Production of Direct-Photons and Neutral Mesons at Large Transverse ...

Fermi National

Accelerator

Laboratory

FERMILAR-Pub-93/007-E E709

Production of Direct-Photons and Neutral Mesons at Large Transverse Momenta by x- and p Beams at 500 GeV/c

G. Alverson et al The E706 Collaboration Fermi National Accelerator Laboratory P.O. Box 500, Batauia, Illinois 60510

January 1993

Submitted #

to Physical

Operated by Universities

Review D

Research Association

Inc. under Contract No. DE-ACOZ-76CH03000

wim the United States Department

of Energy

Disclaimer

This report was prepared as an account of work sponsored by an agency of the United States Gouernment. Neither the United States Government nor any agency thereof nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or seruice by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof: The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof,

Production

of Direct-Photons

at Large Transverse

and Neutral

Momenta

Mesons

by T- and p Beams

at 500 GeV/c G. Alverson,@)

W. F. Baker,c3) G. Balloccbi,(‘o+)

D. Berg,@) S. Blusk,@) C. Chandlee,(“‘)

C. Bromberg,@)

W. Dhgosz,(e)

W. Faissler,(“)

G. Fano~rakis,(‘~)

C. Johnstone,(3)

R. Miller,c4) E. Potbier,

P. Gutierrea,(‘)

A. Sinanidis,(‘)

The Fermilab (1) University

A. Lawa, A. MxI&(~)

J. Mans~ur,(‘~~~)

(3) Fermi National (4) Michigan

(8) Pennsylvania

and M. Zieliriski(‘o)

CA 95616

11 00 07, India Batavia,

East Lansing, Minneapolis,

IL 60510

MI 48824 MN 55455

University,

Boston,

MA 02115

of Oklahoma,

Norman,

OK 73019

State University,

University

of Pittsburgh,

Pittsburgh,

PA 15260

Rochester,

NY 14627

(10) University

of Rochester,

address:

address:

D. Weerasundara,@‘)

Laboratory,

of Minnesota,

(9) University

address:

P. Slattery,(“)

C. Yosef(‘,g)

Delhi

State University,

(7) University

D. Skow,(‘“,f)

- Davis, Davis,

of Delhi,

(6) Northeastern

K. Ruddick,(‘)

E706 Collaboration

Accelerator

(5) University

R. Roser,(“l

N. Va&s,(‘“)

of California

(2) University

(c) Current

J. Huston,(4)

I. Kourbanis,(ef)

G. Wu,(es’) T. Yasuda,(“)

(b) Cument

K. Hartman,@“)

B. M. Rajaram,(2~i)

W. Toothacker,@)

(a) Current

G. Ginther,@)

C. A. Nelson, Jr.,c3) B. Y. Oh,@) D. Orris,@~‘)

P. Shepard,@) R. K. Shivpuci,(‘)

J. Whitmore,@)

L. de Barbam,

D. Garelick,(@

P. Luken~,(~~‘) S. Mani,

E. Prebys,(‘“,h)

L. Smell,@)

T. Chand,@)

S. Easo,(R,d) E. Engels, Jr.,c9)

T. Ferbel,(‘o)

V. Kapoo~,(~)

F. Lobkowicz,(“)

D. Caky,

W. H. Chung,(B)

.I. Dunlea,

G. Glass,(e) M. Glaubman,(“)

C. L&&is,(e)

D. Brown,@)

B. C. Choudhary,(2’c)

W. DeSoi,(“)

R. Ber~son,(‘~~)

CERN,

LeCroy

1211 Geneva 23, Switzerland

Corporation,

University

Park, PA 16802

Chestnut

of California,

Ridge, NY 10977

Riverside,

CA 92521

(d) Current (e) Cumnt

(f) Current

address:

(g) Current

INFN,

address:

University

of Perugia,

address: Stanford University, Fermi

National

address: Michigan

(h) Current (i) Current

address:

Accelerator State Univemity,

address: Princeton

University,

Pemgia,

Stanford, CA 94309 Laboratory,

Batavia,

East Lansing, Princeton,

December

2

28, 1992)

IL 60510

MI 48824

NJ 08544

L. B. and S. B. 5’. College, Saga?, Karnatka (Received

Italy

57’7401, India

ABSTRACT We present

results

clude incident kinematic -0.7

from the initial

ranges of transverse

production,

momentumand

the nuclear

production

dependence

The cross sections are compared

different

rapidity

We have measured

as well as the v/r”

Cu, we have extracted A”.

experiment

with

choices of the QCD momentum

of 3.5 5 pi

Typeset

Using

13.85.Qk,

12.38.Qk,

ratio.

5 10 GeV/c

and

and direct-

parameterized

as

log QCD predictions

for

scales and several sets of parton

25.4O.Ve, 25.8O.L~; 13.85.Ni

3

and span the

From the data on Be and

of x0 production

next-to-leading

REVTEX

The data in-

cross sections for r”

functions.

PACS numbers:

E706.

and p beams at 500 GeV/ c on Be and Cu targets,

r-

5 ycm 5 0.7, respectively.

photon

run of Fermilab

distribution

I. INTRODUCTION The study of inclusive (pi) has been important perturbative with

methods

data.

single-hadron

to the development

functions

a quantitative

on the validity

of the parton

functions

aspects of direct, or prompt, in the literature

of direct-photons

provides

at large transverse

&CD, and can serve as a valuable

momentum

tool for probing

only the two processes shown in Fig. 1 contribute

calculations

and quark-glum

Compton

tons from the processes shown in Fig. directly

in which all particles

be individually

reconstructed

prompt

combination

To first order,

to the direct-photon

cross section,

scattering.

These diagrams capture unambiguous

higher-order

[4]. Since the pho-

at the elementary parton,

of the collision.

originating

photon produc-

structure.

of an outgoing

data are especially

with Deep Inelastic

sensitive

Scattering

vertex

This is in contrast

from the fragmenting

parton

to the gluon distribution

data can be used to map out the gluon distribution

four

diagram,

functions.

production,

function

to

must

parton

state gluon in the Compton

and Drell-Yan

of

the momentum

and then combined into a jet to determine

Because of the presence of an inital

photon

hadronic

1 are produced

reflects the kinematics

jet production, momenta.

of hadrons

[3]. The production

precise tests of theory

and not in the fragmentation

of the photon

ele-

a clean test of perturbative

of the process, and relatively

can be carried out to provide

the interaction,

of QCD matrix

distribution

tion at large pi have been addressed extensively

the essential characteristics

comparison

of partons.

and phenomenalogical

namely qTj annihilation

momentum

of QCD [1,2]. Large pr is a regime where

yield information

ments, and on the effective product Experimental

at large transverse

can be applied to QCD to provide

Such comparisons

and the fragmentation

production

In

direct-photon

over a wide range of

kinematics. Previous

measurements

center-of-mass the highest obtained

energies and incident

In the following photon

beam particles

[5-71. Our measurements results on proton experiments

sections, we describe the experimental

layout

and ?y” production publications

have covered a variety

at the ISR [6] an d in earlier fixed-target

used to extract

QCD calculations. recent

production

energy data for pion beams, and straddle

previously

procedures

of direct-photon

the #

and direct-photon

are compared

with

signals.

predictions

of

provide

interactions [7].

and the analysis

The results on direct-

from next-to-leading

log

The results presented in this paper expand upon more abbreviated of our work

[8], and provide 4

detailed

comparisons

with

QCD

predictions

at the next-to-leading

Fermilab

experiment

II. THE EXPERIMENTAL

SETUP

E706 is a second generation

fixed-target

to measure direct-photon end, a liquid

log level for both ?y”s and direct-photons.

production

and the associated

argon electromagnetic

was used to optimize

calorimeter

the reconstruction

was split into front charged-particle were initiated

shows the layout

of the various

spectrometer

was employed

spectrometer

components

To this

with

photons

simultaneously

(particularly

from

The EMLAC

against hadrons.

showers in the EMLAC

and to reconstruct

components

fine segmentation

signal).

discrimination

system was used to identify

by charged particles,

event structure.

to the direct-photon

and back sections to provide

tracking

designed

(EMLAC)

of overlapping

K’ decays, a large source of background

experiment

interaction

vertices.

that

Figure

of the Meson West spectrometer. by experiments

are discussed in the following

A 2

This

E706 and E672 [9]. The

subsections.

A. Beam Line The Meson West secondary tum recombining mass tagging. 800 GeV/c length

beam line.

It is capable of delivering

The secondary

protons,

aluminum

particles

extracted primary

beam line is a high resolution, utilized

a high intensity

in the experiment

from the Tevatron,

target.

two stage, momen-

incident

The components

beam with

were produced

by

upon an 0.77 interaction

of this beam line are shown in

Fig. 3. For the data presented 29 GeV/c

(RMS

here, the secondary

half-width).

The beam was occasionally

100 GeV/c for detector calibration at an angle of 0 mr to maximize particles,

purposes.

tuned

For positive

primary

proton

The negative secondaries were produced

the production

particle

yield was low4 per primary

beam from entering

per primary

angle was 1.4mr to prevent the intense

the secondary beam line aperture. proton.

to 500 &

to 25, 50 and

the yield, which for this run was 2 x lo-’

proton.

with a 23 set spill, during

beam energy was tuned

The Tevatron

which the experiment

typically

The positive

operated on a 57 set cycle, received 5 x lo7 secondary

beam particles. The intermediate plane.

The momentum

focus of the beam was momentum defining

The final focus, at the experiment

collimator

was located

target,

was momentum 5

dispersed in the horizontal at this intermediate recombined.

focus.

The second

stage of the beam line included were parallel differential

in both Cerenkov

The counter 7psia.

a 50m long straight

the horizontal counter

and vertical

used to identify

used a 42m long helium

planes.

This section

contained

the masses of the incident

gas radiator

a

particles.

in the pressure range from 4 to

Using this device, the negative beam was found to contain 97% x- and 3% K-

while the positive

beam was measured to contain

The beam line contained Their function upstream

five muon spoilers,

absorber,

of the target in the experimental

in Fig. 2) were positioned

[lo].

constructed

of magnetized

steel.

hall.

detectors

utilized

(SSDs),

Its purpose was to absorb hadrons

an analysis

magnet,

tracking

and proportional

wire chambers.

Six

beam tracks.

a transverse

to charged particles,

momentum was followed

wire chamber (PWC).

The SSD system was constructed

of seven X - Y modules,

The three modules upstream thick,

with two planes per

of the target, and the first one downstream,

consisted of 3 x 3 cm2 wafers, while the remaining the SSDs were 250 to 300 pm

of silicon

of the target and were used to reconstruct

plane) of N 450 MeV/c

by sixteen planes of proportional

6600 instrumented

system consisting

The analysis magnet, which imparted

(in the horizontal

to as veto walls

of the target and used to reconstruct

Eight SSD planes were located downstream vertex.

counters (referred

Tracking System

a charged particle

SSD planes were located upstream the interaction

the beam just

beam halo particles.

B. Charged-Particle The experiment

A

between the hadron absorber and the target to veto events

generated by any remaining

module [ll].

and < 2% K+

composed of 900 tons of steel, surrounded

in the beam halo. Two large walls of scintillation

impulse

91%p, 7%r+

was to deflect muons in the beam halo away from the spectrometer.

5m long hadron

strip

section, in which the trajectories

modules had 5 x 5 cm2 wafers. All

and had a pitch of 50 pm. There were a total of

SSD channels.

The PWC system consisted

of four modules,

with four planes per module

[12].

Each of the four modules had one X, Y, U and V view. The wires in the U-view were tilted

at 37”, and in the V-view

at -53”,

relative

to the vertical.

The sense wires

had a pitch of 0.254 cm. The active areas of the four modules were: 1.22 m x 1.22 m, 2.03 m

x

2.03 m, 2.03 m

of the liquid

x

2.03 m, and 2.44 m x 2.44 m, and covered the accepted region

argon calorimeters.

was fed through

a current limiting

The central region high voltage of each PWC plane resistor which reduced the sensitivity 6

of that region

during

high intensity

running.

The desensitized

area varied from 2.54 cm x 2.54 cm

in the first PWC module to 5.08 cm x 5.08 cm in the last PWC module. a total of 13,400 instrumented

PWC channels.

The combined

was designed to operate at a 1 MHz interaction

There were

PWC and SSD system

rate.

C. Liquid Argon Calorimeter The next element in the experiment ter (LAC)

[13], consisting

of an electromagnetic

lengths) followed by a hadronic the calorimeters

was a 3m diameter

(HALAC)

cryostat,

supported

to the beam axis. The calorimeters

line, with the front end of the EMLAC

located

calorime-

section (z 30 radiation

section (z 8 interaction

resided in a stainless-steel

be moved transverse

(EMLAC)

liquid-argon lengths).

Together,

by a gantry that could

were centered on the beam

N 900 cm downstream

of the target.

The active area covered a polar angle range of 1.3” < .9 < 10” as seen from the target. Within

the cryostat,

through

a 40 cm diameter beam pipe, filled with helium gas, was inserted

the center of both calorimeters.

a low density

(0.07g/cm3)

liquid

In front of the electromagnetic

argon excluder.

The purpose of both the excluder

and the beam pipe was to displace argon by materials to reduce interactions

that could complicate

The electromagnetic as shown in Fig. consisting

of substantially

lower density

event reconstruction.

section of the LAC consists of four independent

4. Each quadrant

of an absorber,

and another

section was

was constructed

0.25 cm of liquid

quadrants,

of 66 layers, with

argon, a copper-clad

each layer

G-10 anode board,

0.25cm argon gap. The absorber in the first layer was aluminum

while

for all other layers it was a 0.2cm thick lead sheet.

The lead sheets were made of

98.6% lead, 0.07% calcium

and tin were added to increase

the mechanical

rigidity

and 1.3% tin; the calcium

of the lead sheets.

The G-10 anode boards were octant size, with each octant read out independently. The anode boards had alternating a radial

board.

The radial

focused (in a projective containing

boards consisted

geometry)

radial layer was = 5.5mm. 92 instrumented

an outer region containing

radial and azimuthal

strips,

azimuth

(see Fig. 4). The inner/outer

readout

board)

The width

each subtending

x/192

spatial 7

with

cut to be

of the strips on the first into an inner region,

radians

strips each subtending

separation

starting

of 254 strips,

boards were divided

188 instrumented

allows for improved

of a maximum

on the target.

The azimuthal

strip geometry

in azimuth, r/384

and

radians in

(at a radius of 40.2cm on the first

resolution

at large radii

by subdividing

the outer strips by a factor by reducing

The inner/outer

of two.

radial/azimuthal

correlation

split also aids reconstruction

ambiguities

arising

from multiple

showers

in the calorimeter. Longitudinally,

the calorimeter

sists of 22 layers (about

is read out in two sections.

10 radiation

44 layers.

This front/back

showering

particle

lengths),

The front section con-

while the back consists of the remaining

split is used to measure the direction

and to help discriminate

of incidence

between electromagnetic

of the

and hadronic

showers. D. Readout System for the Electromagnetic Data acquisition using the RABBIT

and trigger-signal (Redundant

Instrumentation

Calorimeter

(LACAMP)

amplifiers a shielded

Faraday

The outputs amount

Group at Fermilab consisted

z 20 amplifiers

of charge deposited

on their

time for signals in the EMLAC

[14]. The Liquid

converters.

the circumference

Argon

of charge integrating

The LACAMPs

were lowithin

of the top of the cryostat.

consisted of voltage signals proportional respective

was N 350ns.

channel was delayed by SOOns, giving or reject the event.

system de-

per crate; the crates were positioned

room surrounding

from the LAC amplifiers

was performed

Transfer)

of 16 channels

and eight channels of time-to-voltage

cated in 20 crates with

for the EMLAC

Analog Bus Based Information

veloped by the Particle AMPlifier

processing

Calorimeter

strips in the calorimeter. The output

the trigger

The rise

from each LAC amplifier

the time required

When an event satisfied the trigger,

to the

either to accept

each channel was sampled

twice, once before the delayed pulse and once at the peak of the delayed pulse. This voltage difference was digitized

by one of two digitizers

per crate, provided

that the

absolute value of the voltage difference was greater than a predefined

threshold

zero suppression

took place be-

window)

stored on the digitizer.

cause the voltage difference fell within

If no digitization

the zero suppression

flagged as not having useful data, and the data acquisition

window,

(the

the channel was

system then proceeded to

the next channel. Timing

information

on the LACAMPs.

Their function

between the beginning this timing amplifier

measurement channels.

was provided

through

the time-to-voltage

converters

was to produce a voltage proportional

of a LAC pulse and the trigger was generated

by summing

signal.

the output

To avoid dead time caused by the arrival 8

(TVCs)

to the time

The signal used for of four consecutive

of a second pulse prior

to the readout

of the TVC, a second (slave) TVC circuit

was enabled while the first

was busy. E. Trigger System The trigger, the EMLAC,

which selected events depositing

high transverse

was designed to operate at lo6 interactions

was formed by weighting

the energy in the radial strips according to the distance from

four steps, which involved trigger

Figure

(pr) in

per second. The trigger pi

the center of this strip to the beam line (E sin 0). The triggering a final trigger.

momentum

defining

a beam particle,

5 shows the symbolic

process took place in

an interaction,

logic diagram

a pretrigger,

for the SINGLE

and

LOCAL

used in this analysis, whose elements are discussed below.

Beam

Definition

of the target

-

Two l/16”

x 1” x 1” scintillation

were used to define beam particles.

counters located upstream

A beam particle

was accepted if

both counters had a signal in coincidence. Interaction

Definition

the target

and covering

interaction.

The logical

beam definition in the definition

-

Four scintillation

the acceptance

counters,

located

of the spectrometer,

downstream

were used to define an

OR of the four counters was taken in coincidence

to define an interaction. of the pretrigger

An interaction

if this interaction

of

with the

strobe was generated and used candidate

satisfied the following

criteria: l

No other interaction

l

The interaction

occurred within

was not initiated

was no signal in the “halo” two beam defining

*60 ns of this candidate

by a particle

scintillation

counters.

interaction.

in the beam halo; that is, there

counter located just downstream

This 5” x 5” halo counter had a 3/S” diameter

of the hole

cut out of its center and was centered on the beam. Pretrieeer of each octant

Definition

Signals were generated for the inner and outer parts

using the sums of the weighted

signals were passed through p&rigger

-

threshold

than would normally

zero-crossing

(ZZ 1.7 GeV/c) be available

discriminators,

which

for the intrinsically a p&rigger 9

These

time (575 ns)

slow signals of the LAC.

zero-cross outputs

were used to obtain

strips.

served to define a

and t 0 set a more precise pretrigger

The logical ORs of the discriminated parts of each octant

energies in the radial

from the inner and outer

signal for that

octant.

The

following

additional

requirements

were placed on the final formation

of a p&rigger

for any octant: l

No pi deposition

in that octant greater than 1.5 GeV/c

within

z 300 ns prior to

the interaction. . Absence of a noise spike generated by the 400 Hertz power supplies for the LAC electronics. l

No incident

muon as defined by the logical AND of the signals from the two veto

XV&.

This p&rigger

signal was used to latch the LAC, PWC and SSD signals. The presence

of a p&rigger

was required

Trigger

Definition

electromagnetic

trigger.

The two main triggers

used to select events containing

showers of high pi were: (1) The SINGLE

the pi in a localized given threshold cal threshold

-

for any subsequent

region of an octant

(x 4 GeV/c),

and (2) the LOCAL*GLOBAL

was simultaneously

trigger,

and a GLOBAL

required

ment suppressed events that contained

(Z 4 GeV/c).

only low-pa photons.

two decay photons

to occupy any single LOCAL

LOCAL*GLOBAL

trigger

also recorded.

In addition,

a reduced

a TWO

local signals, one from any octant ing away-side employed

octants;

threshold

function

of pi.

GAMMA

requirements

(summed

The LOCAL

trigger

of the information

the various electronics

modules.

require-

angle between their (z

A prescaled

2.5GeV/c)

was

was formed by requiring

two

and the other from any of the three correspondtrigger. allowed

threshold

(FZ 2GeV/c)

was

The variety

of trigger

definitions

with

a measurement

of trigger

efficiencies

as a

If a LAC trigger was satisfied, a signal was sent to the online computers the recording

over

The LOCAL*GLOBAL

threshold

a reduced value for the LOCAL

for this TWO

different

GAMMA

in which the lo-

region, would trigger.

GLOBAL

in which

to be above a

threshold

trigger insured that ?y” OF 7 events, that had too large an opening with

trigger,

(16 radial strips) was required

for an octant was z 2 GeV/c,

the whole octant)

LOCAL

for the event. The trigger

Otherwise,

circuitry

z 30~s before being enabled for the next p&rigger. III. DATA ANALYSIS

10

to initiate

a reset signal was sent to

then allowed a settling

time of

During

the 1987 - 1988 Fermilab

0.5(0.8) events/pb corresponded LOCAL

was accumulated

with negative (positive)

of approximately

beam on Be; data on Cu

to about 10% of the Be sample. Only events which satisfied the SINGLE

trigger were included

off-line

fixed target run, a sensitivity

using the Fermilab

in the present analysis.

The events were reconstructed

ACP [15] sy st em. The following

the event reconstruction

procedures

losses due to various inefficiencies

and the methods

subsections

will describe

used to correct

the data for

and biases.

A. Event Reconstruction The event reconstruction track reconstruction procedures

for this experiment

and electromagnetic

in the following

Charved-Particle formed independent and downstream

procedure

included

shower reconstruction.

charged-particle We discuss these

paragraphs.

Tracking

-

The

charged-track

track segments, upstream

of the magnet

reconstruction

of the magnet using the SSD planes,

using the PWC planes.

These segments were then

linked at the center of the magnet to form the final reconstructed The first step in reconstruction

involved

algorithm

forming

straight

track.

line segments in each of

the four views (X, Y, U, V) of the PWCs, and in each of the two views (X, Y) of the SSDs. A “view-track”

was defined in the PWCs when at least three out of the four

available planes had hits associated with the track. with a track it had to lie within dimensional

coordinates,

A maximum space-track

In order for a hit to be associated

f 1.5 wire spacings of the trajectory.

termed “space-tracks”,

were formed using the view-tracks.

of 16 hits (4 planes and 4 views) could be associated had to have a total

Tracks in three

of at least 13 hits within

with a track.

A

?r 1.5 wire spacings of the

trajectory. After reconstruction

of the PWC space-tracks

PWC tracks and SSD tracks were projected magnet. that

After accounting

corresponded

linked

tracks.

spatially

2.5mrad.

The interaction

field, SSD and PWC tracks

at the center of the magnet were considered caused deflections

primarily

tracks in the Y view were also required

The charge and momentum vertex position

Y), and then correlated

by requiring

each of the

to the X-Y plane at the center of the

for the effects of the magnetic

Since the magnet

slopes of the linked

and SSD view-tracks,

candidate

in the X-Z plane, the to match

to within

of each linked track was then calculated.

was reconstructed

separately

in each view (X and

a match in the Z coordinate.

For each SSD view,

11

tracks that linked vertex.

Unlinked

with PWC tracks were preferentially The final interaction

of view vertices that matched Electromagnetic

independently

vertex position

Reconstruction

-

for each quadrant.

was determined

Electromagnetic

The readout

left Rand right R, containing

each octant in that quadrant,

and inner @ and outer a, containing

algorithm

operated

l

in the following

First, a search was performed,

. Third,

con-

the energies in the The reconstruction

within

each view, for energy depositions,

termed

(defined below). of these peaks were calculated

by fitting

to a

shower shape.

photons

A “group”

showers were

sequence:

Second, the energies and positions predetermined

by the pair

the energies in the radial strips of

strips for T < 40.2 cm and for T > 40.2 cm, respectively.

and “peaks”

of linked

in each quadrant

azimuthal

“groups”

number

the

best in the Z coordinate.

Shower

sisted of four “views”:

l

to reconstruct

SSD view tracks were also used if an insufficient

tracks were present.

reconstructed

employed

were reconstructed

was defined as consisting

by correlating

peaks from different

views.

of consecutive

strips, each with energy above

150 MeV, and with at least one strip with energy above 300 MeV, and a total group energy of more than 750 MeV. a group

that

had energy

A “peak”

minima

was defined by a sequence of strips within

(“valleys”)

such that

the maximum

was fitted

to the shape of an electromagnetic

simulations

on both

value was more than

2.5~~ above the valleys. shower determined

value,

Each peak

from Monte Carlo

and tuned to match showers in the data. The results of that fit include

the position

and energy of the peak.

approximately were required

Photons

were formed by correlating

the same energy in the R and @ views. (T > 40.2cm)

could only be correlated

peaks, and a peak in the @ view could only be correlated containing

of uncorrelated

- namely,

views a radial

with inner (outer)

with a radial

Q,

peak in the

that 4 value.

The correlation for complicated

peaks of

Peaks from different

to occupy the same general region of the detector

peak with T < 40.2cm octant

sides of a maximum

routine

in the electromagnetic

events containing

reconstruction

a large number of photons,

energy (not associated with any photon).

12

algorithm

leaving

may fail

a large amount

Such events were eliminated

by requiring

the total uncorrelated

energy for the triggering

quadrant

to be less than

10 GeV. A relative

interaction

time was calculated

for photons

tained from the TVC channels of the amplifiers

based on information

corresponding

to that photon’s

obenergy

deposition. B. Single-Photon Two of the largest potential from bremsstrahlung electromagnetic

and Di-photon

Identification

sources of background

to the direct-photon

signal are

of beam halo muons in the outer regions of the EMLAC

decays of r” and 7 mesons. The majority

(from beam halo) was eliminated veto wall that overlapped

by requiring

that satisfied the trigger.

of the muon background

that no hit register in the section of the

th e region of the electromagnetic

(shadowed)

and from

An off-line reconstructed

calorimeter

vertex was also required

for each

accepted event. Figure 6 displays the 7-y invariant LOCAL r”

triggers,

energy

requiring

asymmetry

mass in the ?y” and v mass ranges for SINGLE

a pi of at least 3.5 GeV/c

is defined

as A = IE,,

for the two-photon

- E,I/[E,,

pairs.

The

where ET1 and

+ E,,],

42 represent the energies of the photons. The lower points, with the dashed error bars, represent the mass spectrum for A < 0.75. For OUT study of # production, a ?y” was defined as any two-photon M,,,

in the range llOMeV/c’

combination

with

A < 0.75, and invariant

< My7 < 160MeV/c2.

Both photons

to occupy the same octant.

An 7 was defined as any two-photon

A < 0.6 (for direct-photon

background

and 450MeV/c2

< Myy

analysis was determined acquisition

period.

signals, sidebands and 7 peaks. distributions

estimates

by the trigger threshold

To account

for combinatorial

pi

within

in the same octant

candidate

to form a no or q with

be discussed in a following

the no and 7

to the trigger

13

signals.

an A < 0.75. The residual

The minimum threshold

from the

if it did not combine with another photon

using our Monte Carlo simulation

section.

data

mass range of the #

the K’ and 11mass ranges to obtain the respective

from ~“6 and I)S was calculated chosen to correspond

under

from these side bands were then subtracted

A photon was a direct-photon

in the

used during the corresponding background

with

was A < 0.75)

value employed

were selected which covered the equivalent

Distributions

were required

combination

this requirement

Th e minimum

< 650MeV/c*.

mass,

background

program,

which will

pi value in the analysis was again

set in each particular

running

period.

We defined a directionality

parameter

where Tf, Tb are the reconstructed zf and z6 are the distances calorimeters

for photons

front and back radial positions

of the front

to the target.

Photons

originating

from the target parallel

have directionality

to the beam line have large

direction&ties.

Directionality

and in-time

dates in order to minimize the veto wall criteria

background

were imposed on all direct-photon

from muon bremsstrahlung vertex was found.

could not be reconstructed,

and have good directionality. out of time with

requirements

when an interaction

time or directionality

Figures

in the veto wall (in the region corresponding

from the target are frequently

to the triggering

Applying

essentially

In cases where the photon

for events with and without

signal in the veto wall produces a large concentration most of these showers, leaving

events that passed

7 and 8 show plots of the single-photon

versus directionality,

also have large directionality.

candi-

they were assumed to be in-time

Showers not originating

the interaction.

arrival time at the EMLAC

that

of the shower, and

and back sections of the electromagnetic

close to zero, while showers from muons traveling positive

as follows:

octant).

of out-of-time

Requiring

a

showers (muons)

the veto wall requirement

only in-time

a signal

showers (photons)

eliminates originating

in the target region. To eliminate nal, we required projected

possible electron

or hadronic

that accepted showers not spatially

to the LAC

(within

lcm

(Et/E,

electrons (and showering

charged hadrons)

in the front

z 8% of the candidate

cross sections.

single-photons

and that

at least

section of the electromagcut serves to reject

and is not an isolation

of direct-photon

sig-

overlap with any charged track

> 0.2). Note that th’ 1s t rack projection

ployed in collider measurements cut eliminates

to the direct-photon

of the center of the shower),

20% of each shower’s energy be contained netic calorimeter

background

cut in the sense emThis track projection

from our sample.

C. Energy Scale Due to the steep pi

dependence

the energy scale of the EMLAC section

at a specific energy.

determined

of inclusive

can produce

cross sections,

large changes in the measured

Because of this sensitivity,

with great care. One possible procedure 14

a small change in cross

the energy scale must be

is to set the energy scale such

that the reconstructed be compromised overlap

n-Omass is centered at the accepted value.

by the sensitivity

of the two photon

of the reconstructed

showers originating

the EMLAC describing

can be calibrated the different

how the momentum The tracking p+pL-.

from ?y” decay.

photons,

using the decays: Ki Ki

Before

we will describe

+ ?T+?T- and Jill,

decays were obtained

curvature,

when these tracks were interpreted

mass distribution

momenta.

events were selected by the E672 [9] dimuon

tices that had two tracks of opposite of a Ki,

charged-particle

system was established.

system was calibrated

by the muon spectrometer.

we set the

from a ?y” decay converted

applied to reconstructed

scale of the tracking

Instead,

Using these events, the energy scale in

from reconstructed

corrections

The J/$J + p+p-

identified

of the magnet.

can

mass value to the degree of

energy scale using events in which one of the photons into an e+e- pair upstream

This method

for events in the Ki

--t

trigger

and

from secondary

ver-

with an invariant

mass close to that

as pions. Figures 9 and 10 show the r+?r-

mass region,

and p+p-

mass distribution

in

the J/y5 mass region. The first energy correction account

of the difference

pedestals

determined

for showers detected

between

the pedestals

from calibration

run. A fit was performed

in the EMLAC

estimated

tasks frequently

carried

out at the start of a

reconstruction

in each channel code) to determine

the mean pedestal levels expected in the data. Special events, requiring definition

(and no interaction),

250MeV

were observed for some channels.

stable over the course of the data taking, shift corrections differences

photon

energies in each octant

kinematic

in the LAC.

Octant-to-octant

as large as

these shifts were found to be

and consequently

a single set of pedestal

This

in the energy scale caused by octant-to-

correction

was implemented

by resealing

by a factor based on the reconstructed variations

had a range of 6%.

the

no mass for

Only ?y” events in a

region where the ?y” decay photons were separated by more than 3 cm were

used to set this relative problems

However,

was for variations

octant

octant.

Pedestal differences

biases

was used for all the data.

The next correction

that

only a beam

were used for this purpose in order to minimize

caused by the presence of energy in the calorimeter.

taking

from the data and the

to the tails of the pulse height distribution

(when no peak was detected by the electromagnetic

involved

energy scale in order to minimize

the sensitivity

to possible

in energy sharing between the photons.

An additional

correction,

which had to be applied 15

separately

to single-photons

and x’s, arose from energy losses at the inner/outer from Monte

Carlo events generated

within

@ boundary.

~I~lOcrn of this boundary,

an empirical

correction

reconstructed

energy and distance to the boundary.

The Monte

for the average reconstructed

Carlo was also used to generate

showering in the material ter.

This correction

function

determined

to the reconstructed

decay in which one photon

the energy determined An empirical

E/P

=

1.0 for all energies. scaled by the relative

converted

a function

energy. The photon

has all the final corrections for the momentum

as a

via elec-

The ratio of

to the momentum

(E)

as a function

of electron

from the requirement

enthat

distribution,

E/P

correction

and

was applied

energy loss in the material

to

in front

calorimeter.

# mass for the three particle

of photon

calorimeter

A similar

to electron

of the active region of the electromagnetic The resultant

and electrons,

into an e+e- pair.

11 shows this corrected

all adjustments. photon

calorime-

showers.

to E was then calculated Figure

its average value, after making photons

for photons

system (P) was obtained

correction

of

for energy lost due to

shower energy was determined

using the electromagnetic

by the tracking

ergy.

energy loss as a function

a correction

independently

of energy, and applied to reconstructed

trons from #

and provided

in front of the active region of the electromagnetic

was determined

The final correction

This was calculated

ye+e-

system is shown in Fig. 12 as

energy used in the calculation

while the e- and e+ momenta

scale of the tracking

system.

of the mass

have the final corrections

The agreement

with

the accepted

value of the ?y” mass is better than 0.5%. A correction reconstruction EMLAC. which

was also applied

that depended on the spatial separation

These corrections

indicated

that

were determined

for energy losses in

of the two decay photons in the

from Monte Carlo studies of #

the electromagnetic

IT” energies for cases of highly uncertainty

to the ?y” energy to account

reconstruction

overlapping

showers.

algorithm

underestimated

This contributed

a systematic

in the # cross section of less than 1%.

Using all the above corrections,

the #

and q masses in the 77 decay modes

were found to be 135.26 f 0.06 MeV/c’

and 545.06 f 1.38 MeV/c’,

within

The difference

0.5% of their

accepted values.

respectively,

between the reconstructed

and 7 masses and their accepted values, combined with the uncertainty ~SSS

decays,

for the K’ (Z 0.5%), and the uncertainty

system (= 0.5%), results in a systematic

in the momentum

uncertainty 16

both ?ya

in the ye+e-

scale for the tracking

in the overall

energy scale of

N 0.9%, which contributes cross sections. provides

Figure

= 10% uncertainty

13 shows the invariant

an independent

the establishment

?y”s and single-photons.

cause r” mesons provide be properly

accounted

The contribution was determined

using electron

7/e-

This consistency

to the systematic

is particularly

important

signal, and must production.

The normalization

of the function

runs at 25,50 and 100 GeV/c.

their beam uncertainties and varying

By varying

(and thus the overall normalthe magnitude

given by the Monte Carlo, an estimate

in the cross section was obtained.

be-

in the cross section from inaccura-

as follows.

calibration

response function),

energy loss function

uncertainty

energy scale between the

to the direct-photon

uncertainty

was estimated

energies within

of the E/P

of a consistent

for to obtain the cross section for direct-photon

function

ization

~“7 mass in the w mass range which

the main background

cies in the E/P these electron

x0 and direct-photon

check on our energy scale. The value of 782.6 f 8.0 MeV/c’

for the w signal confirms reconstructed

to the absolute

of the uncertainty

The result was a contribution

in the 7 cross section of 12%, and 4% in the #

of the relative

to the systematic

cross section.

D. Monte Carlo Simulation The relatively surement

small cross section for direct-photon

particularly

sensitive

to backgrounds

production

from decays of hadrons

photons in the final state. The largest of these backgrounds losses due to the acceptance of the EMLAC, from asymmetric

x’s,

that must be estimated

geometrical

low energy photons

properties

single-photon

of the spectrometer

used the GEANT

a data base of standard

in the standard

at a level

incident

y or electron

eterized

shower simulation

[16] package shapes and

All the essential

of electromagnetic

showers are

program.

parameterization

generate complete

geometrical

that can be used to model a specific detector.

physics processes that take place in the development An empirical

background

by Monte Carlo simulation.

developed at CERN, which contains

were established

arise from:

failure to reconstruct

Such sources provide

The Monte Carlo simulation

represented

which yield

T” OI u decays; and coalescence of nearby photons from symmetric

decays of high-energy

material

makes its mea-

was developed

energy in the appropriate

to simulate strips

the deposition

of the EMLAC.

of the

A param-

was chosen because of the large CPU time required

electromagnetic

by a comparison

showers.

The parameters

of data with the full-shower 17

to

used in the simulation Monte Carlo, and were

tuned by varying teristics

the input

shower shape and parameters

(such as the zero suppression

were established

window

by the Monte

data events.

After

the simulated

events and the data, and the resultant

assess backgrounds simulation

the parameters,

to the direct-photon

Test criteria

Carlo simulation

good agreement

signal.

charac-

was achieved

Monte

with

between

Carlo was then used to

Each of these aspects of the overall

process is described below in more detail.

Simulation

of Detector

Response

shower generation

were determined

using the GEANT

full-shower

rameterized

-

and 100GeV

capability.

The lateral

obtained

simulations

during

values of Er/Et

for the front

calibration

of the detector

was also determined

(Et/E,)

response.

GEANT

simulations

to adjust the front to total energy ratio of individual

energies were modified tuations

Sian distribution

with

to incorporate

window

used during

and the presence of inactive in the simulation. of crosstalk

(
3.0 GeV/c

the momenta exception

were used as the Monte Carlo input

of all observed photons

that two-photon

as individual

photons,

combinations

Carlo equivalents

original

of all particles

then allowing

background

same production

were generated

by a random

the remainder

with

by simultaneously

For each rotating

the

angle around the beam direction,

and

from 7, 7’ and w decays, we assumed the simply replaced by particles

of the event kinematics

n”s in

of the appropri-

unchanged.

This was done

because the measured signals for states other than the r” had large backgrounds. assumption

of similar

data on relative

v/r0

production

the

were not treated

for these mesons as for n’s, so the simulated

the Monte Carlo sample were therefore ate mass, leaving

Carlo,

in their rest frames.

contributions

characteristics

in the Monte

with a mass value of 135MeV/c’.

the K’S to decay randomly

To simulate

events. For each such event,

with mass < 175 MeV/c*

but rather as r’s,

event, five Monte momenta

were included

and

properties

for the 7 and x0 is consistent

yields, which as we shall shortly

The

with our

see, do not appear to depend

on PT.

In order to increase statistics with the original pi

> 4.0GeV/c)

tributed

at the highest-p?

event structure,

values, events were also generated

but with leading particles

boosted in pi by an additional

to the boosted samples yielded

(#,v,

1.25GeV/c.

a result consistent

. . . that had initial The background

with the original

at-

sample

in the region of overlap. To account for the systematic suppression

loss of low energy showers (a consequence of the zero

employed in the LAC readout),

low energy photons

from PYTHIA

[17]

generated # events were added to the event driven Monte Carlo in order to duplicate the particle

and energy distributions

Acceptance reconstruction described

and Reconstruction efficiencies

Efficiency

were determined

above. Except for the geometric

with large inefficiencies real events. abutted

observed in the data.

were excluded

In particular,

steel support

radius of the EMLAC,

Geometric

acceptance

using the sample of Monte acceptance

calculation,

from consideration

regions of the detector

plates),

-

regions near the central

and regions between octants 19

Carlo events

kinematic

for both Monte

near quadrant

and

regions

Carlo and

boundaries

(which

beam hole and at the outer

were not included.

The geometric rapidity.

acceptance for r’s and ys was determined

The ratio of the number

of ?y” decay, found within determined

the fiducial

as a function

for PT. To determine

of single-photons,

region to the number

single-photons

and #s in the data.

and x’s was determined

The reconstruction

in a similar

and reconstruction independent

had considerably

reconstructing

r’s

simulations

was

interpolation

larger

efficiency

for single-photons

manner using the same reconstruction

algorithm

14 shows the average weight for the combined for nos. The single-photon

corrections

and yc,,,, with

of pi

weights were

an average value of about

average weights,

from two photons,

photons had to fall within Comparison

that were generated

The inverse of this acceptance ratio was used to weight

employed to analyze the data. Figure

r’s

pairs in the case

the acceptance at any given pi and ycm, a quadratic

on this grid.

essentially

OI two-photon

of pi and

of pi and ynn, using a grid of 0.1 units for yc,,, and 0.5 GeV/c

was performed

acceptance

as a function

which

reflects the larger

and the fact that

1.25.

The

challenge

to reconstruct

a r”

of

both

the acceptance of the EMLAC.

of Data

and Monte

Carlo

agree with the data, a set of criteria

-

To verify

that

the Monte

Carlo

were used to compare the results of

the Monte Carlo with data; these are discussed below. Partially

reconstructed

or misidentified

direct-photon

background.

An indication

the two-photon

energy asymmetry

subtracted

asymmetry

asymmetry

values is attributed

corrected

account

for about

distribution.

Figure

Monte

Carlo and data.

plot for both

80% of the total

of the loss of n”s can be obtained

source of background

from

15 shows the backgroundThe fall off at large

to losses of low energy photons.

between Monte Carlo and data indicates that this dominant

r’s

The agreement

that the ?y” loss is properly to the direct-photon

simulated,

and

signal can be adequately

on the basis of the Monte Carlo calculation.

Another

important

comparison

front of the EMLAC

relative

not make a direct

contribution

the ratio of reconstructed

to the total (Ef/E,).

can influence

the reconstruction

other neutral

mesons. Figure

Carlo simulation

involves

Although,

to the direct-photon of photons,

16 shows the reconstructed

and data for different

photon

energies.

good over a wide range of energies, with a significant

the value of Ef/E,

background,

and thereby

energy in the

shifts in (Ef/E,)

the identification (Ef/E,)

does

of r’s OI

ratio from Monte

The agreement is relatively

discrepancy

only at the lowest

energies. Another

indication

of the performance

of the Monte Carlo is the comparison 20

of

x2 distributions

for fitting

photons

ference in x2 distributions in the reconstruction for reconstructed

for data and Monte Carlo could imply

suggests that there are no major

(which

This is plotted

contribute

by the comparison

in Fig.

that the Monte

Carlo properly

to the combinatorial

data that were not simulated for by a 2.6 % correction background

in the Monte

Sivnal

the background

in the plot).

photons

The small difference contributions

in the

This difference was compensated -

To determine

that satisfied the direct-photon

events (which contain no direct-photons)

-0.2

single-photon

as (+y/?~)b~k,was used to estimate

single-photons

of pi for four rapidity

as a function

< ycm < -0.2,

criteria,

the number

of

the Monte Carlo

were processed using the same analysis code

the data. The reconstructed

r” yield were then obtained ratio of the simulated

mass

are also in good agree-

to non-Gaussian

Carlo.

invariant

to the # cross section.

single-photons

ycm < 0.7, -0.7

represents

background

to the Direct-Photon

employed for analyzing

of the two-photon

18. These distributions

observed in the tails of the ?y” peak is attributed

Background

differences in reconstruction

Carlo.

measure is provided

ment, indicating

a possible difference

photons from Monte Carlo and data. Again, the agreement between

between data and Monte distribution.

dif-

efficiency between the two. Figure 17 shows the x2 distributions

the two distributions Another

to the assumed shower shapes. A systematic

and the accepted intervals:

-0.7


50%. This method a purely

Monte

of individual

for determining Carlo method,

trigger

corrections

is more appropriate

because these corrections

are based on the structure

events rather than on averages over many events. However, for unusual

decay configurations

this procedure

can occasionally

lead to large event weights.

situations

were handled by using the magnitude

correction

only when it was less than 2, but using a calculated

termined

by implementing

corrections

the trigger

that exceeded this limit.

of the individual

in the Monte

Carlo)

The average correction

this trigger for any given region of the detector, as described

in the previous

account for any variations whether

paragraph.

in the trigger

corrected

automatically

The systematic

variations

for trigger

uncertainty

in two ways. The first method trigger thresholds;

higher trigger trigger

performance

trigger

of pi and

respectively. required

for analysis of

weight was calculated,

weight is then calculated

to

of other octants by determining

the event would have satisfied the trigger in the other octants as well. This

weight accounts for threshold

different

a preliminary

An additional

average value (de-

as a function

When the pr observed in an event exceeded the minimum

Such

event based trigger

for events with

is shown in Figs. 20 and 21 f or r” and single-photons,

rapidity

than using

threshold

threshold

and dead channels in other octants.

variations

introduced involved

at fixed rapidity by the trigger

comparing

We thus

for all octants.

corrections

was estimated

no cross sections from runs with

the cross section in the turn on region for the run with the

was compared to the cross section for the run with the lower

which was nearly

100% efficient in this region.

23

The second method

involved

shifting

to determine paring

the trigger

the sensitivity

runs with

different

pi corresponding atic uncertainty

of the cross section to the threshold trigger

to a trigger

thresholds

efficiency

was parameterized

the pi dependence corresponding

to a trigger

using a function

efficiency

an uncertainty

Com-

of 2.5% at a of this system-

whose shape was determined

corrections,

normalized

of 80%. Although,

of these systematic

the first procedure

indicated

measurement.

of 80%. The pi dependence

of the average trigger

much lower estimates through

turn on curve for a given run by one standard-deviation

uncertainties,

by

to 2.5% at the pi

the second technique the uncertainties

yielded

estimated

were used as upper limits to the actual uncertainties

from

the trigger. Due to inefficiencies

in the pretrigger,

17% for the backward

rapidity

an additional

pi independent

region and 2% for the central

regions was applied to each cross section, which contributed uncertainty

correction

and forward

an additional

of

rapidity

systematic

of 2%. F. Other Corrections to the Cross Sections

In addition invariant l

to the corrections

cross sections also include

Absorption material

of beam particles between

Vertex reconstruction examination

corrections - This

the beam counters

The average correction l

discussed in the preceding for the following

correction

effects:

was calculated

and the interaction

point

based on the in the target.

was 6%. inefficiency

- This correction

was determined

of events that did not have a reconstructed

was found to be linear in vertex position, end of the target,

sections, the reported

vertex.

from a visual The correction

decreasing from 1.09 at the upstream

to 1.04 at the downstream

end, with an average correction

of

6.5%. . Online

veto due to backscatter

which particles thereby

vetoing

from the interaction the event.

target configuration. which

did not include

in which

- Corrections

were emitted

The magnitude

The correction veto-wall

were made for losses of events in backward

of this correction

was determined

depended

using minimum-bias

signals in the trigger.

the veto wall had a signal in time was studied

24

into the veto walls,

The fraction

on the events

of events

as a function

of the

number

of interaction

correction

l

counters

that had signals.

for the number of accidentally

direct-photon

and rr” triggers is comparable

Losses due to the off-line criteria

This correction

to the triggering

of the EMLAC.

Losses from

with good events were corrected by determining

the number

quadrant

by the application

averaged 8.5%. The same correction

Losses due to the uncorrelated above, was used to estimate small uncorrelated

of the veto-wall

was used for direct-photons.

energy requirement

- A similar

procedure

the loss of events from the requirement

energy in the trigger

quadrant.

This

to the

of having

correction

averaged

Losses due to 7 conversions

in material

downstream

of the interaction

- The correction

conversions

downstream

of the vertex

for photon

by tracking

individual

tion lengths

of material

Asymmetry

criteria.

0.6%.

corresponded l

events.

on signals in the veto wall - In order to eliminate

of good R’ events that were eliminated

l

for

by a muon, an event was rejected if either of the veto walls had

random coincidences

approximately

to that for minimum-bias

a

which

to losses of x 3.5%. We assume that the correction

typically

a signal corresponding

l

vetoed events due to backscatter,

corresponded

events triggered

This was used to generate

photon

trajectories

the photon

traveled

to determine through.

the number

vertex

was found of radia-

The average correction

to 7% per photon. cuts applied to rr” and r~ candidates

were applied

to two-photon

pairs in the ““(7)

samples of data for the ?y” and 7 cross sections.

- Asymmetry

cuts of 0.75 (0.6)

mass regions The isotropy

to obtain

clean

of r” and 7 decay

was used to account for these losses. l

Losses due to restrictions and the restriction were applied photon

signal.

l

directionality,

photon

arrival

time, Ef/E,,

that charged tracks not overlap with showers - These criteria

to minimize

background

Only a small number

which a correction from #

on photon

was estimated

decays. The correction

Losses of direct-photons

from muons and hadrons

in the direct-

of photons were removed by these cuts, for

by gauging their effect on photons

originating

was 5.5% for direct-photons.

due to random combinations

with other photons result-

ing in a pair in the x0 or 7 mass band - This loss was estimated 25

from the Monte

Carlo by replacing mentum.

?y”s or 7s in an event with

The events were then reconstructed,

that were lost was determined

as a function

to correct the observed direct-photon

single-photons

of the same mo-

and the fraction of pi.

of single-photons

This dependence

signal; the correction

was used

factor was typically

10%. Table I summarizes both beam polarities, corrections

the corrections

discussed above, for x0 and y cross sections, for

on Be. Except for the overall trigger and acceptance corrections,

to the 7 cross section are similar

The net systematic

uncertainty

to those for x0 production.

in the overall normalization

due to the corrections

listed above, as well as from several smaller effects (at the 1% level) that have not been described,

is N 10%. G. Summary of Systematic Uncertainties

The primary sources: (E/P)

contributions

uncertainty function,

in parameters (including

in the determination

uncertainty

in the magnitude

related to background These systematic

These systematic

are summarized net uncertainty

tables.

in the direct-photon

energy

corrections,

response),

uncertainty cross section

and the overall

have been discussed in the preceding

for the direct-photon in quadrature,

Combining

as given by the

for the direct-photon

of the detector

uncertainties

uncertainties

arise from the following

of the trigger

subtraction

in Table II and, combined

sections in the appropriate

uncertainty

of the EMLAC

such issues as proper modeling

normalization. sections.

to the systematic

and rr” cross sections

are quoted with the cross

these systematic

uncertainties

cross section of N 25% at a pi of 4 GeV/c,

decreases to x 16% at a pr of 8GeV/c;

yields a which

for the x0 cross section, the uncertainty

is

x 11% for all pi values. Based on the analysis

of the final distribution

regions, and results from ye+e-

of 77 pairs in the rr” and n mass

studies, we estimate an upper limit on the uncertainty

in the energy scale of 0.9%. This represents

an uncertainty

contributes

to the cross sections, which has not been

included

an additional

N 10% uncertainty

in the energy scale, and

in the tables.

The central values of the beam momenta were uncertain an uncertainty

(FZ 10%) in th e c al cu 1a t’ion of the theoretical

26

to N 2%. This introduces expectations.

IV. CROSS SECTIONS A. Neutral Mesons Tables III,

IV and V show the invariant

duction

in 500 GeV/c

rapidity

as well as the full rapidity

comprehensive

x-Be

cross sections per nucleon

and pBe interactions,

respectively,

range, as a function

than those given in our previous

of pi.

for r” pro-

for three regions of These results are more

abbreviated

publication

[8]. Figures

22 and 23 show the cross sections for the same data sets averaged over the rapidity range -0.7

< ycrn < 0.7.

The curves in Figs.

22 and 23 represent

by Aversa et al. [18], using parton

based upon code provided for the pion and nucleon are illustrated

next-to-leading-log

from Aurenche

et

al. [19].

for two choices of the scale (Q’

between data and theory x- and proton

QCD calculations distribution

The calculated

functions

cross sections

= p$ and Q* = &/4).

(for the smalIer choice of scale) is reasonable

Agreement for both the

data over the full pi range shown. r

The analogous over the rapidity

results for ?y” production

in K-Cu

and pCu collisions,

averaged

< gc,,, < 0.7, are given in Table VI, as a function

range -0.7

of

PT.

A Dependence

nucleon.

-

However,

The cross sections for rr” and 7 production

effects such as nuclear

nucleus can result in deviations nucleon number.

in the

of the cross sections from linear dependence

on the

The form A”, with a being a parameter

both beams are consistent

is often used to parameterize

on Cu and Be targets.

with one another,

are consistent

with

‘The calculations used preliminary al. via private communication. fits.

in question,

that can depend on rapidity,

Table VII shows the measured values of a determined

for T- and proton induced collisions The results

or secondary

previous

and indicate measurements

x0 fragmentation

E706 x0 production

function,

The average values of a for a small deviation

27

from unity.

by the Chicago-Princeton

functions provided by J. P. Guillet et

data were included in the fragmentation

while the quark fragmentation

by results from e+e- data.

the A-

for r” production

However, data from UA2 and the ISIt dominated the determination

fragmentation

per

scattering

pi, and on the nature of the particle dependence.

shadowing

are reported

function

of the gluon

was constrained primarily

group [20], as well as recent measurements similar

from Fermilab

experiment

E605 [21], for

regions of PT. The curves shown in the Figs. 22 and 23 have been calculated

using a value of cx equal to 1.08. The value of (Y was not well determined inadequate

statistics

direct-photon

for direct-photon

value of a for direct-photon x0 Production

targets,

and are expected to yield a

production.

Ratio

-

The corrections

using methods similar

necessary to measure 7 production

to those already described for nos. The mea-

sured ratio of T/T’

production

is shown in Fig. 24; within

ties no significant

dependence

on pT or on incident

value for the 7/x0

production

ratio is 0.44 & 0.05 f 0.05 for r-Be

0.44 !c 0.06 f 0.05 for pBe interactions. of the same magnitude uncertainties

Systematic

as the statistical

in the trigger corrections

the rather large uncertain-

beam is observed. uncertainties

uncertainties, measurements

actions

26 displays

as a function

and

on these results are reflect residual

1221. Although

our pBe re-

all the data are compatible

uncertainties. B. Direct-Photon

Figure

interactions

and primarily

sult is somewhat lower than those of the ISR experiments, their statistical

The average

and in the energy scale. Figure 25 summarizes

our results and those of selected previous within

due to

for data on Cu. Recent runs of E706 have a far larger sample of

events from Be, Cu and hydrogen

were determined

production

the observed 7/rr” of pT. Photons

regions (with energy asymmetries sample; however, no isolation

Cross Sections production

contributing

ratio for x-Be

and pBe inter-

to 77 pairs in the rr” or 7 mass

less than 0.75) are excluded from the direct-photon

cuts are imposed.

The bands shown in the figure repre-

sent Monte Carlo estimates of the residual contributions

to these ratios from photons

originating

is due to ?y” decays and the

from meson decays; 80% of this background

rest is dominantly uncertainties

due to 7 mesons. The width

on the Monte Carlo calculation.

The background sponding

estimates

shown in Fig.

26 are lower than those in the corre-

figures in our earlier paper [8]. Although

a shift downward

of the midpoint

est values of pi, and essentially moreover

of the bands represent the systematic

the direct

photon

the data points in Fig.

the difference is relatively

of the band by about 30% of its width

small

at the low-

no change at high pi - the effect is systematic,

signal, which is proportional

26 and the estimated 28

background,

to the difference is sensitive

and

between

to even such

small effects. This change, which results from an overestimation from r” decays in our earlier analysis, has been incorporated sections presented below, which as a result are slightly pi

(by 60% of the estimated

systematic

of the contribution

into the inclusive

cross

higher at the lowest values of

uncertainties)

than those presented

in our

earlier paper. Tables VIII, photon

IX and X show the invariant

production

a function

in r-Be

of PT.

Figures

over the full rapidity to-leading-log functions