Memories of a Nobel laureate

I started working on neutrino detection early in my career. I had rather broad interests in other fields of the physical sciences and worked on a number of other projects. Usually, an experimentalist develops a number of skills and applies them to solving new problems that appear interesting. I was fortunate to live at a time when there were many interesting new developments, and fundamental science was well supported.

In 1942, I received my PhD from Yale in physical chemistry, and went directly into the army as a reserve officer. After the war, I decided to look for a research position with a view to applying chemistry to studies in nuclear physics. After two years with the Monsanto Chemical Company in applied radiochemistry of interest to the Atomic Energy Commission, I was very fortunate in being able to join the newly created Brookhaven National Laboratory, which was dedicated to finding peaceful uses for the atom in all fields of basic science: chemistry, physics, biology, medicine and engineering.

I became a member of the chemistry department in 1948 and remained there until my retirement in 1984. In 1948, a scientist at Brookhaven was able to choose an independent research programme consistent with the laboratory’s effort. After reading a stimulating review paper by H R Crane (Crane 1948), I decided to begin by selecting an experiment in neutrino physics, a field of physics that was wide open to exploration, and a suitable one for applying my background in physical chemistry. So, how lucky I was to land at Brookhaven, where I was encouraged to do exactly what I wanted, and get paid for it!

First experiments

My first experiment was a study of the recoil energy of a lithium-7 nucleus resulting from the electron capture decay of beryllium-7. In a beryllium-7 decay, a single monoenergetic neutrino is emitted with an energy of 0.862 MeV, and the resulting lithium-7 nucleus should recoil with a characteristic energy of 57 eV. A measurement of this process provides evidence for the existence of the neutrino, postulated by Wolfgang Pauli in 1931. An experiment of this nature had been carried out much earlier, but the result was inconclusive. In my experiment, the energy spectrum of a recoiling lithium-7 ion from a surface deposit of beryllium-7 was measured. The energy spectrum of the recoiling lithium-7 was found to agree with that expected from the emission of a single 0.862 MeV neutrino (Davis 1952).

Later, I began working on a radiochemical experiment for detecting neutrinos using a method that was suggested by Bruno Pontecorvo in 1946 (Pontecorvo 1946). Louis Alvarez proposed carrying out the experiment at a reactor, but lost interest in the project (Alvarez 1949). Since no-one else appeared interested in attempting the chlorine-argon neutrino detection method, it seemed a natural and timely experiment for me to work on.

The Pontecorvo method makes use of the neutrino capture reaction, n + ³⁷Cl → ³⁷Ar + e^–. The reaction produces the isotope argon-37 that decays back to chlorine-37 by the inverse of the capture process, with a half-life of 35 days. In my experiment, carbon tetrachloride served as the target material. After exposing a tank of this liquid to a neutrino source for a month or two, the radioactive argon-37 atoms produced by neutrino capture were removed and counted in a small Geiger counter. The neutrino capture cross-section is extremely small. Therefore, one must use a very large volume of carbon tetrachloride. To observe the argon-37 decays, it was necessary to develop a miniature counter with a very low background counting rate. There are background effects that must be studied as well, particularly those from cosmic rays.

The technology for carrying out the experiment on a relatively large scale, using 1000 gallons of carbon tetrachloride, was developed with the Brookhaven Graphite Research Reactor as the neutrino source. That reactor did not have a high enough neutrino flux to detect with this target size, so neutrinos were not observed. Furthermore, a reactor emits antineutrinos, and the Pontecorvo reaction requires neutrinos. It was not clear in 1952, however, whether neutrinos and antineutrinos were different particles, nor was it clear how they could differ. After all, there are other instances in nature where the particle is its own antiparticle, for example the photon and the neutral kaon.

So, in 1954, I built an experiment using 1000 gallons of carbon tetrachloride in the basement of one of the Savannah River reactors, the most intense antineutrino source in the world. One can calculate the total capture rate from all fission-product antineutrinos by chlorine-37, presuming neutrinos and antineutrinos are equivalent particles. The sensitivity for detecting neutrinos and the flux at this location was sufficiently high to provide a critical test for the neutrino-antineutrino identity. However, the experiment failed to observe a clear signal from reactor neutrinos. The Savannah experiment demonstrated that the argon-37 production rate was a factor of five below the rate expected if neutrinos and the antineutrinos were identical particles (Davis 1957).

Discovery

While I was at Savannah River doing these experiments, Fred Reines, Clyde Cowan and their associates were performing a beautiful experiment, the first detection of a free antineutrino (Reines and Cowan 1956). Their experiment was a clear demonstration that the neutrino postulated by Pauli was indeed a real particle. They observed antineutrinos being captured by a hydrogen nucleus (proton) producing a positron and a neutron. The measured cross-section was consistent with that expected from Fermi’s theory of b-decay.

In 1957, T D Lee and C N Yang suggested that the neutrino was a two-component particle whose interactions violated the principle of parity conservation. This concept was confirmed in a beta-decay experiment by Chien-Shiung Wu of Columbia and her associates at the National Bureau of Standards. The two-component theory postulates that all neutrinos have spins rotating in a left-handed helicity with respect to their direction of motion. Right-handed antineutrinos will not interact with the chlorine-37 nucleus and produce argon-37 because they have the wrong helicity. The Reines and Cowan experiment should and did observe reactor antineutrinos. All was explained. But this was not the end of the matter.

Pauli and many others did not believe all was as it seemed, and urged that another and more sensitive chlorine-37/argon-37 experiment be performed. To complicate matters, many physicists measured the spins of electrons from many beta-decay sources. They found that the electrons were not necessarily polarized as expected. After a couple of years of improved experiments, these polarization studies ultimately found that the two-component theory was correct.

Don Harmer from Georgia Tech and I set about building a three-times larger chlorine-37/argon-37 experiment. After several years, we obtained a greatly improved result. We found that the argon-37 production rate was a factor of 20 below the expected rate for neutrinos and antineutrinos being identical particles. In this experiment, our sensitivity was limited by the production of argon-37 in our liquid by cosmic rays.

Textbreak=After the Savannah River experiments were terminated, I started thinking about an experiment to measure neutrinos from the Sun. The first step in the plan was to set up one of our detectors as a pilot experiment in a deep mine to measure the background effects and determine the ultimate sensitivity for observing solar neutrinos. The measurements of argon-37 activity could be made more sensitive and specific by using proportional counters. The Sun emits only neutrinos, so the chlorine-argon method was the simplest means of studying solar neutrinos. In 1959, we located a mine near Akron, Ohio, to begin these studies.

Observing the neutrinos from the Sun had the potential of testing the theory that the hydrogen-helium thermal fusion reactions are the source of solar energy. However, the proton-proton chain of reactions in the 1950s was regarded as the principal source of the Sun’s energy, and this chain emitted only low-energy neutrinos from the primary proton-proton reaction. These neutrinos were below the energy threshold of the chlorine-argon reaction. We were saved from this impossible situation, however, by a new development.

In 1958, two nuclear physicists, H D Holmgren and R I Johnston at the Naval Research Laboratory measured the ³He + ⁴He → ⁷Be + g reaction, and found it had a higher than expected cross-section. It was immediately recognized that this reaction could be important in the terminal stages of the proton-proton cycle. Furthermore, the beryllium-7 could react with a proton and become boron-8. These two radioactive products, beryllium-7 and boron-8, would be the source of energetic neutrinos, ones that could be measured by the chlorine-argon radiochemical method. W A Fowler and A G W Cameron immediately relayed to me these developments. They pointed out that the neutrino flux from these neutrino sources could perhaps be easily observed by the chlorine-argon detector! I might add that Fred Reines was also stimulated by the new findings and immediately embarked on a programme of solar neutrino research (Reines 1967). There was also a very active programme in the Soviet Union in solar neutrino research under M A Markov and G T Zatsepin.

Homestake

These events ultimately led Brookhaven, with support from the chemistry office of the Atomic Energy Commission, to build a 100 000 gallon chlorine-argon neutrino detector in the Homestake Gold Mine at Lead, South Dakota. The scale of the experiment was determined by a theoretical estimate of the expected neutrino flux and the neutrino capture cross-sections for each solar neutrino source in the Sun. It was necessary to measure the production cross-sections of the neutrino-producing reactions, and derive their rates in the interior of the Sun. The aim was to forecast as accurately as possible the rate of solar neutrino capture in the Homestake detector.

This great effort was largely carried out at the California Institute of Technology under the leadership of Fowler. During this period, John Bahcall calculated the neutrino capture cross-section to produce argon-37 in excited states (Bahcall 1964). Of particular interest was the analog state in argon-37 and a means of calculating other excited states by studying the decay of calcium-37. The effect of these states was to greatly increase the expected neutrino capture rate of the energetic boron-8 neutrinos with energies extending to 15 MeV. Bahcall and I wrote an account of these activities in a volume of papers dedicated to Fowler. As outlined in the article, very many nuclear physicists and astronomers contributed to the basic physics that supported this early effort in solar neutrino astronomy. Our task at Brookhaven was far simpler, and we (Don Harmer, Kenneth Hoffman and myself) had the fun of building a large detector and making it work.

We were very fortunate that the Homestake Mining Company accepted our project. They assisted us in many ways with the building and operation of the experiment. The great depth of the experiment (4850 feet or 1560 m) turned out to be a crucial element.

The observed neutrino capture rate was much lower than was anticipated from the solar model calculations (Davis et al. 1968). In fact, we did not observe a solar neutrino signal at all, and our results were expressed only as upper limits. The low neutrino capture rate was a result that many theorists found difficult to accept. They believed that there must be some chemical inefficiency in the recovery of a few atoms of argon-37 in the massive Homestake detector. We made numerous tests to check the chemical efficiency, and found that the chemical procedures were reliable.

The main difficulty was experimental: the argon-37 counting needed to be improved to search in a more sensitive way for a solar neutrino signal. Brookhaven electronic engineers Veljko Radeka and Lee Rogers solved this problem by devising a pulse rise-time system to discriminate argon-37 decay events from background events. We began using this new system in 1970. After about a year of observations, a clear solar neutrino signal was observed. The signal was smaller than the earlier limit, but we were convinced that the Homestake experiment would in time make a valid measurement of the solar neutrino capture rate in chlorine-37 to compare quantitatively with the solar model calculations.

The pulse rise-time system development gave the Homestake experiment a new life. The solar neutrino production rate was indeed lower than the solar model predictions by a factor of about three. The most likely explanation, in my view at the time, was that the solar model was in error.

Fred Reines organized an in-depth conference on all aspects of solar neutrino research at the University of California’s Irvine campus in 1972. There was an excellent discussion of the theoretical and experimental matters, and new experiments (Trimble and Reines 1973). This conference was of great importance in defining many basic problems and new directions in this new field. Another conference of a similar nature was held at Brookhaven five years later, in 1978. Trevor Pinch made an interesting sociological and historical study of the reaction of the scientific community to the Homestake experiment in his book Confronting Nature (Pinch 1982). We had to wait 18 years for another experiment, the Kamiokande experiment in Japan, to confirm that the solar boron-8 neutrino flux was low.