Measurement of the high-energy contribution to the Gerasimov-Drell-Hearn sum rule
MMeasurement of the high-energy contributionto the Gerasimov-Drell-Hearn sum rule
M. M. Dalton ∗ , A. Deur ∗‡ , C. D. Keith Thomas Jefferson National Accelerator Facility, Newport News, VA 23606, USA
S. ˇSirca ∗ University of Ljubljana, Ljubljana, Slovenia
J. Stevens ∗ William & Mary, Williamsburg, Virginia 23187, USA
Endorsed by the GlueX Collaboration ∗ Spokesperson ‡ Contact
Abstract
We propose to measure the high-energy behavior of the integrand of the Gerasimov-Drell-Hearn (GDH)sum rule on the proton and the neutron up to 12 GeV. The convergence of the GDH integral will beinvestigated for the first time and to high precision. The validity of the GDH sum rule on the neutronwill be accurately tested for the first time, while for the proton the uncertainty will be improved by 25%relative. The data will allow precision testing of Regge phenomenology in the polarized domain. The a and f Regge trajectory intercepts will be obtained to an order of magnitude higher precision than thecurrent best estimates. The data will also contribute to the determination of the real and imaginary partsof the spin-dependent Compton amplitude, the polarizability correction to hyperfine splitting in hydrogen,and to studying the transition between polarized DIS and diffractive regimes.The experiment will require a circularly polarized photon beam (produced from a longitudinally polar-ized electron beam) with a flux approximately ⁄ of the GlueX-II experiment E12-13-003. The experimentwill run in two configurations which require two different CEBAF beam energies. A new longitudinalpolarized proton and deuteron target will be needed in Hall D. The experiment will require 21 PAC daysat the nominal CEBAF energy and another 12 PAC days at an energy ⁄ to ⁄ of the nominal.1 a r X i v : . [ nu c l - e x ] A ug une 20, 2020Dear Members of the Je↵erson Lab PAC:I am writing to convey the GlueX Collaboration’s endorsement of the proposal titled Mea-surement of the high energy contribution to the Gerasimov-Drell-Hearn sumrule that has been submitted to the PAC by Mark Macrae Dalton, Alexandre Deur, JustinStevens, and Simon ˇSirca.This endorsement implies a commitment of the entire collaboration to operate the detector,sta↵ shifts, calibrate and process the data, as well as provide support for and review of thefinal data analysis. The procedure for obtaining endorsement is defined in the bylaws ofthe GlueX Collaboration (available at ) and is overseen by the GlueX Collab-oration Board. An ad hoc committee was appointed to review the technical feasibility andphysics merit of this proposal. After evaluation of the outcome of this review, the boardrecommended a vote of endorsement to the entire collaboration. Finally, the collaborationgranted endorsement via a unanimous vote cast by a quorum of members. On behalf of theGlueX Collaboration, I would like to express our enthusiasm at the prospect of starting adoubly polarized photoproduction program in Hall D.Sincerely,Matthew ShepherdProfessorGlueX Collaboration Spokesperson ontents ν -dependence of ∆ σ ν -dependence of ∆ σ . . . . . . . . . . . . . . . . . . 273 Systematic uncertainties 30 ν dependence of ∆ σ . . . . . . . . . . . . . . . . . . . . . . . . . 307.2 Uncertainties affecting the absolute normalization of ∆ σ . . . . . . . . . . . . . . . . . . . 307.3 Uncertainties and target spin flip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 f ( ν ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339.4 The intercept of the a Regge trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359.5 Deuteron asymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369.6 Polarizability correction to hyperfine splitting in hydrogen . . . . . . . . . . . . . . . . . . 369.7 Transition between polarized DIS and diffractive regimes . . . . . . . . . . . . . . . . . . . 37
10 Summary 38 Introduction
We propose to measure the high-energy behavior of the integrand of the Gerasimov-Drell-Hearn (GDH)sum rule, i. e. of the doubly polarized total photoproduction cross-section asymmetry. The high-energydomain is where a failure of the GDH sum rule may occur for a number of reasons elaborated below, andsuch behavior would indicate unknown structural features or dynamic processes in the nucleon. The datawill also improve significantly the precision at which the sum rule is tested on the proton, and offer a testof comparable accuracy for the neutron, which is not tested at present.Independently of the sum rule study, the measurement will investigate QCD in an energy domainwhere its phenomenology is unknown when spin degrees of freedom are explicit. The experiment will thusprovide a baseline for the EIC’s study of the transition between polarized deep inelastic scattering (DIS)and polarized diffractive regimes. In particular, our data will help to clarify a discrepancy between fitsof the photoproduction and DIS world data and the corresponding theoretical expectations, which giveconflicting predictions for the power-law dependence of the GDH integrand.The experiment will be sensitive enough to provide, for the first time, a precise measurement of thedeuteron asymmetry in the diffractive regime. Chiral effective field theory will also be tested in a differentregime than that covered by the low- Q JLab spin sum rule program. Finally, our measurements willconstrain the polarizability contribution to the hydrogen hyperfine splitting.We propose to perform the measurement on the proton and neutron in Hall D with CEBAF at 12 GeV.Hall D is uniquely suited for such a measurement thanks to its photon tagger and its high-luminosity, largesolid angle detector. Overall, the experiment aims at providing an absolute measurement of the polarizedcross-section difference at a ≈
5% accuracy, typical for such experiments. However, the key goal of theexperiment—to determine the high-energy behavior of the GDH integrand—does not require absolutenormalization and thus will have significantly reduced uncertainties of about 2%, since only point-to-pointuncorrelated errors contribute.A shorter version of this document was submitted as a Letter of Intent to PAC47 [1]. Based on it,the PAC acknowledged the feasibility of the experiment, considered Hall D best suited for performing it,and encouraged this first step toward a comprehensive doubly-polarized program in Hall D [2]: “
The PACrecognizes the science case for this LOI and recommends preparation of a full proposal with focus on theextraction of the actual value of the GDH integral at high energies. The PAC would be pleased to seethe development of ideas towards a full program with a circularly polarized photon beam and a polarizedtarget in Hall D. ”The proposal is endorsed by the GlueX collaboration, which will support the preparation of the exper-iment, its run, analysis and publications related to the experiment, see the letter of endorsement page 2.
The Gerasimov-Drell-Hearn (GDH) sum rule [3] is a general and fundamental relation that links theanomalous magnetic moment κ of a particle to its helicity-dependent photoproduction cross-sections: I ≡ (cid:90) ∞ ν ∆ σ ( ν ) ν d ν = 4 π Sα em κ M , (1)where ν is the probing photon energy, S is the spin of the target particle, M is its mass and ν = m π (1 + m π / M ) is the threshold energy for pion photoproduction ( m π is the neutral pion mass), and α em is the electromagnetic coupling constant. In our case (proton and neutron) S = 1 / σ ≡ σ P − σ A isthe difference in total photoproduction cross-sections ( γN → X ) for which the photon spin is parallel and5nti-parallel to the target particle spin, respectively. Note that with this relative sign definition the GDHintegral is positive—the opposite convention is also seen in the literature.The sum rule is valid for any type of particle: nucleons, nuclei, electrons, even photons. For the proton,the right-hand side of Eq. (1) gives I p = 204 . µ b, while for the neutron, one obtains I n = 233 . µ b.The experiment has two major thrusts, which we call convergence and validity . The first thrust is toverify that the GDH integral converges to a finite value. It is the high-energy behavior of the integrandwhich determines this. Mathematically, since it is weighted by 1 /ν , σ P − σ A must decrease faster than1 / log ν in order for the integral to converge. We will test this by precisely measuring the high- ν dependenceof the spin-dependent cross-section difference.The second thrust is to improve the determination of the GDH integral itself which will impose a morestringent test on the validity of the sum rule, and thereby improve our sensitivity to physical processesthat may cause a real or an apparent violation of the sum rule.It is illustrative to consider the unpolarized equivalent of the GDH sum rule, (cid:90) ∞ ν ( σ P + σ A ) d ν = − αM . (2)This “rule” itself is clearly invalid, as the spin-independent (total) cross-section is a positive definitequantity (like all cross-sections) while the sum rule sets it equal to a negative number. In addition, theintegral itself does not even converge to a finite value due to the behavior of the integrand at high energy.Fig. 1 shows the spin-independent cross-section as a function of ν . The high- ν behavior can be describedby ( σ P + σ A ) ∝ ν . , which does not result in a finite integral. The empirical observation of this divergencein multiple hadronic interactions, has led to the postulation of the pomeron in 1961. This divergence andthe resulting insight would not become apparent without extending the measurements to sufficiently highenergies. (GeV) n -
10 1 10 ( m b ) A s + P s -
10 1
Figure 1:
Unpolarized total photoabsorption cross-section σ P + σ A for the proton (black symbols) and deuteron(blue symbols) as a function of the photon beam energy. The data are from Ref. [4]. The lines are Regge fits includinga pomeron term proportional to ν . . In contrast to this, the sum rule obtained by using the second order of the low-energy theorem (seeSection 2.1) has an additional 1 /ν factor, yielding the Baldin sum rule [5], (cid:90) ∞ ν σ P + σ A ν d ν = 4 π ( α E + β M ) , (3)6here α E and β M are the electric and magnetic polarizabilities, respectively. The 1 /ν factor overcomesthe convergence problem, and polarizability measurements have verified the sum rule prediction [6, 7]. A failure of the GDH integral to converge would be a startling development and would immediately indicatethat some phenomenology is missing from our understanding. We propose to measure the functional form of the spin-dependent cross-section difference ∆ σ at high photon energy to high precision. This part ofthe measurement does not rely on any existing data and is not sensitive to many systematic uncertaintiesaffecting measurements of absolute cross-sections.Regge theory predicts the cross-section at high energy to be described by the functional form givenin Eq. (14) where the parameters must be determined from the data [8]. Our data will allow a test ofRegge theory well into the region where it is expected to be applicable. Mueller and Trueman [9] assessedthat if Regge behavior holds, the spin-dependent cross-section should drop to zero faster than 1 / log ν ,which would be sufficient for the GDH integral to converge. Section 6.2 discusses the sensitivity of ourmeasurement to the parameters of the presumed Regge dependence.In addition to the sum rule study, our measurement will investigate QCD in its diffractive scatteringregime, where Regge theory is expected to describe the scattering process. It would be the first clean test forpolarized photoproduction. As signaled in the review of Ref. [10], this phenomenology has not been testedwith spin degrees of freedom: “above the resonance region, one usually invokes Regge phenomenology toargue that the integral converges [...] However, these ideas have still to be tested experimentally. [...] the realphoton is essentially absorbed by coherent processes, which require interactions among the constituents suchas gluon exchange between two quarks. This behavior differs from DIS, which refers to incoherent scatteringoff the constituents.” The lack of spin-dependent tests is an important shortcoming also emphasized byBjorken [11]: “Polarization data has often been the graveyard of fashionable theories. If theorists had theirway, they might well ban such measurements altogether out of self-protection.”
This is supported by a starkdiscrepancy that exists between fits of the photoproduction and DIS world data and expectations fromRegge theory, see Section 9.4.
The saturation of the integral beyond a given ν indicates the energy scale at which the characteristic scaleof the object structure becomes irrelevant (or its mass scale scale for a structureless object).For a lepton, at first order in perturbation, ∆ σ ( ν ) is non-zero only at ν in the vicinity of the leptonmass [12] (where it switches sign to ensure that the integral yields zero).For a nucleon, only a single quark participates in a high-energy reaction and, if quarks are structureless, κ q = 0 for the active quark and it does not contribute to the sum rule. A failure of the sum rule to saturateat its expected value by a certain ν imply that there remains an additional contribution at higher energy.Measuring the sum rule to 12 GeV allows one to bound the contributions to the nucleon structure thatcome from energy scales larger than 12 GeV. Other possible causes that could invalidate the sum rule exist,and all involve high- ν phenomenology, see Section 2.Thus, while the nucleon GDH sum gets most of its contribution from the resonance regime, the high-energy part is critical since it may reveal possible substructure or unknown structural processes. Indeed,it is the high energy domain that would expose a failure of the sum rule. It is important to measure ∆ σ on both the proton and the neutron for two reasons.7irst, we would provide two independent tests of the sum rule, since processes in the neutron may bedifferent to those in the proton. In general, it is possible for the sum rule to appear valid for one type oftarget, and invalidated for other ones. For example, it has been shown that within the Standard Model, thesum rule is true for the electron, but this has no bearing on its validity for, e.g., nucleon targets. Gatheringneutron data in Hall D would be especially important because the neutron world data are not as extensiveas those for the proton: they are less precise and extend only to up to ν of 1.8 GeV, see Fig. 2.Second, gathering data on both nucleons allow for an isospin analysis of their high- ν behavior, asdiscussed next. Regge theory suggests that at high ν , ∆ σ ( ν ) ∝ ( ν + M/ α − [8], with α a Regge intercept. For theisovector part, ∆ σ p − n ≡ ∆ σ p − ∆ σ n , α should be determined by the a (1260) meson trajectory, whichis still not well known. For the isoscalar ∆ σ p + n ≡ ∆ σ p + ∆ σ n part, α should be given by the f (1285),which is better known. Thus, an analysis of isospin decomposition of the Regge trajectories requires anaccurate measurement of the proton and the neutron.Independent of the f (1285) intercept value, i.e. of the question of ν -dependence, the absolute normal-ization of ∆ σ p + n (i.e. c in Eq. (14)) is not precisely known. In fact, ∆ σ p + n is at present assumed to bezero in analyses since the measured asymmetry on the deuteron in the diffractive regime is consistent withzero [13, 14]. This experiment will be precise enough to measure clearly, and for the first time, a non-zeropolarized deuteron signal in this regime (at least 10 σ based on Regge expectations). Theoretical dispersion analysis of the measured ∆ σ ( ν ) will yield the complex spin-dependent Comptonamplitude f ( ν ) (see next section and Section 9.3), and thereby tests chiral effective field theory ( χ EFT),the leading non-perturbative approach to QCD at low energy-momentum. This will complement the JLablow- Q spin sum rule experimental program that tested χ EFT and showed that the description of spinobservables remains a challenge for χ EFT [15].Finally, measuring ∆ σ ( ν ) will provide a baseline for some of the Electron Ion Collider (EIC) studies,as well as constrain the polarizability contribution to the hydrogen hyperfine splitting, see Sections 9.6and 9.7. The GDH integrand was measured at MAMI and ELSA for energies in the range 0 . ≤ ν ≤ . . ≤ ν ≤ . . ≤ ν ≤ .
42 GeV and yields the contribution from single π exclusive production γp → π p to the integral of (125 . ± . ± . µ b [16].For the proton, the MAMI measurement covers 0 . ≤ ν ≤ . ± ± µ b. The ELSA measurement covers 0 . ≤ ν ≤ . . ± . ± . µ b [6].The JLab 6 GeV CLAS experiments (E04-102 [17] and E06-013 [18], both part of the CLAS g9 rungroup) had as one of their goals to measure some of the important photoproduction channels contributingto the proton GDH sum. E04-102 measured the single-pion production contribution for ν up to 2 GeV,and E06-013 measured the π + π − contribution for ν up to 3.1 GeV. These data are still under analysis andlimited in their ν to the same range as the MAMI and ELSA experiments. During the CLAS g14 run,8hich used the HDice target, both proton and deuteron data were gathered, with part of the run usingcircularly polarized photons with ν up to 2.5 GeV. The goal of g14, however, was searching for missingresonances and the trigger was not suited for total cross-section measurements. Thus, we do not expectany direct information on the GDH sum from g14 and, in any case, its maximum ν coverage does notextend beyond that of ELSA. Finally, the 6 GeV CLAS experiment E94-117 [19] was approved with anA − rating but did not run due to a delay in the polarized HDice target availability and the termination ofthe 6 GeV program.Further GDH data on the proton and neutron have been acquired at GRAAL at ESRF (Grenoble,France, 0.5-1.5 GeV) and HIGS at TUNL (Durham, USA up to 0.1 GeV). However they are not publishedyet. Another GDH experiment on deuterium is approved to run at HIGS in 2021, with beam energiesbetween 6 MeV and 20 MeV.An experiment with similar goals as this proposal, E159 [20], was approved at SLAC but did not occurdue to termination of the experimental program in End Station A.The GDH sum rule generalized for electroproduction has been the object of active experimental pro-grams at ESRF, JLAB, MAMI, SLAC and TUNL, see Refs. [21, 15] for reviews. A very low- Q GDHprogram has been carried out in Halls A (E97-110 [22]) and B (E03-006 and E06-017 [23]) during the 6GeV era. These experiments measured the inclusive doubly-polarized electron-scattering cross-section onproton and deuteron (eg4) and He and neutron (E97-110). The results can be extrapolated to Q = 0to investigate the GDH sum rule. However, due to elastic radiative tails rising at large ν , the maximum ν value of these experiments was limited to 1.7 GeV at the lowest Q used for the Q → Q GDH program does not investigate the questions discussed in this proposal.To summarize, the proton data are limited to about 3 GeV and the neutron data to 1.8 GeV. TheLEGS, MAMI and ELSA ∆ σ data are published, while the GRAAL, HIGS and CLAS data on specificchannels are yet to be published. n - b ) m ( A s - P s proton n - b ) m ( A s - P s neutron Figure 2:
World data of the spin-dependent cross-section difference ∆ σ on the proton (left) and neutron (right).Data from various experiments are combined and rebinned. For the proton, the contribution from ν = ν ≈ .
145 GeV to 0.2 GeV (from threshold to the start of theMAMI measurement) is estimated at ( − . ± µ b by the MAID2007 parameterization [24]. Including the9easurements from Fig. 2 and the MAID prediction yields an integral over the range ν = ν to 2.9 GeV of(226 ± . ± µ b. Fig. 3 shows the measured running of the GDH integral. To obtain the full integral, theunmeasured ν > . ν ≈ . n b ) m R unn i ng GDH i n t e g r a l ( proton p r e li m i na r y n b ) m R unn i ng GDH i n t e g r a l ( neutron Figure 3: “Running” of the GDH integral data for the proton (left) and neutron (right) starting at ν = 0 . ν ≤ ν ≤ . − . ± µ b and ≈ − µ b, respectively, by the MAID2007 parameterization [24]. The data from various photoabsorptionexperiments have been combined and rebinned. The green horizontal lines show the expected value of the GDH sum.The red points are the recent generalized GDH results from electroproduction extrapolated to Q = 0 for the proton(preliminary, publication in preparation) and at Q = 0 .
035 GeV for the neutron [22], statistical uncertainty (inner)and total (outer) error bars. Nevertheless, a Regge parameterization is assumed for the ν > . − µ b to − µ b [25] are obtained, where therange stems from the uncertainty on the a intercept (parameter α a in Eq. (14)). For combined fits onthe proton and neutron data, the high- ν contribution is − µ b [21] but the fit does not agree well withthe proton data, as seen in Fig. 4. This suggests that a significant systematic uncertainty, which is difficultto quantify, is involved in the high- ν estimate due to the limited range and quality of the existing data.These projections show the full GDH integral lying in the range from 191 µ b to 212 µ b, which brackets theexpected value by significantly more than the statistical uncertainty.The CLAS eg4 electroproduction data extrapolated to Q = 0 yield a preliminary result of I p =(203 ± µ b, with a maximum energy coverage up to ν ≈ . ν part is estimated bya parameterization of spin structure functions g ( Q , ν ) and g ( Q , ν ), including a Regge-based constraintfor the highest ν [23].To summarize, the best estimates are compatible with the proton GDH prediction of I p = 204 . µ bwith a 10% accuracy, this one being dominated by the large- ν extrapolation, whose form is assumed toobey Regge theory, without it being verified for polarized photoabsorption. In fact, as seen in Fig. 4, theextrapolation fits do not describe the data well. This important shortcoming of the current consensus onthe status GDH sum rule can be addressed by the proposed GDH experiment in JLab Hall D. It is expected to be released in summer 2020.
Simultaneous fit of Regge parametrization, Eq. (14), to existing spin-dependent data on proton andneutron. Figure from Ref. [21]
No assessment of the neutron GDH sum rule has been published yet. We formed the neutron GDH runningintegral by using the published world data and show it in Fig. 3. The MAID model is used to estimatethe unmeasured low- ν contribution. Also shown is the GDH integral generalized to electroproduction,measured at Q = 0 .
035 GeV [22] and including an estimate of the high- ν part. How the generalizedintegral evolves to Q = 0 is an unsettled question (the state-of-art χ EFT estimates disagree). However, atsuch low Q , the evolution to Q = 0 is expected to cause only a small change. Regardless of this question,Fig. 3 illustrates the present lack of convergence and validity check of the neutron GDH integral. This proposal will utilize the polarized 12 GeV CEBAF beam to measure the high-energy behavior of theGDH sum on the proton and neutron, with two primary objectives: • The convergence of the GDH integral will be studied through a measurement of the yield difference∆ y ( ν ) ∝ ∆ σ ( ν ). This eliminates uncertainties coming from normalization factors and unpolarizedbackgrounds. • The validity of the GDH sum rule will be studied through a measurement of ∆ σ ( ν ), which will moreextensively test Regge and chiral effective field theories.Hall D, with its high-luminosity photon tagger and its large solid angle detector is uniquely suited forsuch an experiment. The experiment is not feasible in other JLab Halls since they lack the photon taggingcapability of Hall D and a GDH measurement via electroproduction is poorly matched to a study of itslarge ν domain. The measurement would have to be done at low enough Q so that a reliable extrapolationto Q = 0 could be done. However, with an 11 GeV beam, this would require a measurement at scatteringangles smaller than 0.8 ◦ , which no Hall can reach and where the elastic radiative tails are prohibitivelylarge.The Hall D measurement would extend by a factor of 4 the experimental integration range for theproton and by a factor of 7 for the neutron. It will cover the domain relevant to clarify the question ofthe convergence of the GDH integral and the validity of Regge theory for the nucleon spin structure, whileprobing for unknown parton process or structure. 11he proposal is laid out as follows. In Section 2 we sketch the derivation of the sum rule, discuss theno-subtraction hypothesis and discuss potential mechanisms that may cause the sum rule to be violated.In Section 3 we examine the experimental requirements needed for a measurement of the spin-dependentcross-section on the proton and neutron. In Section 5 we describe the simulation of the experiment signaland backgrounds. In Section 6 we discuss the expected statistical uncertainties and the implications forthe analysis of the functional forms. In Section 7 we discuss the systematic uncertainty expected on theabsolute cross-section given conservative assumptions. In Section 8 we summarize the total time requestedfor the experiment. In Section 9 we discuss the impact of the experiment on Regge phenomenology, thespin-dependent Compton amplitude, and diffractive physics. Several methods have been used to derive the GDH sum rule [21, 12]. To elucidate what the sum ruleactually tests, we outline here the derivation using the dispersion relation approach. It starts from theforward real Compton scattering amplitude F ( ν ) and utilizes causality (dispersion relation); unitarity;Lorentz and gauge invariances (low energy theorem). While studying the GDH sum rule tests all thesehypotheses, the latter two are robust and stand at the foundation of quantum field theory. The exception,the “no-subtraction hypothesis”, enters the derivation of the first item, the dispersion relation. Its validitydepends on the observable involved: for a nucleon target, it involves QCD in general and specifically thehigh-energy behavior of F ( ν ). There has been much discussion on whether the no-subtraction hypothesisholds in the context of the nucleon GDH sum rule; see, for instance, Ref. [26].The forward Compton amplitude F ( ν ) depends on the polarization of the incoming and scatteredphotons, (cid:15) (cid:15) (cid:15) and (cid:15) (cid:15) (cid:15) , respectively, and on their momenta which, for forward scattering, obey k k k = k k k ≡ kkk .Five functions f i ( ν ) can then be defined to parameterize F ( ν ) since it is a scalar quantity. For real photons k · (cid:15)k · (cid:15)k · (cid:15) = 0, which reduces the number of parameters to two: F ( ν ) = f ( ν ) (cid:15) (cid:15) (cid:15) ∗ · (cid:15) (cid:15) (cid:15) + f ( ν ) σσσ ( (cid:15) (cid:15) (cid:15) ∗ × (cid:15) (cid:15) (cid:15) ) , (4)where σσσ are the Pauli matrices. The spin-independent amplitude f ( ν ) is used in the (unpolarized) Baldinsum rule derivation [5], while the spin-dependent amplitude f ( ν ) yields the GDH sum rule. Causalityimplies the analyticity of f ( ν ) in the complex plane, yielding the Cauchy relation: f ( ν ) = 12 iπ (cid:73) f ( ε ) ε − ν d ε = 12 iπ (cid:90) + ∞−∞ f ( ε ) ε − ν d ε . (5)The right-hand side equality holds if the Jordan lemmas are valid for f ( ν ), that is, if f ( ν ) vanishes when ν → ∞ . In that case (cid:60) e (cid:0) f ( ν ) (cid:1) = 1 π P (cid:90) + ∞−∞ (cid:61) m (cid:0) f ( ε ) (cid:1) ε − ν d ε , (6)which is the Kramer-Kr¨onig relation [27], ubiquitous to all fields of physics. Crossing symmetry implies f ( ε ) = − f ( − ε ) ∗ which, applied to Eq. (6), yields (cid:60) e (cid:0) f ( ν ) (cid:1) = 2 νπ P (cid:90) + ∞ (cid:61) m (cid:0) f ( ε ) (cid:1) ε − ν d ε . (7)Unitarity gives (cid:61) m (cid:0) f ( ε ) (cid:1) = ε π ( σ A − σ P ) . (8)12 low energy theorem (Lorentz and gauge invariances, and crossing symmetry) can be used to expand f in ν : f ( ν ) = − ακ M ν + γν + O ( ν ) . (9)The derivative of Eq. (9) together with Eqs. (8) and (7) yield the GDH sum rule: df ( ν ) dν (cid:12)(cid:12)(cid:12)(cid:12) ν =0 = ακ M = 14 π (cid:90) ∞ ν ( σ P − σ A ) d νν where we have changed the dummy variable ε to ν in the integral. One of the mechanisms that could compromise the above derivation and lead to a violation of the GDHsum rule is the possibility of a J = 1 pole of the Compton amplitude [28]. Such a pole would invalidatethe Jordan lemma since (cid:60) e ( f ) would not vanish as ν → ∞ , (cid:60) e (cid:0) f ( ∞ ) (cid:1) (cid:54) = 0. But (cid:61) m ( f ) would stillvanish and thus a pole would not affect the overall convergence property of the GDH integral I . It would,however, affect the ν − dependence of ∆ σ ( ν ) since it would add a constant to the GDH relation comingfrom the contribution of circle integration that was assumed to vanish in the right-hand side of Eq. (5).This would lead to a “subtracted GDH sum rule” I ≡ (cid:90) ∞ ν ∆ σ ( ν ) ν d ν = 2 π ακ M − π (cid:60) e (cid:0) f ( ∞ ) (cid:1) . (10)As discussed in [12], a pole would be related to the behavior of a Compton amplitude at high energy.In fact, the current data also indicate that if a pole is present, it would manifest in the high- ν behavior of∆ σ : the data show that the resonance region saturates the GDH sum, (cid:90) ≈ ν ∆ σ ( ν ) ν d ν ≈ π ακ M . Thus, the additional term − π (cid:60) e (cid:0) f ( ∞ ) (cid:1) in Eq. (10) must come from the behavior of ∆ σ at higher ν . Possible causes for a GDH sum rule violation—or its apparent violation when the integral is measured overa finite ν -range—are reviewed in [12]. The ones most often considered are a) the existence of unknownhigh-energy phenomena, such as quark substructure (non-zero quark anomalous moments) [29]. b) Theexistence of a J = 1 pole of the nucleon Compton amplitude [28] as just discussed in Section 2.2; and c)the chiral anomaly [30]. All proposed mechanisms would manifest themselves at high ν . Since there is nolow- ν mechanism that could invalidate the sum rule, and since the convergence can be investigated onlybeyond the resonance region, to truly verify the sum rule, the behavior of ∆ σ at high ν must be measured. Testing the convergence of the GDH integral requires only the shape of the high energy part of the integrand.It therefore suffices to measure the yield difference ∆ y ( ν ) = N + − N − , where N +( − ) is the number of eventsin a bin ν for positive (negative) beam helicity. This quantity is insensitive to normalization uncertaintieswhich are typically dominant in experiments measuring cross-sections. Furthermore, uncertainties from theunpolarized contributions (target dilution) cancel in the N + − N − difference. For the integral to converge,13 ∆ σ/ν | must decrease with ν (baring exotic behavior such as a singular contribution at ν → ∞ ), and thusonly the ν -dependence of ∆ σ must be established in order to assess the convergence. Recall that such adecrease does not occur for | σ | and that the unpolarized equivalent of the GDH sum does not converge.Testing the validity of the sum rule will require normalizing ∆ y ( ν ) into a cross-section by measuringthe beam flux, target density, solid angle, target and beam polarisation, as well as detection efficiencies.The signal of interest is the total spin-dependent yield of photoproduced hadrons, that is simply count-ing events with at least one hadron in the final state and a reaction invariant mass greater than the nucleonrest mass. The three main ingredients needed for measuring the spin-dependent yield are: • a beam of circularly polarized tagged photons; • a longitudinally polarized target; • a large solid-angle detector. Circularly polarized photons are necessary to measure σ P and σ A . They can be generated using CEBAF’spolarized electrons with an amorphous radiator. Their polarization can be approximated by [31]: P γ ≈ P e y (4 − y )4 − y + 3 y , (11)where y = ν/E , E is the electron beam energy and P e is the electron beam polarization. P γ ( y ) obtainedby the approximation (11) and by the exact formula are shown on Fig. 5. Also shown is the effect of usingdifferent radiator materials for the exact formula. The material of the radiator is of little importance froma polarization point of view. simplewith screening E γ / E P γ E =
12 GeV, Al radiator
Al radiatorW radiator E γ / E P γ ( with screening ) P γ ( simple ) E =
12 GeV
Figure 5:
Left: photon circular polarization versus the energy fraction ν/E using the approximation given byEq. (11) (dashed blue curve), and the exact formula for aluminum (full red curve). Right: ratio between the exactand approximate calculations for different radiator materials (blue: aluminum; orange: tungsten). The results hereassumes 100% electron beam polarization and the curves on the left panel need to be rescaled by the actual electronbeam polarization, assumed to be 80% in this document.
In this proposal P e is assumed to be 80%. In terms of the figure-of-merit, the increase of photon beampolarization at higher ν more than compensates the decreasing flux and cross-section. Thus, we expect abetter statistical precision at larger ν , see Figs. 15 and 16 for the projected results.No electron beam polarimetry is presently available in Hall D. The electron beam longitudinal polar-ization can be measured at <
1% level using the injector Mott polarimeter or the polarimeters in Halls A It is relatively easy and of moderate cost to build a M¨oller polarimeter for the Hall D beam. However, this may not be
14r C. Spin precession can be calculated to few degree accuracy for a beam energy known at the 10 − level,the presently known accuracy on the beam energy in Hall D. In order to bound the time variation of thepolarization, Mott polarization measurements at the source will be necessary, if no polarization measure-ment is done in the other Halls. Calculations indicate that the depolarization resulting from synchrotronradiation and energy spread are below 1%, which is confirmed by the high beam polarizations measuredin Hall A and B at 11 GeV [32].There is presently no equipment in Hall D to monitor, record and control the electron beam helicityinformation and its charge asymmetry. Its implementation is straightforward and its cost estimated to beless than $ The photon flux is monitored by the Hall D Pair Spectrometer (PS), which has been calibrated at thepercent level by dedicated runs performed at very low current: A calorimeter is inserted into the beam tomeasure the flux by counting every photon. Two such devices are available, the Total Absorption Counter(TAC) and the Compton Calorimeter (CCAL). This is more than sufficient for the present proposal.As it exists, the PS covers a momentum range of about ± ν -range of the experiment. The energy ranges would be approximately3.0 GeV to 5.6 GeV, 4.2 GeV to 7.7 GeV and 6.6 GeV to 12.0 GeV. Alternatively, one could upgrade thePS detectors so that they cover the low energy photon flux by adding detector paddles at larger angles. Apossibly more attractive option would be to move the PS detectors closer to the PS magnet so that theycover a larger energy bite, with an associated decrease in the energy resolution of the pair.The photon energy is tagged with a resolution better than 0.5% [33] which is more than sufficient forthe present proposal. As with any experiment that measures a beam spin asymmetry at CEBAF, it is necessary to quantifypotential false asymmetries that arise from beam properties that change with the polarization orientation,i. e. helicity-correlated beam asymmetries. The experiment is insensitive to angle and position differencesat the target as it is completely azimuthally symmetric. It is, however, sensitive to potential beam motionat the collimator which may change the transmitted flux. Simulations were done of the transmission ofthe beam through the collimator for two beam sizes which bound the usual size of the beam, 0.5 mm and1 mm at the collimator. Larger transverse beam size has lower transmission but less sensitivity to beammotion. It was found that for a photon beam from an amorphous radiator (bremsstrahlung without acoherent component), both the transmission and the sensitivity to beam motion are independent of thephoton energy.For a beam size of 0.5 mm at the collimator, the transmission is well modeled by T = 0 . − . x where x is the offset of the beam from the center of the collimator in mm.The 12 GeV CEBAF Beam Parameter Tables indicate that we can achieve helicity correlated positionsat the target <
25 nm averaged over an 8 hour period. A 25 nm position difference at the Hall D collimatorwould lead to a 10 − relative change in rate if the beam is centered on the collimator and a 3 × − relative change in rate if the beam is 1 mm off center. This is negligible compared to the size of the trigger warranted solely for the purpose of this single experiment. Although we anticipate that this experiment will initiate a programusing circularly polarized photons in Hall D, we conservatively assume in this document that there will be no polarimeteravailable.
12 GeV CEBAF Beam Parameter Tables × − when dilution from the full butanol molecule is considered.We expect that no correction will be made, nevertheless it would be prudent to measure the electronbeam position in the tagger hall with the DAQ. Values much smaller than 25 nm are routinely achieved forparity violation experiments. In practice, we could tolerate position differences up to 250 nm if the beamis centered on the collimator. For this experiment, we propose to design and construct a frozen-spin polarized target that can serve asthe foundation for a new polarized target program in Hall D. Protons and deuterons in samples of butanoland d-butanol will be dynamically polarized outside the detector in fields up to 5 T and temperaturesaround 0.3 K. Proton polarizations above 90% and deuteron polarizations approaching 90% have beenpreviously demonstrated in a similar system in Hall B. Once polarized, the sample temperature is reducedbelow 50 mK, and the target is retracted from the polarizing magnet and moved into the Hall D detectormagnet. A thin, 0.5 T superconducting solenoid will be incorporated inside the target cryostat to maintainthe polarization while in transit. Data is acquired while the nuclear polarization slowly decays in anexponential fashion characterized by the 1 /e time constant T . In-beam values of T of 2800 h (1400 h)were obtained, with the polarization parallel (anti-parallel) to the holding magnetic field, in Hall B usinga 0.56 T holding field—meaning a polarization loss of 2% or less per day. The Hall B experiments werehalted about once per week to reverse the polarization, which required 4–8 hours. The greater photon fluxenvisioned for experiments in Hall D will produce a warmer sample temperature and decrease T , but thiswill be offset by the higher holding field of the Hall D magnet (1.8 T). For this proposal we assume anaverage polarization of 80%, with a 3% accuracy on the polarimetry.The target sample will be 7 mm in diameter and 100 mm long, and for optimum cooling will becomprised of multiple 1–2 mm beads of frozen butanol (C H OH) or its fully deuterated counterpart.These will be chemically doped with an appropriate paramagnetic radical for dynamic nuclear polarization,the nitroxyl radical TEMPO in the case of protons and the trityl radical CT-03 for deuterons. Assuminga 60% packing fraction for the beads, the target density will be about 0.66 g/cm for butanol and 0.73g/cm for d-butanol.A C foil will be placed upstream of the polarized sample to allow the extraction of the relativeasymmetry ∆ σ/σ and is needed to correct the asymmetry for the dilution by unpolarized target material. Although not needed to achieve the goals of this proposal, forming this asymmetry will permit a fast initialanalysis and offer a thorough verification of the main analysis method. Since the diluting material ismostly carbon and oxygen from the butanol, the dilution is essentially obtained by scaling the C foilrate by (4 + 16/12), correcting it for a small detector acceptance effect, and dividing it by the butanolrate. This factor is then used to correct the raw asymmetry. A 3 mm C foil (0.35 g/cm , or 5% of thepolarized target thickness) should be adequate. As was done in Hall B, the foil can be mounted on a heatshield a few cm downstream of the butanol sample for accurate separation based on vertex reconstruction.Dilution from the He– He bath and beam line windows is determined separately, but in the same way, byempty target runs.The anticipated total photon intensity on the target is 7 × γ/ s, leading to a heat load from e + e − pair production of approximately 14 µ W on the butanol sample. The He- He dilution refrigerator built An average of 82% was achieved during the Hall B g9a run. The carbon foil data will allow to compute the number of counts N from the unpolarized part of the butanol, i. e. thedilution D of the asymmetry. Thus, by measuring the diluted asymmetry ( N + − N − ) / ( N + + N − + 2 N ) ≡ ∆ σ/Dσ with thebutanol target, and D with the carbon foil, we obtain the physics asymmetry ∆ σ/σ . Asymmetry-based analyses are generally easier and faster than absolute cross-section analyses, and often more accurate.However, in the context of this experiment, the cross-section difference method is more accurate. In addition, asymmetryanalyses do not provide information on the absolute cross-section, this one being assumed to be available from world data. He- He coolant. The average temperature of the beads,assuming the heat load is equally shared among all beads in the beam path, can be written as T b = (cid:18) ˙ q πr α + T c (cid:19) / . (12)Here, ˙ q is the heat load on an individual bead of butanol of average radius r = 0 .
75 mm, T c is thetemperature of the coolant bath, and α is Kaptiza conduction between the bath and coolant. The latteris difficult to estimate, but we take a value α = 28 W m − K − based on a survey of the availablemeasurements and find a bead temperature of 0.18 K.The relaxation time of protons in butanol in a 0.56 T field has been measured to be T = 30 hr at aslight lower temperature of 0.15 K. Assuming T ∝ B , we estimate T should be 900–1000 h in the 1.8 Tfield of the Hall D detector. This can be increased to more than 2000 h if the 0.5 T transit coil remainsenergized while in the detector. Additionally, some of the beam heating can be alleviated using a photonbeam hardener to reduce the low energy ( <
100 MeV) portion of the flux [35]. Such hardeners have beenused at SLAC, CEA and DESY to suppress low energy bremsstrahlung photons.Still higher fluxes can be sustained if we replace the 0.5 T transit coil with a slightly stronger coil thatincreases the net field inside the Hall D detector to 2.5–3 T and improves its field uniformity to a levelsuitable for dynamic polarization ( ∼
100 ppm). Doing so would allow us to continuously polarize targetsamples inside the detector and run with one or two orders of magnitude greater photon fluxes (maximumDAQ rate and tagger accidentals permitting). In this case, radiation damage to the target material maybecome the limiting factor, but this can be countered by using more radiation resistant sample materialssuch NH and ND .Based on previous experience, two months will be required to install and test the target in Hall D.Polarization of proton samples requires 4–8 hours, and up to 24 hours is needed to reach the maximumpolarization of deuterons. A similar amount of time is required to reverse the polarization, although wemay pursue RF methods such as adiabatic fast passage to speed this process if frequent reversals aredesired. Otherwise, these will be coordinated to coincide with weekly beam studies and RF recoveries ofthe accelerator cavities. While not absolutely necessary, taking beam on both target polarization directionscan reduce systematic biases in the measured cross-section asymmetry. Replacing one sample with anotheris an additional four hours.The estimated cost of this system is about $ He- He dilution refrigerator will be needed to accommodate thepresent cryogenic capabilities of the Hall D 4 K refrigerator. This can be designed and constructed by theJLab Target Group, who built the FROST refrigerator for Hall B.
Hall D is uniquely suited at JLab to measure the total photoproduction cross-section thanks to its largesolid angle and its tagger. The standard GlueX/Hall D detector package plus the recently commissionedPrimEx- η Compton Calorimeter is assumed in this proposal. The GlueX beamline and detector are shownin Fig. 6 and described in detail in Ref. [39], with the relevant components summarized here.
Trigger
The main trigger requirement is similar to that of GlueX, i.e. set to accept most hadronic eventswhile reducing the electromagnetic background rate. Hence, the main trigger will be similar to the GlueX C.D. Keith,
Polarized Targets in Intense Beams , unpublished. E BCAL +2 E FCAL > Drift Chambers
Use of the existing drift chambers is required in order to do charged particle trackingand PID by dE/dx as is currently done in GlueX. For those forward-going particles that do not hit the StartCounter, tracking is required in addition to the Time Of Flight (TOF) wall in order to provide precisetiming that would identify the tagged photon responsible for the detected event. Tracking informationis required in order to do exclusive channel reconstruction—which is used for systematic studies of thedetector acceptance and efficiency as well as the photon beam tagging. Additionally, potentially ancillaryresults are possible for specific exclusive final states, as described in Sec. 3.4.3.
Calorimeters
The calorimeters provide detection of neutral and charged particles over polar angles from0.2 ◦ to 145 ◦ with a nearly complete azimuthal coverage. The trigger relies only on energy distribution inthe calorimeters, where the location and amount of the deposited energy required for a trigger may betuned. Neutral particles are detected by the Forward Calorimeter (FCAL) between 1 ◦ and 11 ◦ and theBarrel Calorimeter (BCAL) between 12 ◦ and 160 ◦ . The Compton Calorimeter (CCAL) covers forwardangles down to 0.2 ◦ . Section 3.4 shows that the acceptance for inclusive events is high and therefore theexpected acceptance correction will be small. Rates
Imposing a 80 kHz data acquisition limit, the 120 µ b total γp unpolarized cross-section yields atotal hadronic rate of 35.9 kHz , see top right panel of Fig. 7. The Bethe-Heitler and Compton backgroundrates are 35.8 kHz and 8.3 kHz, respectively. We neglected the small target wall and cosmic rate contribu-tions. Section 3.5 gives details on the treatment of the backgrounds. Accounting for the tagger acceptanceand efficiency shown in Fig. 8, the total usable hadronic rate becomes 14.2 kHz. The dilution factor of butanol is about 10 /
74 = 0 .
135 (proton) or 20 /
84 = 0 .
238 (deuteron), and the Carbon foil is 5% ofthe main target thickness, which yields a useful rate of 1.8 kHz and 3.2 kHz, respectively, for the proton and deuteron. (GeV) n Beam Energy 0 2 4 6 8 10 12 b ) m ( T r i g s - e + Bethe-Heitler eHadronicCompton (GeV) n Beam Energy 0 2 4 6 8 10 12 ( k H z / . G e V ) T r i g R a t e (GeV) n Beam Energy 0 2 4 6 8 10 12 b ) m ( T a g + T r i g s n Beam Energy 0 2 4 6 8 10 12 ( k H z / . G e V ) T a g + T r i g R a t e Figure 7:
Cross-sections (left) and rates (right) for the total hadronic photoproduction (black), the Bethe-Heitlerbackground (red) and the Compton background (green). The top panels show the trigger cross-sections (total cross-sections multiplied by trigger efficiency) and rates using the standard GlueX trigger. The flux is chosen so that thetrigger rate equals the present maximum DAQ rate of 80 kHz. The bottom panels show the cross-sections and ratesafter accounting for the tagger acceptance and efficiency.
The data analysis will be done inclusively by counting triggered events matched with a tagged beam photon.Although the detector does not have perfect acceptance and efficiency to detect individual particles, whenanalyzed in an inclusive fashion the combination of acceptance and efficiency for an event approachesunity. Simulation shows that the trigger efficiency for hadronic interactions is ε >
93% for ν > ε >
98% for ν > × − X radiator, produce an accidental background less than 15%,see Fig. 9.Assuming that a fraction f of the events in the prompt (in-time) peak are accidental, we can measurethis contribution using the side (out-of-time) peaks adjacent to the prompt peak. The resultant statisticaluncertainty is given by σ = (cid:114) N p + f N p M = (cid:112) N p (cid:114) fM , (13)19 (GeV) n Beam Energy 0 2 4 6 8 10 12 E ff i c i e n c y TriggerTrigger and Tag
Figure 8:
Simulation of hadronic events in the GlueX apparatus. The blue symbols show the average efficiencyfor triggering on these events with the standard GlueX trigger, plotted versus the photon energy. The efficiencyaccounting for tagging the energy of the photon with the tagger is shown by the red symbols. See Sec. 5 for details. where N p are the prompt events, f N p the number of accidental events in any peak and M the number ofside peaks used in the subtraction. For f = 0 .
15, and using M = 10 side peaks, the fractional increase inthe statistical uncertainty due to the subtraction is 0.7%. Analysis of the integral convergence requires measuring the ν -dependence of ∆ σ ( ν ), i.e. N + − N − , butnot necessarily the absolute normalization. This is achieved most easily by simply counting the numberof triggers. Unpolarized backgrounds will cancel in the difference. Any significant polarized background(from Bethe-Heitler pair production) can be corrected for. In contrast to checking the ν -dependence of ∆ σ ( ν ), performing a measurement of the GDH integral requiresan absolute polarized cross-section measurement, which will be obtained by normalizing the yield by thebeam flux, target density, solid angle, target and beam polarizations, and efficiencies.As verification of the primary analysis method, an asymmetry analysis strategy is also possible. Therelative asymmetry A = ( N + − N − ) / ( N + + N − ) ≡ ∆ σ ( ν ) / σ ( ν ) can be formed and, together with thewell-measured σ ( ν ) shown in Fig. 1, one can obtain the absolute ∆ σ ( ν ).For an analysis of the relative asymmetry –and in contrast to the ∆ σ analysis– dilution by unpolarizedtarget material must be corrected for. To provide the possibility of such analysis, a C foil will be placednear the FROST cell, see Section 3.2. It will allow to estimate the dilution. Empty target runs will alsobe necessary.
In addition to the inclusive analysis, a fully exclusive analysis will also be done to study the backgroundfrom electromagnetic processes and to verify that the acceptance and efficiency are well understood.Only a fraction of the events can be studied exclusively since it requires detecting all the final stateparticles. In an exclusive analysis, the measured energies of the final state particles provide an independentmeasure of the beam photon energy, allowing a careful check of the primary analysis method.20igure 9:
Experimental data for the timing difference between the Hall D start counter and tagger, with a 300 nAbeam on a 1.86 × − X aluminum radiator. The prompt peak compared to the out-of-time peaks indicates thatthe accidentals under the prompt peak contribute just less than 15%. We anticipate studying, at a minimum, the background processes of Bethe-Heitler pair production andCompton scattering. As examples of hadronic reactions, the detection and trigger efficiency for minimallyionizing particles can be studied with γp → pρ ; neutrons with γp → nπ + and photons with γp → pπ . Asa by-product of this analysis, the spin-dependent cross-sections of a number of different exclusive reactionsare likely to be produced for the first time at high-energy. The overall analysis will benefit greatly fromthe extensive work done by the GlueX Collaboration to understand the efficiency and acceptance of thedetector in the time leading up to the experiment. Backgrounds to the signals of interest to this proposal (∆ y ( ν ) and ∆ σ ( ν )) are those that do not cancel whentaking the helicity difference, viz the spin-dependent or “polarized” backgrounds. Unpolarized backgroundscancel in the yield or cross-section difference. They can impair the experiment only by contributing to thetotal DAQ rate: If the DAQ rate limit (presently 80 kHz) is reached due to the extra rate from unpolarizedbackgrounds, it will reduce the statistical precision of the experiment. This is indeed the case but as wewill show, the statistics remain more that sufficient. Two possible sources of polarized backgrounds exist: electromagnetic (Compton scattering, −→ γ −→ e → γe ,and Bethe-Heitler process, −→ γ −→ p → e + e − p ) and hadronic (polarized scattering from non-hydrogen, or non-deuterium, nuclei) backgrounds. However, the spin-dependent part of these backgrounds is not expectedto be significant with the FROST target, as explained below. Compton scattering
There should be no significant polarized Compton background because all theelectrons of the O and C nuclei making the butanol of the target are spin-paired. In addition, thesingle (unpaired) electrons of the hydrogen atoms will also be unpolarized since each is shared with the O of the OH-bond. There will be a small polarization of the electrons of O and C if some of thembecome ionized. The remaining unpaired electrons of the ionized nuclei will be fully polarized in the21irection of the solenoid field. However, the fraction of such electrons is estimated to be at less than the10 − level [34].Furthermore any polarized Compton contribution to the GDH sum, (cid:82) ∆ σ d ν , will be nearly fullysuppressed (and exactly suppressed for an experiment with perfect detector efficiencies and solid anglecoverage) since the GDH sum rule on the electron is expected to be zero and to have fully converged tothis null value by ν = 0 . Bethe-Heitler process
The Bethe-Heitler (BH) background can be spin-dependent. Only the pair-created leptons scattering off the polarized protons or deuterons produce an asymmetry, since the Oand C nuclei are unpolarized. We are thus not concerned here with BH on those nuclei. For a purelynucleonic target the unpolarized and polarized BH cross-section and the corresponding asymmetry can beevaluated explicitly: see [36, 37] and references therein. As an illustration, Fig. 10 shows the calculatedsix-fold differential cross-section for e + e − production on the proton with E γ = 12 GeV, in the simplifiedcase p e − = p e + ≡ p e , as well as the photon-target asymmetry, as a function of the scattering angle θ e − = θ e + ≡ θ e . In the calculation of the inelastic contributions to the polarized BH cross-sections whichwere used to compute the corresponding asymmetries, we have used the recent and stable parameterizationsof g p ( x, Q ) and g p ( x, Q ) spin structure functions published in Ref. [38]. γ p → pe + e − , E γ =
12 GeV θ e [ ◦ ] A = d ∆ σ B H / d σ B H [ % ] − p e = /cp e = /cp e = /cp e = . /cp e = . /cp e = . /cγ p → pe + e − , E γ =
12 GeV θ e [ ◦ ] d σ B H [ µ b / G e V s r ] − Figure 10:
Left: elastic (dashed curves) and elastic+inelastic (full curves) Bethe-Heitler ( e + e − ) unpolarized cross-section for E γ = 12 GeV as a function of the scattering angle. Right: Bethe-Heitler photon-target asymmetry (samecurve notation). Note that while the BH asymmetries appear to be substantial, they need to be multiplied by theunpolarized cross-section to yield the BH contribution ∆ σ ( ν ), yielding with the standard GlueX trigger configurationa (negligible) few tens of nb at most: see Fig. 13. The e + e − BH process is expected to dominate over the analogous µ + µ − BH process. If the BH processleaves the nucleus intact, the background can be corrected nearly exactly since the well-known nucleonelectromagnetic form factors are the only phenomenological inputs necessary.
Hadronic background
There will be no double-spin difference or asymmetry arising from polarizedhadronic background because all nucleons in O and C nuclei are spin-paired.22 .5.2 Spin-independent backgrounds
Although unpolarized backgrounds cancel exactly in ∆ σ , they still affect the experiment as they canconsume some of the 80 kHz DAQ capability. The hadronic background rate is estimated to be 31 kHz,assuming a 120 µ b γN cross-section, a 10 cm target length, a 3 mm thick carbon foil and the taggerefficiency shown in Fig. 8.Spin-independent backgrounds arise from photoproduction, from Compton scattering, and from theBethe-Heitler process on C and O. These have been simulated with
GEANT , see Sec. 5 for details.Compton scattering was simulated for hydrogen and then scaled by the number of protons in the target.The Bethe-Heitler process was simulated for the nucleon and scaled up using the target dilution factor.These rates will be confirmed by analyzing GlueX data on the Kapton ([C H N O ] n ) windows of theunpolarized LH target. In all, the 80 kHz data acquisition limit is shared between a total hadronic rateof 35.9 kHz, a 35.8 kHz BH rate and a 8.3 kHz Compton rate, see Fig. 7. This reduces the number ofhadronic events by a factor of 2 compared to the case where there would be no background. The situationcan be improved by upgrading the DAQ and refining the trigger. Although desirable, this is not necessarysince the present hadronic rate is already enough to provide sufficient statistics within a week (10 days) ofrunning on the proton (deuteron). As will be shown in Section 6, if the experiment is performed only at the nominal CEBAF beam energy,there will remain a substantial gap in ν coverage, between 1.8 GeV and 3 GeV, for the deuteron and thusthe neutron—and therefore also for the very desirable isospin decomposition.We therefore propose to take data with a beam energy between ⁄ and ⁄ of the nominal beam energy.This would bring the lower reach of the experiment down to 1.0–1.5 GeV. Any energy in that range wouldbe suitable. As an example, we will assume an electron beam energy 4 GeV in this document. The benefitsof such a data set would be extensive: • we would smoothly link the existing neutron world data with our Hall D data, with no energy gap; • we would obtain an overlap between our measurements at two different beam energies, in the 3 GeVto 4–6 GeV region using different regions of the tagger and different absolute polarizations of thephoton beam; • it would allow us to study the fourth and even third resonance regions for both the proton and theneutron; • there would be an overlap with existing proton and neutron data to significantly improve on thestatistical and systematic uncertainties on the GDH sum; • we could improve the determination of the Regge parameters even further and determine the minimumenergy at which the Regge phenomenology becomes applicable.While it would facilitate the analysis to take the low energy data within the same run period as thatof the nominal energy data, this might be difficult to schedule. A separated low energy run, e.g. duringsummer as it is done frequently, would achieve the same goal since the low and high energy runs have apartial energy overlap, thereby offering a means to normalize out any change in the detector performancethat would shift ∆ y ( ν ). The experiment has been simulated using the same detailed and well tested simulation chain used in GlueX.Event generators specify one or more primary vertices to be simulated, which are randomized within the23arget with timing that matches the RF structure of the beam. A simulation code, either hdgeant or hdgeant4 , tracks the particles through the experimental setup and records the signals they produce in theactive elements of the detector. The output of the simulation is further processed to account for detectorinefficiencies and resolutions and to overlay additional hits from uncorrelated background events. Thesimulation uses the same geometry definitions and magnetic field maps as used in real events reconstruction.The geometry includes the full photon beamline, from the radiator to the photon dump. The simulatedevents are then processed with the same reconstruction software as used for the real events [39].Simulations have been done of the trigger acceptance and tagging efficiency for various processes,considering whether the standard GlueX trigger would fire for each and whether the electron which radiatedthe beam photon would be tagged. The tagging efficiency below about 7.8 GeV drops to about 45% becausethe tagger is designed only to sample this region of the spectrum rather than detect all electrons. -3 -2 -1 -1 E γ , GeV P ho t op r odu c t i on c r o ss s e c t i on σ , m b Total γ p → p π o γ p → n π + γ p → p π + π - no res γ p → p ρ o γ p →Δ ++ π - γ p → p π o π o γ p → n π + π o γ p → p ηγ p → p π + π - π γ p → n2 π + π - γ p → p π + π - Sum
Figure 2: The total photoproduction cross section (the red solid curve) and the partialcross sections for the reactions used at the energies below 3 GeV. The sum of all thesepartial cross sections (the green dotted curve) matches the total cross section very well,below 2 GeV. At 3 GeV the sum is about 30% smaller than the total cross section. Forthe simulation, all the partial cross sections were normalized to keep their sum equal tothe total cross section.
Figure 11:
For energies below 3 GeV, a model, comprising 11 processes with low multiplicity, is able to describeboth the total cross-section and the individual cross-sections.
Fig. 8 shows the acceptance for the hadronic events that are the signal in this experiment. Thehadronic event generator, called bggen , is a standard and well tested tool in the GlueX analysis. It isbased on Pythia [40], but includes additions that describe the low-energy photoproduction cross-sectionsin the resonance region. Fig. 11 shows the model used for ν < https://github.com/JeffersonLab/halld_sim https://github.com/JeffersonLab/HDGeant4 Diracxx software package. This is ageneral-purpose toolkit for use within the CERN/ROOT framework for computing cross-sections and rateswith all polarization observables under the control of the user for both incoming and outgoing particles. Inthe simulation, the Bethe-Heitler process is treated in a fashion fully consistent with tree-level QED, takinginto account the polarization of the photon and both space-like and time-like form factors of the protontarget. Internal radiative corrections are not currently included, but external radiation is automaticallytaken into account by the Geant4 tracking library. (GeV) n Beam Energy 0 2 4 6 8 10 12 E ff i c i e n c y TriggerTrigger and Tag
Figure 12:
Average efficiency for triggering on Bethe-Heitler e + e − events as a function of photon beam energyaccounting for tagging the energy of the photon with the tagger (red symbols) and without accounting for the taggerefficiency (blue symbols). Fig. 12 shows the efficiency for triggering on and tagging Bethe-Heitler events using the standard GlueXtrigger. Use of the trigger brings the rate down to the same order as the hadronic triggered rate. (GeV) n Beam Energy 0 2 4 6 8 10 12 b ) m ( A s - P s Figure 13:
Spin-dependent total cross-section for Bethe-Heitler e + e − events which fire the standard GlueX trigger,plotted as a function of photon beam energy. Fig. 13 shows the polarized total cross-section for Bethe-Heitler e + e − events which fire the standardGlueX trigger and have the beam photon tagged (red events in Fig. 12). On average these events have a https://github.com/rjones30/Diracxx (GeV) n Beam Energy 0 2 4 6 8 10 12 E ff i c i e n c y TriggerTrigger and Tag
Figure 14:
Average efficiency for triggering on (and tagging) Compton events with the standard GlueX trigger,plotted as a function of photon beam energy.
Fig. 14 shows the acceptance for Compton scattering from atomic electrons, γe → γe . Only about 7%of Compton events trigger the detector and are tagged.These simulations will further be used to study the relative fractions of the various processes in order tooptimize the trigger condition for the experiment to reject background while maintaining high acceptancefor the hadronic events of interest.To conclude, the simulation shows that the polarized background contribution is very small and thus,once corrected with the same tools as used here, will be entirely negligible. ν -dependence of ∆ σ To estimate the beam time necessary for the measurement, we use a total collimated photon flux of 7 × s − . Such flux can be obtained with the currently available 1 . × − X aluminum radiator (1.64 µ m),240 nA electron beam current and the standard 5 mm collimator. To determines the trigger rate, we use: • the flux between ν = 3 and 12 GeV: 1 × s − for the nominal energy run and 2 . × s − between ν = 1 and 4 GeV for the low energy run . • a 80% for the detector/trigger efficiencies above ν = 7 . ν = 2 . • a 80% for the electron beam and target polarizations.For ∆ σ , we use the Regge form σ P − σ A = Ic s α a − + c s α f − , (14) for the low energy run, the beam size contraction due to the Lorentz boost effect is smaller. The consequent lower photontransmission through the Hall D main collimator is accounted for by a reduction of 10/28 in the photon flux compared to thenominal case. s = 2 M ν + M , I = ± is the isospin sign of the proton or neutron, and with values c = − . µb , α a = 0 . c = 209 . µb , α f = − .
66 [21].
The highest available CEBAF beam energy is optimal to study the GDH sum rule. We assume 12 GeVwill be available and we suppose that one week of running on hydrogen is a minimum given the investmentof two months to install the target. If ∆ σ indeed follows Eq. (14), then running 7 days on the protontarget and 10 days on the deuteron target yields a similar statistical precision for the neutron and protondata, see Figs. 15 and 16, and allows for an optimal isospin analysis, see Fig. 17. The neutron informationobtained from the deuteron and proton data can be extracted straightforwardly: at our large ν , in thesmooth continuum region past the resonances, the deuteron binding (2 MeV) and Fermi motion (115 MeV)can be ignored. The usual formula to extract the neutron information, n = D/ (1 − ω d ) − p with ω d = 5 . d -state, is expected to be valid. Hence, no issue regarding nucleareffects is expected for the isospin separation. The expectation for the deuteron is shown in Fig. 18.This simulation yields statistical uncertainties on the intercepts of ∆ α a = ± .
007 and ∆ α f = ± . α a = ± .
23 and ∆ α f = ± .
22 extractedfrom the ELSA data [21]. We are comparing here to results from the best fit to the photoproductiondata. The intercept values can also be obtained from low- Q electroproduction data, and with higherstatistical precision, see e. g. the recent determination α a = 0 . ± .
04 of Ref. [14]. However, systematicuncertainties are associated with such extraction, in particular regarding what should be the highest Q values acceptable for a Regge-type fit, and the assumption that the data are Q -independent. Thus, ourprojected results are expected to significantly improve, statistically and systematically, the intercepts valuesderived from photoproduction and low- Q electroproduction. In addition to taking data at the highest CEBAF energy available, it is also beneficial to run with a lowerenergy beam. This allows to bridge the gap between the data we proposed to take and the existing worlddata. The gap is especially large for the neutron and it is clear from Fig. 3 that in order to test accuratelythe GDH sum rule, closing this gap is important. The Hall D low energy data will also greatly improve theworld data quality and offer cross-check between two fully independent experiments. Finally, it providesan avenue to normalize our relative yield to obtain the absolute cross-section difference ∆ σ in case ofunforeseen issues in the determination of the normalization factor in Hall D. Figures 15, 16, and 18, showthat 10 days of data taking shared between proton and deuteron at beam energy of e.g. 4 GeV will providesufficiently precise data to cross-check/normalize with the ELSA data. ν -dependence of ∆ σ It is important to recognize that the choice of the fit form in Figs. 15–18 is only a working hypothesis basedon the leading theory expectation. Interpreting the data with a Regge-based form is not a requirement.The high precision of the data, the high-density binning and the large ν -range will allow a clean extractionof the behavior of ∆ σ ( ν ) regardless of the actual theory driving its ν -dependence. This is illustrated bythe results in Table 1 where various functional forms are used to fit the data shown in Fig. 17 (generatedassuming Regge behavior). As the χ values reveal, the other forms fail to fit the data satisfactorily forthis isospin analysis. (Again, the form aν b used in Fig. 17 is just a working hypothesis. The analysis canbe carried out regardless of the actual ν -dependence of ∆ σ ). As discussed, correlated systematic uncertainties causing a global offset of the yields do no contribute the total uncertainty. (GeV) σ P - σ A ( µ b ) Hall D, 4 days of 4 GeV beamHall D, 7 days of 12 GeV beamELSA-10-505101520 2 4 6 8 10 12 ν (GeV) σ P - σ A ( µ b ) -6-5.5-5-4.5-4-3.5-3-2.5-2-1.5-1 2 4 6 8 10 12 Figure 15:
Left: ∆ σ on the proton from ELSA high- ν data (squares) and expected results from Hall D using a 12GeV beam (red) and a 4 GeV beam (blue). The plain line is the best fit to the simulated 12 GeV data shown in redand based on the Regge form of Eq. (14). It yields α a = 0 . ± .
009 and α f = − . ± .
037 for the interceptsof the a and f Regge trajectories. Only the statistical uncertainty is relevant to determining the intercept values.The systematic uncertainties, expected to be at the 5% level, are not shown. Right: zoom on the expected Hall Ddata. ν (GeV) σ P - σ A ( µ b ) Hall D, 4+5.7 days of 4 GeV beamHall D, 7+10 days of 12 GeV beamELSA102030405060 2 4 6 8 10 12 ν (GeV) σ P - σ A ( µ b ) Figure 16:
Same notation as in Fig. 15, but for the neutron extracted from deuteron data (statistical uncertaintyonly, the systematic ones being unimportant). The best values for the intercepts of the a and f Regge trajectoriesare α a = 0 . ± .
013 and α f = − . ± . (GeV) I s o l s c a l a r σ P - σ A ( µ b ) ν (GeV) I s o vec t o r σ P - σ A ( µ b ) Figure 17:
Isospin decomposition of ∆ σ . Top: isoscalar part, with best value for the intercepts of the f -mesonRegge trajectory α f = − . ± . a -mesonRegge trajectory α a = 0 . ± . ν (GeV) σ P - σ A ( µ b ) Hall D, 5.7 days of 4 GeV beamHall D, 10 days of 12 GeV beam11010 Figure 18: ∆ σ for the deuteron expected results from Hall D (statistics only.). The best fit to these simulated datais ∆ σ = 450(34) s − . . Examples of functional forms used to fit the simulated data in Fig. 17, and resulting χ /d.o.f. For theRegge case (second row) χ / d . o . f ≈ χ /d.o.f. show that the proposed measurement precision and ν range are sufficient to determine accuratelywhat type of functional form the data may follow. Fit form χ /d.o.f (isoscalar case) χ /d.o.f (isovector case) aν b a + bν a + bν + cν a + b log ν ae bν + c ν -range is reduced by half, e.g. to span only the region from 5 to 9.6 GeV, thendifferent functional forms may describe the data equally well, see Table 2.Table 2: Same as Table 1 but with an experimental ν -range reduced to half. Fit form χ /d.o.f (isoscaler case) χ /d.o.f (isovector case) aν b a + bν a + bν + cν a + b log ν ae bν + c ν dependence of ∆ σ For studying the convergence of the GDH integral, it is sufficient to obtain the high- ν behavior of the yielddifference ∆ y ( ν ) = N + − N − , and since the data at various ν are taken concurrently, an accurate absolutenormalization of σ P − σ A is irrelevant. Thus, the accuracy on this goal of the experiment depends onlyon the uncertainties affecting the ν dependence of ∆ σ . It can be assessed with hdgeant4 , see Fig. 8. Thetagger channel inefficiencies cancel in the flux normalization and thus do not contribute. ∆ σ For studying the validity of the GDH sum rule, an absolute ∆ σ is necessary. Our primary method toobtain it will be by performing a standard cross-section analysis, except that A) the target dilution andunpolarized backgrounds need not to be corrected for as they do not affect the cross-section difference. (Thevery small polarized background contribution can be corrected for, see Section 5), and B), no knowledge ofthe target density is necessary due to a ratio cancellation between target polarimetry and absolute cross-section normalization. Hence, target density uncertainty does not contribute to the total uncertainty. The30ystematics uncertainties associated with this method are: • Beam polarization: δP e = 3%, with 2% due to precession and knowledge of beam energy, 1% duesynchrotron radiation depolarization and 1% from Mott/Hall polarimeters. • Target polarization (without target density uncertainty contribution): δP t = 3%. • The photon flux uncertainty, δφ < • The combination of absolute detector, trigger and DAQ efficiencies is assumed to be known to within2-3%.This yields an estimated total systematic uncertainty of 5.0%.The absolute ∆ σ can alternatively be obtained by measuring the asymmetry A = ( N + − N − ) / ( N + + N − ) and using the well-measured unpolarized cross-section σ to provide ∆ σ = 2 σA . We can use thequicker relative asymmetry method as an on-line analysis method and later as a check of the primarymethod. We assume the following values for uncertainties: • Electron beam polarization: δP e = 3%. • Target polarization (including target density uncertainty contribution)): δP t = 4%. • Target dilution: δD = 3%. • Unpolarized cross-section σ (from world data): δσ = 1%.This yields a total uncertainty of 5.9% slightly larger than the absolute ∆ σ analysis, but comparable. Possible false asymmetries related to the beam and target polarizations can be minimized by flippingthe target spin and reversing the beam helicity assignment with the beam half-wave plate. There is nosingle-spin longitudinal asymmetry for photoproduction reactions (in contrast to electroproduction). Anon-uniform acceptance of the apparatus in the polar direction may induce a bias in the data if the σ P and σ A cross-sections have different polar angle dependence. Reversing the target spin once during theexperiment to get two data sets of opposite raw asymmetry sign will ensure than the above asymmetry (andother possible ones) cancels when the two data sets are combined. Reversing the target spin is relativelyeasy and fast ( ⁄ day) and it will add robustness to the final result. Table 3 summarizes the proposed schedule and beam time request. The incentive for starting the experi-ment with the deuteron is two-fold: 1) it is faster to switch from deuteron to proton (12h) than the reverse(36h); and 2) a spin dance to confirm that the beam precession angle is known should be done as early aspossible, and the deuteron asymmetry is expected to be larger than that of the proton.To obtain a comparable statistical precision for the proton and neutron requires the deuteron run timeto be ∼ √ c and c of Eq. (14) of about 5%. This is comparable to theexpected systematic uncertainty of about 5% on the absolute cross-section, thereby making 7+10 days anoptimal run time for the measuring the absolute ∆ σ . (For the other main goal of the experiment –theintercept measurements– the systematics uncertainty is negligible and their precision will be given by the2-4% statistical uncertainty.) We also request 4 days (proton) + 5.7 days (deuteron) of beam time for thelow beam energy run.To minimize the systematic uncertainties, the target spin will be flipped (12h) once for each target. Thetarget will be repolarized to its optimal value during the spin-flip process and the target NMR polarimetryrecalibrated (additional 12h). For the absolute cross-section determination, it is necessary to calibrate eachof the three Pair Spectrometer configurations that would be necessary to span the full energy range of the31agger. This would require three 4-hour “TAC” runs. Finally, to allow for the possibility of an asymmetryanalysis—for which, in contrast to ∆ σ , the unpolarized background needs to be corrected for—another0.5 day of empty target data taking at 12 GeV and a 0.3 day at 4 GeV are necessary. In all, the aboveprogram requires 29.1 days of beam and 4 days without beam for target and beam configuration changes.Table 3: Beam time requested and overhead, listed chronologically. The beam current for production is 240 nA.
Time (day) Target Goal/Remarks10 Deuteron Main production at 12 GeV0.3 Deuteron Spin dance done during above task1 Deuteron Target spin-flip/repol./NMR calib.No beam, done at middle of production0.5 He For background subtraction.Includes target change overhead1 Deteuron → proton switch No beam. NMR calib.7 Proton Main production at 12 GeV1 Proton Target spin-flip/repol./NMR calib.No beam, done at middle of production0.5 Pair. Spec. converter Absolute flux calib.
12 GeV: 21.3 total time at 12 GeV5.7 Deuteron Production 4 GeV0.3 Deuteron Spin dance done during above task0.3 He For background subtraction.Includes target change overhead1 Deuteron → proton switch. No beam. NMR calib.4 Proton Production at 4 GeV0.5 Pair. Spec. converter Absolute flux calib. total time at 4 GeV Total: 33.1 total experiment time
Studying the convergence properties of the GDH integral in the Regge ( ν > σ ( ν ) data at high ν will improveour knowledge on both the imaginary and real parts of the spin-dependent Compton amplitude f ; itwill provide new information on the poorly known intercept of the a Regge trajectory; it will yield thefirst non-zero polarized deuteron asymmetry in the diffractive regime (assuming current predictions forthe nucleon polarized rates in that regime), thereby providing for the first time a non-zero value for theisosinglet coefficient of ∆ σ ; it will reduce the uncertainty of the polarizability contribution to 1 S hyperfinesplitting of hydrogen; and it will provide a photon-point benchmark to study the transition between thewell-understood DIS dynamics of QCD to the lesser-known dynamics of diffractive scattering that will beexplored with the EIC [41]. These seven items are discussed separately in the following.32 .1 Convergence of the GDH integral If it is found that the data obey the Regge theory, ∆ σ ∝ ν b , in the measured ν -range, one can extrapolatethis behavior to larger ν . The integral will converge if b < | b | ≈ . ν -dependence of ∆ σ ( ν ) is expected above 3 GeV, a definitive statement on the convergence is expectedregardless of whether the data obey the Regge expectation or not. We emphasize that the finding thatRegge theory fails in the spin sector would be very significant by itself. Assuming the validity of the Regge theory (or alternatively of the GDH sum rule) we expect to measurebetween 3 to 12 GeV a contribution to the proton GDH integral I p of about − µ b with negligible statisticaluncertainty and a 1 . µ b systematic uncertainty, see Section 7.2. This would change the current assessmentof I p from: (cid:0) ± ±
12 (syst) ±
10 (large- ν projection) (cid:1) µ b to (cid:0) ± ±
12 (syst) ± ν projection) (cid:1) µ b.Thus, the total systematic uncertainty will decrease from 16 µ b to 12 µ b. The precision of the sum rulewill be reduced from 16 /
205 = 8 % to 12 /
205 = 6 %, a relative improvement of 25% on the precision atwhich the GDH sum rule for the proton is currently tested.
There is presently no assessement on the validity of the GDH sum rule on the neutron. Our data com-plementing those of MAMI and ELSA will offer the first test, with a precision comparable to that of theproton. f ( ν ) The spin-dependent Compton amplitude f ( ν ), also denoted by g ( ν ) in literature, is a complex quantitywhose imaginary part is determined by ∆ σ , see Eq. (8), and will thus be measured directly by the ex-periment. Fig. 19 (top) shows the world data on (cid:61) m ( f ) for the proton, extracted from ∆ σ measured atMAMI and ELSA.It is a lovely feature of the proposed experiment that it allows us to access Compton physics and helpsus to constrain other pertinent unpolarized and polarized observables without resorting to a dedicatedCompton setup. Specifically, once (cid:61) m ( f ) is obtained from ∆ σ , the real part of the spin-dependentamplitude, (cid:60) e ( f ), can be determined from (cid:61) m ( f ) by using Eq. (7) [43]. The reliability of this extractionis shown by the violet error band in Fig. 19, and strongly depends on the quality of (cid:61) m ( f ) (blue errorband). It is clear that both error bands increase as ν reaches the upper portion of the previously coveredenergy region, and will continue to do so at higher ν unless high-quality data will be made available. Ourdata will extend the ν -coverage and permit this symbiosis of (cid:60) e ( f ) and (cid:61) m ( f ) to six times its presentreach. 33 . . . . . Re g Im g MAMI + ELSA .
00 0 .
05 0 .
10 0 .
15 0 .
20 0 .
25 0 .
30 0 .
35 0 . c PT Re g c PT Im g ⌫ [GeV] g [ µ b · G e V ] Figure 19:
The proton spin-dependent Compton amplitude f ( ν ), denoted g in the figure. Top: real and imaginaryparts, the latter fitted to GDH data, the former calculated via dispersion relations. Bottom: comparison to NNLO χ EFT calculation at low ν indicating that the measured (blue band) and calculated (dotted green line) imaginaryparts differ appreciably at energies around 0.25 GeV while the real parts (obtained by integrating the imaginaryparts over the energy domain with ν as the integration parameter) agree perfectly at low ν . This reflects the peculiarfeature of the theory that low-energy quantities are well described, even though they are obtained as loop or dispersiveintegrals which include higher-energy domains where the theory is inapplicable. Figure from [42]. If both (cid:60) e ( f ) and (cid:61) m ( f ) are known precisely enough (and given f , which is well measured), the twocomplex amplitudes can be used to determine d σ/ dΩ and the beam-target asymmetry Σ z in the forwardlimit, i.e., d σ dΩ (cid:12)(cid:12)(cid:12)(cid:12) θ =0 = (cid:12)(cid:12) f (cid:12)(cid:12) + (cid:12)(cid:12) f (cid:12)(cid:12) , Σ z | θ =0 = − (cid:60) e ( f f ∗ ) | f | + | f | , where θ is the Compton scattering angle and −→ z is along the initial photon direction. Of these, Σ z | θ =0 ismost interesting since the asymmetry for circularly polarized photons and nucleons polarized along the z axis, Σ z = d σ P − d σ A d σ P + d σ A , provides information on all four spin polarizabilities appearing in Compton scattering. In particular Σ z and its behavior near θ = 0 are very sensitive to chiral loops [44]. The product of the unpolarized cross-section and Σ z for θ = 0 is shown in Fig. 20 (top) together with its uncertainty, which increases rapidly for ν (cid:38) σ ( ν ) in the ν range covered in Hall D will significantly reducethe uncertainty on Σ z .The analysis [42] was performed for the proton only. In addition to improving it with our higher- ν high-precision proton data, our neutron and deuteron data will motivate the same type of analyses for theseobjects. χ EFT is an important effective approach to QCD that should describe it at low energies andmomenta. However, the dedicated JLab low Q experimental program to test χ EFT with spin observables34 . . . . . .
00 0 .
05 0 .
10 0 .
15 0 . fit I of s abs fit II of s abs c PT .
20 0 .
25 0 .
30 0 .
35 0 . ⌫ [GeV] d s d W l a b S z [ nb ] Figure 20:
Unpolarized differential cross-section multiplied with the Σ z asymmetry for the forward Comptonscattering off the proton, showing (top) two distinctive fits of the unpolarized photoabsorption cross-section and itsuncertainties, and (bottom) the χ EFT calculation. Figure from [42]. is showing that their description is a challenge to χ EFT [15]. Thus, providing further tests of χ EFT withnew spin observables or/and in a different regime is critical and can be achieved with the present proposal. a Regge trajectory
In Regge theory, the high-energy behavior of the isovector (non-singlet) cross-section difference is drivenby the a (1260) Regge trajectory such that ∆ σ ( p − n ) ∼ s α a − , (15)where typically α a ≈ .
4, as obtained from fits to DIS data. A very recent such fit [45] resulted in α a ≈ +0 .
45 while the Regge expectations is α a ≈ − .
34 (see Eq. (16) below). Another recent fit,combining both electroproduction and photoproduction data [14], yields α a = +0 . ± .
04, i. e. alsofinds that the sign of the a (1260) intercept is opposite to the theoretical prediction. The situation issummarized in the Table below. α a DIS fit (approx. values) 0 . . ± . − . α a is partly that a (1260) is the only I G ( J P C ) = 1 − (1 ++ )meson to form a “trajectory”, while the second candidate, the a (1640), has been omitted from the PDGSummary Tables as it still needs confirmation. A precise measurement of ∆ σ at high ν for both protonand neutron targets would help to remove this uncertainty. This is an important question to resolve as theintercept is predicted to be given by α a = 1 − α (cid:48) m a , (16)35here α (cid:48) = 1 / (2 πσ ) ≈ .
88 GeV − and σ is the string tension, which is known to be approximately0 .
18 GeV . If α a were indeed ≈ .
45 as suggested by the present DIS data and the (relatively low- ν )photoproduction data, this would imply α (cid:48) ≈ .
44 GeV − and a string tension more than twice as high asthe value commonly accepted and obtained from hadron spectroscopy.As shown in Section 6, if ∆ σ obeys the presumed Regge behavior, the experiment would determine α a at a level of 2%, an improvement in precision of a factor of 25 compared to the present 54% uncertaintyobtained from the best fit to the world data. Since only null asymmetries have been measured by COMPASS, CLAS and SLAC for the deuteron in thelow Q , high- ν , regime relevant to Regge theory, the deuteron coefficient 2 c (see Eq. (14)) that factors the s -dependence of ∆ σ p + n is assumed to be zero in analyses [13, 14]. From the Regge expectation, a non-zerodeuteron asymmetry, i. e. ∆ σ p + n (cid:54) = 0, should be unambiguously measured by this experiment, see Fig. 18,yielding a clear non-zero 2 c = 450 ± A valuable impact of the measurement concerns the effect of proton structure on the hyperfine splittingin hydrogen. The importance of this topic has been emphasized by the “proton radius puzzle” [46]. Thehyperfine splitting is given by E HFS ( nS ) = [1 + ∆ QED + ∆ weak + ∆ structure ] E Fermi ( nS ) , (17)where the proton-structure correction can be separated into three terms: the Zemach radius, the recoilcontribution, and the polarizability contribution:∆ structure = ∆ Z + ∆ recoil + ∆ pol . (18)The current relative uncertainties of the three terms are 140 ppm, 0.8 ppm and 86 ppm, respectively,which need to be put into the perspective of the forthcoming PSI measurement of E HFS whose precision isexpected to be as low as 1 ppm. Our proposed measurement can contribute to the uncertainty reductionof ∆ pol . It can be written as ∆ pol = α em m π (1 + κ ) M [ δ + δ ] , (19)where m is the lepton mass (muon in the case of muonic hydrogen where the effect is easiest to measure).Here δ involves an integral of the spin structure function g ( x, Q ) over both x and Q , δ = 2 (cid:90) ∞ d QQ (cid:18)(cid:26) · · · (cid:27) + 8 M Q (cid:90) x d x g ( x, Q ) (cid:26) · · · (cid:27)(cid:19) , (20)while δ involves a similar integration of g ( x, Q ) (see Eq. (6.43b) of [43] for full expressions). The GDHintegrand at general values of ν and Q , expressed in terms of the polarized nucleon structure functions g ( ν, Q ) and g ( ν, Q ), is∆ σ = − πα em M ( ν − Q / M ) (cid:18) g ( ν, Q ) − Q ν g ( ν, Q ) (cid:19) . (21)At low Q (and for real photons), g is irrelevant, hence a precise measurement of ∆ σ , such as it will be pro-vided by our experiment, directly constrains δ via g . To compute δ , one indeed needs the Q -dependence36f g , i. e. use input from electron-scattering, but since the integrand, as seen in Eq. (20), is weighted by1 /Q , knowing the value at Q = 0 from our real-photon measurement would be extremely beneficial instabilizing the integration. Such a stabilization is essential, as the 86 ppm uncertainty mentioned aboveneeds to be reduced to ≈ g needs to be improved by two ordersof magnitude. As already quoted from Ref. [10], “above the resonance region [...] the real photon is essentially absorbed bycoherent processes, which require interactions among the constituents such as gluon exchange between twoquarks. This behavior differs from DIS, which refers to incoherent scattering off the constituents.”
Thatis, there is a transition between the DIS regime and the very low- x or real photon regimes of diffractivescattering. Studying this transition has been an important part of the ZEUS and H1 programs at HERA,and it remains a very active field of research [47]. However, it is currently limited to unpolarized scattering.The polarized case and its connection to photoproduction is discussed in Ref. [13] and will be exploredwith the EIC [41]. ➿➿➿➿➿ e - (cid:1) e - qqP ➿➿➿➿ ➿➿➿➿➿ ➿➿➿➿➿➿ g g g ɣ * ℙ (unpolarized) or ℝ (polarized) ɣ * (cid:1) (cid:1) (cid:1) ➿➿➿➿➿➿ (cid:1) (cid:1) P Figure 21:
Diquark picture of low- x electron-proton scattering, from the higher Q hard regime (left) to the low Q soft regime with Pomeron or Reggeon exchange (right). ➿➿➿➿➿ e - (cid:1) e - q qP g ɣ * ɣ * (cid:1) ➿➿➿➿ Figure 22:
Another possible processcontributing to electron-proton scatter-ing.
The usual theoretical description of diffractive scattering is thediquark picture: the hard virtual photon emitted by the scatteredlepton hadronizes into a q ¯ q pair of coherent length 1 / ( xM ). Athigh enough Q , each quark exchanges a gluon with the proton, seeFig. 21, left panel. As Q decreases, gluon rungs on the gluon ladderappear (Fig. 21, central panel), as well as gluons exchanged betweenthe q and ¯ q . At low Q , the interaction between the coherent q ¯ q pair and the proton is summed into pomeron ( P ) and reggeon ( R )exchanges (Fig. 21, right panel). Other processes contributing to P and R exchanges exist, such as the one shown in Fig. 22. Thisdescription connects to the usual DIS parton model, e. g. with thegluons in the left panel of Fig. 21 representing the gluon PDF.The pomeron has the vacuum quantum numbers (isoscalarcharge singlet). Being spin 0 allows P to couple to the proton components irrespective of their helicity. P thus controls unpolarized diffractive scattering. Doubly polarized −→ e (cid:48) −→ P scattering filters out P exchangesto reveal the non-singlet R exchange. This filter will be used for the first time at the EIC. This proposedmeasurement of ∆ σ , expected to be also controlled by Regge theory, will provide a Q = 0 baseline to thisstudy of the transition from the hard dipole partonic picture to the soft R exchange picture.37 We propose the first measurement of the high-energy behavior of the integrand ∆ σ/ν of the GDH sumrule, a fundamental relation of quantum field theory whose validity depends on the internal dynamicalproperties of the particle to which the sum rule is applied. The measurement would be performed in HallD, the only place suited for carrying out a high energy GDH measurement, using a FROST target and alongitudinally polarized electron beam on an aluminum radiator. The high- ν domain is where the sum rulemay fail. In fact, the unpolarized equivalent of the GDH integral does not converge, both for proton andneutron. This could be observed only from high- ν data, ν > ν = 12 GeV, would allow us to study the convergence property of the GDHintegral. This can be achieved by a quick and robust analysis since unpolarized backgrounds cancel in ∆ σ and no absolute normalization is needed. Then, once the absolute normalization is determined, the HallD data added to the world data at lower energy will make a relative improvement of 25% on the accuracyat which the sum rule is tested on the proton, and provide for the first time a test of similar accuracy forthe neutron.In order to fill a large gap of missing neutron and deuteron data, it is necessary to perform a shortermeasurement at lower energy, with the beam energy between 4 GeV and 6 GeV. This is required to extractthe neutron integral without the use of a model or extreme interpolation. This will also allow us toprecisely determine the minimum energy at which the Regge phenomenology is valid, as well as to constrainsystematic uncertainties related to relative polarization of the photon beam and tagger geometry. It willprovide an overlap between the Hall D data and the existing world data and allow ∆ σ to be significantlyimproved in the higher resonance region for the proton.In addition to studying the convergence and sum rule validity and independent of that study conclusion,the data will constrain our knowledge of diffractive QCD, whose phenomenology is unverified in the spinsector. As pointed out in [12], not even a model prediction is available for the magnitude of the J = 1pole effect, due to our absence of knowledge of polarized diffractive QCD. In fact, results from fits ofphotoproduction data and of DIS data independently disagree with the Regge theory expectation for thesign of the Regge trajectory intercept driving the isovector part of ∆ σ . The experiment will clarify thisproblem.Given the Regge theory expectation, the experiment should measure the first non-zero asymmetry signalfor the deuteron in the diffractive regime, thereby providing for the first time a non-null determination ofthe coefficient that factors the s -dependence of ∆ σ p + n of the deuteron.Analyzing ∆ σ ( ν ) using dispersion relation techniques will provide f ( ν ), the spin-dependent forwardCompton amplitude. This will further clarify the convergence property of the GDH sum rule, whichdepends on both the real and imaginary parts of f , and will test χ EFT. The latter is especially importantsince tests of χ EFT with polarized observables by the JLab low- Q spin sum rule experimental programrevealed that currently, χ EFT has difficulties to consistently describes spin observables [15].Furthermore, the experiment will provide a Q = 0 baseline for the EIC data. This will be helpfulin particular for the study of the transition between the DIS regime characterized by partonic degrees offreedom to the diffractive regime characterized by effective degrees of freedom such as the pomeron andthe reggeon.Finally, the data will constrain the polarizability contribution to the hydrogen hyperfine splitting.A first goal of the experiment is to map with high precision the energy dependence of ∆ σ on theproton and neutron. This will determine whether ∆ σ follows the expected Regge behavior and if so, thevalues of the isovector and isoscalar Regge trajectory intercepts will determine if the integral converges.Only point-to-point uncorrelated errors contribute to the Regge intercept uncertainties, which guarantiesa fast and robust analysis. Other goals for the proposal require absolute normalization. The necessary38nformation (e.g. polarization) will be gathered concurrently. However, the convergence test does notrequire the absolute normalization and thus will have much reduced uncertainties compared to the absolutemeasurement.With 27 days of measurement (10 days on deuteron at 12 GeV and 5.7 days at 4 GeV, and one weekon proton at 12 GeV and 4 days at 4 GeV) plus 6 days for systematic studies and target changes, andassuming that Regge behavior is observed, the data will provide the Regge trajectory intercepts at the2–4% level, compared to the 50% uncertainties at which they are presently known.Once a polarized target is available in Hall D, a rich experimental program will open. For example,several possible experiments have been discussed in an earlier LOI [48]. A GDH experiment would initiatesuch a program with a comparatively simple set-up and robust observables. References [1] A. Deur, S. ˇSirca and J. Stevens, Measurement of the high energy contribution to the Gerasimov-Drell-Hearn sum rule, LOI12-19-002 to PAC47.[2] PAC 47 comments on LOI12-19-002.[3] S. B. Gerasimov, Sov. J. Nucl. Phys. , 430 (1966) [Yad. Fiz. , 598 (1965)]; S. D. Drell andA. C. Hearn, Phys. Rev. Lett. , 908 (1966); M. Hosoda and K. Yamamoto, Prog. Theor. Phys. (2), 425 (1966).[4] M. Tanabashi et al. (Particle Data Group), Phys. Rev. D , 030001 (2018).[5] A. M. Baldin, Nucl. Phys. , 310 (1960).[6] J. Ahrens et al. (GDH and A2 Collaborations), Phys. Rev. Lett. , 022003 (2001), [hep-ex/0105089];H. Dutz et al. (GDH Collaboration), Phys. Rev. Lett. , 192001 (2003); Phys. Rev. Lett. , 032003(2004).[7] G. M. Gurevich and V. P. Lisin, Phys. Part. Nucl. , 111 (2017).[8] S. D. Bass and M. M. Brisudova, Eur. Phys. J. A , 251 (1999), [hep-ph/9711423].[9] A.H. Mueller, T.L. Trueman, Phys. Rev. , 1306 (1967).[10] D. Drechsel and T. Walcher, Rev. Mod. Phys. , 731 (2008), [arXiv:0711.3396 [hep-ph]].[11] J. D. Bjorken, NATO Sci. Ser. B , 1 (1987).[12] R. Pantforder, [hep-ph/9805434].[13] S. D. Bass and A. De Roeck, Eur. Phys. J. C , 531 (2001), [hep-ph/0008289].[14] S. D. Bass, M. Skurzok and P. Moskal, Phys. Rev. C , 025209 (2018), [arXiv:1808.03202 [hep-ph]].[15] A. Deur, S. J. Brodsky and G. F. De T´eramond, Rep. Prog. Phys. , 076201 (2019),[arXiv:1807.05250].[16] S. Hoblit et al. (LSC Collaboration), Phys. Rev. Lett. , 172002 (2009), [arXiv:0808.2183 [hep-ex]].[17] Helicity Structure of Pion Photoproduction; E04-102. D. Crabb, M. Khandaker and D. Sober(spokespersons)
Measurement of π + π − Photoproduction in Double-Polarization Experiments using CLAS; E06-013.M. Bellis, V. Crede and S. Strauch (spokespersons) [19] J.P. Chen et al.
CLAS proposal E94-117.[20] P. Bosted et al.