VOLUME 19, NUMBER 15 PHYSICAL REVIE%' LETTERS 9 OCTOBER 1967 LEPTONIC DECAYS OF VECTOR MESONS: THE BRANCHING RATIO OF THE ELECTRON-POSITRON DECAY MODE OF THE RHO MESON J. G. Asbury, * U. Becker, William K. Bertram, *j P. Joos, M. Rohde, and A. J. S. Smith* Deutsches Elektronen-Synchrotron, Hamburg, Germany C. L. Jordan and Samuel C. C. Ting Department of Physics, Columbia University, New York, New York/. (Received 25 July 1967) At the Deutsches Elektronen-Synchrotron (a) BH background must be kept small, as (DESY) 6.2-GeV electron synchrotron, exper- its subtraction contributes to (1) a statistical iments are in progress to investigate the lep- uncertainty as well as an uncertainty caused tonic decays of vector mesons. ' We report by lack of precise knowledge of inelastic car- here a measurement of the branching ratio bon form factors. At symmetry, for a given — pair invariant mass M=2/8 is the momen- B= 1 (p' —e+e )/I'(p' —~+v ), (p turn of the electron and positron, 0 the angle from which the constant be coupling g p may of each with respect to the y beam), the BH obtained directly. To obtain (1), the rates of counting rate varies roughly as 9, whereas the two reactions the p'- e+e pair rate varies as 8 '. y+C-C+ p' (2) Therefore, L e++e- ( p - e+e )/BH = 8' and p' and the BH background can be reduced by ob- y+C-C+ -e+e ~ ~'+~ serving p at large angles. (b) The process e -e e contributes to the were measured under the same kinematical measured e e rate at the ~ mass. However, conditions and with the same apparatus, a sym- by using a spectrometer with a mass resolu- metric double-arm magnetic spectrometer tion comparable with the ~ natural width, one which has been described previously. ' The would see any sizable contribution from this measurement of (3) is the subject of the pre- process as an enhancement in the e+e invari- ceding Letter, s which is referred to as I in ant-mass spectrum. Similarly, a large inter- what follows. ference effect between p and ~ could be detect- To first order, both the Bethe-Heitler (BH) ed, as it would cause a distortion in the shape and Compton diagrams contribute to the reac- of the e+e mass distribution. However, there tion are arguments that the rate of p'- e+e- » rate y+C -C+e++e (4) of e —e+e: Bubble-chamber data show for H2 that e photoproduction =10% of po produc- For the case that the lepton pair e+e has tion. If p and ~ are produced by similar mech- an invariant mass M &600 MeV/c', the Comp- anisms, ~ this holds for carbon as well. Then ton term has been estimated and measured' the SU(6) prediction' of the ratio to contribute less than a few percent to (4) for pair production angles 8 & 7'. Therefore, for y 2:y 2:y '= 9.1:2 (6) p (d cp M. & 600 MeV/c' Reaction (4) can be used as a sensitive test of quantum electrodynamics gives the result that e-e+e is only 10% of at small distances. ' As first suggested by p'- e+e Berman and Drell, ' however, the Compton (c) To obtain an accurate value of the branch- contribution will be greatly enhanced if I is ing ratio (1) both the po production cross sec- equal to the mass of a vector meson (p, tion and the p polarization must be known over or p); hence (4) can also be used to study the the kinematical region for which e+e pairs leptonic decays of vector mesons. are accepted. For this experiment, system- For a measurement of the branching ratio atic errors caused by the p' cross-section mea- (1) to be meaningful, three major error sourc- surement were reduced by measuring both (2) es must be reduced or avoided: and (3) with the same apparatus. Also, the VOLUME 19, NUMBER 15 PHYSICAL REVIEW LETTERS 9 OCTOBER 1967 p polarization was determined indirectly as 11. ~(thear. ) implied in I, the conclusion being that p pro- duction at forward angles proceeds via diffrac- 10. tion dissociation, the p', therefore, being com- 1) Asbury ~taL, PR. L. 18, 65(1967) 2) 8~=9.54, 9 =2.250 GeV/c pletely transversely polarized. The angular 3) 8.=15', A=1,400 GeV/c 4) 8a=l5, PI= 1.950 GeV/c distribution W(8*) in the p' rest system is then determined: For pion pairs, W (8*)=(3/87r ) sin'8*, (7) 7|'7r and for the electron pairs, = W (8*) (3/16m ) (1+cos'8*) (6) i2) 0 (8* is the angle in the p' rest system between 0 200 600 1000 M~ (Me Y/C'J the and di- decay products the incident y-ray = rection). FIG. 1. The ratio 1t (Experimental yield)/(Calculat ed BH yield) is shown as a function of the invariant On the basis of considerations (a)-(c) above, mass of the e+e- pairs. the experimental arrangement was chosen to minimize systematic errors as follows: First, on the basis of ten events, for the spectrom- the largest central pair opening angle consis- eter setting 8, =15', P, =1.95 GeV/c (M, =1020 tent with a reasonable counting rate was cho- MeV/c'). Here the observed e e pair rate sen, namely 30'. The uncertainty in inelastic is =7 times that expected from BH; this gives form factors then introduces an error =5'fo in strong evidence that the decay y - e+e has the BH subtraction. Second, hodoscopes (de- been observed. scribed in Ref. 2) giving a mass resolution of The mass spectrum of e+e events is shown +15 MeV/c' were used to enable &u —e+e con- in Fig. 2(a). The BH background has been sub- tributions to be observed. Finally, as described tracted, and all corrections have been applied. in I, errors in the p' production cross section The spectrum is in agreement with the modi- due to the of nonresonant m+m- back- presence fied Breit-signer mass distribution' of p -m n ground were reduced to +10% by the careful of 1 (1" =130+5 MeV' and M =765 p p MeV/c'). study of Reaction (3) for different target elements, The smoothness and approximate symmetry momentum transfers, and incident peak brems- of the distribution indicates that the cu - e+e strahlung energies k . (The bremsstrahlung contribution is (5%. beam is described in Ref. 2.) The branching ratio (1) is now obtained by = The e+e pair rate at 8, 15' and p, = 1.400 dividing the integrated yield of Fig. 2(a) by the GeV/c (central pair mass Mo =750 MeV/c2) yield of p mesons produced in the same accept- was much larger than that predicted by quan- ed region; this latter yield is shown in Fig. tum electrodynamics (QED). To show that this 2(b). Taking the polarization factors given apparent deviation is caused by an enhancement in Eqs. (7) and (8) into account, we obtain the in the Compton diagram and not by a fundamen- result' tal QED breakdown (modification of Feynman B = (6.5 + 1.4) x 10 propa, gator or vertex), a run was taken at 8, =9.5', p, =2.25 GeV/c (M, also =750 MeV/c'). As a direct application of our knowledge of As shown in Fig. 1, the ratio the behavior of p' photoproduction cross sec- tions on carbon IEq. preceding Letter] Experimental yield (2), R= and B determined in this experiment, the Bethe -Heitler yield p photoproduction amplitude (2) is determined increases rapidly with angle, in a manner con- experimentally to be of the form sistent with Eq. (5), demonstrating that the 2at. ip 2 2 large pair rate for M=750 MeV/c2 can be ex- A =D(k)pe ie /(m -m im 1"), - plained without invoking a fundamental break- p p down of QED. where ie'+ is an arbitrary phase and where Also shown in Fig. 1 is the value of R obtained, D(k) normalizes the amplitude to the observed 870 VOLUME 19, NUMBER 15 PHYSICAL REVIEW LETTERS 9 OCTOBER 1967 zero-degree p' photoproduction cross section. In the region of the p' mass the dominant virtual Comp- — ton scattering amplitude is therefore of the form -A ~ . A —with A —being the AC y~p py-- y~ee+e y ~+e amplitude for timelike y-& pairs. The interference between the BH and Compton processes is then described by the cross section &at 2 g 7W P YP G (q')E E 'G (k) int dE dE dQ dQ , m2 (E E i / 1 1 && 2m I I+21 + I[-,''~ (E E )+k.p E k.g, E ] j (kp kp j + — + — — + ie z /Qe, Qp ~ ——(P P +P P )I — -I D(k)ae (9) ~ +. —. +& -& ip k m 2-m2-zm I p p where k = photon four-momentum; G(k) = brems- strahlung energy spectrum; M = mass of tar- The interference cross section dajnt is asym- get nucleus; Q =k+P p-p (re—coil four-mo- metrical under interchange of four-momenta mentum of nucleus); P = initial four-momen- of electron (p ) and positron (p+), and thus tum of nucleus; and t=(k —P+ —P )'. The met- can be observed by taking the difference between ric chosen is@»=1, g;;= —1 (i =1, 2, 3), and two asymmetrical yields for spectrometer set- the constant deter- tings of opposite polarity.