|
The Birth of the Plasma Source |
|
I joined the Spectrochemical Analysis Section of the National Bureau of Standards in mid-1957. Bourden Scribner, the Section Chief, gave me the rare opportunity to design my own research project in atomic spectroscopy. I was one of Fassel’s early students, but my thesis was on IR spectroscopy. At Iowa State, I knew atomic spectroscopy only from Fassel’s one-semester course, and from regular attendance at Saturday morning meetings of Fassel’s group. I came to NBS from Bert Vallee’s Biophysics Research Laboratory at Harvard Medical School, where my first assignment was to complete the construction of a multichannel flame spectrometer and then use it for research. At that laboratory, there was also a project on a high-temperature cyanogen-oxygen flame as an emission source. That research was done by Keiichiro Fuwa, who later became a Professor at the University of Tokyo. I followed his work with interest, but my involvement was only incidental. When I drove to New York to visit family and friends, I stopped on the way back at the Cyanamid research labs in Stamford, Connecticut, to pick up full cylinder of cyanogen. I carried the cylinder of toxic gas in the trunk of my car, which would surely not be allowed these days. In thinking about what research to do at NBS, I wanted it to have practical use at that laboratory. A major part of the work of the Section was preparing and certifying Standard Reference Materials (SRMs), with emphasis on ones for atomic emission and x-ray fluorescence spectroscopy. These spectroscopic methods depend on well-characterized standards, especially for solid samples, so certification of new standards depended on wet chemical methods and on comparison with previously certified standards. Solution standards could be made from pure chemicals, but the prevailing atomic emission methods for solutions – the rotating disk and porous cup electrodes with a spark – did not offer the kind of precision needed to categorize standard reference materials. A flame is more stable than an arc or spark, but flames aren’t hot enough to give spectra of all metallic elements, especially at low concentrations. I began to think about an electrical plasma in a flame-like form, where one might be able to introduce a liquid sample as a spray. I found examples in the literature of arc discharges that were controlled in size and position by a sheath of cool gas or liquid. As the current in an arc goes up, the arc normally increases in diameter, so that the current density and the arc temperature stay fairly constant. Cooling the outside of the arc prevents the diameter increase. At high enough currents, there is an electromagnetic “pinch” that actually narrows the diameter. An arc of this type was described as early as 19231, but no way was found to introduce elements (for plasma diagnostics) except to insert a wire through the wall of water that constrained the arc – and that soon damaged the device. I experimented with one of these devices, using a 20 ampere arc source that was not otherwise in use. A group of engineers from Perkin Elmer came through at this time, and saw what I was doing. One of them knew enough about these discharges to comment that I didn’t have enough current to get a pinch. He was right, but that wasn’t what I was after. Ordinary arc temperatures were all I needed. But I couldn’t figure out how to bring in a sample. I then turned to a newer form of a gas-constrained arc, the “plasma jet” of Weiss2 and Peters3. The arc burned in a cylindrical chamber, which had an opening at one end through a hole in the cathode. A gas or liquid was brought in tangentially to the inside of the chamber to constrain the arc. Plasma emerged through the hole in the cathode in a flame-like form. Its purpose was plasma physics research, but perhaps it could be adapted as an analytical source. There was a small machine shop, and I’d learned to use these tools in high school. I went to Brooklyn Tech, one of New York City’s selective high schools. Unlike the more famous Bronx High School of Science, Tech’s curriculum was designed to train technicians, so it emphasized shop courses. Having the shop at hand made it easy to experiment with various configurations, and I soon arrived at the design shown schematically in Figure 1. The outer diameter of the chamber was about two inches. The water-cooled parts were machined from brass and soldered together. If I forgot to turn on the cooling water, the solder soon gave way, ending that experiment. The outer chamber was machined from hard rubber. The only tricky part in making the device was inserting the tube through the side to get a smooth tangential gas flow. The atomizer could be the homemade one shown in the drawing, but it was easier to use the atomizer part of a Beckman hydrogen/oxygen flame source. Spraying acidic solutions through the mostly brass atomizer resulted in a short life, but I learned to make repairs more easily than to make a new atomizer. Helium was used as the coolant because it was the least expensive inert gas. Now that I know more about plasma physics, I suspect that argon would have made for a better-behaved discharge because of its lower ionization potential compared to helium. I needed an arc stand to hold the source, and found an old one in one corner of a room. Over time, I modified some parts, and removed some others. It was only when we moved from Washington to Gaithersburg that I learned where that stand came from. On the day of the move, Scribner waited in the new labs and I stayed in Washington to keep an eye on the movers. (It was a good thing that one of us stayed behind, because the movers were there to pick up and move the large spectrographs and spectrometers before they were readied for moving.) I was through with the old arc stand, so I marked it to be left behind. A day or two later, in Gaithersburg, Scribner noticed that the stand was missing, and it was then that I learned that his first task when he came to NBS was to build that stand. An initial set of analyses of two stainless steel SRMs for iron, chromium, and nickel, with standards made from pure metals dissolved in acid, with vanadium added as an internal standard, agreed with the certified values about as well as one could expect with photographic measurements. The calibration curves are shown in Figure 2. The results were encouraging, but they didn’t prove that the plasma jet source could give the precision I was after. The results were in our first publications on the light source4, 5. We learned that Korolev and Vainshtein, in Moscow, were also working on a plasma jet source6. It was expensive to buy translations of Russian papers, so I signed up for a one-year course at NBS to learn to read scientific Russian. The teacher was a white Russian émigré who met and married a U.S. Navy officer in Shanghai, and so came to Washington. She still had her preference for cold weather; in the winter she insisted on opening the classroom windows, so she was comfortable and we bundled up. She taught us about Russian customs as well as the language. One day, she told us that in the old days everyone had a silver holder for their tea glass. I wanted to pass the course, so I didn’t ask whether the servants and serfs also had silver holders. Once I knew enough Russian to read scientific publications, I saw that the Russian plasma jet was similar to mine, but that it wasn’t intended to analyze solutions. The samples were powders in a conventional graphite electrode that served as the anode. There were many questions about our new light source. We couldn’t see how the arc made contact with the cathode, but there were markings on the outer part of the cathode after a run. Scribner suggested that we take high-speed movies. That was an adventure in itself. The film ran through the camera so fast that the last few inches of a reel thrashed around in the camera until it shut down, leaving the inside of the camera full of tiny bits of film that had to be thoroughly removed before the next reel was put in. A high speed video camera would certainly have been easier, if such a thing existed. When the developed film came back, we saw that we had the wrong idea about the shape of the discharge outside of the chamber. There was no flame-like cone of plasma. Rather, a thin arc streamer emerged through the opening in the cathode and moved around randomly, while it doubled back and contacted the top of the cathode. We saw a stable cone of plasma because our eyes couldn’t follow the rapid movement of the streamer. Others were perhaps aware of this instability. Osborn7 and Lou Owen8 placed rods of graphite or thoriated tungsten (welding rod) horizontally above the plasma jet, as the cathode. Our experiments with this design weren’t successful. Graphite rods burned back too quickly, and there were copious white fumes, probably tungsten oxide) from the welding rod. We then tried the design in Figure 3. The external cathode is a ¼” thoriated tungsten welding rod, held in a water-cooled chamber made of brass. Lou Owen came to be involved to some extent with my later development of this source. In 1962, he left his job at the uranium processing plant in Portsmouth, Ohio, and started his own company, Tomorrow Enterprises, whose services - as listed on its letterhead - were “Design and manufacture of custom laboratory equipment”, and “Botanical research and experimental horticulture.” We purchased a dc arc power supply from him, to replace the old unit that had served until then for most plasma jet experiments. Lou got a contract with another U.S. Government agency (the Air Force, as I recall), and was looking for lab space near Washington, preferably free. I thought I would learn from him if he was in my lab, I had some extra space, and the government connection of his project justified allowing him to use NBS facilities. Lou became a frequent visitor to NBS and other labs in the area. He would either stay at a low-cost motel, or find a place to park his Volkswagon minibus that was outfitted with basic living arrangements. At PittCon one year, Owen gave a talk that he began by inviting those who were really interested to gather at the podium for a discussion. He mentioned that some of his work was at NBS, and another scientist from Scribner’s group who was there thought the performance was undignified and reflected badly on NBS. She complained to Scribner, who matter-of-factly mentioned the complaint to me, and that is where it ended. Meanwhile, with the new plasma jet with external cathode, and the old arc supply, I ran more studies of the precision of analysis of a synthetic stainless steel “sample”, on an Applied Research Laboratories (ARL) Quantometer, using iron as the internal standard element. The results were inconsistent. Occasionally, the measured relative standard deviation (rsd) was as low as 0.2%, but it averaged much higher. At times, the precision of measurement of the an intensity ratio was as much as three-fold poorer than the precision of measurement of the internal standard alone9. On closer examination of the data, I noticed that the line intensities in the raw data often drifted during a series of runs. The spectrometer had in it a low-intensity “fatiguing lamp” (intended to stabilize the responses of the photomultipliers), and with that lamp one could conveniently measure the stability of the instruments’ intensity measurement, which was consistently in the range of 0.2% rsd. That seemed to point to the source as the origin of the drift. But then I noticed that two Cr II lines with similar excitation energies, which I had included in the measurements, drifted in opposite directions in some experiments. That seemed to point back to the spectrometer for the drift. Small movements of the exit slits wouldn’t affect measurements with the fatiguing lamp, but would affect measurements of line spectra. In the next experiment, a pair of standards was run at the beginning of the series of measurements, then after every five runs with the sample, and at the end. The intensity ratio for each line pair for the standards was plotted vs. run number. From this, one could plot or compute a calibration for each run of a sample individually, assuming a linear drift between calibration runs. By this tedious process, rsd values of about 0.3% were obtained regularly. This result was mentioned in a talk at the XII Colloquium Spectroscopicum Internationale10, but it was not published in detail. My conclusion was that the plasma jet source was capable of precision of analysis limited by the stability of the spectrometer. By then, my interests were turning in another direction, and others were experimenting with discharges that could be fed with a solution spray. Fassel in the U.S. and Greenfield in the U.K. reported their first work on induction-coupled plasmas at about the same time. Fassel persisted in making improvements to the ICP for over a decade. That justly earned him recognition as the father of ICP spectroscopy. To many analytical chemists, “plasma spectroscopy” is synonymous with atomic emission spectroscopy, and it is listed as a specialty in forms, including the membership form of the Society for Applied Spectroscopy.
References 1. H. Gerdien and A. Lotz., Z. Tech. Phys. 4, 157 (1923) 2. R. Weiss, Z. Phys. 138, 170 (1954) 3. R. Peters, Naturwiss. 41, 571 (1954) 4. M. Margoshes and B. F. Scribner, Natl. Bur. of Standards Report No. 6160 (1958) 5. M. Margoshes and B. F. Scribner, Spectrochim. Acta 15, 138 (1958) 6. V. V. Korolev and E. E. Vainshtein, Zhur. Anal. Khim. 15, 686 (1960) 7. A. B. Osborn, J. Sci. Instr. 36, 317 (1959) 8. L. E. Owen, Appl. Spectroscopy 15, 150 (1961) 9. B. F. Scribner and M. Margoshes, Natl. Bur. of Standards Report No. 7342 (1961)
|