189

DIQUARK EFFECTS IN FRAGMENTATION

David Hanna, Harvard University, Cambridge, Mass., USA (Aachen-CERN-Harvard-Munich-Northwestern-Riverside Collaboration)

ABSTRACT

Some results from an ISR experiment looking at pp collisions with a high-pT at (y = are presented. Forward production in the same hemisphere30° as the 1.tr3)igger indicates that when a valence is removed from the proton, the remaining valence behave as a rather than fragmenting individually. Preliminary evidence for charmed production is also presented.

RESUME Nous presentons des resultats d'une experience aux ISR. Nous y avons etudie des collisions pp dans lesquelles il y avait un pion energetique (y La production de particules petits angles dans le meme hemispherea 30indique° = que1,3), . lorsqu'un quark de valence sorta du proton, ceux qui restent produisent des parti­ cules d'une maniere coherente. Des preuves preliminaires pour la production de charmes sont presentees . 190 The fragmentation of quarks has been extensively studied in e+e- annihila�ion as we ll as in collisions and deep inelastic leptoproduction experiments1l, but data on the "diquark" system which remains when a valence quark is ejected from a is considerably more limited2). In order to understand more about the diquark fragmentation3l mechanism, we have undertaken a study of small-angle hadron production in pp collisions triggered by a high-momentum pion at 30° in the same hemisphere. In the naive quark-parton model, the trigger pion is thought to be the decay product of a quark e ected in a hard scattering process, as shown in Fig. "or j 1.

Y1

TRIGGER

p = -Js12

Fig.

2.5 a pion of sufficiently high pT GeV/c) , the parent is likely to be a valence quark (x 0.2), implying the (>product ion of a forward diquark. Moreover, the bj charge of the> trigger pion should reflect the flavour of the quark (u for n+ ; d for TI-) and thereby give a clue to the flavour content of the diquark (ud or uu) . A morE' realistic calculation using the QCD-inspired mc,del of Feynman, Field and Fox4) indicates that hard may be the origin of many of our triggers . Table shows the fractional composition of a pT 3 GeV/c trigger as given by this calcu­ latior5 ) . = Table 1

h h- r ° 63Origi GeVn, of ig3 GePV/c,trigger 8 = 30 particle' 0.26 Pr = Xbj Fraction of the trigger Parent parton n+ - 1T u quark 0.57 ± 0.03 0.17 ± 0.03 d quark 0.07 0.03 0.37 0.03 ± ± 0.32 0.03 0.40 ± 0.03 ± anti-quark 0.04 ± 0.03 0.05 0.03 ± 191

Most of the apparatus has been described previously6) . Briefly, it consisted of two coaxial spectrometers surrounding Beam just downstream of intersection region I6 (Fig. 2) . The outer spectrometer was used to define and detect the trigger particle, whilst the inner one measured forward fragments from 1° to 6° with respect to the beam axis.

R-606 Generacut throughl layout (Arm (vertical beam ax1 s}

B , Magnet mwp.c U 1 ) IT'Wpc Bock

16 - ,/ �I _ :::-:i ; : } ���9n9:!�s \ I \H EMt 01 81

Im

TF;

chambers EM 1 = Spectrometer magnet B z Dr1It 3 = Lampshade magnet 8, EM 8; c, U1Dt F Sc intillation counters C2 =Cerenkov counters =Shower counters l ��,C, ur 18, S, 018, F ig. 2

The trigger spectrometer was based on an air-core toroid known as the Lamp­ shade Magnet (LSl!) , which had a l/r field dependence with a maximum value of 3 kG and an average field integral of 1.5 kG•m. Twelve coils divided the azimuth into 30° sectors, 10 of which were instrumented with a Cerenkov (C) and shower (S) coun­ ter combination as well as trigger scintillators (TF, TB) . The Cerenkov counters contained C02 at atmospheric pressure, which has a pion threshold of 5 GeV/c. Large drift chambers7) , before and after the magnet, measured the particle trajectories. The six sectors used in the trigger covered the range 45° 135° and 235° 315°, and were equipped with a paddle scintillator (B) bei< ng � < the shower counter< � , <

The trigger was designed to select events in which a fast TI± was emitted at 30° with respect to Beam 1, so a pulse from an LSl1 Cerenkov counter was required. To reduce background from pairs or delta-rays firing the Cerenkov, we demanded single-particle ionization in TB and the shower counter . This trigger rejected , on line, 99% of below pion threshold, and a later software cut eliminated all but a small sample of pT GeV/c triggers which were saved for comparison purposes. At of 1 GeV/c the< 1 trigger particle may only be identified as an Pr "inclusive hadron'', but above Cerenkov threshold (pT GeV/c) the fraction of approaches 1. = 2.5 Th·e data presented here were taken during 100 hour.> of running (at of 63 GeV) , corresponding to an integrated luminosity of � 1.4 pb-1• Track/S recons­ truction and fiducial cuts in the LSM left a total of 1·40,000 positive and 107,000 negative trigger tracks with pT 1 GeV/c. > About half of the selected events were found to have one or more charged par­ ticles in the forward spectrometer, and these were identified using in­ formation from the Cerenkov counter. The K, and p thresholds in freon-114 are TI, 2.6, 9.3, and 17.7 GeV/c, respectively, which makes TI/K separation very difficult over much of the momentum range of interest here. On the other hand, it is easy to distinguish between and baryons since the former are almost non-existent above proton threshold. For our purposes, the structure, qq or qqq, is more funda­ mental than the flavour, so we label forward hadrons as either mesons or , using the notation to denote mesons since pious are the dominant component . TI The data were first examined for evidence of the ¢ correlation between the trigger and the forward particles. This correlation, which has been seen in pre­ ) vious ISR studies8•9 of high-pT phenomena, is related to the intrinsic transverse momentum of the partons . The trigger preferentially selects partons with some motion in the trigger direction. Figure 3 shows the difference in azimuth between the trigger and the forward hadron, for protons6¢ , of x > 0.4 for three ranges

Tr igger < 0.5GeV/c Trigger GeV/c c) Trigger GeV/c a) P,- bl P,- >I P,- >2

2

ltt+f

Fig. 3 193 of trigger P · The data have been subjected to fiducial cuts and have been r weighted by the inverse of the geometrical acceptance, Acc(p1 ,pT ,¢) . The fitted curve is a simp le cosine, and the growth and saturation of its amplitude with trigger P as well as the peaking at 6¢ 180° , are consistent with the notion of r • a recoiling system of partons balancing the= Fermi motion of the trigger parton. Using a simple model, we infer from the amplitude a value of 300 MeV/c for the intrinsic P of the constituents • r . we proceed in analogy with quark fragmentation studies, defining the invariant quantity zdnh f(z) Ndz = where nh is the number of hadrons of type h observed with fraction z of the di­ quark' s momentum, and N is the number of triggers taken. If D� (z) is the fragmen- 1 tation function for a diquark of type i into hadrons of type h and all forward par- ticles came from diquark fragmentation, then

f(z) = zEE.D� (z) , is the fractional population of type1 1 i diquarks . As mentioned earlier, where Ei we expect this picture to be only approximately true, owing to the presence of gluon triggers . Since the diquark is defined to be the proton minus a valence quark, its momentum is given by p - xbj)/S/2 where xbj is the fraction of the incident = (1 proton's momentum carried by the trigger quark. To calculate xbj ' we assume a sub-process like that shown in Fig. 1 and obtain, from simple kinematics, 2 xbj P (eY1 e Y )//S = r + In this experiment y 1 1.3 and y2 is not measured. Since previous measurements8) = indicate that (y2) 0.05 for a similar kinematical configuration, we use y2 0 in our calculation with= the result = x 0.075 bj Pr , where P refers to the trigger parton. Assuming that we trigger on hadrons taking r 80% .to 90% of the jet momentum, we use 0.08 xbj P , = r where P is now the transverse momentum of the triggering hadron. r Figure 4 shows f(z) versus (1 - z) for p, and TI- for positive and negative n+ , triggers in trigger P ranges 1.0-1.5 GeV/c and 2.5-4.0 GeV/c. Diquark effects are r expected to be more pronounced in the higher Pr plot. Arguments based on phase­ space considerations1 0 ) suggest that f(z) should behave as a power of (1 - z) , so we use a log-log scale where such behaviour appears as a straight line. For the 194

al bl •negative tngger •negativeNEGATIVE trigger MESONS •negativePOSI triggerTIVE MESONS cl PROTONS pos1t1ve tngger v positive trigger v positive trigger v 1.0 < PTTR1a < I. 5 GeV/c 1.0< PTTRta < !. 5 GeV/c l.O< PTrRIG < I. 5 GeV/c f

�I

+

el

•di negative trigger •negative trigger 11 NEGATIVE MESONS POSIT I VE MESONS •negative triggerPROTO NS •• v positive trigger v positive trigger v positive trigger 2. 5 < PTrR1a < 4.0 GeV/c 2.5< PT1R1a < 4.0 Ge.Vic 2 ,5 < PT,.,.< 4.0 GeV/c � �� 0.1 i

' +t  -;·:· " 0.01 f * t

0.001 0.1 0.2 t 0.5 0.2 0.5 0.2 0.5 1 - z Fig. 4 trigger 2.5-4.0 GeV/c interval, the forward mesons are consistent with a single Pr (3.5 2.5 0.2-0.7. power for both trigger signs for and for TI+) over the z-range The proton data require two powers, 0.5TI- for the negative trigger and 1 for the positive. The general features of these distributions are: i) protons are produced more frequently and with a harder z-spectrum than mesons; ii) there is a positive correlation between proton production and a negative trigger which increases with trigger pT; iii) a similar but weaker effect exists for TI+ ;

iv) the distributions are independent of trigger sign . - TI I 95 This behaviour supports a picture2) where the diquark behaves coherently1 1) . The diquark and ejected quark set up a colour field which excites quark pairs from the vacuum (Fig. 5a). Eventually, the diquark recombines with either a new quark or one from the sea to create a high-z baryon. Mesons are produced from the remaining quarks and are not correlated to the trigger sign.

a) b) BARYON ��M ESONS Fig . 5 An alternative process, where the two valence quarks recombine independently as in Fig. Sb, would produce fast mesons correlated to the trigger charge, an effect not seen in our data. Note that a trigger charge effect has been seen by another group8) for triggers at[ smaller angles. ] The dependence of the protons on the trigger sign can be interpreted in two ways. The uu diquarks inside the proton are faster than the ud diquarks for per­ haps the same reasons that u quarks dominate over d quarks at high xbj " Alterna­ tively, a uu diquark can transform into a proton or n++, both resulting in final­ state protons, while the ud system has more opportunities to produce final-state either directly or from n's. The weak dependence of n+ on trigger sign could also be connected to baryon resonance production. The probability to produce heavy quark pairs from a colour field should be flavour-independent, depending only on kinematic effects related to the mass of the pair and the effective c.m.s. energy of the separating charges. In this expe­ riment, for a 1 GeV/c trigger quark, we have the following approximate four­ momenta for thePr quark= and diquark (neglecting masses): pq (2.0, 1.0, 0.0, 1.7) pdq (30 , 0.0, 0.0, 30 ) which implies a c.m.s. energy for the system of 4.2 GeV. Since this is above threshold for producing a cc pair, it seems reasonable that charmed baryons should result from some of the triggers. With this in mind, we look for production in this data set. ++ + We have at our disposal uu and ud diquarks, so we look for the �c (uuc) and Ac (udc) states using the decays K-pTI+TI+ and K-pn+, respectively. The kinematics of the charm decays are such that the proton is more likely than the mesons to end up in Arm 1, carrying a sizeable fraction of the momentum, so we require that the proton be positively identified. This means that it must 196

be an Arm 1 particle with charge +l and momentum greate:r than pion threshold

(p � 2 . 6 GeV/c) in a cell where no light was detected. With the proton identified, the pions and are chosen by consistency checks . The pions are positive par­ ticles in Arm or the LSM that are not identified kaons or protons . This means 1 that if the pion candidate is in Arm and over threshold, it must produce a pulse 1 in the Cerenkov counter. Similarly, the kaons are any negative particle, but if in Arm and above pion threshold they must not radiate. 1 The trigger particle is always a spectator since in our picture it never arises from the decay of the obj ects we are looking for. The :iign of the trigger is noted and the charm candidates are binned in separate histograms depending on that sign. This is in keeping with diquark ideas since a positive (negative) trigger should + ++ produce a ud (uu) diquark and hence a /\c o:c ) more easily than a negative (posi- tive) trigger. Figure 6 shows the invariant mass spectra for K-p.<+ combinations in 40 MeV/c2 bins; 6a is for negative triggers and 6b is for positives. To be in this plot, the event must have at least two particles in Arm This amounts to requiring at least one track in addition to the proton, but 1.this extra track is not required to be part of the Kpn combination. It is effectively a multiplicity cut that enhances forward multibody states. It was found that a cut of p < 3 GeV/c for 2 the K improves the signal .

a) b)

40

K-p rr+ Mass - Negative Tr igger K-prr+Mass-- Positive Trigger Fig. 6

The positive trigger plot contains a > 3G bump in the 2.27-2 .31 GeV/c2 bin. The central value for the bin is 2.29 GeV/c2 , so we could claim that we have an 1 indication for the /\+ since some experiments report values near 2.285 GeV/c2 2) c Figure 7 shows the invariant mass for K-pn+n+ combinations, again divided according to trigger signs . No cuts are imposed here except to require that the 197 ' a) b)

�! 30

20

2. 2 2.4 2.6 2.8 2.2 2.4 2.6 2.8

K- p7T+ 7T+ Mass- Negative Tr igger K- p7T+7T+ Mass-Positive Tr igger

Fig. 7

x of the KpTITI system be greater than 0.2. We use 50 MeV/c2 binning because the ++ + statistics are poor and also the zc is expected to be broader than the Ac since it decays strongly. The 2.45 to 2.50 GeV/c2 bin in Fig. 7a has 22 events where 10 are expected, ++ a 4cr effect, so it is tempting to identify this with the Zc . It has no possibi­ lity to be since a charge-2 state coupling to strangeness is exotic in SU(3) . ++ +y* ) The Zc (12 ) has been observed indirectly1 3 and its mass has been inferred to be 2 ++ + ) 2.426 GeV/c , so we may be looking at the zc (3h ) excitation predicted14 to lie 50 MeV/c2 higher. The 3h+ state does not decay electromagnetically to the 1h+ + state because a strong decay to Ac by TI+ emission is allowed . This decay must be seen if we are to believe that we have produced z++,s. ++ + In looking for the decay of zc to Ac + TI+, we must also see how often a fake cascade from KpTITI background to KpTI background will occur. Table 2 shows the actual accounting; but in summary we have , from 33 control events, 10 cascades wh ile from the 22 Zc candidates we have 9. Note that in the cascade study we have used a mass + region 2.260 0.020 GeV/c2 for the Ac . This is because the KpTI mass spectrum for the negative ±trigger KpTITI sample shows a 3cr bump in this region rather than at the 2 + mass 2.290 ± 0.020 GeV/c (see inset in Fig. 7a) . These different Ac masses are a point of some concern since, although the topologies are somewhat different, it is difficult to imagine a systematic effect that would produce a 30 MeV/c2 shift in the same detector. These results are consistent with two hypotheses. The "peak" is just a back­ ground fluctuation (expect six or seven cascades) , or it contains some Zc's (for 6 Zc 's and 16 background we expect 6 + 5 11 cascades) . The Zc hypothesis relies mostly on the bump in Fig. 7 for its credib= ility . 198 Table 2 Kprrrr to Kprr Cascades Mass Kprrrr Mass Kprr Cascades

Control 1 2.35-2.40 2.14- 2.18 1 Control 2 2.40-2.45 2.19-2.23 4 Peak 2.45-2.50 2.24,-2.28 9 Control 3 2.50-2.55 2.29-2.33 0 Control 4 2.55-2.60 2.34-2.38 5 The situation so far seems to conform very well to the naive diquark arguments + ++ presented earlier. We see a possible /\c signal with a positive trigger and a l:c candidate when using a negative trigger. The mass plots for the opposite sign show no enhancement, although one does see a bump near the /\ mass in a negative trigger +,� plot but only if an extra rr+ is present. This would imply that we see a decay pro- ++ duct of a l:c which was formed from a uu diquark. A se·rious problem encountered is the two values of /\+ mass needed to explain t�e data; the "direct /\+" has a mass c ++ + c that is incompatible with the l:c cascade to the /\c . is hard to build a strong ++ + It argument for the l:c and /\c simultaneously. A further problem is the large production rate. After correcting for accep­ tance, efficiency, and branching ratios, we have a suspiciously large number of charm candidates given our limited number of triggers. This could mean that these bumps are merely fluctuations. Work is in progress to understand these pra':>lems, so the results of the charm search should be regarded as preliminary.

I would like to thank all my friends in the ACHMNR Collaboration, as well as Professor Tran Thanh Van and the rest of the conference organizers. I am grateful to R. Field for his helpful calculations. 199

REFERENCES AND FOOTNOTES

1) L. Sehgal , Proc. Int. Symp . on and Interactions at High Energies, Hamburg, 1977 (DESY, Hamburg, 1977), p. 837. 2) For a summary of the experimental information, see M. Fontannaz , B. Pire and D. Schiff; Phys . Lett. (1978) 315 . !.B!_ 3) For our purposes , a diquark is everything remaining after a valence quark has been removed from a baryon. As in the case of hadrons, we expect its gross features to be determined by the valence quarks it contains . The term frag­ mentation is used broadly here to indicate all processes which effect the transformation of a diquark into hadrons. 4) R.P. Feynman, R.D. Field and G.C. Fox, Phys . Rev. D (1978) 3320. � 5) R. Field, private communication. The weak point in the calculation seems to be the amount and distribution of glue in the proton. The errors in the table are designed to show the effects of different gluon parametrizations . 6) K.L. Giboni et al ., Phys . Lett. 85B (1979) 437. L. Baksay et al., Nucl. Instrum.�thods 133 (1976) 219. 7) F. Ceradini et al ., Nucl. Instrum. Methods 156 (1978) 171. 8) CCHK Collaboration: M. Della Negra et al ., Nucl . Phys . Bl27 (1977) 1. D. Drijard et al., Nucl . Phys . Bl56 (1979) 309. 9) M.G. Albrow et al., Nucl. Phys . Bl35 (1978) 461. 10) R.P. Feynman, Photon-hadron interactions (Benjamin, Inc., Reading, Mass., 1972) . R. Blankenbecler and S. Brodsky, Phys . Rev. D 10 (1974) 2973. S. Brodsky and Gunion, Phys . Rev. D (1978) 848. J. _!2 11) The concept of the diquark as a bound sub-state of the nucleon has been used by many authors for a wide variety of applications . For a review, see R.T. Van de Walle, Diquarks, Lectures presented at the "Ettore Majorana" Int . School of Subnuclear Physics, Erice, 1979. 17th Course: Point-like structures inside and outside hadrons. 12) J. Dorfan, SLAC-PUB 2429 (Mark II Collaboration) . J. Peoples, talk given at the 15th Rencontre de Moriond, Les Arcs , France, 1980 . For a review of the other charmed baryon candidates and their masses, see F. Muller, Recent observations of charmed baryons and their implication on hadronic processes (preprint CERN-EP/79-48) , Talk given at the Int. School of Phys ics, Cargese, 1979 (proceedings in preparation) . 13) C. Baltay et al ., Phys . Rev . Lett. � (1978) 73. 14) A. de Ruj ula et al ., Phys . Rev. D 12 (1975) 147.