When CERN saw the end of the alphabet

In 1966/7 Steven Weinberg, Abdus Salam and John Ward proposed a local gauge theory, SU(2) x U(1), for a unified description of electromagnetic and weak interactions, with a Higgs mechanism to give mass to the (weak) field quanta. When I arrived as a student at Johns Hopkins University in 1966, Ward was a professor there. I could understand that something exciting was going on from the discussions at the physics seminars, but could not appreciate the importance it would subsequently acquire.

The most striking feature of the weak interactions is their very short range of less than around 10^-15 cm, i.e. less than 1% of the size of a nucleon. This is compared with a range of around 10^-13 cm for nuclear (strong) forces, and is in stark contrast to the “infinite” range of the electromagnetic force. The short range of the weak interactions implied very massive mediating particles or quanta, W⁺ and W^–, for the charged current, the only known weak interactions at the time. However, the unified description of Weinberg, Salam and Ward had four field quanta, two charged and two neutral, implying that a new type of “neutral current” weak interaction should exist. This would be mediated by the Z⁰ – a particle that is closely related to the massless photon, in fact it is almost identical except for being very massive. The renormalizability of the theory, shown in 1971 by Gerard ‘t Hooft and Martin Veltman, and by the discovery of the weak neutral currents at CERN in 1973, made this unified electroweak scheme appear plausible. But what could the mass of the W and Z particles be?

Where and how?

The observed linear increase of the neutrino-nucleus cross-sections with incident energy up to E_ν ~350 GeV, which was consistent with the (old) Fermi four-fermion point interaction, could not last forever. At a neutrino-nucleon (or rather, neutrino-quark) centre-of-mass energy of the order of 300 GeV the cross-section would reach the S-wave unitarity limit, so the effects of W-exchange had to come in, in order to modify this unacceptable behaviour. The non-deviation from linearity in the measured cross-section indicated m_w > 50 GeV, and was consistent with infinite m_w. Meanwhile, the charged current and neutral current data from neutrino interactions, when incorporated into the Weinberg-Salam-Ward scheme, were giving a weak mixing angle sin²Θ_w ~0.3-0.6, which implied m_W,Z ~60-100 GeV. Subsequently, measurements of sin²Θ_w narrowed its value down to around 0.23, providing by 1982/3 a much better estimate of m_w ~80 GeV and m_z ~90 GeV to within a few GeV. In the late 1970s and early 1980s the forward-backward angular asymmetry, due to γ-Z interference, in e⁺e^– → µ⁺µ^– at the top PETRA energies (√s ~30-40 GeV) also indicated m_z < 100 GeV rather than an infinite m_z. So, the question was where could these W and Z intermediate vector bosons be produced and how could they be detected?

In 1976 CERN’s SPS began operating with particle beams of energies up to 350-400 GeV onto a fixed target, i.e. with centre-of-mass energies of √s ~30 GeV, which was insufficient for W and Z production. The same year David Cline, Carlo Rubbia and Peter McIntyre proposed transforming the SPS into a proton-antiproton collider, with proton and antiproton beams counter-rotating in the same beam pipe to collide head-on. This would yield centre-of-mass energies in the 500-700 GeV range. Provided the antiproton intensity was sufficient, the W and Z particles could be produced through their couplings to quarks and antiquarks, and detected through their couplings to leptons as prescribed by the Weinberg-Salam-Ward model. Then, in 1979, Weinberg, Salam and Sheldon Glashow were awarded the Nobel prize for electroweak unification and the prediction of weak neutral interactions, which implied the existence of the Z particle. (Ward was no doubt of the same class, but the Nobel prize can only be awarded to three people at most.) This indicated that the theoretical community was more convinced of the existence of the W and Z than most of the experimentalists at the time.

The proton-antiproton collider

CERN meanwhile went ahead with the proton-antiproton collider, and by the summer of 1981 the heroic endeavour of transforming the SPS into a proton-antiproton collider had been accomplished, despite the many uncertainties, including unknown unpredictable beam-beam effects. There is no doubt that Carlo Rubbia, with his enthusiasm, power of conviction and charisma, played a key role in this phase of the project. The first proton-antiproton collisions occurred on 9 July 1981, almost exactly three years since the project had been officially approved. Within hours, the first events that had been seen, detected and reconstructed in UA1’s central tracker were shown by Rubbia at the Lisbon conference (UA1 collaboration 1981).

The PS proton beam at 26 GeV was used on a fixed target to produce antiprotons at ~3.5 GeV, creating about one antiproton per 10⁶ incident protons. The antiprotons were then stacked and stochastically cooled in the antiproton accumulator at 3.5 GeV, and this is where the expertise of Simon Van der Meer and coworkers played a decisive role. With a few times 10¹¹ antiprotons accumulated per day, the cooled (phase-space compactified) antiprotons were reinjected into the PS, accelerated to 26 GeV and injected into the SPS, counter-rotating in the same beam pipe with a proton beam. Both beams were then accelerated to 270 GeV and brought into collision in two interaction regions at √s = 540 GeV. Sufficient luminosity remained for about half a day. The initial luminosity in November/December 1981 was about 10²⁵ cm^-2 s^-1, but subsequently increased by a factor of 10⁵ over the following years.

The UA1 (Underground Area 1) detector was conceived and designed in 1978/9, with the proposal submitted in mid-1978. At that time we were in barracks on the parking lot in front of building 168, at the same time and place that CDF was designed, for Alvin Tollestrup was spending a year at CERN. UA1 was approved in 1979, and was constructed and essentially functional – including the reconstruction software – by the summer of 1981 (although part of the tracker electronics was still missing). At the time of approval there was a general incredulity in the particle physics community (although not obviously in UA1) that UA1 could be built – and even less operated – in time when compared with the much more focused design and modest size of the UA2 detector. That this was possible was largely thanks to Rubbia’s enlightened absolutism (or more diplomatically to his unrelenting efforts), and to his unbelievable intellectual and professional capabilities and stamina.

The two detectors

UA1 was a huge (~10 x 6 x 6 m³, ~2000 tonnes) and extremely complex detector for its day, exceeding any other collider detector by far. The design was simple, beautiful, economical and, as it turned out, very successful. In the days of initial construction, the collaboration counted around 130 physicists from Aachen, Annecy, Birmingham, CERN, College de France, Helsinki, London/QMC, UCLA-Riverside, Rome, Rutherford, Saclay and Vienna. There was a large, normally conducting dipole magnet with a field of 7 kG perpendicular to the beamline. The collision region was surrounded by a central tracker – a 5.8 m long, 2.3 m diameter drift chamber with 6176 sensitive wires organized in horizontal and vertical planes. Tracks were sampled about every centimetre and could have up to 180 hits, with a resolution of 100-300 µm in the bending plane. This detector was at the cutting edge of technology; it was the first “electronic bubble chamber” and the reconstruction software was done by ex-bubble chamber track reconstructors. The tracker was surrounded by electromagnetic (27 radiation lengths deep) and hadronic calorimeters (about 4.5 interaction lengths deep) down to 0.2° to the beamline. This almost complete coverage in solid angle became known as “hermeticity”. The central electromagnetic calorimeter – which was to play a key role in the subsequent discoveries – was very effectively and economically designed as a lead-scintillator stack in the form of two cylindrical half-shells each subdivided into 24 elements (gondolas). The entire detector was doubly surrounded by ~800 m² of muon drift chambers with a spatial resolution of ~300 µm. The overall cost of the detector was about 30 million Swiss Francs, and the central ECAL about 3 million – which was probably the best ever investment in particle physics.

While UA1 was designed as a general-purpose detector, UA2 was optimized for the detection of e^± from W and Z decays. The emphasis was on calorimetry with a spherical projective geometry – much simpler than that in UA1. There was full coverage in solid angle, except for 20° cones along the beamlines. There were about 500 calorimeter cells with a granularity of about 10° by 15° in polar and azimuthal angles, with a three-fold segmentation in depth in the central region (40-140°) and two-fold segmentation in the forward regions (20-40° and 140-160°) to allow electron-hadron separation. The central calorimetry was, in total, about 4.5 interaction lengths deep, while the forward one was about 1 interaction length (two sections of 18 and six radiation lengths). There was no central magnetic field, but the two forward regions were equipped with magnetic spectrometers (two sets of 12 toroid coils). In the central part there was a vertex detector made of coaxial drift and proportional chambers to detect charged tracks and the collision vertex. Preshower counters improved electron identification through the spatial matching of tracks and clusters. The collaboration counted about 60 physicists, with groups from Bern, CERN, Copenhagen, Orsay, Pavia and Saclay.

The jet run

The first real physics run was in December 1981. Known as the jet run, it was devoted to the search for jets arising from the hard scattering and fragmentation of partons as expected from QCD. The integrated luminosity was about 20 events per µb. The main initial effort in UA1 was based on the tracker, i.e. the measurement of high-momentum tracks and the correlations in azimuth and rapidity between charged particles. Within the collaboration, not enough attention was paid to the searches based on energy clusters in the calorimeters. The UA2 search, based exclusively on calorimetry, was simpler and gave more telling results. At the Paris conference in the summer of 1982, UA2 had clear back-to-back two-jet events, one of which was particularly spectacular with a total transverse energy (E_t) of about 130 GeV. The UA1 result was somewhat less elegant. The subsequent studies by UA1 and UA2 were based on calorimetric jet algorithms and the data were selected by total E_t or localized E_t depositions. This gave an excellent confirmation of QCD expectations in terms of cross-sections, fragmentation functions, angular distributions, etc. But what about the W and Z particles?

On the trail of the W

In the case of the W particle, both experiments looked for Drell-Yan production – that is ubard → W^–, udbar → W⁺ with the antiquarks, qbar, largely from the valence antiquarks in the incident antiprotons, and the quarks from the incident protons, with a fractional momentum x ~m_W,Z/√s ~0.2. This identification of incident partons was to facilitate the unambiguous identification of a possible resonance mass peak with the expected properties of the W^+/- – namely the spin of 1 and the V-A nature of weak interactions – which should manifest themselves through characteristic forward-backward asymmetries in the decays of the W to a charged lepton and neutrino (W → lν). For the running period at the end of 1982 we expected a luminosity in excess of 10²⁸ cm^-2s^-1 and an experimental sensitivity of > ~10 events/nb – an increase of 1000 compared with the previous run. The theoretical predictions for the cross-section for W → lν were ~0.5 nb, so few events were expected.

In the run in November/December 1982 the collider attained a peak luminosity of 5 x 10²⁸ cm^-2s^-1. UA1 collected 18 nb^-1 of data, with the total number of recorded triggers about 10⁶ for 10⁹ interactions in the detector. The electron trigger in UA1 was two adjacent gondolas or bouchon petals with > 10 GeV, with a rate of ~1 s^-1. The criteria that in December allowed UA1 to select the first five W → eν candidates unambiguously, required an ECAL cluster of > 15 GeV, a hard isolated track of p_t > 7 GeV/c roughly pointing to the cluster, missing E_t > 14 GeV, and no jet within 30° back-to-back in the plane transverse to the electron candidate. This became known as the Saclay missing E_t method. This selection in fact gave six events, five of which turned out to be fully compatible with e^±. In these five events, the electron had an E_t of ~25 GeV in one case and between 35 and 40 GeV in the others, closely balanced event-by-event by the missing E_t. Thanks to the hermeticity of the UA1 design, the resolution on missing E_t in UA1 was 7 GeV in hard/jetty events, so the observed missing E_t was highly significant in each event (> 5σ). The sixth event had 1.5 GeV of leakage in the HCAL and, upon detailed inspection, turned out to be a case of W → τν → π^±π⁰ν.

In the first weeks of January 1983 an independent search – not based on a missing E_t selection, but on stringent electron selection requirements – was performed at CERN. It found the same events, without the tau event, but with an additional event in the endcaps that was below the Saclay/missing E_t selection cuts. These events were announced later the same month at the Rome conference and went in the publication announcing the discovery of the W (UA1 collaboration 1983a). The key to this success was the built-in redundancy of UA1 – which allowed the same events to be found by two largely independent methods, resulting in clean samples with no nearby background events – and the fact that the reconstruction software was ready and working. The already perceptible Jacobian peak behaviour giving m_w = 81±5 GeV clinched the day.

In the same run UA2 had four W → eν candidates (UA2 collaboration 1983a). The electron identification was based on a calorimetric cluster of more than 15 GeV, with longitudinal and transverse shower profiles consistent with e^+/-, track-preshower-calorimetric cluster spatial matching, and electron isolation within a cone of 10°. In the forward-backward regions, where there was a magnetic field, momentum/energy (p/E) matching was enforced but the electron was not required to be isolated. Moreover, events with significant E_t opposite to the electron were rejected. These events also had missing E_t, but the 20° forward openings resulted in poorer resolution, and thus the separation of events from the background was not so good. In fact one of the consequences of UA1’s hermeticity and the selective power it provided for W → lν events, was that the D0 detector at Fermilab, which was designed in 1983/4, was made as hermetic as possible.

Catching the Z

In April/May 1983 came the next run with 118 nb^-1 of integrated luminosity for UA1. This gave an additional sample of 54 W → eν events, giving m_w = 80.3 + 0.4-1.3 GeV – and the angular asymmetry in the W decay due to the V-A coupling was unmistakable. The first W → µν events were also seen, but most importantly the first Z → e⁺e^– events and one Z → µ⁺µ^– were found. An express line selected events with two electromagnetic clusters of E_t > 25 GeV with small HCAL deposition, and also muon pair events, thereby allowing very fast analysis. The selection of Z → e⁺e^– was much easier than the W selection. The additional requirement of track isolation in the tracker, track-cluster spatial matching and < 1 GeV in the HCAL cell behind the cluster, selected four Z → e⁺e^– events with no visible experimental background in 55 nb^-1 of data. At this stage UA1 decided to publish its evidence for the Z. The first mass determination gave m_z = 95.5 ± 2.5 GeV and the cross-section for Z decay to lepton pairs was about one-tenth that of the W, as theoretically expected (UA1 collaboration, 1983b).

UA2 accumulated a comparable integrated luminosity during April/May 1983. In the UA2 selection for Z events, while one electron candidate again had to satisfy the same stringent requirements as in the W → eν search, the requirements on the second electron candidate were much looser, essentially a narrow electromagnetic cluster and a cluster-cluster invariant mass of more than 50 GeV. This procedure selected eight events altogether, all clustering in mass around 90 GeV. For three out of these eight events, the second electron candidate in fact also satisfied all the tight electron requirements (UA2 collaboration 1983b). With results from UA1 and UA2, the Z particle was definitely found.

This period, around the end of 1982 and throughout 1983, was an amazing time from both a professional and personal point of view. It was an unforgettable time of extreme effort, tension, excitement, satisfaction and joy. Subsequent runs allowed us to nail down the properties of the W and Z better and initiate other searches that were not always as successful but still extremely interesting and exciting.

The discovery of the W and Z particles was a definitive vindication of the idea of gauge theories as appropriate descriptions of nature at this level, and the unified electroweak model combined with QCD became known as the Standard Model. In the 10 years of experimentation at LEP, this Standard Model became one of the most thoroughly tested theories in physics, down to the level of a part in a thousand. However, in the SU(2) x U(1) scheme with spontaneous symmetry breaking, one of the four scalars that did not disappear into the W^± and Z masses has still to be found – and the discovery of the Standard Model Higgs, in the ATLAS and CMS detectors at CERN should eventually complete this story. The discovery of the W and Z at CERN also signalled that the “old side” of the Atlantic regained its eminence in particle physics. “…L’espoir changea de camp, le combat changea d’âme….” (Victor Hugo, “Waterloo”.)