II—
A Million Volts or Bust
1—
Some Preliminaries
The naturally occurring radioactive isotopes emit nuclei of helium (a particles), fast electrons (b particles), and penetrating x rays (g rays). The energies of these particles and rays are usually given as multiples of the "electron-volt," the energy of motion an electron acquires in falling through a potential difference of one volt (two-thirds the voltage of a standard flashlight battery). An electron-volt is an absurdly small measure, a little over a million-millionth of the "erg," the basic energy unit of the physicist; and the erg itself amounts to less than the footfall of a fly. To fix ideas and notation, an alpha particle from the naturally occurring isotope polonium with atomic weight 210 (Po210 ) has an energy of 5.3 million electron-volts (5.3 MeV) and a velocity of 1.6 billion centimeters a second (1.6·109 cm/sec), about one-twentieth the velocity of light. The individual polonium a particle, although projected with tremendous speed, has a negligible mechanical effect on the surface that stops it. A great many of them, however, are noticeable.
A noticeable quantity is 37 billion, roughly the number of disintegrations occurring each second in a gram of radium; this rate of collapse, called a curie, is the basic measure of radioactivity. If each of these billions of disintegrations resulted in an alpha particle as energetic as polonium's, all together they would carry away
some 3.5·105 erg, or a little less than a hundredth of a calorie, each second. An ampere of alpha particles is the equivalent of almost a hundred million curies (108 Ci). A milliamp (mA) or even a microamp (µA) of alpha particles greatly exceeds the flux from available natural isotopes: 1µA = 80 Ci, 1 mA = 8·104 Ci; a microamp playing on a surface would heat it at the rate of almost one calorie a second, a milliamp at almost a thousand.
Since the nucleus occupies about as much of an atom as the earth does of the sphere whose radius is its distance from the sun, the chance that an alpha particle will strike a nucleus head on before it comes to rest is very small. But only in such collisions can the particle force its way into the nucleus; hence, if forced entry is the experimenter's goal, he would do well to furnish himself with a microamp rather than with a curie of projectiles. The µA has an even greater advantage than the preceding numbers indicate; the radiation from a natural source goes in all directions, whereas the outpouring from an artificial source may be tightened into a directed beam.
In 1919 Sir Ernest Rutherford unintentionally introduced alpha particles into nitrogen nuclei and, what was much more difficult, recognized what he had done. His initial objective was to study collisions between alpha particles and other light nuclei. As source he used a descendent of radium, RaC (Bi214 ), with a maximum strength of 0.08 Ci. That was strong enough to make the effect under investigation—knocking on, or driving forward, one nucleus by another—easy to detect. The knock-on particles were caught on a screen coated with a material that glowed where hit; and the source could be brought so close to the detector that as many as 40 or 50 flashes a minute could be spotted in the field of a microscope pointed at the screen. In this way, with hydrogen gas as target, Rutherford found swift knock-on protons, and, with oxygen and nitrogen, somewhat slower bumped ions, as he had expected.[1] A slight anomaly occurred in nitrogen, however. Despite every precaution against traces of hydrogen in the experimental space, protons even swifter than knock-on hydrogen ions persistently obtruded. It appeared to Rutherford that a few alpha
particles from RaC, making particularly intimate contact with nuclei of nitrogen, had driven out protons from the heart of the atom. He believed rightly that he had disintegrated the nucleus and wrongly that the process resembled a successful shot at marbles, in which both the impinging and the struck balls end up outside the target area.[2] He continued this work in collaboration with James Chadwick, an excellent experimenter trained at the Cavendish Laboratory, who had perfected his physics in Germany as a prisoner of war and who had returned to Cambridge, where he became in time the associate director of the laboratory. By enlarging the field of their microscope and viewing the disintegration protons at right angles to the direction from source to target, Rutherford and Chadwick showed that alpha particles could knock protons from most light elements and that the yield increased sharply with the energy of the bombarding particles.[3]
In these experiments perhaps ten alpha particles in a million made a collision that resulted in a detectable disintegration proton. Since the average source was 0.05 Ci and the solid angle at the microscope about 0.0002 steradians at most, the maximum number of countable protons was one a second, or, taking the efficiency of the screen into account, around one a minute per mCi. That sufficed to detect disintegration protons but not to give satisfactory quantitative measurements of their speeds, penetration, angular distribution, or yield as a function of the energy of bombardment. Until late in 1924, and probably for some time afterwards, Rutherford and Chadwick continued to suppose that they were engaged in a game of marbles, and tried to fit their meager quantitative results to irrelevant billiard-ball mechanics. But that year, in one of those triumphant syntheses that distinguished the Cavendish, one of its senior members, P.M.S. Blackett, succeeded in photographing in a Wilson cloud chamber (an invention of the laboratory) the trajectories of the particles participating in productive collisions of the type discovered by Rutherford. In Blackett's beautiful pictures—he obtained records
of eight disintegrations in 23,000 photographs presenting 400,000 tracks—no trace appeared of an alpha particle leaving the site of a productive collision (plate 2.1). The mental picture required that it continue on; the optical evidence indicated that it disappeared, swallowed by the target nitrogen nucleus. If so, the resultant nucleus, after expulsion of the swift proton, should be an isotope of oxygen.[4] In keeping with the convention used in the 1930s, we shall write the reaction as N14 (a ,p)O17 and refer to it as an (a ,p) transformation.
The problem of accounting for the energy in nuclear transformations became pressing and frustrating. Here another bit of Cavendish pioneering enriched and complicated the proceedings. Following up work he had done before the war as assistant to J.J. Thomson, Francis Aston perfected his system of separating the isotopes of the light elements (in very small quantities, to be sure!) by electric and magnetic fields. From the trajectories of the ions of the various isotopes he could calculate their masses; which, by 1930, he was reporting to a ten-thousandth of the mass of a proton (mp ). Now 0.001 mp is about 1 MeV: hence Aston's measurements of isotopic masses appeared to be just accurate enough to be used in working out the energy balance in nuclear transformations provoked by the input of a few million electron volts. The upshot: the numbers obtained by Aston, Blackett, and Rutherford and Chadwick did not agree, and prompted the dispiriting hypothesis that the normal internal states of nuclei of the same isotope differ energetically.[5]
If only the statistics were better! If only the sources were not so weak! If only the geometry of the experiments could be improved! Laments of this character enliven the pages of Rutherford and Chadwick.[6] To improve upon the work of nature appeared to
require a machine capable of producing at least a microamp of light positive ions and of accelerating them to a few million volts; if a nucleus propelled an alpha particle at 5 MeV, would it not be necessary to hurl it at 5 MeV to return it? Already before the war Rutherford had judged the creation of machines operating at the highest possible voltage to be "a matter of pressing importance" for the study of beta rays;[7] but for many years it proved impractical to make insulators that could hold much more than seven or eight hundred thousand volts (700 or 800 kV) or accelerating tubes that could withstand even half that much. Subsequent improvements in electrical equipment, x-ray tubes, and insulating materials, the closer ties between industry and academy forged during the war, and the newly important field of nuclear transformation gave new urgency to the matter.
Rutherford properly took the lead in promoting development of million-volt accelerators. In 1927, in addressing the Royal Society of London as its president, he challenged his audience to fulfill his long-time wish for "a copious supply" of projectiles more energetic than natural alpha and beta particles. His appeal received wide attention and many proposals for realizing it came forward. Progress was slow at the laboratory level. In 1930, while his associates struggled to make a source of a few hundred thousand volts, Rutherford asked big electrical industry for help in raising the "puny experiments in the laboratory" to nature's scale. "What we require [he said, at the opening of a new High Tension Laboratory at Metropolitan-Vickers Electrical Company] is an apparatus to give us a potential of the order of 10 million volts which can be safely accommodated in a reasonably sized room and operated by a few kilowatts of power. We require too an exhausted tube capable of withstanding this voltage. . . . I see no reason why such a requirement can not be made practical."[8]
2—
High Tension
Were the million volts the only criterion—were size of current and steadiness of operation of little importance—the quest would have been over soon after it started. Several techniques for obtaining very high, fleeting voltages across discharge tubes succeeded by or before 1930. None gave rise to a useful beam of positive projectiles. Nonetheless they are worth attention, since some of the problems they raised, and workers they employed, recur in our story. In addition to these impulsive methods, two others for obtaining steady high potentials directly were under development in 1930. For some purposes they held an advantage over the cyclotron.
The Impulsive Way
Nature provides big voltages gratis to anyone bold enough to play with lightning. In the summers of 1927 and 1928, three members of the physics institute at the University of Berlin hung an antenna between two mountains in the Italian Alps 660 meters apart. In their definitive arrangement (fig. 2.1), a string of heavy-duty insulators, provided free by their German manufacturer, kept the antenna and the probe line attached to it from ground; the potential reached by the antenna was controlled and measured by the air space between a metal sphere dangling from the probe line and an earthed sphere hanging beneath it. During thunderstorms, the antenna rose to a very high potential over ground, sending sparks between the spheres across as much as 18 meters of pure Alpine air. The intrepid experimenters calculated that, in that case, they were dealing with 15 million volts. The two who survived the experiment, Arno Brasch and Fritz Lange, returned to Berlin to construct a tube that might stand up to a good fraction of this voltage.[9]
In the comparative safety of the Allgemeine Elektrizitätsgesellschaft's research laboratory in Berlin, Brasch and Lange
Fig. 2.1
Brasch and Lange's lightning catcher. E and H are the
spheres between which the discharge occurs; AE, the
antenna; a,a, insulators; b,b, conductors; d, a grounded
wire. Brasch and Lange, Zs. f. Phys., 70 (1931), 17.
tested designs for a sturdy discharge tube with the help of an impulse generator able to reach 2.4 MV. Its principle, the "Marx circuit," may be clear from figure 2.2. A transformer at the center of the hexagon to the right of the diagram delivers rectified direct current via the indicated diodes to the string of n capacitors C , which charge in parallel each to the potential V . When V suffices to drive a spark across the gaps F , the capacitors connect briefly in series, the voltage on the last plate rises to nV , and a spark jumps to the top electrode of the constantly pumped special discharge tube figured on the left.[10] Such generators served the electrical industry to test the characteristics of insulators and other equipment during flashovers. The grandest ever made, which could manage 6 MV, was built by General Electric in 1932.[11]
Another industrial high-voltage instrument adapted to nuclear physics was Southern California Edison's million-volt cascade transformer at Caltech. This object did not satisfy Rutherford's requirement of convenient size: it filled a room 300 square feet in area and 50 feet high.[12] As we know, it was adapted to the
Fig. 2.2
Brasch and Lange's discharge tube and impulse generator. Voltage from the
transformer Tr multiplied by the string of capacitors discharges across the
constantly pumped laminated tube. Brasch and Lange, Zs. f. Phys., 70 (1931), 30.
production of x rays and ion beams by C.C. Lauritsen, who had been so inspired by a lecture by Millikan that he gave up making radios and enrolled at Caltech in 1926, at the age of thirty-four, as a doctoral student. He developed a very fine experimental technique, precise in measurement and simple in style. European visitors judged him to be one of them, free from "the technological extravaganzas that Americans like so much." Brasch, who had reason to know, rated Caltech's engineer turned physicist "an uncommonly good experimenter" with outsized electrical apparatus. Lauritsen's first big challenge as a superannuated graduate student at Caltech was similar to what Brasch and Lange faced at the same time: to make a tube that could stand up to a million-volt generator.[13]
A third group—Lawrence's chum Tuve and his co-workers at the Department of Terrestrial Magnetism at the Carnegie Institution of Washington—had come to a similar situation with still another approach to high potential. Tuve's Ph.D. thesis, completed at Johns Hopkins in 1926, was a measurement of the height of the ionosphere via radio pulses, using a method devised by Carnegie's theorist Gregory Breit. Tuve and Breit continued this work; but they also wanted something novel to mark Tuve's arrival, and decided to try to force the ordinary Tesla induction coil up to several million volts in the service of atomic and nuclear physics. The rationale offered the officers of the institution for this undertaking in the department's report for 1926/27 was that the Tesla coil might be driven to 30 MV to make cosmic rays in the laboratory. Further formal justification for the development of a program so obviously alien to the department's mission was found in the consideration that, as Tuve put it, "the problems of terrestrial magnetism will never reach a really satisfying solution until we possess an adequate understanding of the basic phenomena of magnetism itself." How else secure this understanding than by knocking the nucleus apart? On this flimsy pretext, the department allied itself with the U.S. Navy, which loaned it transformers, condensers, and a glass blower, and steamed forward to create one of the most important laboratories in the world for nuclear physics during the 1930s.[14]
The Carnegie's coil in its definitive form consisted of a primary of a few turns of copper tubing wrapped into a flat spiral around a secondary of many thousand turns of fine insulated wire wound in a single layer on a pyrex tube (plate 2.2). The primary was driven by the discharge across a spark gap of a very big condenser the navy had used in an old wireless transmitter (fig. 2.3). If tuned to the frequency of the oscillating primary discharge, the secondary could acquire a greater peak potential than the air around it could sustain. By immersing it in oil under pressure, however, Breit and Tuve arranged that the secondary held all the primary delivered,
some 5.2 MV (they thought) with a secondary half a wavelength long. Although this estimate appears to have been too high, whatever they had would have driven an effective beam of alpha particles had they had any way to apply their coil. And so, like Lauritsen and Brasch and Lange, having solved the problem of high tension to their satisfaction, they set out, in 1928, to make a tube to withstand it.[15]
Fig. 2.3
Schematic of the installation of the Carnegie Institution's 5=mV Tesla coil.
Breit, Tuve, and Dahl, PR, 35 (1930), 56.
Lauritsen fielded the first effective tube, which required a scaffolding fourteen feet high of good California redwood for its support (fig. 2.4). Its fundamental features were adapted from the high-potential x-ray tubes developed at General Electric by W.D. Coolidge, who reached some 350 kV in 1926 by shielding the glass walls with a copper tube (fig. 2.5). In this way he defeated the buildup of charge in the walls from electrons driven into them; and hence also the discharges to the walls that punctured the tubes and made the main obstacle to increasing the voltage of x rays. To go beyond 350 kV, Coolidge recommended cascading two or more tubes (fig. 2.6), passing the electron beam through thin windows that would act as anodes for one tube and cathodes for the
Fig. 2.4
Cal Tech's high-voltage x-ray installation. The corona
shields are attached to the metal rings at the joints
between the gas-pump cylinders constituting the
tube; the very long cathode reaches almost to the
tube's bottom. Lauritsen and Bennett,
PR, 32 (1928), 852.
next and stop positive ions that might otherwise gain high energy and make difficulty. Lauritsen's tube had four segments, to correspond with four cascaded transformers; each tube was twenty-eight inches long and twelve inches in diameter, and made, appropriately to its location, from the glass cylinders then used in pumps in gas stations. The segments joined at steel rings, which also supported the internal wall shields and, externally, circular slips of tin foil to protect against corona losses. The high-potential end of the cascade transformers fed the long central electrode of the tube (a three-inch steel pipe extending to within an inch of the earthed target) through a water resistor. The tube itself was continually pumped to retain a good vacuum. By August 1928 Lauritsen and Bennet could put 750 kV across the tube with no difficulty, and obtain x rays capable of penetrating over 2 cm of lead.[16]
Fig. 2.5
Coolidge's first design for a high-voltage x-ray tube. Its chief feature is the
metal shields around the electrodes, k and t, which prevent buildup of charge
on the glass. Coolidge, JFI, 202 (1926), 696.
Lauritsen patented the design. General Electric had supported Coolidge's work in the hope that very penetrating x rays might be especially effective against cancer. A cheap and efficient high-tension plant had commercial as well as humanitarian possibilities, whence the considerable interest of both industry and medical philanthropy around 1930 in large numbers of volts. As Lauritsen wrote in his patent application that year, his tube, then able to operate at a million volts, could serve "as the full equivalent of
radium in the treatment of disease, or for therapeutic purposes." A gram of radium then cost $60,000 to $70,000; an x-ray plant of moderate potential, about $30,000. Rewards could be high. And hopes. Many victims of cancer basked briefly in the million volt x rays of the Kellogg Radiation Laboratory.[17]
Fig. 2.6
Coolidge's cascade. Two tubes of the type shown in fig. 2.7 are joined together
so that the anticathode of one becomes the anode of the other. Coolidge,
JFI, 202 (1926), 720.
Meanwhile Brasch and Lange were adapting Coolidge's design for use with the AEG impulse generator and the Alpine lightning factory. Late in 1929 they had a porcelain tube over seven feet long, with walls an inch thick studded with 300 nickel rings. They solved the problem of stray electrons not with complete shielding of the walls, as Lauritsen provided, but by enough rings of sufficient capacity to prevent large buildup of unwanted voltages. This great studded stick—made so long in order to space out the longitudinal potential drop—could stand a surge of 1.2 MV. Nine months later, in August 1930, they miniaturized to three feet by immersing the tube in oil to cut out corona discharges from the now high-potential external walls into the air. Careful study showed that the hurdle to going higher was a creep of electrons along the walls between the metal rings. They nipped the creep in a new tube composed of alternating rings of paper, aluminum, and
rubber of different widths. The spacing of the metal disks remained close, while the creeping distance between them increased enough to discourage the most persistent electron.[18]
Although the vapor pressure of the rubber prevented them from achieving a very high vacuum, Brasch and Lange got spectacular results: the laminated heap withstood the maximum impulse of the generator, 2.4 MV, with transient currents of 1,000 amps; the zippy cathode rays thus produced made x rays that could penetrate 10 cm of lead, and offered a new medical possibility in themselves. Rather than apply very hard x rays against deep cancers, why not try short bursts of million-volt cathode rays, which deliver most of their energy as they come to rest? "Then the irradiation can work very effectively deep within the body without doing so much damage to the parts near the surface." Brasch and Lange's main goal, however, was to adapt their tube to the acceleration of positive ions. They managed to obtain a stream of hydrogen ions at 900 kV, which they thought too low to provoke the transformation of elements. They returned to the Alps to continue their alchemy.[19]
The Department of Terrestrial Magnetism was most impressed by "the spectacular performance" of Brasch and Lange's laminated tube. "They are to be congratulated without reserve," wrote Tuve and his associates, whose own handiwork did not perform quite so well. They had stayed close to Coolidge's design, multiplying segments and decreasing size by immersing the whole in oil. By 1929 their Tesla coil was driving a tube with six segments at 850 kV and one with fifteen segments (all seven feet of it) at 1.4 MV; later, by heatworking the glass to remove bubbles, they operated a twelve-segment tube about a yard long at what they thought was 1.9 MV.[20] Just before Tuve's group learned about the spectacular performance in Berlin, they succeeded in detecting cathode rays and x rays from the tube and fixing their energies at
1,250 to 1,500 kV. It remained to accelerate protons, and also to develop a current in the tube sufficiently strong to permit measurement of the intensity, as well as detection of the existence, of the penetrating radiations. The Tesla coil as used in Washington, with its very brief duty cycle, was not a competitor for the medical purse.[21]
In the report of their work for the year 1929/30, Tuve's group pointed out that the tube perfected with the aid of the Tesla coil already outperformed it; and they called for a new generator "in order to adapt this new tool effectively to the studies in atomic and nuclear physics for which it was developed."[22] The tube won them a prize from the AAAS and some attention: page 1 in the New York Times and an advertisement in Time that with their x rays—reported as equivalent in gamma radiation to $187 million worth of radium—they might split atoms and cure cancer. Lawrence congratulated them on "such a fine recognition," meaning the prize, not the puff.[23] Their disappointment with the performance of their Tesla coil evaporated in this sunshine. They tried to accelerate protons, succeeded on December 8, 1931, and directed the beam overambitiously to shattering the elements. Nothing detectable happened; so small was the beam that "further attempts seeking evidence of nuclear disintegration with this set up [were] discontinued." Lawrence again offered encouragement. The tube, he reminded Tuve, was "a terribly important thing!!!—as I have emphasized to you many times (and you try modestly to give Coolidge the credit for)." It remained to find a way to make the tube produce something other than a prize.[24]
The Old-Fashioned Way
And one was provided. The inventor, Robert Jemison Van de Graaff, conceived a grand idea at the onset of his career, as a Rhodes scholar at Oxford, where he arrived with an engineering degree from the University of Alabama and professional experience with the Alabama Power Company. From Oxford, where he earned his Ph.D. in physics in 1928, he went to Princeton as a National Research Fellow, and soon had a prototype accelerator working at 80 kV. Its principle would have been plain to Benjamin Franklin. An endless belt of a good insulating material running vertically between two pulleys picks up electricity from a point discharge at the bottom and delivers it, again by point discharge, to a large insulated spherical conductor at the top (fig. 2.7). The lower electrical spray comes from any rectified source. The upper spray continues, irrespective of the potential attained by the sphere, which exerts no electrostatic force at its internal surface, until the field at the external surface suffices to break down the air. In a later refinement (fig. 2.8), a second set of points adroitly placed removed electricity of the unwanted sign from the sphere on the belt's downward journey and eliminated the rectified source by connections that allowed the amplification of any slight charge present on the belt. Since the practical limit to the potential of the sphere is the dielectric strength of the air, the way to millions of volts was to increase the sphere's radius (and so lessen its external field at a given potential) and the dielectric constant of the surrounding medium (and so raise the field at which breakdown occurs). To make good on the second possibility, the entire machine must be encased in a vessel that can be evacuated or filled with a gas or fluid under pressure. The first small pressurized model operated in 1932.[25]
By mid August 1931 Van de Graaff could charge a brass sphere mounted on a glass stick to about 750 kV. Between two such spheres, one positive and one negative, an inspiring potential difference of 1.5 MV could be maintained. It, and the trivial cost,
Fig. 2.7
Principle of the Van de Graaff generator. Charge
sprayed on the endless silk belt at the bottom leaves
by corona discharge at the top; it is derived in the
first instance from a transformer. Van de Graaff,
Compton, and Van Atta, PR, 43 (1933), 152.
about $100 for the entire outfit, inspired Karl Compton, who brought Van de Graaff to MIT as a research associate (he became associate professor in 1934) and arranged some heady publicity. The newly formed American Institute of Physics held a dinner for scientists and journalists at the New York Athletic Club; the machine, in an alcove in the dining room, looked like "two identical rather large floor lamps of modernistic design." Van de Graaff demonstrated for his supper, and also for Paramount and Pathé news; he allowed that he saw no difficulty going to 10 MV with two balls each 20 feet in diameter on towers 20 feet tall. Compton misguessed that this big machine would provide alpha particles in a current "so enormously larger than that from radium, that the experiment opens up the possibility of transmutation of the elements on a commercial scale;" and he miscalculated that it could be done for a few hundred dollars.[26]
Fig. 2.8
An improved Van de Graaff generator. The points are
arranged so that the belt charges the sphere when
going down as well as when going up; the system
works with any stray charge, no transformer being
required. Van de Graaff, Compton, and van Atta,
PR, 43 (1933), 153.
There remained what Lawrence, who recognized Van de Graaff as a competitor, called the "old problem of a high vacuum tube." Those who thought they had solved the problem regarded the matter differently. As one of Lauritsen's students wrote him after witnessing one of Van de Graaff's demonstrations, "His scheme is really very good and actually works . . . [and] would make a very fine combination with one of your tubes." Lauritsen eventually did build a Van de Graaff machine.[27] Tuve's group rushed to do so. In September 1931 Tuve drove Van de Graaff and his easily portable equipment to Washington and hooked it up to the segmented tube. With a charging current of 40 µA and 600 kV on the spheres hooked in parallel (plate 2.3), the tube carried a
proton beam of not quite a millimicroamp (mµA), not enough to burn a hole in cardboard, but enough to make tracks in a cloud chamber. Tuve studied the working apparatus closely and got a spark to his nose for his curiosity. It did not discourage him from reaching higher, for 1.4 MV, half the theoretical value of the maximum potential restricted by the dielectric strength of normal air surrounding a sphere two meters in diameter. (A useful rule of thumb: the theoretical maximum in MV equals the radius in feet.) Simultaneously, Coolidge, supported by GE, planned a high-current version at 1 MV, and Van de Graaff, seconded by the Research Corporation, went forward with one requiring two 15-foot spheres.[28]
The two-meter sphere, which cost $700, worked well with the million-volt tube insofar as it could be tested outdoors, where it sparked and fluttered under bombardment by bugs and dust and threw lightning bolts that reduced its redwood base to splinters. The Carnegie Institution's Department of Terrestrial Magnetism had no place for the wonder it had built. We read in its annual report for 1931/32: "A highly satisfactory equipment for the production of high energy particles, particularly of high-speed protons, is thus ready for use as soon as operating space becomes available."[29] During the late fall of 1932, while awaiting the construction of suitable housing off-site (the Carnegie Institution's executive committee approved a building fund in January 1933), Tuve and his associates made a version about one meter in diameter for the space that had belonged to the Tesla coil. With this machine, their desire to do nuclear physics was at last requited. It consisted of two hollow hemispheres of aluminum joined by a short cylindrical section containing the belt, one pulley, an ion source, and the high potential end of a segmental discharge tube. The belt brought about 180 to 200 µA; the sphere held 400 to 600 kV; the tube transmitted as much as 10 µA of proton beam, constant in energy to perhaps 3 percent, to the target. The x rays
incidentally produced drove the experimenters into a hut outside the twelve-inch concrete walls of their laboratory, where they worked by remote control. (Their conservative value for the safe tolerance of radiation was ten times the dose from cosmic rays.) In the operation of the tube, Tuve's group had some advice from Lawrence, whose own investigations had by then acquainted him with the ability of coaxial cylindrical electrodes to focus a beam and with the excellences of certain "Apiezon" oils made by Metropolitan-Vickers as the working fluid of vacuum pumps. With this setup the Department of Terrestrial Magnetism was able to set right some sloppy results reported by Lawrence's Radiation Laboratory.[30]
In 1933 the two-meter machine found a home and Van de Graaff's 15-foot giant threw its first sparks in a disused blimp hanger in Round Hill, Massachusetts. Tuve's photogenic apparatus (plate 2.4), with four belts and two concentric shells (the inner, one meter in diameter), reached 1.2 MV under favorable conditions. Lawrence visited it and was impressed. "I must say that Tuve's apparatus is performing better than I expected," he wrote the Research Corporation after his inspection. "Seeing Tuve's apparatus perform makes me much more enthusiastic about van de Gr[a]aff's outfit than I was before."[31] By then, November 1933, the cyclotron could give more volts, but at far less current, than Tuve's "outfit." As for Van de Graaff, his 1.5 MV model gave a charging current almost a million times Lawrence's beam, as his patron Compton liked to observe, and his giant one held promise of another factor of ten (plate 2.5 and fig. 2.9).
"Experience to date indicates that there is in sight no unsurmountable obstacle to the construction of [10 MV Van de Graaff] generators." When Compton spoke these words—which were realized long after the war, by a technology not available in the
Fig. 2.9
An insider's view of the 15-foot generator. It delivered 1.1 mA to the accelerating
tube under a tension of 5.1 MV. Van Atta et al., PR, 49 (1936), 762.
1930s—Van de Graaff's original model was showing off at Chicago's Century of Progress Exposition, "producing millions of volts for the enlightenment of the visitors."[32] But neither this nor any other million-volt plant was the first to accomplish the purpose of all, and bring that enlightenment to physicists vouchsafed by the disintegration of the atom.
The English Way
Like Van de Graaff and Lauritsen, John Douglas Cockcroft began professional life as an engineer, in his case in 1920, with a degree from the University of Manchester. He spent the next two years as a college apprentice at Metropolitan-Vickers, working with large transformers and strong insulators. And then, like his counterparts, he changed his field and place of study, to mathematics and Cambridge. In his second year there he began to frequent the Cavendish Laboratory, where he did postgraduate work after gaining a second bachelor's degree in 1924. Metro-Vick continued to give him partial support, on the understanding that he would do some research work for them; and, reciprocally, he served the Cavendish as a "spare-time, honorary electrical engineer."[33]
In 1926 Cockcroft's experimental space was invaded by another man from Metro-Vick, T.E. Allibone, who had been inspired by Blackett's demonstration of disintegration and Coolidge's design for high tension to try his hand at artificial sources. With the advice of colleagues at the High Voltage Laboratory at Metro-Vick, he chose the Tesla coil; and it appears to have been the progress he had made with it during 1926/27 that prompted Rutherford to speak optimistically of the prospects of artificial sources before the Royal Society. Another would-be atom splitter then arrived, Ernest Walton, who came from Trinity College, Dublin, with a degree in mathematics. Walton tried two methods of a type that will occupy us presently. Both failed. Nor did Allibone
fulfill Rutherford's hopes, although he did succeed, with the help of a transformer loaned by Metro-Vick, in developing tubes that could stand 450 kV in air and 600 kV under oil. By then, the end of 1928, Cockcroft had finished the research on molecular beams that constituted his doctoral work. Notwithstanding the failures with which his room was strewn, he decided to try to shatter nuclei himself.[34]
Cockcroft was a businesslike man. He took up atom splitting because he knew it to be practicable. In contrast to all the other would-be splitters, he kept his attention fixed upon the goal: to make particles with energies sufficient to penetrate nuclei, not with energies above a million volts. In estimating the minimum requirement, he followed calculations in a manuscript that circulated in the Cavendish in December 1928. Its author, George Gamow, a young Russian working in Niels Bohr's Institute of Theoretical Physics in Copenhagen, explained that, according to the then new wave mechanics, a charged particle making a head-on collision with a nucleus has a chance of entering even if it does not have as much energy as would be required to do so by the older physics, on which the estimate of millions of volts had been based. Cockcroft deduced from Gamow's equations that protons would be better agents than alpha particles and that a proton of 300 kV would be about one-thirtieth as efficient against boron (in fact, beryllium) as an alpha particle from polonium would be against aluminum. In January 1929 Gamow came to Cambridge to talk, Rutherford approved Cockcroft's project, and Cockcroft and Walton teamed up to produce protons at 300 kV in sufficient quantity to overcome the low probability that any of them would effect a disintegration.[35]
Cockcroft chose 300 kV as his first goal because he judged it to be within easy reach of the art of vacuum tubes. The arrangement he and Walton devised is shown in plate 2.6. Metro-Vick provided much of the technology: the 350 kV transformer, shown
schematically at the left, a model custom-made (but later marketed for x-ray plants) to fit the cramped space of the Cavendish; the rectifiers A , designed by Allibone; the pumps, constantly at work on the rectifiers and discharge tube, invented by Metro-Vick's C.R. Brush and run on his Apiezon oil of miraculously low vapor pressure. F is a 60 kV transformer that energizes the little canal ray tube (atop the discharge tube) that Cockcroft and Walton used as a source of protons; the entire transformer F stands at 300 kV above ground. The discharge tube itself, a bulb with two steel pipes as electrodes, terminated in a small experimental space and a hookup to the vacuum system. About 1 µA of 280 kV protons survived the trip down the tube to slam into targets of lead or beryllium. "Very definite indications of a radiation of a non-homogeneous type were found," Cockcroft and Walton wrote in August of 1930, without saying what they indicated.[36] Very probably they had disintegrated beryllium. Not knowing what evidence to look for—they expected to find gamma rays—they concluded that Gamow was mistaken and sought higher potentials.[37]
A move to a larger room with a higher ceiling allowed them to build new apparatus to a plan of Cockcroft's. The design multiplies voltages by an intricate set of condensers and switches. In the setup of figure 2.10, where all condensers have equal capacities, the action begins with the closing of the dotted switches, S1 , S2 , S3 . That links K3 and X2 in parallel, at the potential E of the constant source. Next the dotted switches open and the solid ones close, causing X2 to divide its charge with K2 and bringing each to E /2. The switches are again reversed, leaving K3 and X2 at E , K2 and X1 at E /4. Reverse connections again: we have K3 at E , K2 and X2 at 5E /8, K1 and X1 at E /8. And so on until the upper plate of K1 is 3E above ground. In practice Cockcroft and Walton used a low-frequency alternating current for E and rectifying diodes as switches; since they could now make rectifiers for 400 kV, and since each rectifier had to stand twice the voltage E , they used a 200 kV transformer as source, four rectifiers, and four con-
Fig. 2.10
Principle of the voltage multiplier
constructed at the Cavendish. Cockcroft
and Walton, PRS, A136 (1932), 620.
densers, in the hope of attaining a final drop of 800 kV. The experimental tube (fig. 2.11) came in two segments; the middle electrode, maintained at half the total potential, carried a little diaphragm, G, to stop stray electrons. The successful proton beam, 10µA at 710 kV, passed from the evacuated tube through a window of mica into the experimental space.[38] What happened there started an era in nuclear physics.
Metro-Vick continued to play a part. The firm patented Cockcroft's voltage multiplier, although, as it turned out, a German, Heinrich Greinacher, had anticipated him, and only the British observed his rights. Cockcroft's circuit could power a high-potential x-ray plant, and in hard x rays, as we know, there was hard cash. What Metro-Vick had in mind in patenting the voltage multiplier appears from a request from George McKerrud, lieutenant to A.P.M. Fleming, the company's director of research, to Cockcroft, to entertain F.L. Hopwood, a member of the Grand Council of the British Empire Cancer Committee. "Will you please see Allibone and fix up to have a really good show going,
Fig. 2.11
Accelerating tube and target arrangement of
the Cockcroft-Walton machine. The source
is at D; C is a metallic ring joint between the
two sections of the constantly pumped tube.
The mica window closes the evacuated
space. Cockcroft and Walton,
PRS, A136 (1932), 626.
because . . . the Cancer Research people . . . are, as you know, a very rich organisation so that we hope that they will contribute towards the support of the work of the future and probably order some tubes to be made." Hopwood left Cambridge convinced that
Metro-Vick could produce "the super x ray tube." McKerrud congratulated Cockcroft on his successful soft sell. "The lunch at St John's was an invaluable detail in the scheme." Hopwood's institution, St Bartholomew's Hospital in London, commissioned Metro-Vick to make it a million-volt x-ray plant based on Cockcroft's patented voltage multiplier.[39]
3—
Magnum Per Parva
The high-tension accelerators stretched the power of insulators and the nerves of physicists to the breaking point. They also taxed the finances and furnishings of the few institutes in which they were developed: they demanded unusual electrical service, special apparatus and safety precautions, and, above all, space, a commodity more precious even than money in the laboratories of the time. An obvious way to relieve the tension on men, material, and money was to accelerate particles in several steps, each requiring only a moderate electrical force. The high-voltage energy would be accumulated on the particles, not on the apparatus.
The idea, we say, was obvious. It occurred to several, probably to many, physicists and electrical engineers during the 1920s. But between theory and its realization stood the usual malevolence of the inanimate and the conservatism of persons who are not inventors. Three examples will illustrate the variety of problems encountered in realizing the principle of acceleration by steps: the proto-betatron, the proto-linac, and the cyclotron.
The Electrical Vortex
The beam of charged particles undergoing acceleration constitutes a current that can be considered the secondary ciruit of a transformer. Recall the watchword of the faith of the electrodynamicist: F = evH / c , where F signifies the force exerted on a particle carrying a charge e and moving with velocity v by a
magnetic field of strength H , and c stands for the speed of light.[40] For the formula to hold, H must be perpendicular to the plane defined by v and r . Since F stands at right angles to v , it can neither speed up nor retard the motion of the particle; instead it pushes it constantly toward a fixed point, forcing the particle to describe the arc of a circle. The situation appears in figure 2.12. The greater the velocity v , the larger the radius r of the circle must be so that the tendency of the particle to fly off at a tangent can be countered by an inward magnetic push. The balance of tendency (mv2/r ) and push (evH / c ) gives the most important relation we shall have to consider, the "cyclotron equation,"
where m is the mass of the particle.
Fig. 2.12
The force on an electron circulating in a
magnetic field. The field H is perpendicular to
the plane of the paper; the force ( ev / c )H is
perpendicular to the velocity v and directed
toward the orbit's center.
In October 1927, an electrical engineer named Rolf Wideröe presented his doctoral dissertation at the Technische Hochschule, Aachen. In it he described experiments based on the cyclotron equation and the transformer principle. By increasing H in time, Wideröe developed an electric force in the direction of v (the transformer principle); and by constructing specially shaped poles for the electromagnet producing H , he made it possible for particles to remain in circular orbit with constant radius (the cyclotron equation). As he observed, to obtain the necessary centripetal and tangential forces from the same electromagnet, its poles must provide a field that is half as strong at the electrons' orbit as within it—a condition easier to state than to realize.[41] Wideröe called this device a beam transformer; it can be applied practically only to electrons , which, because of their small mass, can attain useful energies under the electric force produced by the time change in H. Evidently the entire acceleration must be accomplished during that part of the cycle of the alternating current energizing the electromagnet in which the electromagnetic force acts in the desired sense.
That is the theory. In practice even the partial cycle was more than Wideröe could use, since he did not succeed in holding his electrons to circular orbits within the evacuated glass doughnut in which he tried to accelerate them. The best he could do, even with several extra coils arranged to compensate for unevenness in the H field, was to guide the electrons around a circuit and a half before they ran into the walls of their doughnut.[42] He thus confirmed the prediction of his former professor, Wolfgang Gaede, an expert in vacuum technology, who had refused to allow the project in his institute at the Technische Hochschule in Karlsruhe on the ground that it was sure to fail. That had driven Wideröe to Aachen, to the more optimistic Walter Rogowski, an expert on cathode-ray tubes and oscilloscopes. As Wideröe later explained
his failure to implement his idea, which had come to him as early as 1922, "The theory of the stabilizing forces acting on the orbit had not yet been developed sufficiently."[43]
No more did the electrical vortex succeed in Cambridge, where Walton tried to implement Rutherford's suggestion of acceleration in the "electrodeless discharge." Here electrons in an evacuated tube without electrodes serve as carriers of the secondary current, the primary being generated in a coil wrapped around it. A supplementary magnet supplied the additional H field, which, according to Walton's detailed and correct calculations, would hold the electrons to a tight circular path within the tube. Walton charged a condenser to 40 kV and discharged it through the coil, setting up oscillations that produced the required varying magnetic field. If all went as calculated, the electrons would attain an energy of 536 keV in a quarter of a cycle. They declined to be regulated, ran into gas molecules, and scattered into the tube wall. Walton gave them up and joined Cockroft in multiplying voltages.[44]
The idea was so good and so obvious that several other physicists flirted with it before Cockcroft and Walton's success with straight tubes and high voltages in 1932. The friendly rivals from South Dakota each tried his hand. Working with Breit, Tuve set up an apparatus similar to Wideröe's, but independently of Wideröe. Tuve and Breit made an important improvement by injecting the electrons into their accelerator at high speeds via an electron gun and claimed to have obtained an acceleration to 1.5 MeV, but they nevertheless ended no more successfully than Wideröe; "No provision has been made [they wrote] to repeat the process very often." Lawrence thought that he could correct the fault of Walton's design, which did not make sufficient provision for axial focusing, with extra coils to create a field that would drive errant electrons back to their orbital plane. He had his assistants realize his design. On June 10, 1931, he tried it, with no better luck than his predecessors. The same success would have attended the efforts of Leo Szilard, had he attempted to make flesh the transformer-accelerator for which he applied for a
German patent in January 1929.[45] There were other patents as well. One of them, secured by Max Steenbeck, a physicist at the laboratories of the Siemens electrical company in Berlin, perhaps referred to a machine that worked.[46]
The method can be made to work. After a decade's experience in focusing beams of fast particles in other sorts of accelerators and in electron microscopes, physicists managed to construct a magnetic system capable of steering and speeding electrons on circular orbits within evacuated tubes. Szilard proposed to have a try in 1938, in collaboration with a physicist at the Clarendon Laboratory at Oxford, James Tuck; their unimplemented design was perhaps the most promising put forward before the first conspicuous success. That was achieved in 1939 and, on a larger scale, in 1940 by Donald Kerst at the University of Illinois.[47] He had the assistance of a former student of Oppenheimer's, Robert Serber, in computing the motions of the electrons. The business continued at General Electric's research laboratory, where a machine for 22 MeV and then for 100 MeV came into existence during the war. Just after the war, Kerst's "betatron" had an important influence on machines built in Berkeley.[48]
Resonance Acceleration
Wideröe had a second failure to announce in completion of his thesis. This one involved the acceleration of heavy ions by modification of a technique proposed, but not implemented, by a Swedish physicist, Gustav Ising. In Ising's plan, positively charged particles fly down a straight evacuated glass tube through a series of hollow cylindrical electrodes, each separately connected to one pole of a spark gap in an oscillatory circuit (fig. 2.13). The particles are drawn from the source into the first electrode when
Fig. 2.13
Ising's proposal for a linear particle accelerator. The high-frequency field is
supplied by a discharge across the spark gap F; K is the cathode; a1 , a2 , a3 ,
connections to the drift tubes. Ising, Kosmos, 11 (1933), 171.
its potential has the correct sign and size; while in it they feel no electrical force; on escaping from it they are again accelerated, the direction of the field in the gap between the first and second electrodes having meanwhile altered to the correct sense. The business in principle can be continued through any number of gaps and electrodes.[49]
Wideröe turned to Ising's method after failure of his protobetatron in the hope of having something that worked to describe in his thesis. He replaced the old-fashioned spark vibrator with an up-to-date vacuum-tube oscillator, and theory with practice. He contented himself with only two accelerating steps, which did indeed give the ions twice the energy they would have gained from falling through the maximum potential across the oscillator.
Nevertheless Wideröe regarded his process as unpromising. He was not interested in particle accelerators, but in what he called "kinetic-voltage transformers," devices for generating currents of high-energy particles useful to engineers. He got currents so small, however, that he despaired of realizing anything much above a milliamp, which ruled out his invention as "a technical generator for high direct voltages."[50] But so puny a flow, of no consequence to engineers, was enough and more than enough for physicists, for, as we know, a milliamp of the right sort of particles exceeds the output of alpha rays from eighty kilograms of radium.
In Wideröe's linear accelerator (fig. 2.14), or "linac" to use the later term of art, sodium or potassium ions from the heated filament at A fall through 20 kV across the gap I when the oscillating potential Ub reaches its maximum negative value on the metal tube BR. In the time the ions float through the tube, which shields them from electrical forces, the potential on it switches to maximum positive value, and the emerging ions drop through 20 kV across gap II. On emerging from the grounded tube S they move under an electrostatic force between the plates at K to strike the fluorescing screen P a distance a below the axis. From measurement of a , Wideröe could confirm that the ions at P had energies of 40 keV. Much higher potentials could be reached, he thought, by multiplying the number of steps, increasing the voltage Ub , and using heavier ions.
This last consideration, which is perhaps not obvious, played an important role in the early history of accelerators. Wideröe's primary improvement over Ising, using a vacuum-tube oscillator to generate the high-frequency potential Ub (the spark gap in fig. 2.14 is for calibration, not generation), avoided the wide band and inconstant energy of the spark circuit, but brought a limitation of its own. It could not operate effectively above a frequency of 107 Hz, or about ten million oscillations a second. It was this limit that directed Wideröe's attention to heavy ions rather than to protons. A singly charged ion of mass AmH (mH is the mass of the hydrogen atom) has a velocity of about 
cm/sec at 1 MeV.
Fig. 2.14
Wideröe's linac for acceleration of heavy ions. The high-frequency field is
supplied by the circuit containing the tube SR. Livingston, Development , 102.
In the last steps of its acceleration it would travel 
cm during one period at 107 Hz, near the frequency limit of the oscillators then available. To accommodate protons (A = 1), Wideröe would have needed an evacuated acceleration tube many meters in length filled with electrical equipment and maintained at a pressure of less than a millionth of an atmosphere (10–7 mm Hg). These specifications were technically impractical. For cesium ions, which Wideröe proposed as particularly favorable for attaining 1 MeV, the apparatus would be about a meter in length.
Wideröe's thesis was published in the issue of the Arkiv für Elektrotechnik of December 17, 1928. On the very same day—the resonance is scarcely credible—Szilard applied for a patent on a similar device at the Reichspatentamt in Berlin. In contrast to the engineer, the physicist explicitly intended his machine to accelerate particles that might disintegrate atoms; as Szilard later explained himself, he rightly saw that the nucleus was the frontier in physics and wrongly guessed that its disintegration would lead quickly to "practical application of very great importance." But he was also busy preparing an application for a British patent on a scheme of refrigeration that he and Einstein had invented, and he
devoted little care to the practical details of his linac. Its principle appears from figure 2.15: ions accelerated between the anode 13 and the perforated cathode 11 enter the accelerator tube 9 sectioned by the grids 1, 2, . . . 6. At S is an oscillator, one of whose leads supplies the odd-numbered grids and the other the even-numbered, which are therefore 180° out of phase with one another. In the ideal case, an ion passes through a grid only when all the grids instantaneously reach zero potential and when the potential is rising on the grid just passed and falling on the grid next in line. The ions are not shielded from the ac field by metal cylinders, as in Wideroë's version, and no doubt the wires in the grids would thin the beam to uselessness long before it had passed the 100 grids that Szilard specified for acceleration to 2 MeV. He planned to work at a hundred million cycles per second, but did not say where he would procure an oscillator. The patent for the linac was not granted.[51]
Fig. 2.15
Szilard's linac. The rf oscillator S is at the far right; the source is roughly
indicated by the circuit 11, 12, 13. Szilard, CW, 1 , 553.
Another inventor, perhaps independent, of the Wideröe linac was Jean Thibaud, a young physicist who worked with classic radioactive substances in Maurice de Broglie's private laboratory in Paris. Impatient with what nature and radiochemistry provided, Thibaud hoped to reach 10 MeV in small steps; he managed to get to 145 keV by pushing positive ions through no fewer than 11 successive gaps between cylindrical electrodes with an oscillator going at 3·106 Hz. To proceed to alpha-particle energies, however, to compass "the liberation of intra-nuclear energy, an outcome of undoubted general utility," Thibaud recognized that he would need a tube ten meters long and a very powerful and dangerous oscillator, capable of handling 100 kW. "To avoid these difficulties," he said, "I have tried another method."[52] It was the method of the cyclotron. Thibaud did not disclaim credit for its invention or for making it operate independently of Lawrence.[53]
The cyclotron solved the problem of tube length and made possible the resonant acceleration of protons and other light particles. In it the ions spiral out from the center of an evacuated shallow drum, suffering a kick each time they cross a gap between electrodes arranged along a diameter of the drum (fig. 2.16). Between kicks the particles are maintained in circular orbit by a magnetic field. Several people say that they thought of some such scheme before Lawrence did, but for various reasons did not publish it. Wideröe recalled that one of Rogowski's assistants asked whether the ion beam could not be curled around and made to spiral repeatedly through the same electrode gap. "I answered him, as I remember quite well, that I thought that it might be possible but it seemed relatively impracticable to me and I thought the possibility of hitting the gap again could not be very great."[54] Denis Gabor, later a Nobel prizewinner for something else, thought of the
Fig. 2.16
Simplified schematic of a cyclotron, showing the spiral path
of an accelerated ion. Livingston and Blewett, fig. 6.4.
cyclotron as early as 1924, or so he said. But he had other things to do. "My Dr.-Ing. study [on oscilloscopes like Rogowski's] was already in progress, and I let it go." He had another opportunity a few years later. On January 5, 1929, the irrepressible Szilard filed for a patent on, yes, a cyclotron, which he constructed, on paper as usual, by bending the accelerator tube of his linac into a drum sectored by grids. The accelerated particles were cycled back through the grids by a magnetic field. Szilard recalled late in life that he had offered the exploitation of his inventions to Gabor, who then served the Siemens electrical empire, but neither Gabor nor Siemens appears to have seized the opportunity.[55]
The most circumstantial account by a near inventor we owe to Max Steenbeck, who finished up an adventurous career in physics as head of the board for scientific research (Forschungsrat) of the German Democratic Republic. When a doctoral student at the University of Kiel in 1927, Steenbeck conceived the idea of a cyclotron while trying to illustrate certain properties of electrical forces to a particularly stupid student. He worked out a numerical example—acceleration of protons to MeV energies in an orbit of
20 cm under a magnetic field of 14,000 gauss—exactly the experiment done four years later in Berkeley. It was not tried in Kiel. An assistant to Walther Kossel, one of Steenbeck's professors, said it would be too costly; the other professor, Hans Geiger, said that it would be too risky. Before finishing his degree, Steenbeck went to work for Siemens in Berlin. He told his co-workers what he had in mind. They insisted that he publish the idea, no doubt as a precaution for possible future patent applications. Steenbeck agreed, but reluctantly, since the calculation was so elementary. His chief returned his manuscript with a note that he wished to speak further about it. Steenbeck interpreted the note as dissuasion (the chief had probably intended only to request a patent search) and, following his inclination, dropped the matter. Siemens lost the cyclotron again.[56]
These several anticipations of practical resonance accelerators do not detract from, but rather enhance, the achievement of Ernest Lawrence. As Szilard said of Lawrence's success, "The merit lies in the carrying out and not in the thinking out of the experiment."[57] Lawrence came across Wideröe's paper around or before April 1, 1929, while glancing through current journals. (The relevant issue of the Archiv für Elektrotechnik reached the University library on January 23, 1929.) This inspirational encounter may seem, and is said to have been, a matter of chance. The library had just opened a subscription to the Archiv , Wideröe's issue being only the fourth received; the Archiv catered primarily to "engineers working scientifically at electrotechnology" and to physicists only in so far as they took an interest in technical problems; and the Archiv was written in German, a language that Lawrence did not understand. The historian, who seeks the causes of things, prefers to think that Lawrence sought the journal for a definite purpose. Lawrence and Beams had cited an article in the Archiv by Rogowski and others in their paper on Kerr cells; similar references later appeared in theses by Lawrence's students and in a paper he wrote with one of them.[58] It may well be that Lawrence
asked the library to subscribe to the Archiv . He later told Wideröe that he had picked up the issue with the fateful article to pass the time at a boring meeting. Having this succor to hand was not a matter of chance.[59]
Lawrence later gave another explanation. He said that in 1928, when he moved up, and far, from assistant professor at Yale to associate professor at Berkeley, he decided to shift his attention from the effete and played-out photoeffect to the robust new field of nuclear physics. His interest in apparatus drew him to the problem of high-voltage machines. He was already familiar with Tuve's struggle with the Tesla coil and had decided that a million volts could be reached directly only with great difficulty and large apparatus. He would have looked through the Archiv for exactly what he found there, a low-potential way to high energy. Lawrence would soon be building machines rated by his contemporaries as gigantic; but he recommended his first magnetic resonance accelerator as needing only "relatively modest laboratory equipment."[60]
It is sometimes said that Lawrence's cyclotron is Wideröe's two-step bent into a circle. The proper analogy to Wideröe's apparatus, however, is a single ring, within which the ions circulate under a magnetic field H (which Wideröe did not require) as they accelerate from periodic knocks, from the high-frequency field F . To shield the ions between knocks, each of the electrodes has to fill almost half the ring. As the ions speed up, they come to the accelerating gaps between electrodes at shorter and shorter intervals; to remain in step with them, the frequency of the field must increase. Similarly, H must grow to keep the ions on their circle. Such a system was then not practicable. It became the dominant method after the war when implemented in, among other machines, the Bevatron, in the sixth generation of Berkeley cyclotrons.
Lawrence saw that a looser analogy to Wideröe's ion accelerator might be possible with constant control field H and oscillator frequency f . The possibility appears from the cyclotron equation (2.1): the frequency of rotation of an ion of charge ei , mass mi , in the field H is independent of the radius of the orbit :
It appears that nature conspires in favor of cyclotroneers. Taking, then, not an annular tube, but a shallow drum or cylinder, as the playground of the ions, Lawrence worked out, on paper only, the scheme indicated in figure 2.17. The iconographer might consider it a combination of Wideröe's circle for electron acceleration with his mechanism for kinetic voltage transformation, a creative misunderstanding abetted and even made possible by Lawrence's inability to read Wideröe's text. "I merely looked at the diagrams and photographs," Lawrence said, in accepting the Nobel prize for his invention.[61]
In the figure, A and B represent hollow half-cylinders of metal, which are alternately charged to a maximum of around 10 kV by a radio oscillator. The close connection with Wideröe appears in, among more obvious features, Lawrence's retention of the word "tube," which Wideröe used appropriately for his straight glass accelerating vessel, for the cylindrical, metallic tank of the cyclotron.[62] "Cyclotron," perhaps Lawrence's first term for his invention, also echoed the notion of tube, in analogy to "radiotron," "thyratron," and "kenotron." (In formal parlance, Lawrence insisted upon "magnetic resonance accelerator;" "cyclotron" rose from slang to the official name in 1936.)[63] To return from the name to the substance of things: Ions liberated in the gas above a filament near the center of the cyclotron enter the gap between the
Fig. 2.17
Schematic of the cyclotron. The circle shows the
meridian plane; ions enter at a, accelerate as they cross
the gap at b, c, etc., between the two dees A, B. The
inset at the top shows the vacuum chamber or "tube"
in elevation; H the magnetic field, "fil" the ion source;
the dotted lines indicate particle orbits. Lawrence
and Livingston, PR, 40 (1932), 23.
cylinders when the electric force there can pull them in the direction shown by the arrow at a . On entering A , which shields them from the radio-frequency (rf) field F , the protons revolve in a semicircle ab under H , which is perpendicular to the plane of their orbit, and reemerge at b just when F has reversed itself to pull them into B . The acceleration across b brings the ions into a wider semicircular orbit bc within the shielding electrode B , whence they emerge at C again in step with the rf field. They gradually spiral out under the guidance of H , acquiring 10 or 20 kV from F at each crossing of the gap, and so reach the circumference after a hundred or so turns with an energy of a million volts.[64]
Thus the theory. There were reasons to doubt the possibility of its implementation. Only an inventor could think that the beautiful synchronization could last, that the ions could be kept going for one hundred turns without colliding with other molecules or flying from the median plane into the walls of the electrodes or going astray in crossing the gap. Lawrence planned to put his hollow electrodes, or dees as they were later christened, inside a very good vacuum, which would reduce the average distance between collisions to about the length of the spiral path; but that brought the problem of maintaining vacuum seals under the stresses created by the radio-frequency field and the guiding electromagnet. As for keeping the ions in the median plane through the dees, that appeared to be, and for a time was, the greatest problem of all. But here again nature unadorned favored the cyclotroneer.
Lawrence did not move quickly to implement his plans. No doubt the difficulties just described, and the refusal of his friend Thomas Johnson of the Bartol Foundation, then teaching in Berkeley, to join him in the work, gave him pause. But he had a strong positive reason for shelving the idea. He had spent the time since his arrival at Berkeley in the late summer of 1928 building up apparatus to continue his studies of the photoeffect. He had three doctoral students, "a relative importance . . . that I could never have attained at Yale in years," and a need to get results, to justify the reputation for clever and productive experiments that he had brought, and that had brought him, West.[65] One of his students, Niels Edlefsen, confirmed a curiosity earlier studied by Lawrence, that the probability of ionization of potassium vapor increases with the frequency of the light after reaching a local maximum at the series limit. Lawrence tossed the problem of explaining the rise beyond the limit over to Oppenheimer, Berkeley's new part-time theorist; it fell to his graduate student Melba Phillips, who took the problem of photoionization in potassium vapor for her thesis. The business was too much for Oppenheimer's people too, a "pandora's box," according to its historian, "of ultraviolet photoabsorption, photoionization, and ARPES [angle-resolved photoelectron spectroscopy]."[66] Lawrence
continued the experiments and found excitement, if not hope, in the box. "The longer I am in scientific research work," he wrote his parents, "the more fascinating it becomes."[67]
In January 1930 Lawrence began to look more favorably on his "proton merry-go-round." He had had two useful stimuli, one reassurance that it might work, the other notice that he might be anticipated. The reassurance came from Otto Stern, the world's expert in the handling of beams of hydrogen atoms, then providentially visiting Berkeley from his base in Hamburg. Stern thought the spiral beam had a chance and urged Lawrence to take it. Lawrence acknowledged the debt: "Probably if you hadn't urged me so strongly I would not have started the development of the method until some time later." And some time later: "It was your enthusiasm one evening during dinner . . . that stimulated me to try the development of the method." The stimulus from competition came from Tuve and company, who in the Physical Review for January 1, 1930, announced that they could operate their Tesla coil under oil under pressure at 5 MV with 120 sparks/sec. Should they succeed in driving alpha particles along at 5 or 10 MeV and at such a rate, they said, they would have radiation equivalent to that from 2,600 grams of radium. They had found the way to high energy, and found it "without much difficulty."[68]
Lawrence did not work alone. It was one of his strengths. From the time of his arrival at Berkeley, he published only two independent research reports, in 1933 and 1935. He accordingly did not attempt to build the merry-go-round himself, but enlisted Edlefsen, who had just finished his thesis on photoionization of alkali vapors; Edlefsen had a teaching assistantship in the Physics Department and reluctantly stayed on to help make the protons spin. By the end of February 1930, work had begun. "I have started an experimental research based on a very interesting and
important idea," Lawrence wrote his parents. "If the work should pan out the way I hope it will it will be by all odds the most important thing I will have done. The project has fascinating possibilities."[69] In March he was working "night and day on . . . producing million volt protons without using high potentials. If this turns out it will of course constitute an important piece of work."[70] But it was not turning out.
During the summer of 1930 Lawrence lazed playing tennis while his graduate students, now half a dozen or more, were "making up for my laxity in activity along research lines."[71] Then a breakthrough came. Edlefsen left to become assistant professor of irrigation investigations in the University's Agricultural Experiment Station.[72] In a carefully worded statement, he and Lawrence described their method, implied that it had succeeded, and proposed parameters for an instrument capable of accelerating protons to a million volts. There is considerable doubt that Edlefsen achieved any resonance acceleration at all. But Lawrence wanted to believe and his natural optimism resonated with his enthusiasm. At the end of their statement, which Lawrence read to a meeting of the National Academy of Sciences on September 19, 1930, he let all qualifications go: "Preliminary experiments indicate that there are probably no serious difficulties in the way of obtaining protons having high enough speeds to be useful for studies of atomic nuclei."[73]
4—
Double Play
In the fall of 1930 Lawrence found what he needed to make accelerators: two graduate students, very capable, hardworking, and dependent on his good opinion. One, David Sloan, we have already met; the other, M. Stanley Livingston, who came to the Physics Department with an M.A. from Dartmouth in the fall of 1929, sought a subject for a doctoral thesis. Both had financial support: Sloan his Coffin Foundation fellowship, Livingston a teaching assistantship. Livingston took up Edlefsen's problematic cyclotron; Sloan, Wideröe's proven, but useless, ion accelerator. Their working during 1930/31 demonstrated the practicability not only of both devices, but also of Lawrence's as yet tentative method of organizing work: simultaneous development of complementary instruments or parts by graduate students or postdocs, each of whom had responsibility and considerable independence on his own project. They were inspired to outstanding work by a spirit of cooperative competitiveness and by the bittersweet satisfaction of laboring at the edge of technology. "It is rather remarkable," Lawrence wrote one of his old professors, "how physics is attracting the best much as engineering used to do."[74]
The straight way proved the quicker. Sloan and Lawrence used mercury ions in a tube with eight electrodes driven by a 75-watt oscillator, with whose robust personality Lawrence had become acquainted during his summer at GE. They immediately got the expected eightfold amplification, 90-keV ions with the oscillator operated at 11 kV, some five times its rating. They extended the tube with thirteen more electrodes and got ions of 100 keV, then of 200 keV, exactly what they expected, with less than 10 kV on the oscillator (fig. 2.18). They reported this performance to the National Academy of Sciences in December 1930 and recommended their installation for its simplicity and economy. "An entirely practicable laboratory arrangement," viz., an evacuated tube 2.3 meters long containing electrodes run at 25 keV would give mercury ions of 125 kV.[75] To what purpose? Lawrence and Sloan
suggested studies of the properties of the high-speed rays, but did not stop to undertake them themselves. Instead, Sloan built a bigger tube, with thirty electrodes (plate 2.7). By the end of May 1931 he had passed the great artificial barrier and could boast of 0.01 µA of mercury ions with energies exceeding a million volts.[76]
Fig. 2.18
Sloan's linac. Mercury ions from the source at the bottom
accelerate through the canal ray tube and then through the 21
accelerating tubes to the collector at the top. Lawrence and
Sloan, NAS, Proc., 17 (1931), 68–9.
Although the long linear accelerator had only a short life at Berkeley during the 1930s, it contributed much more to the Laboratory than the inspiring achievement of a million volts. It gave experience in working with a powerful commercial oscillator that could put as much as 90 kV on an electrode; in the definitive version of the 30-electrode tube, Sloan got ions of 1.26 MeV, or an average gain of 42 kV per electrode. It also forced attention on the problem of the synchronization and focusing of the beam. Synchronization—keeping the beam in step with the rf field at the first short electrodes—was resolved by trial-and-error adjustments, which gave important information about the fates of ions that do not enter a gap when the peak voltage spans it. Focusing turned out to be easy: it was another case where nature favors the particle accelerator. Figure 2.19 shows the lines of force between neighboring cylindrical electrodes. Besides suffering acceleration along the tube, an ion is driven toward the axis of the accelerating system during transit of the first half of a gap and away from it during transit of the second. If it enters the gap at the optimum time, near the peak voltage, it will speed up and spend less time in the second than in the first half of the gap. The net result of the minuet is motion toward the axis: the openings of the electrodes act as lenses, which inhibit ions from leaving the accelerating system.[77]
The discovery and exploitation of automatic focusing opened the possibility for ever longer linear resonant accelerators. Sloan started on a new tube, with thirty-six electrodes, each able to hold 80 kV, to be driven by two oscillators working at a wavelength of 27 m. He expected to have mercury ions at 4.5 MeV. If successful, he would go to the next mystical threshold, 10 MeV. That would require an accelerating system forty feet long fed by eight power oscillators.[78] Such a system would no longer be the convenient laboratory accelerator advertised by Sloan and Lawrence, but a big project in radio engineering. Sloan did not proceed immediately to quicker quicksilver. By November 1931 he had begun to construct an x-ray tube of novel design, which was to
Fig. 2.19
Electrostatic focusing across the dee gap. AB is the median plane; the curves are the
lines of electric force at potential difference 2 V between the dees; the errant hydrogen
ion traverses the path indicated back toward AB. Lawrence and Livingston,
PR, 40 (1932), 29.
play an important part in the technique and financing of the Laboratory.[79] Sloan finished the 36-electrode tube with the help of a graduate student, Wesley Coates, in 1932. It gave mercury ions with energies up to 2.85 MeV. Coates loosed these ions on various targets and discovered that soft x rays left the collision sites.[80] A similar investigation by another graduate student, Leo Linford, disclosed that each impacting mercury ion drove out around ten electrons, a result not uninteresting to designers of high-voltage vacuum tubes.[81]
While Sloan pushed Wideröe's technique to a million volts, Livingston struggled to reproduce the resonance experienced by Edlefsen. The earliest experiments recorded in Livingston's notebook took place on September 20, 1930. He had built a cylinder of metal 4 inches (10 cm) in diameter, about the size of Edlefsen's instrument, to serve as vacuum chamber; and he had access to a small magnet, capable of a maximum of 5,500 gauss, to control the spiralling ions. During October and November he found only feeble indications of resonance, and they came at values of the field H that depended upon the potential V of the oscillator. That violated the sound doctrine of equation 2.1. Livingston decided that his predecessor's success had been an illusion of faith.[82]
The obstacles Livingston had to overcome may best be indicated by following the path of ions that succeed in completing the spiral course. Obstacle 1. The source must give off an ion current iS strong enough that the survivors will constitute a detectable current iD at the collector (fig. 2.20). Livingston assumed that fewer than one in a thousand ions would run the course; to get an iD of a few microamps he would need an i S of a few milliamps. He did not know how to produce milliamps of protons. Lawrence wrote to the Forschungsinstitut of the Allgemeine Elektrizitätsgesellschaft, which he was told had perfected a proton source, but insufficient information returned and Livingston had to work with hydrogen molecule ions 
. These he generated in the center of the chamber by firing electrons from a radio-tube filament into the hydrogen gas that filled the chamber.[83]
Obstacle 2. The pressure P within the chamber must be small enough to make collisions between ions infrequent and large enough to accumulate an adequate i S . In practice a vacuum pump worked continuously, maintaining a pressure around 10–5 mm Hg,
Fig. 2.20
Livingston's cyclotron. Livingston, Production (his doctoral thesis), fig. 3.
at which the average distance between collisions equalled the length of the spiral course.[84]
Obstacle 3. The grid across the entrance to the dee, which Lawrence and Livingston wrongly thought necessary for shielding, must not soak up a sensible fraction of the beam; the final arrangement used parallel slits, which opposed less metal to the ions than a rectangular screen.[85]
Obstacle 4. An electrostatic force D must be established between a special plate and the dee at the periphery of the spiral to deflect the beam (if any) into the collector. Since the value of the deflecting force measures the energy of the ions, it is
important that the passage into the cup be so restricted that only ions in a narrow range of energies can be focused on it by D ; on the other hand, the passage must be wide enough to admit an iD capable of registering itself, and, ultimately, of inducing nuclear transformations.
On December 1, 1930 Livingston recorded his first success: with a newly designed detector, he had killed the dependence of H on V . "At last we seem to be getting the correct effect."[86] He then tried what values of oscillator frequency f , dee potential V , and pressure P gave the sharpest and strongest rise in iD as he brought the electromagnet through the value H = 2pfme / c calculated for resonance. With f = 2.5 ·106 and V = 300, he got a rise around H = 3300, where theory placed it. If the rise did record the presence of resonantly accelerated 
ions, they each had an energy W of 6 keV:
which, for and R = 4.8 cm, the radial distance to the entrance to the collector, is 9.6·10–9 erg or 6,000 eV. These ions had accelerated in twenty steps over a spiral path of ten complete turns.[87] That was most encouraging if true. Over the Christmas holidays Livingston borrowed a magnet capable of almost 13 kilogauss (kG), over twice the limit of 5.5 kG with which he had been working. On January 2, 1931, he got good resonance at 12.4 kG, V = 1,800 volts, for a calculated energy W of 70 keV attained in forty-six steps. With the magnet at its maximum, 12.7 kG, W = 80 kV, Livingston detected resonance at this setting with a little less than 1,000 volts on the dee, indicating eighty-two crossings of the dee gap, or forty-one entire turns. That was enough for a thesis, if not for disintegration.[88]
Although Livingston marshalled his evidence very well, it did not amount to a demonstration. The resonances came sharply at the predicted places, to be sure; but i D showed other bumps too, which often exceeded the resonance peak. Figure 2.21 gives a typical best case; but as appears from figures 2.22 and 2.23, the curve of iD against H depended sensitively on P and D , and on V as well. In the drawings D and B are resonance peaks, the former for ions that travel through a complete circle in one period of the rf field, the latter for ions that go through a half turn in a period and a half. Livingston explained A and C as consequences of background ionization, photoeffects, partially accelerated ions, scattering, and so on, and supported his special pleading by arguments that need not be repeated. What was needed was a clean measurement of W ; but to obtain the value of the deflecting potential D (= W ) that made iD a maximum required a collimation so tight
Fig. 2.21
Livingston's thesis results, 1: the ordinate is iD , the abscissa the magnetic field
H . Livingston, Production , fig. 5.
Fig. 2.22
Livingston's thesis results, 2: iD against H at several pressures in the
"vacuum" chamber. Livingston, Production , fig. 8.
Fig. 2.23
Livingston's thesis results, 3: iD against H for various values of the deflecting
potential D. Livingston, Production , fig. 11.
that the beam entering the collector became too weak to measure.[89]
The point about inferring rather than measuring the energy of the particles bothered Lawrence. "We can make them spin around alright," he wrote Swann in January, "but we have not been able to determine how many times and therefore what speeds we have been able to produce."[90] Although by April 1931, when it was time to declare results, the difficulty had not been overcome, Lawrence and Livingston had no doubt that they had demonstrated the way to high energy for light ions. Their declarations differ subtly, however, in accordance with their places, personalities, and ambitions. Livingston, concluding his thesis: "There appear to be no fundamental difficulties in the way of obtaining particles with energies of the order of magnitude of one million volt-electrons." Lawrence, concluding a talk to the American Physical Society: "There are no difficulties in producing one million volt ions in this manner."[91]
A test was already being implemented. In a move typical of his later practice, Lawrence had started on the next bigger machine as soon as experiments with the one in hand gave the first indications of success. On January 9, a week after Livingston's ions had appeared to reach 80 keV, Lawrence wrote around for advice about building a magnet with the specifications he and Edlefsen had given: R = 10 cm, H max = 15 kG. He received useful information from Kenneth Bainbridge, then at the Bartol Research Foundation, who had just built a similar magnet to practice mass-spectroscopy in Aston's manner. According to Bainbridge, materials would cost around $700.[92] Lawrence turned to the University for money, an advance, he called it, on (indeed almost over) his next year's research allowance. The University's Board of Research, still headed by Armin Leuschner, recommended the advance, which the new president of the University, Robert
Gordon Sproul, approved, setting a pattern for the early development of the Laboratory. Both had recently had the opportunity to take Lawrence's measure. The preceding semester, with the advice and prompting of G.N. Lewis, they had defeated an attempt by Northwestern University to lure him away. Lawrence had emerged from the negotiations with a full professorship, at a high salary and a low age, and with easy access to the chief financial and administrative officer of the University.[93]
Another pattern began then too: the estimated budget fell far below the true cost. Lawrence acquired the additional $500 he needed from another familiar source, the National Research Council, from which he had had a grant-in-aid for his photoelectric studies.[94] His second cyclotron still served the express purpose of bringing physicists to high energy at low cost: no unusual funding was required. Manufacture of the magnet was entrusted to Federal Telegraph, which brought it to the Physics Department on July 3, 1931, just after Lawrence had returned from the symposium on the production of high-energy particles at the American Physical Society meeting in Pasadena, where he heard Tuve describe the acceleration of protons and ions to a million volts by a Tesla coil. Tuve had depreciated that great barrier as a "moderate" voltage. Lawrence went back to Berkeley hoping to set protons spinning in less than two weeks.[95]
He did. He wrote on July 17: "The proton experiment has succeeded much beyond our fondest hopes!!! We are producing 900,000 volt protons in tremendous quantities—as much as 10–8 ampere. The method works beautifully." The published announcement followed familiar lines. "With quite ordinary laboratory facilities proton beams having great enough energies [to effect disintegration] can readily be produced." All it takes is a magnet with pole pieces nine inches in diameter, a maximum field
of 15 kG, and a few accessories, like the 500-watt short-wave power oscillator Lawrence borrowed from Federal. One gets ions of over 500 keV, an amplification of at least 100, and full conviction that a million volts are around the corner and ten million less than a dream away.[96] In August 1931 Lawrence was awooing in New Haven. He received a telegram from the secretary of the Department: "Dr Livingston has asked me to advise you that he has obtained 1,100,000 volt protons. He also suggested that I add 'Whoopee'!"[97]
There remained the problem of direct measurement. By November Lawrence and Livingston had obtained a current iD strong enough to allow the necessary collimation by applying to their instrument the very important discovery made by Sloan and Lawrence that grids worsen the focusing of the circulating beams. Lawrence alerted two main doubters of cyclotronics, Lauritsen and Tuve, to his latest accomplishment. "Recently we have put our experiments on a very sound basis by proving quantitatively by electrostatic deflection that the particles we are measuring are the expected high speed protons or H2 + ions."[98] To get high final voltages, Livingston had to put a relatively high potential on the dee. He could not go beyond an amplification of seventy-five steps. According to Lawrence, Livingston thought that he had struck the relativistic limit, where increase of particle mass with velocity destroys the synchronization expressed in equation 2.1. In fact, imperfect regularity in the controlling magnetic field was the culprit. Lawrence suggested placing small soft-iron shims shaped as in figure 2.24 where they would do the most good. "A little work with this led to the most gratifying results," he wrote Tuve soon after the new year. On January 9, 1932, Lawrence and Livingston attained protons at 1.22 MeV with only 4,000 volts on the dee, an amplification of over 300.[99]
Fig. 2.24
Lawrence's earliest designs for cyclotron shims. Lawrence, "Notebook,"
6 Jan 1932 (39/2).
It was time to write up. One knows the format. The principle of magnetic resonance acceleration had proved itself with "quite modest laboratory equipment," which functioned best with nature's cheap gridless focusing and a few shims that looked like tears. With a magnet costing $1,200 and with a 20-kW power oscillator, protons could be spun around 300 times to an energy of 1.22 MeV. To be sure the final current was small, only 0.001 µa. But by improving the source, multipying the shims, putting in two dees with 50 kV between them, and using a magnet giving 14 kG between pole pieces only 114 cm in diameter, it would be "entirely feasible" to go to twenty-five million volts.[100] The technical achievement was mainly Livingston's; the inspiration, push, and, above all, the vision of future greatness, were Lawrence's. His professorship gave him the place to stand, and Livingston gave him the lever, to move the world of physics. Compare the case of Thibaud, who, also in 1932, had made a little cyclotron, which he ran in the gap of a 10-kG magnet with 20-cm (8-inch) poles; he took as his ion source a discharge tube that fed the vacuum chamber through a narrow canal. Thibaud therefore could maintain his ion source at a pressure considerably higher than the pressure in the chamber and he obtained a proton current a thousand times larger than Livingston's. It is not clear that he caused his ions to resonate; it is certain that he had no one like Lawrence to push him along; and no further progress was made in cyclotroneering in France for several years.[101]
When this account was completed, in February 1932, Lawrence had long since sought a magnet to raise his energy by a factor of ten. It is almost superfluous to add that he had begun raising money for it the preceding July, as soon as protons resonantly accelerated to 500 keV appeared in the collector of the second cyclotron. It was "practically certain," he said then, that he could get to 10 MeV, or even to 20 MeV, if he could put his hands on a suitable magnet and oscillator. He reckoned he needed $10,000, or possibly $15,000, for the purpose. That took him out of the range of University research funds and NRC grants-in-aid. Another angel had to be found, he wrote the man he hoped would act the part, "if we are to proceed immediately towards the goal of 20,000,000 volts."[102]