oxide, CL₂O₃, which is unstable so it can decompose again and by doing so it releases the CLO back into the reacting system. So that it affects the kinetics.

Norton:

This led to Melina misinterpreting the results because…?

Cox:

He recognized that there was a difficulty, and he could not explain it in one case. But when he moved on to look at the CLO [???]. He did not recognize that it was the same problem. We talked about it and said what in the hell is going on? I talked to Mario several times and in the end it is all sorted out now. To continue the story really about the ozone-hole and the interpretation, you are asking what made me change my mind. I am thinking back and the thing that really changed my mind was in 1986 when Susan Sullivan did these measurements in Antarctica before the air born campaign where she…

Norton:

[Inaudible].

Cox:

That is right. Where she observed this chlorine dioxide molecule, OCLO, OCLO₂ whatever you want to call it, present in the stratosphere substantially concentration and as soon as I read that I knew that this heterogeneous reaction theory had to be. There had to be something that was converting the chlorine out of its reservoir, and almost certainly you had to appeal to these heterogeneous reactions for that.

Norton:

What about the other theories of what was proposed to try and explain the whole?

Cox:

I listened to them when I went to numbers of meetings where these were sort of presented, but I was never convinced there was anything other than a chemical explanation of this phenomenon.

Norton:

There is one thing that I am interested in. I talked to [???] Foreman [?] and Brian Gardner [?] and all of them and they were pretty much convinced right from the get go that it had to be a chemical explanation. They had the data going back to 1956 and there was not any eleven-year cycle in their data like the sum of proton theories, and there wasn’t any change in temperatures of the hundredth millibar level which would be required by dynamical explanation. So for them the data indicated that it could not be dynamical or solar explanation so it had to be chemical. Is that sort of reasoning convincing to you, or were there other reasons for thinking that it had to be…?

Cox:

Yes. I listened to talks on the ozone layer for ten years, and as far as I could see, there did not seem to be any way that you could suddenly remove the ozone layer. I am not a meteorologist or an atmospheric physicist, but it always seemed to me that those people were advocating a dynamical theory were always having to push the boat out as far as their science was concerned.

Norton:

Contortionist.

Cox:

Yes, just to explain it. It just seemed to be much simpler to believe, but observations such as those Susan made, as soon as the atmosphere was perturbed there has to be a chemical explanation.

Norton:

What about other measurements of atmospheric constituents that had been done prior to that? I know there was the one that New Zealanders had done on the measurements over the Antarctic and they had shown that there were extremely low levels of NO₂.

Cox:

Yes it was the so-called Noxin Cliff [?] theory. I think that there was some credibility in those, and it was probably those reports and those papers that actually led to our cautionary note in the 1985 evaluation that maybe in some parts of the atmosphere the heterogeneous chemistry could be important.

Norton:

You actually were cognizant of that when you were writing that paper.

Cox:

I was aware of the Noxin Cliff. It was something that clearly a detail that was not explained by the chemistry. Do not forget that the amount of observational data that you had to deal with was actually rather small at that time. It is like there was no whole picture, and yet these guys were actually talking about large scale phenomenon occurring.

Norton:

Observations being taken here.

Cox:

That is right but John Noxin had a global view of how the atmosphere worked and knew that when he took measurements in these parts of the atmosphere that he could connect those to the polar circulation. He thought, “I am getting these big changes here, there must be something going on, on a big scale.” Many people like me who came to this from the air pollution side did not have this big view of the global atmosphere and how it worked. It took a long time for people to understand that.

Norton:

You have been doing the heterogeneous experiments here at Cambridge since 1985?

Cox:

Yes, that is right.

Norton:

What is the state of the chemical understanding of what is going on?

Cox:

We are getting a little bit more of the picture, and I think the aim that I have and others in this game is to try and build a framework so that these processes can be represented accurately in the atmospheric models. The idea to do that is to understand the details of the processes to derive physical chemical parameters that are needed that can be measured in the laboratory and which can be transferred to the model, so that you can have essentially a fundamentally based representation of these processes. Hopefully that can be reduced to a sufficiently accurate parameterization that can be carried in a large model. It is not altogether clear whether that even be done, but we think it is worth striving for.

Norton:

You work with John Pyle closely then, providing them with laboratory data that they can use on their models?

Cox:

Yes we did.

Norton:

You are at the level where you are carrying out experiments getting the hard data that provides constraints for the parameterizations they are using in their models?

Cox:

Yes, it provides the basis of the parameterization they are using hopefully.

Norton:

What is your feeling on how reliable the models have become over the years? They are actually using 2-D and 3-D models in that aren’t they?

Cox:

They are definitely much better in terms of describing the real world then they used to be. That has really come about by having observational data sets that they can test. In the sort of general practice, modeling is not a good way of [inaudible]. The modeling work essentially provides you with a numerical picture of the atmosphere, and the eventual aim is to predict how the system will respond to certain changes. But first you must actually show that it is really represented what is there today and possible what was there yesterday. You test the model by simulating observations. Certainly the models are very, very much more accurate in being able to do that. One must conclude that as far as predicted capability, it is much better. It is very difficult to say from the result of this sort of model calculation whether you really got the whole story. I think that is where there will be continual refinement and there will be continual updating of the models, and they will of course be asked harder and harder questions as the observational data gets…

Norton:

That seems to be maybe one of the fundamental problems with a lot of the modeling and is why a lot of people see them as being unreliable, is because they keep asking things of the models which they can’t quite do yet.

Cox:

Right.

Norton:

It always seems like they are just not quite up to answering the questions. You were describing the models as more predictive than it seems that a lot of people like to think of models.

Cox:

Yes. That is not the only thing that drives people to do modeling. It is satisfying to describe the real world. In my own sort of experience, it has been the actual requirements of being able to say something about the future. In other words, the users input into these programs as being a major sort of driver for the models. Right from the start it was the questions from the government regulative bodies and whatever well is this idea true, and then what will happen.

Norton:

I know the first computer models you have done at Princeton were numerical simulations of the atmosphere and all of that. But it seems like a lot of the modeling that is done can actually be sort of described as being used as probes. What you are doing is you construct some model of the atmosphere and you compare it to your data, then you see where there is a disagreement, and those disagreements allow you to probe the data to see where your understanding is coming up short and where you need to put things in. What do you think about that sort of…?

Cox:

It is a little bit outside my personal experience because I am not modeler. Generally my use of models in practice has been having an idea generated from doing laboratory experiments and coming up with some new data that you think is significant in describing some atmospheric process, and then putting that data in a model and say compare the model without that data with the model with that data to see whether it has a significant effect on some diagnostic that you are looking at in the atmosphere. I think that many people working in the lab in atmospheric chemistry always ask themselves that question. You know, what does this mean for the atmosphere? They should be thinking about it if they are atmospheric chemists! And many people do. That is really how chemists have actually contributed a lot to this whole business.

Norton:

So you try to design your experiments so that the quantitative results can be used.

Cox:

Yes. The prime consideration is basically getting experimental data that is relevant for the right conditions in the atmosphere — the right pressures and temperature and so on.

Norton:

You do a lot of chemical experiments in the lab. To what extent do you use models for your experimental work?

Cox:

We have a model and we use them all the time in all our experiments.

Norton:

You have to bring some sort of model before you …

Cox:

You have to have a model of your experiment.

Norton:

Can you say something about what you think the differences or similarities are between the sort of modeling you do in your lab work as opposed to modeling that goes on in a computer?

Cox:

The main difference is that in the laboratory you hope to reduce the number of variables. Physical variables actually are much better defined, both in terms of time dependence. You’ve got more control over the physical variables. On the other hand, to come back to something that we were talking about some time ago, is in the laboratory experiments you cannot avoid having surfaces around. So you have to take care that you can overcome the problems of surfaces, whereas in the atmosphere you may be able to completely disregard the surface for many problems that you are dealing with. The other main difference in the atmosphere is that you are looking at a fluid dynamical system. You have air motion that you have to take into account, whereas in a laboratory experiment you actually set up your own air motions, the gas environment, to suit your experiment.

Norton:

What you are saying is that with the actual lab experiments you get better control over the situation than you do with the computer simulations?

Cox:

Yes, you can set your boundary conditions easier.

Norton:

There is less uncertainty?

Cox:

There is less uncertainty in the boundary conditions.

Norton:

Can’t you get simple computer models where you get just a much control as you do in the lab? Say you perform an experiment in the lab and you understand it is sufficient and you get a lot of data and you can transfer that pretty much into a computer so you can perform the experiment in the computer. I guess one thing is if you had a computer model for like a pendulum or something, in the computer you can actually completely eliminate friction whereas in the lab you can’t. In that sense you have more control over the computer than you do in the lab. What do you think about that? Ultimately it is based on…

Cox:

The possible interference of surface reactions. You could have a model of your laboratory experiment, and you can choose to put in a surface reaction or take it out. You could then test how important that surface reaction might be. But generally speaking, you are going to have to be making some guess work about that surface reaction. The best way of dealing with that in a laboratory situation is to actually devise a surface that is unreactive, so that your interpretation is not complicated by uncertainties associated with postulated interference. The ultimate objective of the laboratory work has been to really obtain transferable parameters that you can use for describing processes in the atmosphere. If you want a transferable parameter, you want to really understand what that parameter is actually representing and to make sure its value is consistent with a body of knowledge of theory of gas phase or heterogeneous reaction kinetics, so that you have confidence in the number that you can then transfer to the atmosphere. Then it becomes not an adjustable parameter in an atmospheric model.

Norton:

So you want the outputs of your experiments to be compatible with not only the larger theory body of knowledge and working with it, but also so that it will fit into the computer models, which of themselves have to be consistent with the larger body of knowledge.

Cox:

Yes. This process of essentially validating and gaining the confidence in your laboratory experiments may involve the computer model of your experiments. You may need a numerical model to actually retrieve the data, or you might just use one to do statistical analysis to get the best value and understand the uncertainties.

Norton:

The way you are describing it is that your laboratory work is not completely divorced from models.

Cox:

No. There is a commonality of the interface. An example might be that some type of experiments that you do in the lab you can consider your reaction cell, or whatever you are looking at, as a box, and you are looking at the time dependent evolution of chemistry in that box. You can derive some fundamental data from that. To do that you would have what might be called a zero dimensional box model. You can use a zero dimensional box model to try and sort out some atmospheric problem, provided that you have satisfied yourself with the transport terms, that is, the things that are affecting the chemistry in your box, are not being disturbed by dynamical influences. In fact, in some situations some very useful work has been done in the atmosphere by using box models.

Norton:

The simplest form would be the aranouses [?] model for the atmosphere for doubling CO₂.

Cox:

Yes something like that or some of the earlier photochemical smog models that are just basically considered a box in the boundary layer and looked at the chemistry.

Norton:

Susan Sullivan’s first model for trying to explain the ozone-hole was a box model, wasn’t it?

Cox:

Yes, exactly. Now the box model comes into rather good use because the understanding more about transport in the atmosphere they can now affect the boxes around. You can have a box model that actually can move along and you have to make certain assumptions about mixing in and mixing out. The basic concept of the box model in a Lagrangian calculation is quite respectable and reputable and useful.

Norton:

It is going to be an example of a more general principle; you have to reduce or make the situation as simple as possible in order to get any understanding out of it. Because if you are creating a model that represents the atmosphere in every single aspect, what is the point, because you have the atmosphere.

Cox:

Yes.

Norton:

Can you comment on the extent to which those sort of simplifications, certainly not with respect to the models but stuff in the laboratory that you do that allows you to make interpretations of your data? I am thinking of examples where, as you were saying, that you have this box model where you have something going on in there, and you are not really concerned so much exactly what is going on in that box, but you are getting other parameters out of it or other data or something. You talked about the example of the stew-pot experiments where you threw the stuff in and you get some data out of it. The data does depend ultimately on the causal relationships between the chemicals and what is going on. But at some level you don’t need to know exactly what is going on in there in order to get reliable information at another level.

Cox:

With the stew-pot chemistry, yes, it works in two ways. You basically can use stew-pot chemistry to understand or to measure reaction rate parameters, provided you have an accurate model description of the parameters that you are actually trying to deduce. You need to know the mechanism of the chemistry which is relevant to the parameters that you are deducing from this type of experiment. The other stew-pot chemistry approach is more like the sort of thing of coming back to Roland’s first experiments, is you put some chemicals in a jar, and you think you know what is going to happen. Then you make measurements and you find something different, and then it is up to you to actually deduce why you are getting an unexpected answer. That also can be a useful type experiment, but is often more difficult to interpret and more difficult to convince others that your interpretation is unique. This type of stew-pot experiment is used in a number of ways, but it can in fact give very accurate data of relevance, provided the mechanism of the relevant chemistry in the system is well known.

Norton:

What you are saying is that you have to have an understanding of the important mechanisms.

Cox:

Yes. You have to have a model of the important chemistry in there if you are going to get accurate data actually.

Norton:

Because I was asking a sort of question — There are certain physical mechanisms that are operant in the stew pot that you do not have to have any knowledge of at all in order to get useful information.

Cox:

Provided they don’t influence the data, is the point.

Norton:

Okay, I guess that is pretty much it.

Cox:

I hope it has been useful.