Ahmed Zewail: ...and actually this is fabulous on the femtosecond time scale. The shortest pulse nowadays that you can generate in the laboratory, and this is done in Europe, is four femtoseconds. So we have four femtosecond pulses, we can clock them to better than three femtoseconds, and therefore any molecular system can be studied in this way.

I'd like to show you the laser arrangement we have used. I've made it sound so simple, but of course when you try to do the first experiments... I'll show you one of the first illuminating experiments: two atoms and a spring, nothing fancy in this. If I solved for the Schrödinger wave equation from 1926, it would tell you that you must calculate the probability distribution function for how to find the spring in space. And so, if this is the distance, you'll find that this probability is everywhere; that would be the Schrödinger wave equation. Simply put, the probability of finding the spring anywhere is essentially the same. According to wave mechanics, in the microscopic system, that is the quantum mechanics of that system, delocalized over all of space.

On the other hand, if we give this spring an impulse of a femtosecond, then what you see is the solid bell shape of the curve, the wave packet I told you about. What this means is that the two atoms at that time are localized in space at 2.7 Angstroms. If I waited there for a long time, this bell shape will move into the other side, so the spring stretched. Waiting longer, it will come back to the compressed shape. Therefore, by simple logging of all these, just like Young's experiment, we are able to see this as if the system follows classical motion with the spring compressing and stretching. This, of course, is what we saw experimentally - stretching, compressing, stretching, you can even see the anharmonicity of it, and you can measure it with accuracy of about 10 femtoseconds. [2:50]

Of course, we wanted to move into the real world of chemistry, as John told you. In 1931, Henry Eyring and Michael Polanyi (the father of John Polanyi) made a prediction about the energy landscape of how chemical bonds will break and form. Of course, this entire journey occurs in about 50 femtoseconds. We wanted to freeze the system as the atoms of a hydrogen molecule get together to make a chemical bond and break a chemical bond, to freeze them in the transition state, which would have been a dream in 1931.

Again, a paradigm case that illustrates this point is two atoms in the process of separating from each other, breaking a chemical bond. I was stimulated by Linus' book, The Nature of the Chemical Bond. He had this example where he compared the alkyl halides in terms of the covalent potential and the ionic potential. Pauling's point was that depending on the metal and the halogen, these two curves can cross or not cross. If they cross, the bond will be a mixture of covalent and ionic character; if they do not cross, and the ionic is too far, then the bond takes on one of the two. So in the femto age, we can prepare this structure far from the equilibrium - Pauling was interested in the equilibrium, static structure - putting a wave packet here in the covalent region. But as the spring tries to break, it will move from the covalent to the ionic, with a chance of breaking the bond, but it might also reform the bond again, and so on.

This is history now: the two atoms here, electron jump, some go out and break the bond, others like each other very much and will bounce back, and we can see this resonance motion in the laboratory as a function of time in a beautiful, classical way. In fact, the beauty of this experiment is that it was done with about 10 million molecules, but what you see is the behavior of a single molecule. By using these very short pulses, we were able to prepare all of these atoms together coherently, as if all 10 million are glued together, moving, breaking and remaking chemical bonds all in harmony. If all 10 million are behaving slightly differently in time, we would never see this resonance motion; that is the beauty of this experiment, which is the analogue of Linus' thinking about the nature of the chemical bond in sodium chloride. [6:19]

Of course, the field has progressed so much that I don't have time to go through many examples, just a few, and end up with where we're going in the future. For example, this is an experiment that was published from our lab on the analogue of hemoglobin, done by synthesizing these picket fence structures because we wanted to see how the molecular structure is going to influence the dynamics of oxygen-binding into the heme. This was an interesting experiment in its own right, because we can study this under physiological temperature, at room temperature, in water. The other ones I showed you were fundamentals, so we had to isolate the molecules from the rest of the world in molecular beams and so on.

Here, we study this in water and the idea was to see why the hemoglobin binds oxygen and at what time scale, and what the recombination time, the probability of this oxygen rebinding again, is. We used the femtosecond resolution to start the motion, kick out the oxygen, and then probe the recombination and the dynamics. This was published in Angewandte Chemie and the results were very striking; the electrochemistry was also done in collaboration with Fred Anson, my colleague, and this is a very valuable system because the peptide bonding and the hydrogen bonding help to understand the relationship between the structure and dynamics.

I'd like to show you a transient femtosecond in a complicated system; here is the picket fence, porphyrin, and we study the electron transfer into the metal. The time scale is 250 femtoseconds, with no oxygen and no base; with hemoglobin as well, you need to have a base, amidazole, which comes from the other end. Not only is porphyrin important for this oxygen binding and separation, but the amidazole as well. Here is the one with no oxygen and no base, and a case with base; it doesn't change that much. But once we bubble oxygen at room temperature, just as it would happen in the body, we see that the transients are completely different. In fact, the oxygen leaves the heme in 2 picoseconds. The first charge is in 500 femtoseconds, and the oxygen leaves in two picoseconds. Furthermore, we couldn't find the oxygen recombining at all in up to 15 picoseconds. In an experiment that was just finished, we see it recombine in 15 microseconds. The issue, then, is why oxygen does this in such a molecular assembly, called supra-molecular chemistry, and how we relate that structure into the dynamics, but this is not the subject to dwell on here. [10:06]

This is a recent experiment published in PNAS, and I'm very intrigued by this problem, as I think Linus was from the structural point of view. But I am so fascinated by this idea of "molecular recognition." I do believe that you will see exciting things in the next ten years or so, because I don't think that the answer is simply in the molecular structure of the two entities. How does a protein recognize a molecule, which is a drug in this case, and how do the drug molecules go into the DNA? These are all very weak forces, including the hydrophobic forces, which we are trying to shed some light on, meaning that water molecules around this cavity help the molecule to move around and go into the protein. We just published a couple of papers trying to elucidate this process, because they all happen at the femtosecond time scale.

Finally, in your genetic material for example, an electron can move into DNA, and there are claims in the literature that this might damage the DNA and cause all kinds of problems, including cancer. What is interesting about the chemical and physical aspects of it is there are some claims that if an electron comes here it might move inside the DNA as if it's a free metal; in other words, a conductor. DNA would be conducting electrons at very high rates, and that would mean it's acting like a metal. We decided to collaborate with my colleague Jackie Barton, who does a beautiful system here in terms of genetically modifying bases inside the DNA duplex. We did a series of femtosecond experiments where we can excite one of the bases in DNA to allow an electron to move, measuring the speed at which it does this. Of course, we can genetically modify this so we can study it as a function of distance, as a function of the base pair, and as a function of the driving energy, and all of this has been done.

Just to show you something dramatic, we took a duplex of DNA; here is the amino purine and next to it is guanine. You can see that as we add adenine one base pair between the two, or three, or four, how the femtosecond dynamics drastically change as we add one more base pair between the two bases we are studying. Electron transfer happening inside the DNA is efficient, but it is also highly localized to three to six or so, and we can give the dynamics while not believing that it's an efficient molecular wire. This is what we concluded in the last paper, but there are interesting dynamics taking place, and it relates to the motions of the base pairs, the structure of the base pairs, and the driving force of the base pairs. [13:45]

So where is this going? In 1980, I conjectured - because we didn't have femtoseconds in those days - in a Physics Today article, that if we have sufficiently brief and intense radiation that we may be able to fulfill a chemist's dream to break particular bonds in large molecules. The idea here is that if we can localize a nuclei on the length and timescale of the actual chemistry, maybe we can intervene in that critical moment and modify the chemical outcome.

Now, it turns out that we, and others, have demonstrated this in small systems, but that is not what I am interested in. The issue is whether you can do it in really large molecular systems. A couple of years ago we published an article in Science where we used a femtosecond pulse to look at the chemistry in the Norrish reaction, known to chemists, and varied the molecular size, but it was exactly the same reaction. We showed that if we can use femtoseconds to deposit the energy, and then probe it, that in theory you should expect a decrease in speed by four orders of magnitude for this size [largest] compared to this size [smallest]. However, at most you see a factor of two or so. The point here is that in the future we might be able to beat the so-called "statistical behavior of molecules" by at least three or four orders of magnitude, and we are continuing to study this general area in the control of molecular chemistry. If we can do that, I think we will be on the way to something very exciting.

There is another area that my group is focusing on, and in fact in the January 19 issue of Science there is an article with the latest on this subject; that is, x-ray crystallography and electron diffraction can be used to give you static molecular structures. Our drive, or interest, right now is to figure out if we can get into much more complex biological systems, more so than chemical systems, in order to understand problems that I'm intrigued by, such as protein folding - there is a lot of conjecture in the literature right now and it's an active area, but we really don't understand it. Can we probe on the level of the atoms and molecules? What happens, for example, when a protein goes from one state into the other? Can we learn about the landscape, about the enormous selectivity that exists in biological systems that does not necessarily come into play in chemical systems?

In order to open all of these areas, my group's thinking is to try to find new techniques that will allow us to see very complex systems, to understand some of the principles of these dynamics which might not be as simple as forming chemical bonds. In so doing, we decided to combine femtosecond lasers with the Einstein idea of the photoelectric effect to create an electron pulse, focused very quickly into a beam of molecules. Now, rather than taking a static picture of the molecular structure, we're taking a snapshot. If we can get the pulse to be especially short, we should be able to see this snapshot at that particular time, just as we did in the femtosecond domain.

I had a brave graduate student at the time by the name of Chuck Williamson - who is here today - who was told when he joined my group that this was a hopeless experiment to do, and as he told me once it's a "no to the power of ten" experiment. I think one of our first diffraction images taken was lousy, at a time scale of 10 or 20 picosecond pulses, barely showing diffraction images.

This is also a message to students, because whenever you start something new all of the experts will tell you it's not going to work, it's useless, and so on. But don't listen, just keep going. There were many experts in diffraction who told us that this is hopeless, that we wouldn't get anywhere, and that they knew all about it. We started to barely see diffraction images, which were improved during the tenure of Chuck, as well as two outstanding post docs working on this. The rings became much better, and one of the beautiful things about diffraction is that you can really fit the radial distribution function and scattering function theory into the experiment, and compare using iteration methods until you get some good information about the molecular structure. We are now getting the molecular structure at that particular instant in time. [20:05]

One of the systems that we studied recently is the illumination of halogens from ethane to form ethylene, its fluorinated derivative. In the process of doing this, you can break the first bond and then the second bond, and in so doing the literature conjectures that the intermediate formed is this bridge structure. The electronic structure during the course of the chemical reaction will change, and this atom, the iodine in this case, will be between the two carbons. So we thought that this [bridge structure] is a possibility, but so is this structure, that the whole problem is a dynamics problem; namely, that you can break the first chemical bond very quickly, but the second one might break on a time scale that is shorter or longer depending on the rotation around the carbon-carbon bond. This will determine what you will get out in terms of stereochemistry and the earlier studies that people have done.

We followed this with diffraction, focusing on a given bond - in this case, the carbon-iodine bond - to see how this would evolve with time. I'll just show you the latest, which is beautiful in my opinion. You can get the molecular scattering and the radial distribution function. This is the intermediate, assuming that there is a rearrangement of the electronic structure. There is no agreement between theory and experiment - theory is green, experiment is black. On the other hand, if we just simply say that the first bond breaks, then the second, on a time scale that is shorter than the rotation around a carbon-carbon bond, we get very good agreement between theory and experiment, establishing that this is the intermediate.

As we can learn from diffractionists, we can refine more and more by iteration and are able to find the bond distances and the angles for this particular intermediate; this is now not a stable structure at infinite time, but this is taken at five picoseconds after initiation. It is also very interesting that we can find the bond distances in this intermediate. You try to form a pi bond and break a sigma bond, and therefore you should see this bond in the intermediate become shorter compared to the initial parent while this one becomes longer, which is what we see experimentally. Of course, after you do all of this you look at theory, which is in better agreement with what you got experimentally.

We just published a paper by Hyotcherl Ihee and myself in Angewandte Chemie. It's a beautiful piece, because this is the reaction known for iron pentacarbonyl; people sometimes stretch it and use it as a hemoglobin model for liberating these ligands to make five CO's plus iron. They are all intermediates which we can freeze and see at a given time, which Hyotcherl did with Jim Cao, a post doc at Florida State. You see here that we can trap the FeCO₄, measuring the diffraction data very accurately, showing that the structure and the state proposed for the generation of this intermediate is not correct. For example, you can see the A1 state and the B state, which doesn't agree at all with the B2 triplet, but it beautifully agrees with the singlet state and we can get the molecular geometry of this molecule. There are probably about 50 to 100 papers on the state and structure of this. [24:49]

I want to show you what I consider a new breakthrough; this is the one that came out in Science. This is a brand new apparatus; after Chuck left we had to build a new one because he knew all the tricks with the old one, and we couldn't get it to work as efficiently. We decided to build this brand new one, and have five outstanding people working on it, which you will see in a minute. We call this apparatus the Third Generation Diffraction Machine; in addition to the usual diffraction - here the electron is generated in this way, sent to a CC camera here - we also have a mass spectrometer which we can use to identify the whole species and fragments that are coming out in the chemical reaction. There is also a femtosecond system, and the remarkable thing is that we have achieved 500 femtoseconds with a very good electron density, but we have published 1 +/- 0.2 picoseconds in Science. You can see clearly that these are the electron pulses coming out, and we can measure the separation and pulse widths precisely.

The beautiful thing about this apparatus is its reliability; we can now look at structures that we previously couldn't even dream of seeing. We can see for the first time that these diffraction images change from the two-dimensional CCD camera in front of your eyes. This is not computer enhanced. In these six images, you can start to see the rings form as a function of time. This is an incredible enhancement of sensitivity and reliability that allows us to study quite different molecular structures. The most recent article in Science studied a molecule without heavy atoms - we usually look at molecules with heavy atoms - but we just studied, for example, a cyclic hydrocarbon, and there will soon be a paper with very exciting complex molecular structures that are being trapped at very short times. Again, connecting to Linus; he was very interested in electron diffraction at Caltech, and the Caltech group was the first to use these new techniques in the United States. We are now just following in his footsteps, but adding the element of time to try and understand what is happening.

Time has been part of the theme in this lecture. It is amazing that none of us have thought of the usefulness of time resolution in areas such as medicine and biology; the highest atomic clock in the world is achieved using femtosecond pulses, and you can get an accuracy of 10^-15 through femtoseconds. I have been told that the most femtosecond lasers sold are not to chemists or physicists, but to hospitals, because now they can image these very fine tumors using femtosecond lasers. In the world of microelectronics you can use femtosecond pulses to lift a piece of material on the micron scale without dissipating heat into the microchip, in the same way that you can deposit energy into a molecule without letting it leak. There are all kinds of interesting applications for this, but when we started we hadn't really thought about them.

I was intrigued by reading this book recently. Unlike scientists, who are very conservative and work hard, here is our friend [Newt] Gingrich who said, making a prediction about the future, "The first settlements beyond our galaxy will have occurred by 2500. We will have left our solar system by 2200. We will have permanent colonies on the moon and Mars by 2100." Fantastic! He really knows what we are going to do in the future.

One is fortunate to be at a place like Caltech, where Linus told me personally that he really enjoyed his scientific work and interacting with outstanding people. It is also very fortunate to have more than 100 people from all over the world interacting on this idea that I have tried to convey to you in one hour; namely, a voyage through time. Thank you. [31:28]

John Westall: Thank you very much for the voyage through time. I believe we do have a discussion section this afternoon at 5:00, but we have a few minutes right now if you would like to entertain any questions related specifically to this talk.

Audience Question #1: For the reaction of diomethane, is that a gas phase reaction?

Ahmed Zewail: We wanted to do it in the gas phase so that we could compare it with quantum mechanics equation, but if you do it in the gas phase it is much more difficult because we have to do it in a molecular beam. So it was gas phase, and that is the reason the experiments are extremely difficult.

Audience Question #2: How many oxygens leave the heme in your experiments? I was wondering if you could speculate on how far it goes away, and why it takes so long to go back.

Ahmed Zewail: Very good question. When I told you 1.9 or 2 picoseconds for the oxygen, I was actually hiding something; I don't know yet what the real time scale is for the bond to break between the oxygen and the heme. That could be about 100 or so femtoseconds, but the molecule then undergoes a torque, rotating into the pocket structure itself, and then might have moved fast to another location before recombining very slowly. So now we do a polarization experiment to reorient the configuration to see how much of the two picoseconds are due to that initial torque. This way, we might be able to separate the time scale for the first breakage from what it will take for the motion. But we do know the time for the recombination is extremely long.

Audience Question #3: [unintelligible]

Ahmed Zewail: It is very possible. It is very well known in solids that you can get one dimensional and two dimensional energy and electron transport, and people have studied this for years and years. The issue is, when you consider DNA as a polymer, would you really get an efficient one-dimensional electron transfer within it? The chemical idea there is that pi-stacking would help you to go through. But if you think about it from a biological point of view, would that be good? Do you want to transfer damage over very long distances, for example? Then the question becomes a physics question: can we measure how fast the carriers move from one spot on the DNA to another spot over a given distance? When we measure the rate at which these carriers are moving, and you know the voltage, you can find the mobilities just do not compare. It is very possible that a protein localized at a given point will be efficient at this point but not at others, and we have not examined that.

John Westall: I believe that all the examples you have shown were unimolecular systems. Of course it would be interesting to apply the same techniques to bimolecular systems, but then there are issues establishing the coherence necessary to make the whole thing work.

Ahmed Zewail: Right. We have done that, John, and many people have followed this, but I just didn't have the time. Let me just show you an example of what we did. The problem in molecular recognition, which we are now interested in, is that if you have a drug, for example, trying to recognize DNA, most of the kinetics that have been measured - and thermodynamics and everything - are actually dominated by a diffusion-controlled process. As you know, in a bimolecular process you will be taking too much time, a millisecond or so, but the actual process could be in the femtosecond range.

To get around this - we have done this with the late Dick Bernstein - in order to not be held hostage by this millisecond and microsecond diffusion of reagents, you expand two reagents together in the molecular beam. Using a precursor such as HI or HCl or HCr, and CO₂, and mixing the two to dissociate HI, for example, you can liberate hydrogen. Now you are looking at the hydrogen reaction with CO₂.

What we did was to expand these two together, liberate the hydrogen by using a femtosecond pulse to remove the iodine, and then you are looking at these bimolecular collisions between hydrogen and CO₂. We can see the transient complex, and in fact we measured this as a picosecond for the birth of the hydroxyl radical and the carbon monoxide. This is a theoretical calculation that has been done recently to try and understand the fundamentals. People have also used ethylene, and other systems have been studied like this.

Audience Question #4: [unintelligible]

Ahmed Zewail: Right, and in fact this is what stimulated us about the oxygen recombination, was this system of Mulliken, which took 50 years to find the nature of what happens in the encounter between iodine and benzene. This is truly a bimolecular, and you can inject an electron there, and it recombines and you break a bond. Mulliken suggested that the I₂ on the benzene was lying down, the iodine was lying down, but every textbook thought it would be axial. We did the experiment to find the femtosecond dynamics, but we also did an orientation experiment, and in fact, the iodine is pointing essentially into carbon, and not the x-axis; 32 degrees is the angle, and you actually suggested that to me a long time ago, Jack.

John Westall: Well thank you very much for your attention, thank you Professor Zewail for a very interesting lecture. [40:41]