HERMES at HERA: Quark-Gluon Spin Structure of the Nucleon Introduction The year 2002, marked the 75th anniversary of Dennison s discovery that the proton, just like the electron, carries spin. The electron is in our present understanding a point-like particle without internal structure, and spin is one of its intrinsic properties. In the case of the proton, however, it was subsequently established that this particle consists of further subcomponents, the quarks and gluons. The nucleon is a complex system with three valence quarks (2 u and 1 d for the proton, 2 d and 1 u for the neutron), a sea of short-lived quark-antiquark pairs where also strange quarks (s) appear, all bound together by the exchange of gluons. A legitimate question is, then, how the spin of the proton is derived from the properties of its quark-gluon structure. The most successful tool to study the quark structure of the nucleon has been deep inelastic scattering (DIS) in which a high-energy lepton (muon, electron, or neutrino) scatters off the individual quarks. To measure properties related to the spin of the proton using DIS, both a polarized target and polarized lepton beam are required. Due to the technical difficulties in performing such polarized DIS experiments, the answer to this question was for a long time unknown. And when at the end of the 1980s the first such experiments were performed, the answer turned out to be more complex than one could naively expect. In fact, even today we still have only partial answers. In principle there are several sources of angular momentum in the proton that can contribute to its spin: the intrinsic spin of the quarks (1/2) and of the gluons (spin 1 particles) and the orbital motion of these particles. The results of the very first polarized DIS experiments (E80 at SLAC, EMC at CERN), in which only the scattered lepton was detected, indicated that the spin of the quarks, which was expected to be the major fraction, contributed little or nothing to the proton spin. As a followup to these pioneering experiments a second generation of experiments sprang up with greatly enhanced target and beam technologies (SMC at CERN; E142, E143, E154, and E155 at SLAC). They determined that the contribution of the quark spins to the proton spin is indeed limited to about 25 30%. All these experiments were primarily designed to determine the total quark spin contribution with good accuracy. Several questions remained unanswered: If the quark spin contribution is so small, what then are the main contributions? What is the contribution of the different quark flavors? Is there a polarization in the quark sea? It was to find an answer to these questions that the HERMES experiment was started in the beginning of the 1990s. It would use a new polarized target technology at the polarized electron storage ring of HERA at DESY/Hamburg. After the polarization of the beam was proven (in 1992, and finally in 1994) the HERMES spectrometer was constructed in record time. The commissioning run in 1995 immediately gave results on polarized DIS on the neutron, using an optically pumped polarized 3 He target. Since the summer of 1995 HERMES has taken data whenever the HERA accelerator provided 27.5 GeV electrons (or positrons). The Experiment The Polarized Beam The HERMES experiment is located in the hall East, one of 4 experimental stations on the HERA accelerator. HERA is a proton-electron collider where the 920 GeV proton and the 27.5 GeV electron beams are brought into collision in the ZEUS and H1 experiments. In HERMES the beamlines are separated by about 75 cm. The proton beam passes unhindered through a vacuum pipe in the midplane of the HERMES spectrometer. The electron beam crosses the polarized gas target described below. In a high-energy storage ring like HERA the electrons become transversely polarized through the emission of spin-flip synchrotron radiation. In HERA this polarization builds up in about 30 minutes to a level of typically 50 60%. Spin rotators in front of and behind HERMES transform this into the longitudinal polarization necessary for the experiment [1]. The Polarized Target Polarized protons (or deuterons) are produced in an atomic beam source and injected into a 40cm-long open-ended storage cell through the center of which the HERA electron beam is steered. The presence of this storage cell increases the target density by some two orders of magnitude, resulting in luminosities Vol. 13, No. 4, 2003, Nuclear Physics News 19
of about 10 31 cm -2 s -1 A small fraction of the gas in the target cell is sampled through a thin exit tube to the Breit Rabi polarimeter and the target gas analyzer, where the polarization of the target atoms, and the fraction of molecules present, is determined. Polarizations of up to 85% for both protons and deuterons were obtained (in 2000, data were also taken with tensor-polarized deuterium). Overview of Spectrometer The HERMES spectrometer [2] consists of two identical halves installed above and below the HERA beamlines. At the center of the spectrometer a dipole magnet with integrated field strength of 1.3 Tm momentum analyzes the produced particles and determines their charge. Tracking detectors in front of and behind the dipole magnet determine the track of charged particles. In total, 57 planes of tracking detectors determine the trajectory of the particles. The resulting momentum resolution is better than 2.5%, while the angular resolution is better than 1.5 mrad. The separation of leptons and hadrons is done using a combination of an electromagnetic calorimeter, a transition radiation detector, and a pre-shower detector. The electromagnetic calorimeter, with a thickness of about 18 radiation lengths, also serves to detect high-energy photons. A schematic overview of the spectrometer and its components is presented in Figure 1, while Figure 2 shows a recent picture of the apparatus. The RICH An important addition to the original HERMES spectrometer was made in 1998, when the threshold Cherenkov counter was replaced by a full Ring Imaging Cherenkov (RICH) detector [3]. The HERMES RICH uses two radiators: 5 cm of silica aerogel followed by about 1 m of C 4 F 10 gas. The photons are reflected off a spherical mirror onto a matrix of 1934 photomultipliers (1.92 cm diam.). Due to the different index of refraction of the radiator materials (1.03 for the aerogel, 1.0014 for the gas) the rings of Cherenkov light produced by the different particle types can be used to identify pions, kaons, and protons over a particularly large momentum range. Identification is achieved between 1.5 and 15 GeV, thus covering almost completely the kinematic range of hadrons generated in the experiment. Results of Run I (1995 2000) Inclusive Scattering When only the scattered electron is detected, i.e., the case of inclusive scattering, polarized DIS determines the helicity spin structure function g 1. From this it is possible to deduce the total contribution of all quark spins to the nucleon spin. This was already done in the polarized DIS experiments of the first and second generation, and Figure 1. Schematic side-view of the HERMES spectrometer. 20 Nuclear Physics News, Vol. 13, No. 4, 2003
facilities and methods contribution. These direct measurements confirm the results of QCD-analyses of inclusive scattering and give for the first time a complete picture of the polarization of the individual sea-quark flavors. Taken together, all these measurements now definitely show that the spin of the quarks generates far less than half of the spin of the nucleon. The quark spins which do contribute come almost exclusively from the valence quarks. This may explain the remarkable success of the early quark models in predicting the relative size of the nucleon static magnetic moments. These so-called anomalous moments were in fact one of the first pieces of evidence that the nucleon was not a fundamental particle. Figure 2. The HERMES spectrometer and target (right-hand side of picture) during maintenance work. HERMES has added new, more precise data in its kinematic domain to the world data set. This is particularly true for the deuteron target. It is important to note that the latest data, from SLAC and HERMES, have been taken at a significantly lower momentum transfer than the older EMC and SMC data. This allows a global analysis of the scaling behavior of the structure functions from the entire world data set using perturbative QCD techniques. From this analysis one can find the polarization of the quarks and have a first idea of the polarization of the gluons. A recent analysis of this type was performed by HERMES and this results in a value for the polarized quark density of 0.201±0.103±0.034. Semi-Inclusive Scattering However, the basic drawback of the inclusive reactions is still that they sum over all quark flavors. Hence it is virtually impossible to determine separately for each quark flavor what its contribution to the nucleon spin is. To do that one has to turn to semiinclusive reactions. HERMES was specifically designed to study these. The quark which absorbed the virtual photon leaves the nucleon and will fragment into hadrons. The observation of (part of) these hadrons can be used as a filter on the flavor of the original quark. This is the so-called flavor tagging method. For example, when a high-momentum kaon is detected in coincidence with the scattered electron, then there is an enhanced probability that the original quark which scattered the electron was an s-quark from the sea. Using this technique the HERMES experiment has recently determined the polarization of different quark flavors in the nucleon. These are shown in Figure 3. It is clear that the largest contribution to the nucleon spin comes from the u-quarks, while the d-quarks give a contribution with opposite sign. The sea-quarks, among them the strange quarks, give a very small Gluon Polarization The QCD-analyses of inclusive reactions mentioned above indicate that a large (and positive) contribution to the nucleon spin can come from the polarization of the gluons. This is to be directly tested by the COMPASS experiment at CERN and by the polarized pp experiments at the RHIC accelerator at Brookhaven. Some time ago the HERMES experiment reported the first evidence for such large and positive polarization [4]. Exclusive Reactions The last unknown piece in the puzzle of the spin content of the nucleon is the possible contribution of orbital angular momentum. This has been hotly debated over the last years. It was realized in the middle of the 90s that the framework of generalized Parton distributions (GPDs) might give access to this orbital angular momentum contribution. Quite apart from that the study of these offforward extensions of the standard parton distribution functions can provide a wealth of new information on the quark structure of nucleons. Vol. 13, No. 4, 2003, Nuclear Physics News 21
imaginary part of the interference amplitude, and beam charge asymmetries, caused by the real part of the interference amplitude. Strategy for Run II (2001 2006) In the run from 1995 through 2000 the emphasis was on the determination of the helicity structure of the nucleon using longitudinally polarized targets and beam. In the present Run II of HERA (expected to end by 2007), the HERMES collaboration hopes to complete two new major measurements: the first measurement of the nucleon transversity distribution and a broad program of detailed exclusive reaction measurements using a new recoil detector. Figure 3. The polarization of different quark flavors in the nucleon. The GPDs enter into the description of exclusive reactions through the socalled handbag diagram. In an exclusive reaction the target nucleon does not fragment but instead remains in its ground state (or close to it). The cleanest example of such reaction is that of deeply Virtual Compton Scattering (DVCS). This type of reactions was identified for the first time in 2001 at DESY by the H1 [5], by ZEUS and HERMES [6] experiments, and by Hall B at JLAB [7]. Since then, a large body of data on DVCS has been collected. HERMES is here in a rather fortunate position: because of the relatively low energy of the electron beam the DVCS cross section is usually of the same order of magnitude, or less, than that of the pure QED Bethe Heitler process, where the photon is radiated from the incoming or scattered electron. Since both reactions lead to the same final state, they can interfere, which leads to observable asymmetries. HERMES has now established the existence of single beam-spin asymmetries, caused by the Transversity The transversity is the last of the leading twist distributions (the other ones being the unpolarized quark density and the helicity distribution) and is a measure of the polarization of the quarks along the proton spin when the proton is polarized transversely to its momentum (as opposed to the case of helicity). At the present time, it is completely unmeasured. These groundbreaking measurements require detection of hadrons in the final state in coincidence with the scattered electron, something HERMES was specifically designed to do. In this case one uses a transversely polarized nucleon target, and there are observables corresponding to both polarized and unpolarized electron beam. Just the possibility of a measurement has created a flurry of theoretical interest which has resulted in a serious revision in the theoretical treatment of quark distributions and fragmentation functions (e.g.,hep-ph/ 0303034). The first preliminary results on this distribution are expected before the end of 2003. 22 Nuclear Physics News, Vol. 13, No. 4, 2003
4. A. Airapetian et al., Phys. Rev. Lett. 84 (2000) 2584. 5. C. Adloff et al., Phys. Lett. B517 (2001) 47. 6. A. Airapetian et al., Phys. Rev. Lett. 87 (2001) 182001. 7. S. Stepanyan et al., Phys. Rev. Lett. 87 (2001) 182002. Figure 4. Schematic view of the recoil detector under construction at HERMES. E. KINNEY University of Colorado, Boulder HERMES Spokesperson 1997 1999 Recoil Detector At the energies typical of the HERMES experiment it is not really possible to ensure full exclusivity of, e.g., DVCS events by observing only particles in the forward spectrometer (in the case of DVCS, these are the scattered electron and the hard gamma). To improve the exclusivity of the data sample several groups within the HERMES collaboration are now constructing a recoil detector that will sit closely around the target. This will detect the slow-moving recoil proton in coincidence with the electron and gamma. It is expected that this new device employing Silicon detectors and scintillating fibre trackers, will be installed sometime in 2004 or 2005. With the new detector, a much cleaner study of the kinematic dependence of the exclusive reactions can be performed which, in turn, will allow more precise testing of the new models of nucleon structure contained in the GPDs. Eventually one may be able to make a true three-dimensional picture of the nucleon. Conclusion Our understanding of the proton has changed dramatically in the last 75 years, from a basic building block of matter, an atom in the sense of Democritus, to a complex whirl of colored quarks and gluons interacting according to QCD. The understanding of the nucleon s spin has proved to be a very exacting and challenging task, both experimentally and theoretically, and it is a task which is yet to be completed. Not only has the HERMES experiment confirmed the earlier work at CERN and SLAC, but it is also leading the way into the future with its precision measurements of the quark flavor dependence of the proton spin, the first measurements of the transversity, and its measurement of exclusive reactions at high energies. References 1. D. P. Barber et al., Phys. Lett. B343 (1995) 436. 2. K. Ackerstaff et al., Nucl. Instr. Meth. A417 (1998) 230. 3. N. Akopov et al., Nucl. Instr. Meth. A479 (2002) 511. E. STEFFENS Universität Erlangen HERMES Spokesperson 1999 2001 D. RYCKBOSCH Universiteit Gent, HERMES Spokesperson 2001 2003 Vol. 13, No. 4, 2003, Nuclear Physics News 23