I obtained my doctorate after five years of intensive study in what is generally recognized as one of the most difficult types of experiment to perform in any scientific field: coincidence spectroscopy. The experiments are difficult to set up, difficult to run and most of all, difficult to interpret. It's only now that I have some distance from having performed them that I recognize just how tough my doctoral work was.
I'll try to break down a coincidence experiment into comprehensible steps.
- Choose a chemical that is formed temporarily during a reaction. These are called "transition states" or "transient species". You can choose a system with very simple atoms, which will make the theoreticians happy because they can generate a complete dynamical model of the reaction. Alternatively, choose a system that has some important implications for atmospheric chemistry, such as clusters of NO2 molecules, or combustion chemistry, such as oxygenated radicals formed when hydrocarbon fuels are burned.
- Find a way to stabilize that chemical. This is not easy. There are many transient species formed during reactions, and they only live for a matter of microseconds, or sometimes nano- or picoseconds. That's really not long enough to make a measurement of its properties. The method I used for my PhD research was to create a negative ion of the chemical. Of course, we didn't always know if this was going to work in advance, so sometimes a few different combinations of precursor chemicals would have to be used.
Once you have a reliable source of the chemical that can be made into a molecular beam, you can proceed to the next step.
- Measure the chemical's properties by dissociating it in a controlled fashion. This means using a laser. Your laser should be at a wavelength that's appropriate for the kind of environment that your chemical will experience during a reaction. For atmospheric reactions, this generally means somewhere in the visible or near-ultraviolet. You want to pulse your laser so that it intersects the chemical beam periodically. Then you must measure all of the products of the chemical simultaneously. You want to know the masses and the velocities of the products so that you can completely reconstruct where the energy from the laser went into the chemical. You need sophisticated detectors for this measurement. Because you don't want to count more than one event per laser pulse and you want a statistically significant number of events, it often takes days of continuous running to get enough data to analyse a given chemical. Since the experiment is large and complicated, with lots of bits that are prone to drift and require frequent adjustment, this takes a long time. If you've been patiently tending an experiment on a particular chemical for four days, going home only to shower and eat, and suddenly there's a power outage or your data acquisition computer crashes, it is very tempting to go temporarily insane and smash things. With fists, not lasers.
- Determine the chemical's properties using computational analysis. Once you've finally gotten all of this to work, you must now interpret the results. You have to try to figure out how the energy from the laser pulse was distributed in the chemical. Where was the electron that was knocked off of the chemical, and how much energy did it take away? Which bonds did the laser pulse affect and cause to vibrate? Can you see any rotation taking place? If you are comparing several types of similar chemicals, like alcohol derivatives, why did some of them dissociate and others not? Finding answers to these questions can sometimes be aided by performing what are called ab initio ("from first principles") calculations on the chemical. Then you can use the bond lengths, angles and molecular orbitals to create simulated spectra to compare to the experimental ones.
- Marshal this information into some sort of order; attempt to publish. The fun isn't over yet. Just because you understand your results doesn't mean that anyone else will. And with experiments this complex, simplifying the explanation is nontrivial.
Even though it's been nearly eight years since I ran a coincidence experiment, I know I could go back into my old lab and run it now. It's that ingrained in my memory. In fact, I think I could do it even better than I did before, as I've picked up a much clearer understanding of signal processing and analysis from the work I've done since I left physical chemistry. And I think that's pretty awesome.
Now I think you should tell me why you're awesome.