%Paper: 
%From: FORDEN@uazhe0.physics.Arizona.EDU
%Date: Fri, 11 Jun 1993 09:00 MST
%auto-ignore

Introduction.
	Quantum Chromodynamics (QCD) has successfully explained the properties of jets
observed during proton proton and proton anti-proton collisions over a large
range of
center of mass energies as well as jet transverse momentum(1).  The gross
features of the
observed hard interactions can be explained by tree level graphs with small
modifications
by next to leading order (NLO) corrections.  The DZERO detector(2) at the
Fermilab
TEVATRON has searched for events which indicate that there was a net zero
exchange of
color between the colliding proton and anti-proton.  Such processes, if QCD in
origain, can
only come about during the exchange of at least two gluons and are inherently
higher order
processes.
	This search was originally motivated by suggestions from Dokshitzer et al.(3)
and
Bjorken(4) in searches for W+ W- boson scattering in the production of heavy
Higgs
bosons at the SSC and LHC.  In this process a quark from each proton in the
collision
radiates a real W boson which then annihilates to form a Higgs.  The unique
feature that
was pointed out by Bjorken and others is that since there is no color exchange
between the
two hadrons (the W bosons being color neutral) the "underlying event" should be
separated
into two distinct halves with a "void" between the two tagging jets.
	This can be understood, at least qualitatively, by considering the Lund
picture of
hadron-hadron collisions(5).  The two hadrons after the low Pt interaction
which dominates
minimum bias events observed at the TEVATRON and the SPPbarS are viewed as
consisting of a quark (anti-quark) and diquark (anti-diquark).  However, there
has,
presumably, been an exchange of color between the two hadrons and the color
singlets
now consist of the quark (anti-quark) from the proton (anti-proton) and the
anti-diquark
(diquark) from the anti-proton (proton.)  These color singlets are connected by
two strings
which stretch across the event.  This is illustrated in Figure 1a. When the
strings fragment
they produce the particles for the underlying event.  As the scale of the
interaction increases
this picture makes a natural transition to larger Pt jets, with the added
complication that the
drawing of the color lines involving gluons become somewhat ambiguous.  There
is
always, however, at least one color line "crossing the event".  If the
interaction involved
zero net color exchange, as the WW scattering example does, then there would
not be any
color connection between the two "halves" of the event and we would expect to
find a void
between them.
	At the Fermilab TEVATRON, with a center of mass energy of  s**1/2=1800
GeV/c2, the dominant high Pt processes with zero net color exchange would
either be
photon exchange between quarks (and anti-quarks) in the two colliding hadrons
or multiple
gluon exchange, illustrated in Figures 1b and 1c respectively.  It is still an
experimental
question which of these two processes dominates.
Selection of Events.
	Jet events were selected using three different criteria in order to
continuously cover
a wide range of pseudo-rapidity separations of jets.  All three triggers used
the standard
DZERO Level 1 trigger towers.  These consist of calorimeter cells forming
pseudo-
projective towers which are 0.2 x 0.2 wide in Delta eta x Delta phi.  The two
single jet
triggers each required a single such tower with transverse momentum greater
than 7 GeV/c
at Level 1 and then at least 1 reconstructed jet in Level 2 (an on-line
software trigger which
fully unpacks the calorimeter cells) with greater than 30 GeV/c transverse
momentum.  One
of these trigger chains, which will be referred to here as the single forward
jet trigger,
required that both the Level 1 tower and the reconstructed Level 2 jet have
absolute pseudo-
rapidities greater than or equal to 2.0.  There were no other requirements at
the trigger level
that there be other jets or other particles present for either of these two
triggers.  In fact,
most of the events recorded with these triggers were either two or three jet
events. The
kinematics of the parton scattering lead to most of the events for both these
triggers to have
relatively small separations between jets.  The single forward jet trigger had
the "other jet"
being produced predominately on the same side of the event as the trigger jet
(thus the event
center of mass was boosted in one direction or the other) as illustrated in
Figure 2 which
shows  the pseudo-rapidity distribution of the other jet.  The other single jet
trigger did not
make any requirements on the pseudo-rapidities of the jets.  Again the parton
distributions
made events with central jets predominate and hence most of the events obtained
from this
trigger had relatively small separations in pseudo-rapidity (eta.)  However,
both these
classes of events had significant numbers of events with large (>4.5)
separations of jets in
eta.
	The third jet trigger used was aimed at selecting events with large
separations
between jets and will be referred to as the gap trigger.  The gap trigger
required events to
have at least two Level 1 trigger towers, each with 7 GeV/c or more transverse
momentum.
Furthermore, the event was required to have two such trigger towers to have
absolute h's
greater than 2.0 and a difference in eta between them of greater than or equal
to 4.0 (i.e. on
opposite sides of the event.)  The Level 2 trigger then required two jets with
reconstructed
transverse momentum greater than 25 GeV/c, absolute h's greater than 2.0 and,
again, a
separation in eta greater than 4.0.  No requirement was made on the presence or
absence of
jets or particles anywhere else in the event for this trigger.  Thus an event
with the two
triggering forward jets and a central jet would have the same probability of
triggering the
detector as an event with the same forward jet kinematics but not the central
jet.
	The events accepted with these triggers cover the complete range of
pseudo-rapidity
separations of jets from zero to more than 6.4.  The single forward jet trigger
provides a
continuous transition to the gap trigger.  None of the triggers are biased
toward events
without an underlying event.

Calorimeter Response to Single Particles and the Definition of a Void
	The calorimeter provides a very uniform hermetic detector for single particles
as
well as jets and is used instead of the tracking system for the detection of
the presence of
single particles.  This insures an excellent efficiency for the detection of
electromagnetic
showers.  The tracker was used, however, to help determine the response of the
calorimeter to the low energy charged hadrons expected to comprise the majority
of the
underlying event.  Figure 3 shows the Monte Carlo determined efficiency for
reconstructing single tracks superimposed on the Monte Carlo expectation for
the single
charged  particle
spectrum for minimum bias events.  The actual spectrum can not be determined by
DZERO
since it lacks a central magnetic field.  However, the spectrum is a steeply
falling function
of energy and the tracking efficiency is a very steeply rising function with a
sharp turn on at
approximately 200 MeV/c2.  Hence  most of the particles seen in the tracker are
approximately 200 MeV/c2 in energy.  The showers associated with tracks in
minimum
bias events were examined to determine the efficiency for particle detection
(as opposed to
making energy measurements.)  Figure 4 shows the single electromagnetic
calorimeter
tower response to good central tracks.  It is clear that a calorimeter
requirement of greater
than or equal to 200 MeV/c2 in a single tower would detect a significant
fraction of charged
particles with energies greater than 200 MeV/c2 as well as most electromagnetic
showers
with energies greater than 200 MeV/c2.  This single tower cut is currently
forced on this
analysis since the standard DZERO data reduction chain does not retain full
calorimeter
information in a readily accessible manner for large numbers of events.  It is
expected that
this situation will change this summer when the data production scheme is tuned
for this
analysis.
	The rapidity gap between jets is defined as the distance in pseudo-rapidity
from the
inner edge of a jet to the inner edge of the next, closest, jet.  This is
illustrated in Figure 5.
A gap is called a void if there are less than or equal to one electromagnetic
tower with
transverse energy greater than 200 MeV/c2 in this gap.  The effect of varying
this definition
of a void is discussed in the next section on results.
Results
	Figure 6a shows the probability of finding a void, as defined above, as a
function
of the gap between jets.  The events selected with the single forward jet
trigger are used
here for gap widths less than or equal to 2.6 in pseudo-rapidity (hence jet
separations less
than or equal to 4.0) and the gap trigger there after.  There is unit
probability of finding
such a void when there is zero separation between jets
independent of the production mechanism.  As the gap width increases the
probability of
finding a void decreases, being dominated, presumably, by fluctuations in the
fragmentation of the underlying event.  At a gap width of approximately 2.1
there is a
transition from the exponential fall off to a constant fraction of events with
a void.  Figure
6b shows the same plot but with the unrestricted single jet trigger used for
gap widths less
than 2.6.  The effect of the definition of the gap is illustrated in Figure 6c.
 The definition
of a void has been loosened to include events with fewer than five
electromagnetic module
towers with transverse momentum less than 200 MeV/c2 .  A plateau with constant
probability has appeared out past gap widths of 2.8.  This can be interpreted
as the region
over which the underlying event can fragment to a maximum of four particles
between the
jets without a significant effect on the probability of such fragmentation.
However, there is
still a region of large gap widths which have approximately the same
probability of being
voids as with the tighter definition of a void.  Finally, Figure 6d shows the
void probability
of finding a void as a function of gap width for events with a lower jet Pt
requirement (30
GeV/c), single forward jet trigger, as opposed to the 40 GeV/c requirement used
above.
The presence of the region of constant probability of a void indicates that
this is not being
caused by some trigger bias which is Pt dependent.  It is still possible,
however, that an as
yet undetected systematic effect could be causing this plateaue.  One
possibility is that the
efficiency for detecting single particles in the electromagnetic calorimeter is
(detector) eta
dependent.  We are continuing to search for such detector effects.
Discussion
	The presence of a transition from an exponentially falling to a constant
probability
of producing events with voids seems to indicate that there is a new production
mechanism
becoming dominate for large voids.  If this process is the exchange of a
photon, as has
been suggested as a possible source, then the participating partons must be
charged quarks.
The two jet subsample, shown in Figure 2, may be used to
indicate the relative population of initial partons in the interaction.  Simple
kinematics
relates the incident partonUs fractional momentum to the final state jets
rapidities and
transverse momentum.  This relationship is illustrated in Figure 7a which shows
the
Bjorken x of the partons as a function of the "other" jetUs eta.  These can
then be used to
estimate the parton densities as a function of the "other" jetUs eta, as shown
in Figures 7b
and 7c where Morfin Tung(6) lowest order parton density set has been used(7).
	The probability of a reaction in p pbar collisions is determined by the
product of the
parton densities and the subproccess cross section.  Figure 7b and 7c  show
that, even
before weighting by the parton coupling constants, the vast majority of events
in this
kinematic range are caused by either gluon-gluon or gluon-quark scattering.
When the
strong and electromagnetic couplings are included it appears very unlikely that
photon
exchange could account for even the largest void events (where quark-quark
scattering is
suppressed by at least a factor of four to five in the parton densities.)  In
summary, with
our experimental definition of a gap, we observe that about 1% of events with
%%jets
separated by a large rapidity distance are consistent with zero particles
produced between
the jets.

References
1.	R. K. Ellis, in 7th Meeting of the Divison of  Particles and Fields of the
APS,
            10-14 Nov.  1992 Batavia IL, 1992),
2.	R. J. Madaras, in Proceedings of the
	Division of Particles and Fields Batavia IL, 	1992),
3.	Y. L. Dokshitzer, V. A. Khoze, T. Sjostrand, Phys. Lett. B274, 116-121
(1992).
4.	J. D. Bjorken, Phys. Rev. D47, 4077-4087 (1992).
5.	M. Bengtsson, T. Sjostrand, Compt. Phys. Commun. 46, 43 (1987).
6.	J. Morfin, W.-K. Tung, Z. Phys. C52, 13-30 (1991).
7.	H. Plothow-Besch, in 3rd Workshop on Detector and Event Simulation in High
		Energy Physics K. Bos,  B. v. Eijk, Eds. (NIKHEF-H, Amsterdam, The
                       Netherlands, 1991), pp. 148-163.

* The DZERO Collaboration consists of: Universidad de los Andes (Colombia),
University
of Arizona, Brookhaven National Laboratory, Brown University, University of
California,
Riverside,Centro Brasiliero de Pesquisas Fisicas (Brazil), CINVESTAV
(Mexico),Columbia University, Delhi University (India), Fermilab, Florida State
University, University of Hawaii, University of Illinois, Chicago,Indiana
University, Iowa
State University, Korea University (Korea),Lawrence Berkeley Laboratory,
University of
Maryland, University of Michigan,Michigan State University, Moscow State
University
(Russia), New York University, Northeastern University, Northern Illinois
University,
Northwestern University, University of Notre Dame, Panjab University
(India),Institute
for High Energy Physics (Russia), Purdue University, Rice University,University
of
Rochester, CEN Saclay (France), State University of New York, Stony Brook, SSC
Laboratory, Tata Institute of Fundamental Research (India),University of Texas,
Arlington,
Texas A&M University


