Exotic Heavy Flavor Hadrons Stony Brook seminar P

Exotic Heavy Flavor Hadrons Stony Brook seminar P. Grannis Apr. 17, 2017

The quark constituents of hadrons In the original quark model paper, Gell Mann said “… baryons can _now be constructed from combinations _ (qqq), _ _ (qqqqq) etc. while mesons are made out of (qq), (qqqq), etc. …” Zweig made a similar comment regarding his ‘aces’. Until rather recently no direct evidence for such exotic states existed, although in_some cases there are more candidate states than can fit into the available qq multiplets, implying the need for more complex structures. Also the masses and decay patterns of some states could be hard to accommodate in the ordinary quark model. The spectroscopy of mesons containing heavy quarks (c and b) can provide clearer and more direct evidence for exotic mesons (or baryons) due to their distinctive decays. 39

The quark constituents of hadrons Luciano Maiani, in the 2013 Erice school, talked on exotic mesons and recalled the quote from the 1967 Erice School: “… then the data and masses and quantum numbers were examined in detail in order to determine “what is certain, what remains to be confirmed and what should probably be forgotten. ” L to R: Bruno Zumino, Sidney Coleman, Nino Zichichi, Nicolo Cabibbo, Sheldon Glashow, Murray Gell-Mann) In 1967 these words pertained to ordinary mesons, but are valid for exotic mesons today. Many candidates for exotic states exist, but experimental observations are often contradictory and theoretical explanations are controversial. Hadron taxonomy in non-perturbative QCD is difficult! 38

Types of multi-quark states Multiquark exotic mesons may have more complex structures than ordinary mesons, due both to the color combinations and to the binding mechanism. _ u c p _ u c c u gg _ _ u c c _ A pair of colorless qq states bound loosely with a colorless pion in van der Waals force (analogous to the deuteron). Expect such a ‘MOLECULAR STATE’ to have a mass __quite close to the sum of the two colorless hadrons (here D 0 and D 0 ) A colored diquark and a colored anti-diquark pair, bound with a multicolor the diquark _ gluonic state. The color force binding _ (color 3 or 6) and anti-diquark (color 3 or 6) is quite strong so __ D 0 this ‘TETRAQUARK’ state mass can be quite far below the D 0 threshold. A ‘HYBRID’ state of colored quark, antiquark and gluon in an overall color singlet …. or a ‘GLUEBALL’ with no quarks in it And of course more complex structures can be envisioned with Fock space wavefunctions built of superpositions of two or more structures 37

Light quark exotic states The nonet of JPC = 0++ mesons displays a peculiar mass spectrum that is inverted from the pattern for the familiar 1 -- or 0 -+ nonets: mass - - uu±dd _ __ _ _ss- (f) _ _us- (K*) _ __ _ud- (r, w) JPC=0++ _ JPC=1 -- a 0(980) f 0(980) k(800) f 0(500) This could be explained (Jaffe) if the 0++ states were tetraquarks with a color _ 3 diquark bound to a color 3 anti-diquark, giving quark content as shown _ __ _ Maiani, et al. assigned a 0 and f 0 as tetraquarks and predicted analogs with c and -- _ With this content, the mesons with higher mass are those containing the more massive strange quark. (su)(ds) -etc. (su)(us) -etc. (sd)(ud) -(ud) 36

First heavy quark exotic state The first meson to be widely recognized as exotic was the X(3872) seen by BELLE in 2003, originally in B → K X 0 → K (p+p- J/y). It has been seen in both e+e- and hadron colliders by BELLE, Ba. Bar, CDF, D 0, LHCb, CMS, BESIII and ATLAS. Decays to w J/y, g _ J/y, and D 0 D*0 have been observed. BELL E X(3872) mass _ is essentially equal to the sum of the D 0 and D*0 masses. Width is consistent with resolution, so GX < 1. 2 Me. V LHCb measured JPC as 1++ Although the e+e- colliders first saw X as a decay product of B’s, CDF and D 0 have shown that prompt production of X(3872) is more copious (84%). 35

Interpreting X(3872) _ Only unassigned cc charmonium JP= 1++ state was the first radial excitation of cc 1 but the mass splitting from known cc 2’ is wrong. The width of X is_ too small to be charmonium; also the r J/y decay violates isospin _ _for a cc state. So left with exotic 4 quark interpretations with quark content ccqq (q=u or d). X(3872) as a molecular state? The very small binding energy suggests this, but it is hard to see how such a weakly bound object could be directly produced copiously in violent high p. T hadron collisions. X as a tetraquark? Where are the missing partners (would expect both --neutrals cucu and cdcd as well as charged partners such as-cucd) that are not seen. _ _ Hybrid cc /DD* molecule? ? May explain the various puzzles by invoking a state with a more complex Fock space components – e. g. superposition of molecular state and ordinary quark antiquark configurations. The X(3872) puzzles typify the difficulties in deducing the structure of the exotic meson candidates. 34

XYZ states Many exotic XYZ mesons have been sighted. Often, some experiments see it, others do not, so details of production, backgrounds, interferences etc. seem to be important. So m e of th e ca nd ida te XY Z’ s (The nomenclature itself is murky!) More or less: Y= neutral 1 -- states seen in e+e- s-channel production Z= charged states X= all the others (PDG uses X(mass)) 33

Another example: Y(4140) CDF In 2009, CDF announced a J/y f resonance, Y(4140), in B+ → Y K+, close _ _ to the J/y f threshold. The quark content is thus ccss and the state is manifestly exotic, likely a molecular state. CDF also had evidence for a higher state at 4274 Me. V. Shortly thereafter, LHCb, Belle, Ba. Bar and BESIII failed to detect this state. However CMS and D 0 did confirm it at the 3 and 5 s level and saw the higher mass state. D 0 Subsequently, D 0 presented strong evidence that Y(4140) is also produced inclusively (not only through the cleaner B decay channel), so interpretation as a reflection not likely. Very recently LHCb, with a larger data set and a sophisticated amplitude analysis confirmed Y(4140) with JPC=1++, Y(4274) as well as two higher states. The pattern of non-confirmation, then confirmation, ambiguous interpretation, dependence on production modes etc. is typical. 32

D 0 Bs p± D 0 was a multipurpose _ detector at the Tevatron pp collider (ran 1992 – 2011) + D 0 did a search _ _ _ for _ a resonance in Bs p p (quark content for an analog of known states _ _ bsuu/bsdd) of csuu/csdd. Nothing was seen but a side look at the Bs p± mass distribution (Bs → J/y f, J/y → m+ m - f → K+ K- ) showed a peak slightly above threshold of 5506 Me. V. Due to rapid mixing, _ one cannot distinguish between Bs and Bs. The analysis does not separate p+ and p-. The quark content of a state in this channel can be: _ _ __ _ _ bsud, bsud or bsud, Such 4 -flavor (charged) mesons have not been seen before, and are manifestly exotic. 31

D 0 Bs p± hadronic channel Phys. Rev. Lett. 117, 022003 (2016), ‘hadronic’ refers to the decay Bs → J/y f as opposed to the semileptonic decay Bs → Ds+m- X to be discussed later. Dotted lines indicate particles with lifetimes too short to detect. 30

Bs. D 0 p± hadronic: B s p± Selections J/y: p. Tm > 1. 5 Ge. V; M(mm) = (2. 92, 3. 25) Ge. V; selected dimuons are constrained to the PDG J/y mass. f: p. TK > 0. 7 Ge. V: MKK = (1. 012, 1. 030) Ge. V Bs candidate: M(m+m- K+K-) = (5. 304, 5. 425) Ge. V; Lxy/s(Lxy) > 3 Additional p±: add a track assumed to be a p with p. Tp>0. 5 Ge. V; IPxy<0. 02 cm, IP 3 D<0. 12 cm p. T(Bsp) > 10 Ge. V and an optional cut on the angular separation of Bs and p± DR = √Dh 2 + Df 2 < 0. 3 (the “cone cut”) The Bs p mass resolution is improved by using M(Bs p) = Mobs(m+m- K+K- p) – Mobs(m+m- K+K -) + M PDG(Bs) Bs p candidates are restricted to 5506(threshold) < M < 5900 Ge. V 29

Bs p± hadronic: Backgrounds Identify two classes of background in J/y f distribution: a) Real Bs modelled by MC Bs production (Pythia) b) ‘False’ Bs from combinatorial mm KK. Model these by sidebands >5 s from Bs peak, with center of gravity at M(Bs) c) Obtain Bs p background from combination of Bs (or J/y f sideband candidate) and random p± in same event ‘real’ Bs ‘false’ Bs sideband s Combine real and false backgrounds in appropriate proportion (71%/29%). However the real Bs (data points) and fake Bs (histogram) background distributions are nearly the same. 28

Bs p± hadronic: Background and signal model Fit the background to a product of polynomial and exponential: FBKD = (C 1 + C 2 m 2 + C 3 m 3 +C 4 m 4)*exp(C 5 + C 6 m + C 7 m 2) Fit is good both with and without the cone cut. The DR < 0. 3 cone cut suppresses the events at high mass: The signal X is assumed to be a relativistic s-wave Breit-Wigner with a mass-dependent width G(m) = GX (qm/q 0) where qm is the rest frame momentum of Bs at M(Bs p)=m and q 0 is the momentum at the central MX value. The BW is convoluted with the Gaussian resolution, s=3. 8 Me. V Fit data (or MC pseudo-expts) to NX • BW(MX, GX) + (Ntot-NX) • FBKD 27

Bs p± hadronic: Fit result MX = 5567. 8± 2. 9 Me. V With DR < 0. 3 cone cut, fit for MX, GX and NX, taking mass resolution s = 3. 8 Me. V, obtain: Non-zero width → strong decay GX = 21. 9 ± 6. 4 Me. V NX = 133 ± 31 events Probability for null hypothesis to reproduce the yield gives local significance of 6. 6 s. significance = √(-2 Lb/Ls+b) Using prescription of Gross and Vitells for Look Elsewhere Effect (LEE) get 6. 1 s global (statistical ) significance. If the X decay is s-wave (it is close to Bsp threshold of 5506 Me. V), JP = 0+ and X is counterpart to a 0(980). If this bump were really X→ Bs*p± with Bs* → Bs + unseen g (48 Me. V) , would have MX =5616. 5 Me. V. If this decay is s-wave, JP = 1+. 26

Bs p± hadronic: Fit result 10< p. T(Bs)<15 Ge. V 15< p. T(Bs)<30 Ge. V Fits in two p. T(Bs) intervals returns similar X parameters, whereas the background shape changes Fraction of Bs that come from X decay is r=8. 6± 1. 9 ± 1. 4%. (This seems like a large fraction!) 25

Bs p± hadronic: Cross checks To test whether the X is really due to candidates with genuine Bs, we fit for the Bs signal yield in a series of bins of J/yf p mass. The resulting Bs yield as a function of M(J/y f p) gives a peak and width compatible with the primary analysis and NX = 118 ± 22 events. (In this analysis there is no non-Bs background. ) 5568 Me. V known B 1 Other checks: v Look at Bd p± to see if a similar peak is generated: (none seen) v Look for peaks in Bsp or Bs. K (none seen) v Alternate background functions & binning, v Compare p+ and p- (no difference) v Two versions of Pythia Bs background 24

Bs p± hadronic: Systematic uncertainties With the systematic uncertainties (and LEE), the significance is 5. 1 s. _ _ With the strong decay and the bs content of Bs and ud content of p+, the X flavor content is busd – the first four-quark state built from 4 distinct flavors. 23

Bs p± hadronic: No cone cut Remove the cone cut, keep the functional form of the background (different parameters) and refit: Fix MX and GX to values for fit with cone cut and obtain NX = 106 ± 23 events, with corresponding local significance of 4. 8 s. The difference from the 133 events with cone cut is not well explained as a statistical fluctuation. The lower yield without the cone cut led PRL editor to insist on “Evidence” in the title, rather than “Observation”, despite the 5. 1 s global significance with the cone cut. We note that the yield is sensitive to the choice of background function in the high mass region, 5. 7<M<5. 9 Ge. V and alternate choices give higher yields. We also note that there may be states such as Bc → Bs np (n>1) that are not in our MC. The inclusion of added high mass backgrounds would cause the yield in the no-cone-cut case to increase. 22

Bs p± hadronic: vary cone cut DR<0. 2 Can the cut create a peak? DR<0. 3 DR<0. 5 DR cut efficiency Bs p mass Fit data with a series of DR cuts from 0. 2 to 0. 5 (and no cut). Fitted mass is stable within statistical uncertainty. We see no evidence of the cut sculpting the peak. 21

Bs p± hadronic: Other experiments LHCb (pp collisions at 7 & 8 Te. V [PRL 117, 152003 (2016)] sees no evidence for X(5568). Fraction of Bs that come from X is less than 2. 4% (95% CL), compared with D 0 (8. 6%). CMS (pp at 8 Te. V) set limit at fraction of Bs from X less than 3. 9% (95% CL) (ICHEP 2017, CMS-PAS BPH-16 -002) CDF and ATLAS have yet to report results. So, is the X(5568) dead? Or are there differences due to production channel, energy? ? 20

Bs p± semileptonic channel In preliminary analysis for Moriond QCD 2017, D 0 undertook to search for X(5568) in a complementary decay channel: X(5568) → Bs p± with Bs → Ds- m+ n (+X) D s - → f p - → K + K - p. X = possible other particles (& charge conjugates) from Bs Selections: Ø K’s: p. T>1. 0 Ge. V, opposite charge, MKK = (1. 012, 1. 030) Ge. V Ø m: p. T = (3, 25) Ge. V; opposite charge to p from Ds; good vertex with Ds Ø Ds and Bs decay vertices well separated from primary vertex Ø p in Bs p decay: p. T=(0. 5, 25) Ge. V (same selection as in hadronic channel) Ø p. T(m. Ds) > 10 Ge. V Ø 4. 5 Ge. V<M(m. Ds)<M(Bs) to minimize effect of final state n Ø No cone cut is imposed https: //www-d 0. fnal. gov/Run 2 Physics/WWW/results/prelim/B/B 68/ 19

Bs p± semileptonic data D 0 MC M(m. Ds)>4 Ge. V M(m. Ds)<4 Ge. V Restricting M(m. Ds) to be large (En low) improves mass resolution for Bs p To further improve mass resolution, define: M(Bsp)=m(m. Dsp)-m(m. Ds)+m. PDG(Bs) M(Bsp) D 0 preliminary Data Bs p mass distribution M(Bs p) After correcting for efficiency/acceptances, expect ~20% more semileptonic X(5568) than hadronic. 18

Bs p± semileptonic – background We generate Pythia Monte Carlo to simulate background, modified to use EVTGEN for b/c hadron decays. Data events with no interactions are superimposed to simulate the effects of pileup. MC events are reweighted in p. T(Bs) and p. T(m) to agree with data kinematic distributions. Data MC no RW MC w/ RW 17

Bs p± semileptonic – background modeling Fit reweighted MC background with baseline function: Fbkg = (C 1 D+C 2 D 2+C 3 D 3+C 4 D 4) exp(C 5 D+C 6 D 2) where D =m(Bs p) – mthreshold Three alternate fits of MC background: 1) Fbkg = (C 1+C 2 D 2+C 3 D 3+C 4 D 4) exp(C 5+C 6 D +C 7 D 2) (the functional form used for the hadronic channel analysis) 2) Fbkg = m[ (m/mthreshold)2 -1 ]C 1 exp(C 2 m) (the ‘ARGUS’ function for resonances near threshold 3) The MC histogram itself, smoothed by ROOT. 16

Bs p± semileptonic – background cross check Also consider two alternate background shape choices to the MC: (a) Data sidebands in m(fp±) distribution (b) Data same (wrong) sign Ds± m± events sideband SS MC background shape reproduces the sideband wrong sign shapes well. points: Ds sideband data histogram: Monte Carlo points: wrong sign data histogram: Monte Carlo 15

Bs p± semileptonic – signal modeling The signal is modeled with the same relativistic Breit Wigner s-wave shape as used for the hadronic channel, convoluted with a Gaussian mass resolution whose (mass-dependent) width is taken from MC. This resolution is dominated by the missing n. D 0 MC Resolution + n smear n only 3. 8 Me. V mass resolution due to track curvature resolution is convoluted with the smearing due to missing n. MC of X(5568) signal including resolution effects and the effects of the unseen neutrino shows a narrow peak with a broad flat tail due to an association of a Bs with an incorrect p± 14

Bs p± semileptonic – fit Data are fit to the sum of background and signal distributions with the background parameters fixed to the values from fitting the MC distribution. The signal mass MX and width GX, and signal and background normalizations are obtained from the fit. We test the closure of the fit and background model by injecting signals of various sizes in MC ensembles with M=5568, G=21. 9 Me. V and fitting for MX, GX and NX. D 0 MC Output NX and MX agree with inputs; Fits tend to have small GX for small Ninput. 13

Bs p± semileptonic – systematics Vary energy scale and resolution by 1 s; fit signal to p-wave BW; vary resolution smearing due to n. Dominant uncertainty due to background shape, estimated by comparing baseline and alternate background models. 12

Bs p± semileptonic – results Fit results: MX = 5566. 7 GX = 6. 0 NX = 139 +3. 6 +1. 0 Me. V -3. 4 -1. 0 +9. 5 +1. 9 -6. 0 -4. 6 Me. V +51 +10. 9 -63 -31. 5 Statistical significance 4. 5 s; significance with syst: 3. 2 s We have evidence for X(5568) in the semileptonic channel, with consistent MX and GX as hadronic channel, Fraction of Bs from X(5568) (p. T(m. Ds)>10 Ge. V) is 7. 6± 2. 9% in SL channel. Compare this with the hadronic fraction of 8. 6± 2. 4% for p T(J/y f)>10 Ge. V. Finding a resonance in final state with n is novel, enabled by the cut 4. 5 Ge. V<M(m. Ds)<M(Bs) and the fact that X(5568) is close to Bsp threshold 11

Comparison of Bs p± hadronic and semileptonic channels Semileptonic channel The hadronic and semileptonic channels agree on mass and width of X(5568), so it makes sense to combine the two. Hadronic channel, no ΔR cut 5568 10

Combination of Bs p± hadronic and semileptonic channels Since the semileptonic and hadronic fit parameters are consistent, we do not need the look elsewhere effect (LEE) for the semileptonic channel. Can combine hadronic channel with/without cone cut Inputs to combination and or 9

Combination of Bs p± hadronic and semileptonic channels Because of the n, different triggers, and different backgrounds the systematic uncertainties for hadronic and semileptonic channels are nearly independent. For the combination of two uncorrelated measurements with p-values p 1 and p 2, the combined p-value is: pcomb = p 1 p 2 (1 – ln(p 1 p 2)) For phad = 3. 8 x 10 -7 (with cone cut) and p. SL = 6. 4 x 10 -4, we obtain pcomb = 5. 6 x 10 -9 corresponding to 5. 7 s significance Combining the hadronic result without cone cut and the semileptonic result, we get a significance of 4. 7 s. 8

Bs p± summary D 0 sees a strongly decaying (G>0) meson in two quite different channels with different backgrounds etc. The combined significance is 5. 7 s (4. 7 s when using the hadronic channel without cone cut. ) The X(5568) would be the first meson that includes four quark flavors. The X(5568) lies 200 Me. V below the Bd K± threshold making a pure Bd K± molecular state unlikely. If tetraquark, there should be companion exotic mesons. No other experiment has yet observed X(5568), but LHC and Tevatron production processes might differ. We await the Tevatron analysis from CDF. 7

Coda on Pentaquarks _ _ Just as exotic mesons _ can be constructed from qqqq, we can have exotic baryons such as qqqqq (pentaquarks). In 2015, the LHCb collaboration reported two structures, Pc(4450)+ and Pc(4380)+ in the decay Lb→ Pc+ K- with Pc+ →J/y p The higher mass state is narrow (G=39 Me. V) and the lower one broad (G=205 Me. V). Detailed analysis reveals the Argand diagram phase variation appropriate for resonances, and JP=3/2 - and 5/2+. R. Aaij et al. , LHCb Collaboration, PRL 115, 072001 (2015) These Pc states are within 400 Me. V of the J/y p threshold. The minimal quark content is uudcc, manifestly an exotic pentaquark. _ 6

D 0 Pentaquark search Motivated by the LHCb result, D 0 searched for strange analogs of the LHCb states: e. g. Lb→ Pc+ f with Pc+→J/y L , and J/y →m+m-, L→pp(and charge conjugates). https: //www-d 0. fnal. gov/Run 2 Physics/WWW/results/prelim/B/B 69/ Such Pc states could come from decays of Lb 0 (as in Figure), Xb-, etc. Take search window up to 500 Me. V above J/y L threshold, based on the LHCb states and theoretical models. Event selection is based on that for the D 0 Lb lifetime measurement; extended tracking version to increase acceptance for low momentum p from L decay. 5

D 0 Pentaquark search Divide J/y L sample into ‘prompt’ and ‘non-prompt’ subsets based on J/y decay length significance. See clear Lb signal in non-prompt sample, so focus on this subset. Lb ? 500 Me. V search region 4

D 0 Pentaquark search Perform a series of fits of the form where Fsig is a Gaussian of width s. X and MX is incremented in 10 Me. V steps from threshold to 4. 7 Ge. V, and Fbgr is the background function Pentaquark significance from scan max local Signif. = 3. 58 at MX =4. 32 Ge. V 3

D 0 Pentaquark search Fit result at the step with maximum significance. MX = 4319. 3± 6. 6 Me. V s. X = 15. 6 ± 7. 7 Me. V NX = 142 ± 66 (mass resolution here is 5 Me. V) Local significance = 3. 45 s Perform 6400 pseudoexperiments to determine the probability of a 3. 45 s or higher fluctuation anywhere in the search range. Global significance = 2. 8 s. Use of alternate signal model, alternate background function, or J/y or L sidebands increase the observed yield, so systematic uncertainties do not diminish the significance. Thus no evidence for strange analog of LHCb pentaquark in J/y L, but at least a pointer on where one might look. 2

Summary Exotic mesons and baryons are now definitively observed but their construction is still not well understood D 0 has added a new exotic meson, X(5568) to the zoo with sightings in two channels. LHC experiments do not observe it. The recent pentaquark baryons Pc seen by LHCb should have close family relations. The D 0 search was negative, but suggests a place to focus attention.