IL I~UOVO CIMEI~TO VoL. XX I I I , N. 3 1 o Febbraio 1962
Search for the Decay tx+-~e++e-+ e + (*).
S. ~)ARKER ~nd S. PEI~AS~ (**)
The Enrico Fermi Institute for Nuclear Studies The University o] Chicago - Chicago, Ill.
(ricevuto iI 4 Settembre 1961)
Summary . - - A search for ~+~ e+e--~-e + was conducted using counter techniques. The number of mesons stopped was 109; when multiplied by the efficiency for detecting a three-electron decay, the number became ].7.107. Five events were seen and are probably due to the decay mode lJ.+~ e+e-+e+-}- ~-?~. Even assuming them to be due to ~+-+ e++e--~e +, a new upper l imit of 5.10 -7 is set for the branching ratio of this mode.
1. - In t roduct ion .
Al though normal muon decay is a we l l -understood Fermi in teract ion , there
are several puzz l ing quest ions that e~n be raised. One is the absence of the
dec~y scheme ~+ -~ e++ e- e +, which would obey lepton conservat ion and, wi th in the f ramework of p resent ly unders tood theory , should be as common
as the customary dec~y mode. Un iversa l i ty and conservat ion of fermions alone
do not expla in its absence. What is needed is ~ change in the theory , such
as the addi t ion of a special selection rule (x), a rest r ic t ion to charged currents (~),
(*) Research supported by a joint program of the Office of Naval Research and the U. S. Atomic Energy Commission.
(**) Now at Columbia University, New York. (1) K. N~sHIJI~tA: Phys. Rev., 108, 907 (1957); T. ADACI~I and S. NAKAI: Progr.
Theor. P/~ys., 22, 889 (1959). (2) t~. P. F~Y~.~rAN and M. G~LL-MAsN: Phys. Rev., 109, 193 (1958); E. C. G. SU-
DARSHAN and R. E. MARSHAl(: Phys. l~ev., 109, 1860 (1958); J. J. SAKURAI: 2(UOVO Cimet~to, 7, 649 (1958); J. SCHwIN(~: A'~n. o/ Phys., 2, 407 (1957).
486 s. PARKER and s. PENMAN
or a different assignment of lepton numbers (~). I t has been suggested that weak interactions take place through the coupling of a vector boson to the lepton fields (4). Unobserved decays such as ~->e+e-+e +, K-+7:+ + ~++ ~-, K -+ 7 :- /e -+ e +, and ~+ A p -~ e-/A ~, would be forbidden if the boson were charged. At present, there is no experimental evidence requiring the existence of such a boson, and the absence of an observable rate for the process ~-+e+y militates against it. But this last observation is not con- clusive because the neutrinos associated with muons and electrons may not be identical.
In the absence of a clear understanding of the origin of these restrictions,
it was considered desirable to establish their firmness to as high a degree as possible. There is some reason to believe that the branehing ratio of ~+-~ e++e-+e + may not be zero, even if weak interactions take place only through a charged current. Just as the intermediate boson should give rise to the process ~ ~+e~-y, so should it imply existence of the three-electron decay process by internal conversion. (See Fig. l-a), b), c) (5)). Since the 7
is not physical, it is pos- e ~ sible that the three-elec-
e- tron decay would not be suppressed to the same
. de ee as e Some W 9 unsuspected mechanism
a) b) may be responsible for the nonappearance of
-+ e,;. e+'[/~e_ If the branching ratio
e + ~- ~., ~ -+ e@e-~-e is nonzero, ~ then the branching ratio
~ "-9 ~ ~ for the other should not
c) d) be larger or smaller by a Fig. 1. - Feynman diagrams for ~+-+ e++e-~-e +. factor greater than about
137. This is so because
diagrams for one can be formed from diagrams of the other by the addition of an e-~ vertex. Our present knowledge of weak interaction theory does not permit us to say which of the two decay modes would be more probable.
(3) E. J. KONOPINSKI and H. M. IV~A]-IMOUD: Phys. Rev., 92, 1045 (1953); I. SAA- VEDRA: Nucl. Phys., 10, 6 (1959).
(4) 1~. p. FEYNMAN and M. G~LL-MAN~-: see footnote (2); E. C. G. SUDA~SI~AN and R. E. MARSHAK: see footnote (2); j. SC~WING~R: see footnote (2).
(5) ~[. BANDER and G. FEINBERC~: Phys. t~ev., 119, 1427 (1960).
SEARCH FOR THE DECAY tJ,+-~e++e-+e + 487
There is also the possibility of second order weak interactions. In this process the neutrinos emitted in the ninon decay annihilate each other, pro- ducing an electron pair. This pair, in conjunction with the electron from the muon decay, would give rise to a three-electron decay (see Fig. l-d)). At first glance it would appear that a ~( doubly weak )~ interaction could not possibly give rise to a physically observable process. However, an approximate calculation of the second order process in muon decay gives rise to a badly divergent result because a large number of intermediate states are available. Whether such a calculation has any meaning whatsoever is debatable. If a cut-off is applied, the branching ratio predicted is 4.10-1~.(eut-off energy/nu- cleon mass) ~ (6). Since the rate is directly proportional to the fourth power of the cut-off energy, a different choice could lead to an observable process. There is, of course, no reason to suppose the nucleon mass characterizes the process in any way, and the whole question is perhaps best considered as not understood.
With these considerations in mind, a search for the three-electron decay mode of the muon was undertaken. Previous work has been done with bubble chambers and emulsion (7), but in the 7.10 s decays that have been scanned, no ~- . ed +e-d -e + events have been seen. The present experiment utilized counter techniques. Its form was dictated by a great increase in the duty factor at the University of Chicago 450 MeV cyclotron, an increase that re- sulted from the recent adoption of a vibrating target system (s).
2. - Exper imenta l set-up;
The primary function of the apparatus was to measure the rate of triple coincidence between charged particles emerging from a target in which muons were decaying.
A 65 MeV ~+ beam emerging from the main shielding was magnetically analysed, and directed first through a two foot thick lead wail and then into an iron and lead pipe. The exit of this pipe was covered with beam scintil- lator No. 1 (see Fig. 2), After passing through a carbon moderator and beam scintillator No. 2, the pious were stopped in a circular target scintillator.
(6) H. CREw: private communication. (7) G. LYNCH, J. OREAR and S. ROSENDORFF: Phys. Rev. Lett., 1, 471 (1958);
I. GUREVICH, B. NH;OL'SKIJ and L. SURKOVA: Sowiet Physics JETP, 10, 225 (1959); J. LEE and N. P. SAMOS: Phys. Rev. Lett., 3, 55 (1959); R. R. CRITTJ~NDEN and W. D. WA~KEn: Phys. Rev., 121, 1823 (1961); Y. KRESTINIKOV: 2gintl~ Annual Con]. o~ High Enerqy Physics (Kiev, 1959), unpublished.
(s) j. ROSEN: Nevis Report no. 92.
488 S. PARKER ~nd S. PENMAN
Three counter telescopes whose center lines bisected ~11 equilateral trinngle were arranged around the target.
"II t eam/ ,cmfillotors X
Lead [---7 CarbOQ I ron
5ecbon 4-4 i
Beam pipe on and Lead)
. . . . . . . . . . . . . . , , , , . . . . . . . . . . , . . . . . . . . . i ,43
f I A~ C3
~End view Y
Fig. 2. - Counter geometry.
The electron telescopes each consisted of two counters in coincidence with c~rbon absorber between them. A third lurge anticoincidence counter was on
SEARCH :FOR THE DECAY n+-~e++e +e+ 489
the outside of the array. A lead and iron absorber placed between the anti- coincidence counter and the rest of the telescope prevented electrons from a true three-electron decay from tripping the anticoincidcnce counter. As will be described later, the efficiency of the telescope for electrons in the energy range of interest was quite high. Outside the anticoincidence counter was a plate of lead two inches thick. Its purpose was to convert y-rays originating outside of the apparatus and to insure that they tripped the anticoincidcnce counter. If they converted in the absorber between the antieoincidenee counter and the second electron counter, they could give rise to a shower which would appear to be a three-electron decay. A fine time resolution among the three telescopes was required since the principal source of background is given by acci- dental coincidences. Because good time resolution and high efficiency tend to be antagonistic, the principal time resolution was accomplished by photographing the counter pulses on an oscilloscope. The scope was triggered by an electronic system designed to meet the requirement of a high and predictable efficiency. The method of photographing scope traces also permitted the recording of a great deal of other information about each coincidence.
The oscillography system was, therefore, the heart of the experiment. Two scopes were photographed simultaneously: a four-beam fast scope and a two- beam Tektronix 551 (see Fig. 3). Fast coincidence between telescopes was
+1oo ~4Hz ' , - " / I anb ~
7 ns/cm 1.1 ps/cm
Fig. 3. - Osc i l loscope t races ~or ~+~e+d-e -d -e +.
accomplished by displaying the pulse from the first counter in each electron telescope on each of three traces. Only one counter was displayed, since the technical difficulty of adding counter pulses without degrading rise time and introducing reflections was sufficient so the additional information did not warrant the effort. In particular, by applying the output of the 56AVP photo- tube directly to the deflection plates of the 4-beam scope tube, sufficient pulse height was obtained to obviate the need of additional amplification. Such would not have been the case if mixing networks had been used. Displaying the counter pulses on separate traces permitted arranging delays so that all the pulses arrived at the same time. As a result, sweep speed variations were
490 s. PARKER and s. P~NMAN
not important, as they are in the usual serial display. The counter signals were delayed with approximately 400 feet of l~G63U.
The 4-beam fast scope was adapted from a Lawrence Radiat ion Laboratory design (9) and used a Dumont K1479P11M 4-beam tube. This tube is char- acterized by a very high writing rate and a moderately good vertical de- flection sensitivity. As in the Lawrence Radiation Laboratory design, all operating voltages were taken from the Tektronix 517 on which the 4-beam tube was mounted, but an additional sweep amplifier was added to achieve a sweep speed of 7 ns/em.
The target used to stop inconfing pions was a piece of scintillator 3 inches in diameter and ~ inch thick. The output of the phototube viewing the target was displayed on the fourth fast oscilloscope trace. On the same trace, but with opposite polarity, a summation signal from all the anticoineidence coun- ters was displayed. Distr ibuted amplifiers were used in the display system. The gain in this system was made very high so that pulses much too small to operate the electronic circuits would still be displayed. This could result in a slight reduction in efficiency due to random tube noise in the anticoinci- denee counters, but measurements showed this reduction to be negligible. In all, 311 pictures of accidental coincidences and background events were col- lected during the experiment and in no case was an anticoineidence pulse present. This indicated that the electronic circuit threshold was low enough to reject all events arising from a particle entering from outside. The gain of the anticoincidence display circuit was set to saturate the amplifiers wher~ a few hundred kilovolts of energy were deposited in the anticoincidenee scin- tillator.
The slow displays yielded information about the previous history of the beam, and thus permitted the rejection of events where too many particles had entered within a few muon life-times. The sweep speed of the 2-beam scope was 1.1 izs/cm. Pulses from the target counter were shaped, delayed by about 12 ~s, and displayed on the upper trace. After pulse shaping and delays, pulses from the first counter in the incident beam telescope were dis- played on the lower trace, in the downward direction. Because of the elaborate collimation, no charged particle could emerge from the cyclotron shielding and enter the apparatus without passing through this counter. The pulse pair resolution t ime on the incident beam and target counter pulse shapers was in the order of 100 ns. The rise t ime of the delay cable used (Columbia I tH 2000) was slower than this, but pulses occurring within the rise t ime of the cable appeared as a larger pulse. When the vibrating target was used, instantaneous rates became sufficiently low that delay cable resolution was not a problem.
(9) H. G. JACKSON: Rev. Sci. Inst,r., 29, 527 (1958).
SEARCH I~'OR THE DECAY F+~e+e-+e + 491
The vibrating target yielded mesons in two bursts. The first, lasting 600 ~s, corresponded to the burst with a stat ionary target but with no more than 30% of the intensity. The second, lasting nearly 8 ms, contained between 60% and 70% of the beam. Thus two different instantaneous rates charac- terized the incident flux. To identify the t ime period during which a par- ticulal" event occurred, a marker pulse was placed on the upper trace. A pulse which gave the sum of all the outer electron coincidence counter signals was shaped, delayed, and presented on the lower trace, in the upward direction. This permitted the exclusion of events where a pion had previously scattered out of the beam and into the absorber between the outer electron counter and the anticoincidence counter. Such particles could eventual ly give rise to electrons which could trip two counter telescopes, making an accidental triple coincidence possible with only two electrons. Also, such an electron could conceivably shower and i l luminate all three electron counters, although that never happened in the actual experiment.
tri( from l vi/~ ~ting $
~rig. +gate A1 j
_ _ _T rl,cope I I A3+B3+C3 Idelayl Ida
A2 3 B1 B2
T4- I %-, tT t ,
(shaDers) Fig. 4. - Simplified electronics block diagram.
CI C2 ~-3
ex t. trig.
The electronic system (see Fig. 4) consisted of coincidence circuits for each of the electron telescopes and a coincidence circuit to nlonitor the 1 2 and 1 2 T rate on the incident beam telescope. The electron telescope coincidence
492 S. PARKER ~nd S. PENMAN
circuits were slightly modified versions of a Lawrence Radiation Laboratory design and were set to have high efficiency and a flat-topped timing response about 10 ns wide. The necessity for high efficiency stems, of course, from the fact that the efficiency for a three-electron decay is the product of the efficiencies of the individual telescopes. Under system checks a method for determining the coincidence efficiency is described.
The electron coincidence circuits were followed by transistorized post-discri- minators which fed a slow (60 ns) coincidence circuit. This utilized saturated transistors and was designed to yield 100 /o efficiency with only modest t ime resolution....