Z. Phys. C-Particles and Fields 35, 143--150 (1987)
Zeitschfift
Patriots
Eir Physik C
and F Ls
9 Springer-Verlag 1987
First Measurement of the --, E- -, A n-)
Branching Ratio*
S.F. Biagi 6", M. Bourquin 3, R.M. Brown 7, H.J. Burckhart 4b, Ch. Dor6 s, p. Extermann 3, M. Gailloud 5,
C.N.P. Gee 7, W.M. Gibson 1, R.J. Gray
7
p. Jacot-Guillarmod s, P.W. Jeffreys 7, W.C. Louis 7 c,
P. Miihlemann 3 d, R.C. Owen 1, J. Perrier 3, K.J.S. Ragan 3, Ph. Rosselet s, B.J. Saunders 7, p. Schirato 3 e,
H.W. Siebert 4, V.J. Smith 1, K.-P. Streit 2 f j.j. Thresher 7, R. Weill 5, A.T. Wood ~, C. Yanagisawa 7 g
1 H.H. Wills Physics Laboratory, University of Bristol, Bristol BS 8, 1 TL, UK
2 CERN, CH-1211 Geneva, Switzerland
3 Universit6 de Genrve, CH-1200 Geneva, Switzerland
4 Physikalisches Institut, Universidit Heidelberg, D-6900 Heidelberg, Federal Republic of Germany
5 Universit6 de Lausanne, CH-1015 Lausanne, Switzerland
6 Queen Mary College, University of London, London E1 4NS, UK
7 Rutherford Appleton Laboratory, Chilton, Didcot OX11 0QX, UK
Received 23 January 1987
Abstract.
In an experiment performed at the CERN
SPS charged hyperon beam, we have observed the
radiative decay ~- --, S- 7. From a sample of 11 can-
didates, containing an estimated background of i.6
events, the branching ratio is found to be:
F(~- ~22-7)/(~-~An-)=(2.3+_ 1.0) x 10 -4.
1 Introduction
The interest of the radiative decay ~--~ S 7 (as well
as the decay f2---*Z-7) has been emphasized by
many authors. Within the quark model framework,
two-quark transitions (Fig. 1 a), where a W boson is
exchanged between two quarks of the initial baryon,
are believed to play an important role in 22 + ~PT,
A --* n 7 and other radiative decays of hyperons. How-
* Work supported in part by the UK Science and Engineering
Research Council, the Swiss National Foundation for Scientific Re-
search and the Bundesministerium fiir Forschung und Technologie,
Federal Republic of Germany
Present address:
" Now at University of Liverpool, UK
b Now at CERN, Geneva, Switzerland
Now at Princeton University, N J, USA
d Now at ZeUweger-Uster, Hombrechtikon, Switzerland
e Now at Hrpital Cantonal Universitaire, Geneva, Switzerland
f Also at Physikalisches Institut, Universitfit Heidelberg, FRG
g Now at State University of New York at Stony Brook, NY, USA
a)
U
w
u ~ ~ d
u,d,s m, -~ u,d,s
B)
w
/\
s u,c~,tg u
--
d
,c,t
u,d,s ~-~ usd, s
u,d,s = m, u,d,s
c)
Y
u,d,s ~ ~ ~s
u,d,s ~ ~
~S
d)
T~-
y
_-=--
A
Fig. ! a-d. Diagrams used to calculate the hyperon radiative decay
amplitudes: a two-quark transition, b Penguin diagram (gluon ex-
change) - the photon can be emitted either from an external quark
line or from the internal quark loop. c One-quark transition, d Pole
model with two-particle intermediate state
144
ever, it is clear that these transitions should be much
less important in 2---+ 27-7 or in O---, E-y, since
neither of the initial baryons contains a valence u
quark. Contributions to the Z- and s radiative de-
cay amplitudes can result from "penguin" diagrams
(Fig. 1 b) which involve gluon exchange, or from sin-
gle-quark diagrams (Fig. 1 c). It has been noted that
QCD corrections (one gluon loop added to Fig. 1 c)
are important for single-quark diagrams. Once they
have been taken into account, the contribution of
these diagrams is much larger than that of the pen-
guin diagrams for the Z---+ X-7 decay [1-4].
The decays Z --+27-7 and (2---+Z-7 have also
been studied theoretically within the framework of
the pole model, with two-particle intermediate states
(Fig. 1 d), and lower bounds for their branching ratios
have been derived from the unitarity principle [5]. The
theoretical predictions are summarized in Table 1.
It is interesting to notice that in the case of the if- --+
27-7 decay, the lower limit from unitarity significant-
ly exceeds the sum of the two contributions calculated
from the quark model.
On the experimental side, only upper limits on
the ~-~Z-7 and (2-~-7 branching ratios have
been published so far (see Table 2). The result on
the 3 radiative decay mode comes from a bubble
chamber experiment at the Brookhaven AGS [6],
which observed no candidate in a total sample of
8150 Z- produced by incident K- at 1.75 GeV/c. The
information on the (2- radiative decay mode was ob-
tained by two counter experiments in the CERN SPS
hyperon beam, with incident O- of 115 GeV/c [7]
and 131 GeV/c [8]. Upper limits were derived from
the observed distributions of the ((2 -Z-) missing
mass combined with information on photon energy
and impact position, in events with an identified
(2- --+ Z- + X decay.
Table 1. Theoretical predictions for the branching ratios of radiative
decays of E- and ~- hyperons
Models Ref. E --+2;- 7 ~2 --+~-7
Penguin [2] 10-8-10 -v 10-5-10 -4
Single quark [4] 1.8 x 10 -5 6.8 x 10 -5
Pole Model [5] 1.7 x 10 -4 1-1.5 x 10 -s
Unitarity [5] >l.0x 10 -4 >0.8 x 10 5
Table 2. Experimental upper limits on branching ratios (at 90%
C.L.)
Authors Ref. 2- --+ 2; - ~ ~- --+ Z- 7
Yeh et al. [6] < 1.2 x 10 -3
Bourquin et al. [7] < 3.1 x 10 -3
Bourquin et al. [8] <2.2 x 10 -3
We have recently reported [9, 10] measurements
of the branching ratios of the radiative decays 27 +
P 7 and A -+ n 7 in two experiments carried out at the
CERN SPS hyperon beam. In the second of these
experiments, the A were produced by the decay of
incident Z- hyperons, and the final state of interest
consisted of a neutron, a photon and a 7c-meson.
The process (Z- --+ Z-7, L'- --+ nrc-) is characterised
by precisely the same initial (N-) and final (n, 7, ~-)
particles and, if it exists, can therefore be found in
the data gathered for the A ~ n 7 experiment. We have
indeed observed this process and we report, in the
present article, the first measurement of the corre-
sponding branching ratio (Z --+ 27- 7)/(Z- --+ A 7c-).
2 The Experiment
The experimental set up is shown in Fig. 2 (for a
detailed description of the apparatus, see [9, 10]). The
incident beam particles had a mean momentum of
116 GeV/c. The ~- hyperons were identified by a dif-
ferential Cherenkov counter (DISC); their trajector-
ies were measured in two clusters of multi-wire pro-
portional chambers (labelled A in Fig. 2). Trajectories
and momenta of charged decay particles were mea-
sured in a magnetic spectrometer consisting of a mag-
net (SMI), and clusters of MWPCs (B, C, D, F) and
drift chambers (DC). The scintillator hodoscopes H 2
and H 3 were used in the trigger logic.
Photons emitted at large angles were detected by
a lead glass array (LG) consisting of a lead converter
(2.7 radiation lengths), a MWPC, a scintillator ho-
doscope and a lead glass wall (21.2 r.1.). The MWPC
measured the impact point of photons which con-
verted in the lead plate. The pulse height measured
in the scintillator hodoscope was used to correct for
shower energy deposited in the lead converter and
also to improve the discrimination between photon
and hadron showers.
Neutrons and photons emitted at small angles
were detected by a liquid argon detector (LAD) at
the downstream end of the apparatus. The discrimi-
nation between photons and neutrons was made pos-
sible by the internal structure of the detector: it was
divided in three parts along the beam direction, the
first two made of 40 and 80 lead plates respectively
(4.5 and 9.0 radiation lengths), and the third one of
104 copper plates (2.18 absorption lengths). The
charges produced by ionisation were picked up by
collector strips oriented alternately in the vertical and
horizontal directions.
The purpose of the last magnet (SM2) was to
deflect beam particles to an insensitive region in the
LAD, to reduce accidental background. Helium bags
X
Yvertical
Z
DC
C
Di ---"~'--
--'""' "~176 NIHI
2 t t
Vl Tl T2 M
I
7-/777,,
H2
DC
SM2
I
Dl•e
ca.y 7
region
,H3
/
ch IL i
I
E
//////,4
2m
LAD
-1
i
-2
145
0 5 110 115
2'0
25 3t0 3; rn
Fig.
2. Schematic layout of the apparatus: A, B, C, D, E, F: clusters of MWPCs; DC: clusters of drift chambers; M: multiplicity counter;
SM 1 and SM 2: magnets; H 2-H 3 : scintillator hodoscopes; LG: lead glass array; LAD: liquid argon detector; Ch: threshold Cherenkov
counter, N- : incident hyperon beam; DISC: differential Cherenkov counter. Vt, Tt, T2 : scintillation counters. Coordinates are measured
in a frame where X is horizontal, Y vertical and Z along the nominal beam axis with the origin at the centre of the last quadrupole
magnet of the beam
were installed in front of the LAD, between the last
two E chamber clusters, and inside the two magnets,
to reduce photon conversion probability. The multi-
plicity counter (M) and the threshold Cherenkov
counter (Ch) were not used in the present experiment.
The trigger demanded an incident Z- (defined by
beam scintillators and the DISC) and only one
charged decay particle (defined by a single hit cell
in H2). Up to two cells were allowed to fire in H 3
to allow for possible back-leakage of showers in the
LG array. These trigger conditions accepted the ra-
diative decays Z- ~ Z- ? and ~'- ~ A re-, A ~ n 7 (ex-
cepted those in which a photon converted upstream
of H 2), but also the far more abundant decays S-
A ~-, A--* n~ ~ A very small proportion of the 22
hyperons in the beam triggered the DISC acciden-
tally; they constituted about 1% of our triggers. The
trigger conditions rejected the Z- ~ A ~ -, A ~prc-
decays if the A decay occured upstream of H 3.
For a total intensity of 1.6 x 106 particles per SPS
burst, 200 N- were tagged by the DISC. A total of
1.44 x 106 events which satisfied the trigger require-
ments were recorded on tape during 10 days of run-
ning.
3 Geometrical Reconstruction and
Initial Event Selection
In order to have enough lever arm to measure the
incident ~'- and the outgoing rc =, we looked for
events where both the ~'- and the Z decayed be-
tween the last cluster of beam chambers and the sec-
ond cluster of drift chambers. For this category of
events, a Monte Carlo simulation shows that the pho-
ton strikes the LAD in about 60% of the cases and
the LG in 25% of the cases, whereas the neutron
always hits the LAD. The geometrical reconstruction
of the events was done in two main steps: reconstruc-
tion of the tracks of the charged particles in the spec-
trometer and reconstruction of showers in the calo-
rimeters. In each step, a first selection of events was
carried out based on the topology of the reconstruct-
ed tracks and showers.
The incident particle track was reconstructed in
the beam telescope. We required a single beam track,
with measured angles within _+ 0.25 mrad of the mean
beam direction in both horizontal and vertical projec-
tions in order to reject triggers associated with =-
decays occuring upstream of the last beam chamber.
146
The momentum of the incident particle was deter-
mined to within ___ 2% from its direction in the hori-
zontal plane, using the characteristics of the beam
optics.
The secondary track was reconstructed in the drift
chambers upstream and downstream of SM1. The
two segments were then associated, allowing a mea-
surement of the particle momentum with a typical
precision of ___2%. We required a single negatively
charged particle in the spectrometer, with measured
trajectory and momentum. The track reconstruction
procedures are described in [9] and [10]. At our beam
energy, the maximum angle between the =- and 2;-
trajectories is 1.1 mrad, and our angular resolution
was in most cases insufficient to establish the presence
of a kink on the primary track. We therefore made
no attempt to reconstruct the intermediate (S-)
track.
In order to reject events in which the S- or the
Z- decayed downstream of SM1, we required the
measured momentum of the secondary particle to be
smaller than 50 GeV/c, (the momentum of the zr-
from S- or 27- decay cannot exceed 39 GeV/c) and
the angle between the incident and the secondary par-
ticle tracks to be larger than 1 mrad.
The coordinates of the point of closest approach
between the incident and secondary tracks were com-
puted and a loose cut applied on the distance of ap-
proach (10 mm) to reject accidental associations. This
point of closest approach is obviously not the vertex
(27--r~-) but is close to it, since in most cases the
(27--zr-) angle is between 10 and 16 mrad, thus
much larger than the (~- -27-) angle. The z coordi-
nate of this point was required to lie in the range
from 3.70 to 7.50 m ("decay region" in Fig. 2). At
this stage, 32% of the initial sample of events re-
mained.
The shower reconstruction methods and, in par-
ticular, the procedures to identify photon and hadron
showers, are extensively described in [9] and [10].
The calibration, resolutions and efficiencies of the
LAD, given in [9] and [10], are also valid for this
experiment, since the photon energy ranges are simi-
lar. The spatial resolution (a) was found to be 4 mm
for photon showers and 8 mm for neutron showers.
The energy resolution for photons was well repre-
sented by the formula cr(E)/E= 20%/1/~ (E in GeV).
The neutron energy could not be measured, since on
average only 60% (with large fluctuations) of this
energy was deposited in the LAD. The efficiency for
neutron detection was (88 _+4)%. The reconstruction
efficiency for photon showers varied with energy,
rising from about 50% at 2 GeV to close to 100%
above 6 GeV. This efficiency was significantly de-
graded when the neutron and photon showers over-
lapped, in particular when their distance in projection
was less than 50 mm. For the LG, the resolutions
of the~hoton energy and position were a(E)/E=
25%/]/E (E in GeV) and cr(x)=a(y)=5 mm in the
energy range of 2 to 10 GeV.
We first selected the events having a single hadron
shower in the LAD (the re- were always deflected
away from the LAD). We required this shower not
to be associated with a charged particle track, in
order to suppress events due to E-~An-, A ~pn-
in which the A decayed downstream of H 3. We then
required a photon shower, which may have been ei-
ther in the LAD or in the LG. Events with an addi-
tional photon shower in either detector were rejected
to suppress 3-~ A re-, A ~ n zc ~ ze~ 7Y decays. At
this stage, we were left with 80 092 events which repre-
sented 5.6% of the initial sample. For 65 537 events,
the photon was observed in the LAD, for 14555
events it was detected in the LG. From this point
of the analysis, these two subsamples were treated
separately, because of the different performance char-
acteristics of the two detectors.
For the events with a photon in the LAD, the
problems of pattern recognition were identical to
those met in our previous analysis [10]. Therefore we
applied the same criteria to reject events which might
have presented reconstruction problems. Thus, we re-
quired the photon energy to be larger than 4 GeV
(55345 events left), the distance in both projections
(x, y) between the neutron and photon showers to
be larger than 60 mm (39286 events) and the differ-
ence between the photon energies measured in the
x and y projections to be smaller than 2 GeV (32701
events).
The energy spectrum of the showers reconstructed
in the lead glass was strongly peaked below 2 GeV.
This may be due to genuine low energy photons (from
re ~ decay or bremsstrahlung) or to random counts.
On the other hand, the energy spectrum of the pho-
tons from the decay S- ~ 2;- y does not extend signif-
icantly below 2 GeV. Therefore we required the pho-
ton energy to be larger than 2 GeV (5851 events left).
To suppress possible hadronic showers, we further
demanded signals with a minimum pulse height in
the corresponding scintillator and in the associated
region of the proportional chamber (3350 events left).
In these conditions, the detection efficiency for pho-
tons was estimated to be (82• A cut was also
applied at 25 mm on the distance between the shower
centre and the edges of the lead glass array, to avoid
problems of energy leakage. This reduced the LG
sample to 2873 events.
A detailed Monte-Carlo simulation of the experi-
ment was necessary in order to assess the efficiency
of the apparatus and determine its acceptance. The
147
resolutions and efficiency functions mentioned above
were parametrized and introduced into the M.C. pro-
gram. This program already used in our previous
analyses [9, 10] is extensively discussed in these refer-
ences.
4 Selection of S- --, 22- 7 Candidates and Evaluation
of the Background
For both subsamples of events, we computed the
missing mass between the incident S- and the outgo-
ing n-. The resulting mass ..spectra are shown in
Figs. 3a, b. These spectra show a peak centred at
the A mass. This demonstrated that most of our
events were in fact 3-~An-, A--*nn ~ decays in
which one photon escaped detection, either because
of the incomplete solid angle coverage or because of
..10'
10 ~
~
10 2
">'1o
1
II I I I I I I i
1.000 1.100
MISSING MASS (McV/c 2)
1.200
10 t
10 ~
10
l.iJ
1.000
1.100 1.200
b MISSING MASS (MeV/(: 2)
Fig. 3 a, b. Distribution of the (~- --n-) missing mass: a Data sam-
ple with the photon detected in the LAD. b Data sample with the
photon detected in the LG
inefficiencies. The width of the peak (7 MeV/c 2
FWHM) is consistent with the resolution expected
at the A mass.
The next step of the analysis made use of the
kinematics of the decays S- --* 22-7, 2;- ~ nn-. Using
the following measured quantities:
-
the momentum and direction of the S-,
-
the coordinates of the impact point of the photon
and its energy,
-
the momentum and direction of the n-,
-
the coordinates of the impact point of the neutron,
and under the assumption of a ~---* Z-7 decay, we
calculated the angle between the N- and photon tra-
jectories, and the coordinates of the N- decay point.
The momentum and direction of the 2;- were com-
puted using four-momentum conservation. The 27-
decay vertex was then defined by the point of closest
approach between the 2;- and n- trajectories. These
calculated decay points were restricted to the decay
region defined above (within the errors), which re-
duced the LAD sample to 6455 events and the LG
sample to 1 894 events.
By setting the (n-n-) effective mass to the 22-
mass, we calculated the neutron energy and deduced
the global energy balance A E. In Figs. 4a, b, we show
the observed distributions of the quantity A E. Quali-
tatively, genuine 3----,27-7 decays are expected to
peak around A E= 0. Indeed, the spectrum of A E for
the LAD events (Fig. 4a) shows a cluster of about
30 events centred at 0. Most of the events fall at
the sides of the plot, outside the interval between
-20 GeV and + 20 GeV. This feature is well repro-
duced by the M.C. simulation of the decays Z-~
An-, A ~ n n ~ The A E spectrum for the LG events
(Fig. 4b) shows the same shape, with less statistics
(about 10 events around A E= 0).
The M.C. simulation of genuine (~- ~ 27-7, 22-
n n-) events predicted a peak with a width (o-) of
4 GeV centred at 0 GeV, plus tails extending up to
20 GeV. A cut on A E at _+ 8 GeV, which defines the
"central region" of the A E plot, was predicted to
retain 80% of the good events while rejecting all the
background due to the Z---* An-, A ~nn ~ decays.
It was important, however, to check whether badly
reconstructed events of the type ~- ~ A n-, A --* n n o
could simulate the (S- ~ 27-7, 27- --* nn-) kinematics,
thereby populating the central region of the A E plots.
For this purpose, we selected a sample of events
with two photon showers, which consisted almost en-
tirely of A ~nn ~ events. This sample was subjected
to the selection procedure described above, each event
being treated as two single-photon combinations by
ignoring alternately one of the photon showers. The
resulting A E spectra are shown in Fig. 5 a (4089 com-
binations with one photon in the LAD) and Fig. 5 b