Arxiv:2005.05892V3 [Hep-Ex] 9 Aug 2020 2.2.3 Charmonium and Bot- Tomonium Productions in 2 Υ(1S, 2S, 3S) Studies at E+E− Colliders4 Υ(1S, 2S, 3S) Decays

Experimental review of the Υ(1S, 2S, 3S) physics at e+e− colliders and the LHC

Sen Jiaa, Xingyu Zhoub, Chengping Shena,∗

aKey Laboratory of Nuclear Physics and Ion-beam Application (MOE) and Institute of Modern Physics, Fudan University, Shanghai 200443, China bSchool of Physics and Information Engineering, Shanxi Normal University, Linfen 041004, China

Abstract

The three lowest-lying Υ states, i.e. Υ(1S), Υ(2S), and Υ(3S), composed of b¯b pairs and below the BB¯ threshold, provide a good platform for the researches of hadronic physics and physics beyond the Standard Model. They can be produced directly in e+e− colliding experiments, such as CLEO, Babar, and Belle, with low continuum backgrounds. In these experiments, many measurements of the exclusive Υ(1S) and Υ(2S) decays into light hadrons, which shed light on the “80% rule” for the Okubo-Zweig-Iizuka suppressed decays in the bottomonium sector, were carried out. Meanwhile, many studies of the charmonium and bottomonium productions in Υ(1S, 2S, 3S) decays were performed, to distinguish diﬀerent Quantum Chromodynamics (QCD) models. Besides, exotic states and new physics were also extensively explored in Υ(1S, 2S, 3S) decays at CLEO, BaBar, and Belle. The Υ(1S, 2S, 3S) states can also be produced in pp collisions and in collisions involving heavy ions. The precision measurements of their cross sections and polarizations at the large hadron collider (LHC), especially in the CMS, ATLAS, and LHCb experiments, help to understand Υ production mechanisms in pp collisions. The observation of the sequential Υ suppression in heavy ion collisions at CMS, LHCb, and ALICE is of great importance for verifying the quark-gluon plasma predicted by QCD. In this article, we review the experimental results on Υ(1S, 2S, 3S) at e+e− colliders and the LHC, and summarize their prospects at Belle II and the LHC. Keywords: Υ(1S, 2S, 3S), hadronic decay, radiative decay, exotic states, new physics, cross section, polarization, quark-gluon plasma

Outline 2.2.1 Study of gluon fragmentation in Υ(1S, 2S, 3S) → ggg and Υ(1S, 2S, 3S) → γgg .....4 Contents 2.2.2 Exclusive Υ(1S, 2S) decays into light hadrons ...... 4 1 Introduction3 arXiv:2005.05892v3 [hep-ex] 9 Aug 2020 2.2.3 Charmonium and bottomonium productions in 2 Υ(1S, 2S, 3S) studies at e+e− colliders4 Υ(1S, 2S, 3S) decays . . . . .7 2.1 The Υ(1S, 2S, 3S) datasets at e+e− colliders ...... 4 2.2.4 Search for exotic states in Υ(1S, 2S, 3S) decays . . . . . 14 2.2 Hadronic and radiative decays of Υ(1S, 2S, 3S)...... 4 2.2.5 Search for new physics in Υ(1S, 2S, 3S) decays . . . . . 17 ∗Corresponding author Email address: [email protected] (Chengping 2.3 Transitions between Υ(mS) and Shen) Υ(nS)(m > n)...... 20

Preprint submitted to Frontier of Physics August 11, 2020 3 Υ(1S, 2S, 3S) measurements at the LHC 21 3.1 The pp, pP b, and P bP b datasets at the LHC ...... 21 3.2 Υ(1S, 2S, 3S) productions in pp collisions ...... 23 3.2.1 Υ(1S, 2S, 3S) cross sections . 23 3.2.2 Υ(1S, 2S, 3S) polarizations . 26 3.3 Sequential Υ suppression in heavy ion collisions ...... 28 3.3.1 Υ suppression in P bP b collisions ...... 28 3.3.2 Υ suppression in pP b collisions 33

4 Summary and prospects 35 4.1 Summary at e+e− colliders and prospects at Belle II ...... 35 4.2 Summary in pp, pP b, and P bP b collisions and outlook at the LHC . . . 37

2 1. Introduction etc. The bottomonium spectroscopy has also been greatly enriched in the past decades. The first Υ resonance, namely Υ(1S), was dis- In the hadronic or radiative decays of covered in 1977 in the bombardment of a beam Υ(1S, 2S, 3S), the searches for exotic states of high energy protons to a stationary nuclear tar- + − + − and new physics (NP) are very promising at get [1,2]. Soon, in the e e collisions to the µ µ B-factories and CLEO. The ample gluons in final state, Υ(1S) was confirmed, and Υ(2S) and the hadronic decays of Υ(1S, 2S, 3S) provide an Υ(3S) were also observed [3,4,5]. The discovery entry to many potential exotic states, including of Υ(1S, 2S, 3S) opens a door to the study of the the glueballs, charmoniumlike states (so-called bottomonium spectrum. In the following decades XYZ states), and stable six-quark states. The many different experiments, at lepton, hadron, and Υ(1S, 2S, 3S) decays also provide a good platform ion colliders, reported the fruitful results on these to explore the dark sector physics. In the Standard Υ states. Model (SM), invisible decays of Υ(1S) involving Because b quark is almost 3 times heavier than c neutrinos in the final states are produced with quark, the b¯b system is more non-relativistic than B[Υ(1S) → νν¯] ∼ 10−5 [21]. Low mass dark matter the cc¯ system. Since non-relativistic systems are (DM) particle (χ), if it exists, should enhance the easier to describe theoretically, the Υ system as invisible decays of Υ(1S)[22]. The χ can also be a representative of bottomonium family plays an produced in interactions of SM particles through important role in the studies of strong interac- the exchange of a CP -odd Higgs boson A0, which tions. Although the tt¯ system is completely non- is part of the Next-to-Minimal Super-symmetric relativistic, experimental studies of strong interac- Model (NMSSM) [23]. Such A0 states in the final tion phenomena with tt¯ are very difficult because states of χχ (invisible), µ+µ−, τ +τ −, and hadrons the t quark decays by weak interactions. have been searched for in the radiative decays of Unlike the Υ states above the BB¯ threshold, Υ(1S, 2S, 3S) at Belle, BaBar, and CLEO. the intrinsic widths of Υ(1S, 2S, 3S) are quite nar- Historically, the Υ(1S, 2S, 3S) productions in pp row [6], merely 20 ∼ 50 keV. This can be explained collisions were poorly understood. Several effective by the Okubo-Zweig-Iizuka (OZI) rules [7,8,9, 10]. QCD models, such as the color-singlet model [24, For Υ(4S) and other Υ states with higher masses, 25], the color-octet mechanism [26], and the color- they are allowed to decay via a typical strong inter- evaporation model [27], predict different cross sec- action into a pair of heavy quark mesons: (b¯b) → tions and polarizations. A precise measurement (bq¯)(q¯b)(q = u, d, c), e.g., Υ(4S) → BB¯, which of Υ(1S, 2S, 3S) production cross sections is thus results in a large natural width. Below the BB¯ crucial for distinguishing these models. Thanks to threshold, Υ(1S, 2S, 3S) can not decay via this the good momentum resolutions of the detectors, OZI-favored way, thus leading to a narrow natu- the experiments at the LHC, especially the CMS, ral width. However, they can still decay by strong ATLAS, and LHCb experiments, have the ability interactions with the b¯b pair annihilation via three to distinguish the three Υ(1S, 2S, 3S) structures gluons into hadrons. in the dimuon decay channel. The Υ(1S, 2S, 3S) In the process of e+e− → γ∗ → Υ(1S, 2S, 3S), cross sections in pp collisions have been provided the entire center-of-mass (C.M.) energy of the ini- by CMS, ATLAS, and LHCb at different C.M. ential e+e− turns into the rest mass of the Υ state. ergies, within the complementary rapidity and mo- The C.M. energy was matched to the resonance mentum coverages. The Υ(1S, 2S, 3S) polarizations mass; thus only one Υ resonance was produced at in pp collisions have also been explored at the LHC. a time. Accordingly, a particularly clean environ- QCD predicts that strongly interacting matter ment for studies of the properties of the Υ states at a critical temperature undergoes a phase tran- is provided by e+e− colliders at B-factories (Belle sition to a deconfined state, often referred to as and BaBar) and CLEO. The hadronic and radia- quark-gluon plasma (QGP), in which quarks and tive decays of the Υ(1S, 2S, 3S) resonances have gluons are no longer bounded within hadrons in been extensively studied at B-factories and CLEO the medium. If QGP is formed in heavy-ion col- to test various theoretical predictions by potential lisions, it is expected to screen the confining poten- models [11, 12, 13, 14, 15, 16], effective field ap- tial of heavy quark-antiquark pairs, leading to the proaches [17, 18], lattice gauge calculations [19, 20], melting of charmonia and bottomonia [28]. The 3 dissociation of the quarkonium states depends on where qb = −1/3 is the electric charge of b-quark. the temperature of the medium, and is expected Therefore, one can estimate the strong coupling to occur sequentially, along the increasing values of constant αs according to the formula with Rγ from their binding energies [29]. For example, the loosely experimental measurements. bound states Υ(2S) and Υ(3S) are more likely to The exclusive productions of ggg and ggγ in the unbind than the tightly bound state Υ(1S). Up to Υ(1S, 2S, 3S) decays were studied by CLEO [32, now, the pattern of QGP in Υ(1S, 2S, 3S) states 33, 34], ARGUS [35], and Crystal Ball [36]. The has been established at CMS, LHCb, and ALICE recent results of Υ(1S, 2S, 3S) → ggγ were given by observing a sequential suppression of their yields by CLEO [34] in the measurements of the direct in P bP b and pP b collisions. photon momentum spectrum. The signal yield + − In this review, studies on Υ(1S, 2S, 3S) at e e of Υ(1S, 2S, 3S) → ggγ is determined in the xγ colliders and the LHC are summarized. With re- = pγ /Ebeam spectrum after excluding all possible + − gard to e e colliders, we discuss the hadronic and backgrounds, where pγ is the momentum of the iso- + − radiative decays of Υ(1S, 2S, 3S), especially those lated photon in the e e C.M. frame, and Ebeam is for exotic states and NP, and the transitions be- the beam energy. Two main backgrounds are pho- tween the Υ(1S, 2S, 3S) and other Υ states with tons that come from initial state radiation (ISR), higher masses. Several effective models of QCD which are the dominant background at the highest were compared with and challenged by these mea- photon energy (xγ > 0.65), while at lower ener- surements. The Υ(1S, 2S, 3S) cross sections and gies (xγ < 0.65) the dominant background comes polarizations in pp collisions were studied at the from photons resulting from π0 decays. The ISR LHC. The QGP formation predicted by QCD was background is well simulated by the jetset . [37] probed by comparing the Υ(1S, 2S, 3S) yields from event generator. A data-driven method is used to different types of collisions. estimate the background contribution from π0 decays. Continuum processes have been subtracted using an off-resonance data sample. By import- 2. Υ(1S, 2S, 3S) studies at e+e− colliders ing the branching fractions of Υ(1S, 2S, 3S) → ggg from PDG [6], the Rγ ratios are obtained to be + − 2.1. The Υ(1S, 2S, 3S) datasets at e e colliders Rγ (1S) = (2.70 ± 0.01 ± 0.13 ± 0.24)%, Rγ (2S) = (3.18 ± 0.04 ± 0.22 ± 0.41)%, and Rγ (3S) = (2.72 ± Table1 summaries the data samples collected 0.06 ± 0.32 ± 0.37)%, where the first, second, and at the Υ(1S, 2S, 3S) resonances in the CLEO, third uncertainties are statistical, systematic, and BaBar, and Belle experiments [30]. From this table, theoretical model dependent [38, 39, 40, 41], re- Belle owns the largest Υ(1S) and Υ(2S) samples, spectively. Hereinafter if there are two and more and Babar has the biggest Υ(3S) sample. These uncertainties in the formulae, the first one is sta- datasets provide a solid platform for investigating tistical and the second is systematic. The above the Υ(1S, 2S, 3S) decays. values of Rγ (1S), Rγ (2S), and Rγ (3S) imply the strong coupling constant αs = 0.1114 ± 0.0002 ± 2.2. Hadronic and radiative decays of Υ(1S, 2S, 3S) 0.0029 ± 0.0053, 0.1026 ± 0.0007 ± 0.0041 ± 0.0077, and 0.113 ± 0.001 ± 0.0007 ± 0.008 at the Υ(1S), 2.2.1. Study of gluon fragmentation in Υ(2S), and Υ(3S) resonances, respectively [34]. Υ(1S, 2S, 3S) → ggg and Υ(1S, 2S, 3S) → Considering the strong coupling constant αs can be written as a function of the QCD scale parameter γgg Λ , defined in the modified minimal subtraction ¯ MS Below the BB threshold, the Υ(1S, 2S, 3S) states scheme [6], we can determine the value of ΛMS fur- decay in the OZI-suppressed manner. To be spe- ther. cific, they can decay via three gluons (ggg) or two gluons plus a photon (ggγ). The ratio of these two decay rates was predicted by perturbative quantum 2.2.2. Exclusive Υ(1S, 2S) decays into light chromodynamics (pQCD) [31]: hadrons

0 Γggγ 38 2 αem The OZI-suppressed decays of J/ψ and ψ to Rγ = = qb [1 + (2.2 ± 0.8)αs/π], Γggg 5 αs hadrons proceed via the annihilation of the charm- 4 Table 1: Υ(1S, 2S, 3S) datasets at e+e− colliders [30]. Υ(1S) Υ(2S) Υ(3S) Experiment fb−1 106 fb−1 106 fb−1 106 CLEO 1.2 21 1.2 10 1.2 5 BaBar - - 14 99 30 122 Belle 6 102 25 158 3 12 anticharm pair into three gluons, or two gluons to- two-dimensional maximum likelihood fits to the in- gether with a photon. For both cases, pQCD pre- variant mass distributions for K∗(892)0 candidates ¯ ∗ 0 dicts [42, 43] and K2 (1430) candidates; they are 42.2 ± 9.5 and 32±11 with statistical significances of 5.4σ and 3.3σ Bψ0→hadrons Bψ0→e+e− for Υ(1S) and Υ(2S) decays, respectively. Qψ = = ≈ 12%, B B + − J/ψ→hadrons J/ψ→e e For three-body and four-body final states, a which is referred to as the “12% rule” and is ex- requirement on√ the energy conservation variable pected to apply with reasonable accuracy to both XT = ΣhEh/ s is imposed to extract the signals, inclusive and exclusive decays. However, the rule where Eh is the energy of the final-state particle h in + − √ was found to be severely violated for ρπ and other the e e C.M. frame, and s is the C.M. energy. + − Vector-Pseudoscalar (VP) and Vector-Tensor (VT) The XT distributions for Υ(1S, 2S) → φK K , ∗ 0 − + 0 + − + − 0 0 final states [44, 45]. This is the so-called “ρπ puz- K (892) K π , KSK π , and π π π π are 0 + − zle”. None of the many existing theoretical ex- shown in Figs.2 and3. For Υ(1 S, 2S) → KSK π + − 0 0 planations that have been proposed is able to ac- and π π π π , unbinned simultaneous maximum commodate all of the measurements reported to likelihood fits to the XT distributions are performed date [46, 47, 48]. A similar rule can be derived to extract the signal and background yields in the for OZI-suppressed bottomonium decays, in which Υ(1S, 2S) and continuum data samples, as shown case we expect by the open histograms in Fig.3. For Υ(1 S, 2S) → φK+K− and K∗(892)0K−π+, unbinned simulta- + − BΥ(2S)→hadrons BΥ(2S)→e+e− neous maximum likelihood fits to the K K and QΥ = = = 0.80±0.08. + − ∗ 0 BΥ(1S)→hadrons BΥ(1S)→e+e− K π invariant mass spectra for φ and K (892) candidates are applied to extract the signal and This rule should hold better than the “12% rule” for background yields after requiring events within the charmonium decays, since the bottomonium states XT signal range of [0.985, 1.015]. The statistical have higher mass, pQCD and the potential models signal significances for Υ(1S) (Υ(2S)) → φK+K−, ∗ 0 − + 0 + − + − 0 0 should be more applicable, as demonstrated in the K (892) K π , KSK π , and π π π π are calculations of the b¯b meson spectrum. 8.6σ (6.5σ), 11σ (6.4σ), 6.2σ (4.0σ), and 7.1σ To verify the “80% rule”, Belle measured (7.4σ), respectively. The branching fractions for the exclusive Υ(1S) and Υ(2S) decays into these decays are obtained for the first time. light hadrons, including two-body VT, VP and With 1.09 fb−1 Υ(1S) and 1.28 fb−1 Υ(2S) Axial-vector-Pseudoscalar (AP) final states, three- events of CLEO, the authors in Ref. [51] measured body final states, and four-body final states [49, the branching fractions for a number of exclusive 50]. Among these decay modes, the evidences decays into different final states consisting of 4 – 10 of both Υ(1S) and Υ(2S) are found in the light hadrons, pions, kaons, and protons [51]. Sig- ∗ 0 ¯ ∗ 0 + − ∗ 0 − + K (892) K2 (1430) , φK K , K (892) K π , nificance strength (significance > 2σ) is found in + − 0 0 0 + − π π π π , and KSK π final states, as demon- 17 decay modes for both Υ(1S) and Υ(2S), with strated in Figs.1,2, and3. Note that charge- branching fractions ranging from 1.3 × 10−5 to conjugate modes are implied throughout this re- 109.5 × 10−5. view. The detailed branching fractions and the ratios ∗ 0 ¯ ∗ 0 In the K (892) K2 (1430) mode, the numbers of QΥ corresponding to them [49, 50, 51] are listed in signal events are extracted by performing unbinned Table2, and displayed in Fig.4. From Fig.4, the 5 ∗ 0 ¯ ∗ 0 Figure 1: The invariant mass distributions for K (892) candidates and K2 (1430) candidates from Υ(1S) and Υ(2S) decays from Belle [49]. The open histograms show the results of the two-dimensional fits, the dotted curves show the total background estimates, and the grey histograms are the normalized continuum contributions, which are determined from the data at √ s = 10.52 GeV, and extrapolated down to the Υ(1S) and Υ(2S) data.

+ − ∗ 0 − + Figure 2: Scaled total energy, XT , distributions from Υ(1S) and Υ(2S) decays to φK K and K (892) K π from Belle [49]. The red dots with error bars are from resonance data, the yellow-shaded histograms are from the normalized continuum contributions described in the text, and the cyan-shaded histograms are from the normalized inclusive Υ(1S) and Υ(2S) MC events. The blue arrows show the required signal regions.

6 0 + − + − 0 0 Figure 3: The ﬁts to the scaled total energy XT distributions from Υ(1S) and Υ(2S) decays to KS K π and π π π π from Belle [50]. Solid dots with error bars are from resonance data. The open histograms show the best ﬁts, the dashed curves are the total background estimates, and the grey histograms are the normalized continuum background contributions.

QΥ values measured by Belle [49, 50] are close to CLEO analyzed Υ(1S) → J/ψ + anything using the “80% rule” line. However, most of the QΥ val- its Υ(1S) data sample [56]. To determine the dif- ues from the measurements using CLEO data [51] ferential branching fractions, MC and data samples are below the “80% rule” line. More precise mea- are partitioned according to the scaled momentum ∗ √1 2 ∗ surements are needed to understand the large dis- x = pJ/ψ/( 2 s × (s − mJ/ψ)), where pJ/ψ is the crepancy. momentum of the J/ψ candidate in the e+e− C.M. frame, and mJ/ψ is the J/ψ nominal mass [6]. The value of √1 × (s − m2 ) is the value of p∗ for 2.2.3. Charmonium and bottomonium productions 2 s J/ψ J/ψ in Υ(1S, 2S, 3S) decays the case where the J/ψ candidate recoils against a massless particle. The branching fraction for Υ(1S) → J/ψ + anything in the whole momentum • Υ(1S, 2S) → charmonium + anything region is measured to be (6.4 ± 0.4 ± 0.6) × 10−4, which is consistent with the predictions of both the Although the cc¯ systems have been studied for color-octet and color-singlet mechanisms [52, 53, 54, decades, their production mechanisms, especially 0 55]. The feed-down contributions of ψ , χc1 and in gluon-rich environments, have not yet been fully χc2 to Υ(1S) → J/ψ + anything are B(Υ(1S) → understood. Several theoretical papers suggested 0 0 ψ /χc1/χc2 + anything)B(ψ /χc1/χc2 → J/ψ + that J/ψ can be produced abundantly in the Υ(1S) anything)/B(Υ(1S) → J/ψ + anything) = (24 ± decay via the color-singlet [52] or color-octet mecha- 6±5)%/(11±3±2)%/(10±2±2)%, which are larger nism [53, 54, 55]. In both mechanisms, the branch- than the predictions in either the color-singlet [52] ing fraction of Υ(1S) → J/ψ + anything is pre- or color-octet model [53]. The differential cross −4 dicted to be about a few times 10 . However, sections in x for Υ(1S) → J/ψ + anything from the momentum distributions given by the two sce- CLEO is shown in Fig.5(a) [56]. With a larger narios are significantly different. In the color-octet Υ(1S) data sample, Belle reported B(Υ(1S) → mechanism, the J/ψ momentum is expected to be J/ψ + anything) = (5.25 ± 0.13 ± 0.25) × 10−4 accumulated near the kinematic end point. In con- and B(Υ(1S) → ψ0 + anything) = (1.23 ± 0.17 ± trast, the process with color-singlet mechanism in- 0.11) × 10−4, with substantially improved preci- herently results in a soft J/ψ momentum spectrum sions [57]. The differential branching fractions of because of the two additional charm quarks in the Υ(1S) inclusive decays into J/ψ and ψ0 are shown final states. 7 Table 2: The branching fractions for 22 exclusive light hadron decay modes of the Υ(1S) and Υ(2S) and the corresponding QΥ values from Belle [49, 50] and measurements using CLEO data [51]. Here, the uncertainties include statistical and systematic uncertainties. B(Υ(1S) → hadrons) B(Υ(2S) → hadrons) Measurements Mode Q (×10−5) (×10−5) Υ ∗ 0 ¯ ∗ 0 K (892) K2 (1430) 0.30 ± 0.08 0.15 ± 0.06 0.50 ± 0.24 K∗(892)0K−π+ 0.44 ± 0.08 0.23 ± 0.07 0.52 ± 0.19 from Belle [49, 50] φK+K− 0.24 ± 0.05 0.16 ± 0.04 0.67 ± 0.22 0 + − KSK π 0.16 ± 0.04 0.11 ± 0.03 0.71 ± 0.27 2π2π0 1.28 ± 0.30 1.30 ± 0.28 1.02 ± 0.32 4ππ0 6.10 ± 0.88 1.29 ± 0.56 0.21 ± 0.10 8ππ0 55.49 ± 9.16 12.45 ± 3.36 0.22 ± 0.07 2p2K4π2π0 22.58 ± 5.22 5.23 ± 2.68 0.23 ± 0.13 2K4ππ0 30.81 ± 4.97 7.42 ± 3.19 0.24 ± 0.11 2p6π2π0 49.55 ± 10.08 13.34 ± 5.14 0.27 ± 0.12 0 0 KSK5π2π 101.43 ± 19.63 28.26 ± 13.19 0.28 ± 0.14 2p4ππ0 13.54 ± 2.27 3.90 ± 1.76 0.29 ± 0.14 2K4π2π0 61.67 ± 11.20 18.80 ± 6.46 0.30 ± 0.12 using CLEO data [51] 2p4π2π0 22.68 ± 4.32 7.12 ± 3.34 0.31 ± 0.16 2p6ππ0 32.82 ± 5.91 10.58 ± 3.18 0.32 ± 0.11 2K6π2π0 109.53 ± 21.84 36.18 ± 12.36 0.33 ± 0.13 2p8π 7.69 ± 1.68 3.21 ± 1.17 0.42 ± 0.18 2p2K4ππ0 15.03 ± 3.07 6.29 ± 2.10 0.42 ± 0.16 2K6π 13.93 ± 2.52 5.92 ± 2.12 0.42 ± 0.17 0 0 KS3K3π2π 29.36 ± 6.78 15.68 ± 7.47 0.53 ± 0.28 2K2ππ0 5.43 ± 0.91 3.19 ± 1.15 0.59 ± 0.23 2p2K2ππ0 5.18 ± 1.04 3.54 ± 1.12 0.68 ± 0.26

8 Figure 4: The QΥ values for 22 exclusive light hadron decay modes of Υ(1S) and Υ(2S) from Belle [49, 50] (blue dots with error bars) and measurements using CLEO data samples [51] (black dots with error bars). The band shows the theoretical prediction and its uncertainty.

9 in Fig.5(b). From Fig.5, the scaled momentum is fully reconstructed, and the other charmonium spectra for Υ(1S) → J/ψ +anything from both ex- state X is searched for in the recoil mass distri- periments are relatively soft, peaking around x ∼ bution of the fully reconstructed J/ψ or ψ0. The 0.3, which favor the expectation of the color-singlet evidence for Υ(1S) → J/ψ + χc1 was found with a process [52], and are in sharp contrast to the predic- signal significance of 4.6σ, as indicated by the dots tion of the color-octet model [53, 54]. The study of with error bars in Fig.6. The measured branch- Υ(1S) → charmonium + anything therefore pro- ing fraction is (3.90±1.21±0.23)×10−6. For other vides a powerful platform for distinguishing the cases, considering the significances are less than 3σ, roles of color-singlet versus color-octet mechanisms. the 90% confidence level (C.L.) upper limits on the Besides these charmonium productions in Υ(1S) branching fractions are determined; they are mostly at the level of 10−6. The results are basically con- decays, Belle also studied the χc1 and χc2 productions in Υ(2S) decays, and corresponding branch- sistent with the theoretical calculations using the NRQCD factorization approach [70, 71]. ing fractions B(Υ(2S) → χc1 + anything) and B(Υ(2S) → χc2 + anything) were measured to be • Υ(1S, 2S) → γ + charmonium (2.24±0.44±0.20)×10−4 and (2.28±0.73±0.34)× −4 10 [58], respectively. In addition, it is worth men- The χcJ (J = 0, 1, 2) and ηc productions tioning that, similar to Υ(1S) → χc1 + anything, in Υ(1S, 2S) radiative decays were studied by the scaled momentum spectrum for Υ(2S) → χc1 + Belle [73, 74]. The χcJ states are reconstructed anything also peaks around x ∼ 0.3 [58]. via their E1 transitions to J/ψ. The ηc is reconstructed using the following five hadronic decay modes: K0 K+π−, π+π−K+K−, 2(K+K−), • Υ(1S) → double charmonia S 2(π+π−), and 3(π+π−). To suppress the ISR back- + − 0 The double charmonium production at B- ground e e → γISRψ → γISRγmissχcJ (→ γJ/ψ) factories is still a matter of debate. The cross sec- in Υ(1S, 2S) → γ + χcJ , the square of the miss- + − tions of the processes e e → J/ψηc, J/ψηc(2S), ing mass of the photon from χcJ and lepton pair 0 0 0 2 4 ψ ηc, ψ ηc(2S), J/ψχc0, and ψ χc0 measured by is required to be between −0.5 GeV /c and 0.5 2 4 Belle [59, 60] and BaBar [61] exceeded the non- GeV /c since this background has at least two relativistic quantum chromodynamics (NRQCD) missing photons (γISR and γmiss) and the miss- calculations by approximately an order of mag- ing mass tends to be large. Figure7 shows the nitude [62, 63, 64, 65, 66, 67]. Taking into ac- γJ/ψ invariant mass distributions and invariant count the next-to-leading order (NLO) correction to mass spectra for ηc candidates in the Υ(1S, 2S) NRQCD, the discrepancy between theory and ex- data samples. No significant χcJ or ηc signal is periment can be largely removed [68, 69]. Inspired observed in the analyses. The upper limits at 90% by the unexpectedly high double-charmonium pro- C.L. on the branching fractions for Υ(1S, 2S) → −6 duction in e+e− annihilation, interest has turned γχcJ are within a range of (3.6-100)×10 and for −6 to the double charmonium states produced in bot- Υ(1S, 2S) → γηc within (2.7-5.7)×10 , which are tomonium decays. Comprehensive studies of the consistent with the NRQCD predictions [75]. exclusive decay of Υ(1S, 2S, 3S) into double char- Very recently, Belle reported the first observa- monia have been performed using the NRQCD tion of the radiative decay of Υ(1S) into a char- factorization approach [70, 71]. The branching monium state, i.e., Υ(1S) → γχ [76]. Unlike fractions are predicted to be of order 10−6 for c1 0 the previous Belle analysis using the Υ(1S) data Υ(1S, 2S, 3S) → J/ψ(ψ ) + ηc(ηc(2S)) [70]. For sample [73], the authors use the Υ(2S) data sam- the J/ψ + χc0/χc1/χc2 decay modes, the branching ple and tag Υ(1S) via the Υ(2S) → Υ(1S)π+π− fractions are calculated at the lowest order; that of transition. Although the number of tagged Υ(1S) Υ(1S, 2S, 3S) → J/ψ + χc1 is the largest, about a −6 events is several times smaller than the number of few times 10 , while that of J/ψ + χc2 is only of −7 directly produced Υ(1S) events used in the previ- order 10 [71]. ous analysis, the tagging procedure drastically sup- Belle searched for double charmonia J/ψ(ψ0)+X, presses the backgrounds, especially those from the where X is the ηc, χcJ (J = 0, 1, 2), or ηc(2S), in processes with ISR or final-state radiation (FSR), Υ(1S) and Υ(2S) decays [72]. To increase the signal which have an event topology similar to that of the detection efficiencies, only the J/ψ or ψ0 candidate signal. Moreover, two extra pion tracks increase a 10 (a) (b)

Figure 5: (a) The diﬀerential cross sections in x for Υ(1S) → J/ψ + anything from CLEO [56]. (b) The diﬀerential branching fractions of Υ(1S) inclusive decays into the J/ψ and ψ(2S) from Belle [57].

the initial vector state with emission of a photon.

• Υ(1S, 2S, 3S) → γ + bottomonium Long-lived b¯b states are especially suited for testing lattice QCD calculations [78], potential models [79, 80, 81], and effective field theories [82, 83, 84, 85]. Electron-positron colliders can directly produce the narrow S-wave states Υ(2S) and Υ(3S), whose radiative decays provide access to the triplet P -wave states χbJ (1P ) and χbJ (2P ) with J = 0, 1, 2. The precise measurements for Υ(2S) → Figure 6: The recoil mass spectrum against J/ψ in Υ(1S) γχ (1P ) and Υ(3S) → γχ (2P ) were performed decays from Belle [72]. The red solid curve is the nominal fit bJ bJ and the blue-dashed curve shows the total background. The by CLEO [86]. The signals were searched for in fitted normalized continuum contribution is represented by the inclusive photon spectra in decays of these nar- the cyan-shaded histogram. row resonances. The energy spectra of photons in Υ(2S) and Υ(3S) decays are shown in Fig.9, where the bottom shows the distributions in data after trigger efficiency for low-multiplicity final states of subtracting the backgrounds. Three peaks are ob- the charmonium decay. In order to estimate the vious, indicated as χb2(1P )/χb1(1P )/χb0(1P ) and statistical significance of the observed signal, a si- χb2(2P )/χb1(2P )/χb0(2P ) (from left to right), in multaneous unbinned likelihood fit to γJ/ψ mass Υ(2S) and Υ(3S) decays. The measured branching fractions for Υ(2S) → γχbJ (1P ) and Υ(3S) → spectra in Mrec (the mass recoiling against a pion γχbJ (2P )(J = 0, 1, 2) transitions are listed in Ta- pair) signal, and J/ψ and Mrec sidebands regions is performed, as illustrated in Fig.8. The fit yields the ble3. These branching fractions are basically con- +2.5 sistent with previous measurements [87, 88, 89, 90, number of signal events to be 5.0−1.9, and the estimated background contribution in the signal region 91, 92, 93, 94]. It is worth mentioning that the transition Υ(3S) → γχbJ (1P ) with J = 0, 1, 2 was also is less than 0.1. The significance of the χc1 signal is 7.5σ. Finally, the branching fraction is calculated studied in Refs. [86, 89, 95, 96]. In comparison with +2.4+0.4 −5 the Υ(nS) → γχbJ [(n − 1)P ] for n = 2 and 3, the to be B(Υ(1S) → γχc1) = (4.7−1.8−0.5) × 10 . The obtained result is slightly higher than the pre- measured branching fraction of Υ(3S) → γχbJ (1P ) vious upper limit [73] and much higher than the the- decreases by one order of magnitude. oretical expectation [75]. However, the recent ob- With the substantial productions of χbJ states + − servation of production in the process e e → γχc1 in the Υ(2S, 3S) radiative decays, the exclusive with a large cross section [77] perhaps indicates a decays of the χbJ states into light hadron final similarity of the mechanism of χc1 formation from states and double charmonia can be studied fur- 11 (a) (b)

Figure 7: The γJ/ψ invariant mass distributions in the (a) Υ(1S) and (b) Υ(2S) decays, and invariant mass spectra for ηc candidates in the (c) Υ(1S) and (d) Υ(2S) decays from Belle [73, 74]. The shaded histograms in (a) and (b) are from the normalized J/ψ sidebands, and the shaded histograms (not normalized) in (c) and (d) are from the continuum contributions.

Table 3: The branching fractions for Υ(2S) → χbJ (1P ) and Υ(3S) → χbJ (2P )(J = 0, 1, 2) transitions from CLEO [86].

B(Υ(2S) → χb0(1P )) (3.75 ± 0.12 ± 0.47)%

B(Υ(2S) → χb1(1P )) (6.93 ± 0.12 ± 0.41)%

B(Υ(2S) → χb2(1P )) (7.24 ± 0.11 ± 0.40)%

B(Υ(3S) → χb0(2P )) (6.77 ± 0.20 ± 0.65)%

B(Υ(3S) → χb1(2P )) (14.54 ± 0.18 ± 0.73)%

B(Υ(3S) → χb2(2P )) (15.79 ± 0.17 ± 0.73)%

ther. CLEO presented the observations of 14 exclusive final states of light hadrons in both χbJ (1P ) and χbJ (2P ) decays [97]. Later, Belle used 74 hadronic final states to reconstruct χbJ (1P ) states, where 41 modes are observed with at least five Figure 8: The J/ψγ invariant mass spectrum in the Υ(1S) standard deviation significance [98]. These mea- data from Belle [76]: a) M (the mass recoiling against a rec surements enriched the understanding of the χ pion pair) signal window, b) 20 times wider J/ψ mass side- bJ states. Belle searched for the first time for dou- bands, c) 20 times wider Mrec mass sidebands. Histograms are the background expectation from the MC simulation ble charmonium decays of χbJ states (Υ(2S) → 0 0 0 + − + − 0 γχ , χ → J/ψJ/ψ, J/ψψ , ψ ψ )[99]. No sig- from: b) Υ(1S) → γFSRµ µ , c) e e → γISRψ . The bJ bJ solid lines show the result of the simultaneous fit to all these nificant χbJ signal is observed in the double char- distributions. The dotted line in a) shows the upper limit monium mass spectra. The upper limits at 90% on the χc2 signal yield at 90% C.L. C.L. on the branching fractions of χbJ decays were determined to be at the level of 10−5, except for 12 (a) (b)

Figure 9: The energy spectra of photons in (a) Υ(2S) and (b) Υ(3S) decays from CLEO [86]. The dots represent the data (top plot) or the data after subtracting the backgrounds (bottom plot).

0 −4 B(χc0 → J/ψψ ) < 1.2 × 10 . These upper lim- ment with the recent lattice NRQCD calculation its are much lower than the central values pre- (5.4 ± 1.8) × 10−4 [108]. dicted by the light cone formalism [100, 101] and pQCD calculation [102], but are consistent with calculations using the NRQCD factorization approach [103, 104].

Besides the triplet P -wave states χbJ (1P ) and χbJ (2P ), another interest in Υ radiative decays is to search for the spin-singlet pseudoscalar partner — ηb(nS) state. Measurement of the hyperfine mass splittings between the triplet and singlet quarkonium states is critical for understanding the role Figure 10: Υ(3S) inclusive photon spectrum after subtract- of spin-spin interactions in quarkonium models and ing the non-peaking background from BaBar [105]. The dots for testing QCD calculations. BaBar reported the with error bars show the inclusive photon spectrum after subtracting the non-peaking background, and the solid, dotted, first observation of ηb(1S) in the radiative decay and dashed lines show the χ (2P ) peak, ISR Υ(1S), and of Υ(3S)[105]. The signal of Υ(3S) → γηb(1S) bJ is extracted from a fit to the inclusive photon en- ηb(1S) signal. ergy spectrum in the e+e− C.M. frame, as shown in In addition to the searches for the η (1S) Fig. 10. A peak is observed at E = (921.2+2.1±2.4) b γ −2.8 state in the inclusive radiative transitions of MeV with a significance of 10σ, which corresponds Υ(2S, 3S)[105, 106, 107], the exclusive decays of to an η (1S) mass of (9388.9+3.1 ± 2.7) MeV/c2. b −2.3 η (1S) and η (2S) to 26 different hadronic final The branching fraction for this radiative Υ(3S) de- b b states were also studied [109]. The cascade decay is obtained to be (4.8 ± 0.5 ± 1.2) × 10−4. cays Υ(1S, 2S) → γηb(nS), ηb(nS) → Xb¯b (Xb¯b = In the Υ(2S) radiative decay, the searches for 4, 6, 8, 10 π±,K±, p/p¯) were studied in Ref. [109] ηb(1S) were also performed by BaBar [106] and using CLEO data samples. Figure 11 shows Belle [107]. The first observation of Υ(2S) → the related distributions of ∆M ≡ M(Xb¯bγ) − 2 γηb(1S) was reported by Belle [107]. The branch- M(Xb¯b). The enhancement around 70 MeV/c in +0.6+0.9 ing fraction for Υ(2S) → γηb(1S) is (6.1−0.7−0.6) × the Υ(1S) data corresponds to the ηb(1S) state 10−4 [107]. It is compatible with the BaBar re- with a mass of (9393.2 ± 4.1) MeV/c2, and that +1.1 −4 2 sult (3.9 ± 1.1−0.9) × 10 [106] and also in agree- around 50 MeV/c in the Υ(2S) data corresponds 13 to the ηb(2S) state with a mass of (9974.6 ± 3.1) states include the X(3872), X(4350), Y (4140), 2 ± ± MeV/c . The evidence for ηb(2S) and the tran- Y (4260), Y (4360), Y (4660), Zc(3900) , Zc(4050) , ± ± ± sition of Υ(1S) → γηb(1S) were reported for the Zc(4200) , Zc(4430) , and Zcs [6, 112, 113]. They first time. The products of the branching frac- are studied in the final states that contain a J/ψ(ψ0) ± ± tions B(Υ(1S) → γηb(1S))B(ηb(1S) → Xb¯b) and and up to two charged light hadrons (K /π ). B(Υ(2S) → γηb(2S))B(ηb(2S) → Xb¯b) are mea- No significant signal is found in any of the stud- +33.5 −6 +29.7 sured to be (30.1−7.4 ±7.5)×10 and (46.2−14.2 ± ied modes and 90% C.L. upper limits on the prod- 10.6) × 10−6, respectively. uct branching fractions, i.e., B(Υ(1S) → XYZ + anything)B(XYZ → J/ψ(ψ0) + hadrons), are set, With a much larger Υ(2S) data sample, Belle which are at or below the level of 10−5. also searched for ηb(2S) in the 26 exclusive hadronic final states as above [110]. Figure 12 shows the Considerable efforts in theory have been de- corresponding ∆M distribution [110]. No evi- voted to interpreting the charged charmonium-like dence for ηb(2S) was found, and a 90% C.L. up- states (Zc) as tetraquarks, molecules, hybrids, or per limit on the branching fraction B(Υ(2S) → hadrocharmonia [46, 47, 114, 115, 116, 117]. To −6 γηb(2S))B(ηb(2S) → Xb¯b) < 4.9×10 was ob- distinguish among these explanations, experimen- tained, which is an order of magnitude smaller tal input is needed, especially that on the double ± + − + − than the result in Ref. [109]. This result disfa- Zc production in e e annihilation. For e e → + − + − vors the observation of ηb(2S) in Υ(2S) exclusive Zc Zc , the dependence on the squared e e C.M. decays in Ref. [109]. Previously, Belle reported energy, s, of the electromagnetic form factor, 3 the evidence for η (2S) in the h (2P ) → η (2S)γ F + − , is 1/s for a Z state with tetraquark struc- b b b Zc Zc c −1 transition using a 133.4 fb data sample col- ture or 1/s for a Zc system of two tightly bound di- lected at energies near the Υ(5S) resonance [111]. quarks [118, 119]. However, it remains unclear from +2.8 The measured mass of ηb(2S) is (9999.0 ± 3.5−1.9) which values of s onwards this scaling is applicable. MeV/c2, which is far away from the mass reported Belle searched for doubly charged charmonium-like in Ref. [109]. Therefore, the enhancement observed state productions using the largest data samples of around 9974.6 MeV/c2 in Ref. [109] is likely at- Υ(1S) and Υ(2S)[120]. No significant signals are tributed to statistical fluctuation. observed in any of the studied modes, and the 90% + (0)− C.L. upper limits on B(Υ(1S, 2S) → Zc Zc ) × + + 0 B(Zc → π + cc¯)(cc¯ = J/ψ, χc1(1P ), ψ ) are in 2.2.4. Search for exotic states in Υ(1S, 2S, 3S) de- −6 the range of (1 − 50) × 10 . Here, Zc refers to cays Zc(3900) and Zc(4200) observed in the πJ/ψ final state, Zc1(4050) and Zc2(4250) in the πχc1 final state, and Z (4050) and Z (4430) in the πψ0 final • Search for XYZ states in Υ(1S, 2S) radiative c c state. and inclusive decays The radiative decays of the Υ states below open- • Search for glueballs and light tetraquraks in bottom threshold are used to search for charge- Υ(1S, 2S) decays parity-even charmonium-like states. Searches for the X(3872), X(3915), and Y (4140) states in the The existence of glueballs, with a rich spec- Υ(1S, 2S) radiative decays were carried out by troscopy and a complex phenomenology, is one of Belle [73, 74]. The X(3872) is reconstructed in the early predictions of QCD [121]. However, de- the π+π−J/ψ and π+π−π0J/ψ final states, and spite many years of experimental efforts, none of X(3915) and Y (4140) are reconstructed in ωJ/ψ these gluonic states has been established unambigu- and φJ/ψ. No X(3872), X(3915), or Y (4140) sig- ously. In the experimental aspect, the most out- nals are observed, and the corresponding produc- standing obstacle is the isolation of glueballs from tion rates of the π+π−J/ψ and π+π−π0J/ψ, ωJ/ψ, ordinary hadrons. or φJ/ψ modes are found to be less than a few times Recently Belle utilized the data samples of 102M 10−6 at 90% C.L.. Υ(1S) and 158M Υ(2S) events to search for 0−− Belle reported a search for some XYZ states glueballs (G0−− ) with quantum numbers incom- in Υ(1S) inclusive decays using the world’s patible with quark-antiquark bound states [58]. largest Υ(1S) data sample [57]. The XYZ Two 0−− glueballs are predicted using QCD sum 14 Figure 11: Distributions of ∆M ≡ mΥ(1S,2S) − M(Xb¯b) in Υ(2S) data (top row) and Υ(1S) data (bottom row) are shown with both linear (left column) and logarithmic (right column) scales using CLEO data [109].

Following the above searches for G0−− , Belle utilized the low side of the recoil mass spectra of χc1, f1(1285), J/ψ, and ω from Ref. [58] to search for Xtetra in Υ(1S, 2S) → χc1/f1(1285) + Xtetra and χb1 → J/ψ/ω + Xtetra [124]. Here, the Xtetra is a four-quark state with J PC = 0−− or 1+− calculated by Laplace sum rules and ﬁnite-energy sum rules using tetraquark-like currents [125]. No evident signal is found below 3 GeV/c2 in these pro- Figure 12: The distributions of ∆M in Υ(2S) data from cesses, and 90% C.L. upper limits on the branching Belle [110]. Dots with error bars are the data, the blue solid fractions are set with Xtetra masses from 1.16 to curve is the result of the ﬁt for the signal-plus-background 2.46 GeV/c2 and widths from 0.0 and 0.3 GeV/c2; hypothesis, and blue dashed curve is background compo- those are found to a few times 10−4 or 10−5. nent, respectively. The three χbJ (1P ) components indicated by the red dotted curves are here considered as part of the signal. The inset shows an expanded view of the ∆M distri- • Search for six-quark states in Υ(1S, 2S, 3S) de- bution in the region of [0.035, 0.065] GeV/c2. cays

While the vast majority of known hadrons can be rules [122] with masses of (3.81 ± 0.12) GeV/c2 and described as either quark-antiquark, three-quark, (4.33±0.13) GeV/c2, and a 0−− glueball calculated or three-antiquark combinations, other possibilities using the dynamical holographic QCD model has a are allowed by QCD. Among those, the six-quark mass of 3.817 GeV/c2 [123]. Belle searched for such configuration uuddss is of particular interest, as G0−− in Υ(1S, 2S) → χc1/f1(1285) + G0−− and its spatial wave function is completely symmetric. χb1 → J/ψ/ω+G0−− processes [58]. No evident sig- Generic arguments imply that it should be the most nal is found at the predicted masses in all the stud- tightly bound six-quark state [126]. This property ied processes, and 90% C.L. upper limits are set on was already noticed by Jaffe in 1977 [127]. He their branching fractions. As an example, Fig. 13 predicted the existence of a loosely bound uuddss shows the 90% C.L. upper limits on the branching state with a mass below the 2mΛ threshold by 2 fractions of Υ(1S)/Υ(2S) → χc1 + G0−− as a func- about 80 MeV/c , dubbed the H dibaryon. Re- tion of the G0−− width. Interestingly, a signal at cently, the motivation to search for H has been 2 3.92 GeV/c is evident in Υ(1S) → f1(1285)+G0−− strengthened by lattice QCD [128, 129, 130] and with a significance of 3.7σ, which is consistent with chiral constituent model calculations [131]. With the prediction in Ref. [122] within uncertainty. its mass approaching the 2mΛ threshold from be- 15 Figure 13: The upper limits on the branching fractions for (a) Υ(1S) → χc1 + G0−− and (b) Υ(2S) → χc1 + G0−− as a function of the assumed G0−− decay width from Belle [58]. low (above), H would behave more and more like a ΛΛ analog of deuteron (dineutron), independent of its dynamical origin [132]. If its mass is below 2mΛ, H would predominantly decay via ∆S = +1 weak transitions to the Λn,Σ−p,Σ0n or Λpπ− final states. If its mass is above 2mΛ, but below 2 mΞ0 + mn (2mΛ + 23.1 MeV/c ), it would decay via strong interactions to ΛΛ completely.

Belle reported a search for H in the inclusive processes Υ(1S, 2S) → H + anything, H → Λpπ− Figure 14: Upper limits at 90% C.L. for B(Υ(1S, 2S) → and ΛΛ [133]. The signals are extracted by a se- H + anything) for a narrow (Γ = 0) H dibaryon ver- quence of binned minimum fits to the distributions sus MH − 2mΛ are shown as solid horizontal bars from − Belle [133]. The +1σ (+2σ) values from the fitted signal of M(Λpπ ) − 2mΛ and M(ΛΛ) − 2mΛ. In the fits, the signal peak position is confined to a 4 MeV/c2 yields are shown as the dotted (dashed) bars. (For some window that is scanned in 4 MeV/c2 steps across mass bins, these are negative and not shown.) The verti- − cal dotted line indicates the MH = 2mΛ threshold. The the ranges (m + m + m − ) ≤ M(Λpπ ) ≤ 2m Λ p π Λ horizontal dotted line indicates the average PDG value for and 2m ≤ M(ΛΛ) ≤ (2m + 28 MeV/c2). None Λ Λ B(Υ(1S, 2S) → d¯+ anything). of the fits exhibits a positive signal with the significance greater than 3σ. The fit results are translated into 90% C.L. upper limits on the signal yield, then used to determine upper limits on the inclusive product branching fractions (B(Υ(1S, 2S) → H + anything)B(H → Λpπ−/ΛΛ)), as shown in Fig. 14. The reported results are some of the most stringent constraints to date on the existence of H. charged and neutral particles. The mass of the un- seen S can be identified with the square of the mass A stable sexaquark, denoted as S, with the six- recoiling against the ΛΛ system. As a result, no sig- quark uuddss component was proposed to be dis- nificant signal is observed, and 90% C.L. upper lim- covered in Υ → SΛ¯Λ¯ decays [134]. If S really exists, its on the Υ(2S, 3S) → SΛ¯Λ¯ branching fractions, then S is a good DM candidate. BaBar investigated scanning S masses in the range 0 < mS < 2.05 such a state in Υ(2S) and Υ(3S) decays [135]. Con- GeV/c2 in steps of 50 MeV/c2 (approximately half sidering the S is undetected, the total energy of the signal resolution), are derived, which are shown clusters in the electromagnetic calorimeter not as- in Fig. 15 for the Υ(2S, 3S) datasets, as well as the sociated with charged particles, Eextra, is required combined sample assuming the same partial width. to be less than 0.5 GeV to remove hadronic events These results set stringent bounds on the existence containing several strange baryons and additional of the S state. 16 with zero. The dot-dashed line shows the expected signal for B(Υ(1S) → χχ) = 6.0 × 10−3 [22]. The resulting B(Υ(1S) → invisible) < 2.5 × 10−3 was obtained at 90% C.L. This result disfavors the prediction in Ref. [22] for the Υ(1S) decay to a pair of DM particles lighter than the b quark.

Figure 15: The 90% C.L. upper limits on the Υ(2S, 3S) → SΛ¯Λ¯ branching fractions for the Υ(2S) and Υ(3S) datasets, as well as the combined sample assuming the same partial width from BaBar [135].

2.2.5. Search for new physics in Υ(1S, 2S, 3S) decays Figure 16: Recoil mass distribution against two pions, M recoil , from Belle [136]. The solid curve shows the result π+π− of the fit to the data, the shaded area shows the total back- • Invisible decay of Υ(1S) ground contribution, the dashed line shows the combinato- rial background contribution, and the dot-dashed line shows In the invisible decay modes, the final state parti- the expected signal for B(Υ(1S) → χχ) = 6.0 × 10−3 [22]. cles interact so weakly with the detectors that they are not observable. In the SM, the invisible decay CLEO [137] and BaBar [138] also measured the of the Υ(1S) meson into a νν¯ pair is predicted to branching fraction of the invisible Υ(1S) decay. −6 have a branching fraction of (9.9 ± 0.5) × 10 [21]. To provide a clean sample of Υ(1S) decays, the If the observed branching fraction is significantly events in the transitions of Υ(2S) → π+π− and larger than the SM prediction, physics beyond the Υ(3S) → π+π− are selected by CLEO [137] and SM will be implied. For instance, the low-mass DM BaBar [138]. The invisible Υ(1S) decay is also candidates could couple weakly to the SM parti- searched for in the mass recoiling against the di- cles to enhance the branching fraction to be about pion system. The resulting limits at 90% C.L. are −3 6.0 × 10 [22]. B(Υ(1S) → invisible) < 3.9×10−3 and B(Υ(1S) → Belle reported a search for the Υ(1S) invisible invisible) < 3.0 × 10−4 from CLEO [137] and decay via the Υ(3S) → π+π−Υ(1S) transition us- BaBar [138], respectively. Notably, the limit from ing the Υ(3S) data sample [136]. For the selection BaBar is almost an order of magnitude closer to of the invisible decay candidates, two oppositely the SM prediction than the limits from Belle and charged tracks π+π− are required in the event. CLEO. Considering the products from Υ(1S) decay may go outside of the detector acceptance, the recoil • Υ(1S, 2S, 3S) → γ + A0 mass distribution against two pions in Υ(3S) data, recoil Mπ+π− , still peaks at the Υ(1S) mass and becomes A low mass DM particle, χ, is predicted to be a peaking background to the invisible decay signal. produced in interactions of SM particles via the ex- Expected numbers of peaking background events change of a CP -odd Higgs boson A0 [139, 140], based on MC calculations for Υ(1S) → µ+µ−, which is part of the NMSSM [23]. Transitions e+e−, τ +τ −, and νν¯ are 77.3±12.0, 50.3±8.2, 5.2± Υ(2S) → π+π−Υ(1S) and Υ(3S) → π+π−Υ(1S) 1.0, and 0.4±0.1 assuming the world-average values offer a way to cleanly detect the production of of B(Υ(1S) → µ+µ−/e+e−/τ +τ −) from PDG [6], Υ(1S) mesons, and enable searches for A0 in the and B(Υ(1S) → νν¯) = (9.9 ± 0.5) × 10−6 [21]. radiative decay of Υ(1S). The B(Υ(1S) → γA0) recoil −4 Figure 16 shows the Mπ+π− distribution. The ex- is predicted to be as large as 5 × 10 , depend- tracted signal yield, 38 ± 39 events, is consistent ing on mA0 and couplings [141, 142]. For the 17 UL + − multi-body Υ(1S) → γχχ decay, the branching by N /[NΥ(2S) × B(Υ(2S) → Υ(1S)π π ) × ε], fraction is suppressed by O(α) in comparison with where N UL an upper limit at 90% C.L. on the sig- B(Υ(1S) → χχ), and the range 10−5 − 10−4 is ex- nal yield and ε is the signal efficiency, as shown in pected [143]. The SM process Υ(1S) → γνν¯ has Fig. 18. the same final state as the signal, but is predicted to BaBar also searched for the single-photon decays have a branching fraction of the order of 10−9 [143], of Υ(1S): Υ(1S) → γA0 and Υ(1S) → γχχ [145]. which is three orders of magnitude below our exper- Two kinematic variables are used to extract the sig- imental sensitivity. nals: the recoil mass of π+π− and the missing mass Very recently, Belle searched for A0 in the fi- squared of π+π−γ. The yields of signal events are nal states with a single photon and missing en- extracted as a function of mA0 (mχ) in the interval 0 2 2 ergy [144]. Since the mass of A is unknown, 0 ≤ mA0 ≤ 9.2 GeV/c (0 ≤ mχ ≤ 4.5 GeV/c ) two processes are considered: the on-shell process by performing a series of unbinned extended max- Υ(1S) → γA0 with A0 → χχ; and the off-shell pro- imum likelihood scans. No evidence for the single- cess Υ(1S) → γχχ. The Υ(1S) is tagged via the di- photon decay Υ(1S) → γ + invisible was found, pion transition Υ(2S) → Υ(1S)π+π−. Both signal and the 90% C.L. upper limits on B(Υ(1S) → processes produce only three detectable particles: γA0)B(A0 → invisible) and B(Υ(1S) → γχχ) two charged pions, which have low transverse mo- were obtained; see Fig. 18. The results from menta, and a photon, which has the highest energy Belle [144] and BaBar [145] improve the existing in the C.M. frame. Two observables are used to limits by an order of magnitude or more, and sig- extract the signals: the invariant recoil mass of the nificantly constrain light Higgs boson [141, 142] and dipion system (Mrecoil) and the energy of photon light DM [143] models. The limits from Belle are ∗ in the Υ(1S) frame (Eγ ). Distributions of Mrecoil more stringent than BaBar, especially for the off- ∗ 0 and Eγ are shown in Fig. 17. Irreducible back- shell case. Instead of A → invisible, BaBar and grounds can be categorized into four event types: CLEO searched for A0 via the µ+µ− [146, 147, 148], tau-pair productions Υ(2S) → τ +τ −, continuum, τ +τ − [146, 149, 150], and hadronic final states [151, leptonic decays Υ(1S) → `+`−, and hadronic de- 152], but did not see any significant signal in cays Υ(1S) → γhh. Tau-pair productions from these final states. The most conservative upper the Υ(2S) and continuum backgrounds do not peak limits at 90% C.L. on B(Υ(1S)/Υ(2S)/Υ(3S) → either in the recoil mass distribution nor in the γA0)B(A0 → µ+µ−/τ +τ −/hadrons) are summa- photon energy spectrum; therefore, they are com- rized in Table4[146, 147, 148, 149, 150, 152]. These bined, as shown by the cyan dashed curve in Fig. 17. measurements improve the constraints on the pa- The background contributions from leptonic decays rameters of the NMSSM and similar theories with and hadronic decays of Υ(1S) are predicted to be low-mass scalar degrees of freedom. 20.0 ± 2.8 and 1.2 ± 0.7 events, which are combined as shown by the magenta dashed curves in Fig. 17. • Tests of lepton flavor universality in An unbinned extended likelihood fit in the two- Υ(1S, 2S, 3S) decays ∗ dimensional (Mrecoil, Eγ ) space is performed to estimate the yields of different event types. The fit is The measurements of the ratios Rτµ(Υ(nS)) = + − + − repeated for each value of MA0 or Mχ in the mass B(Υ(nS) → τ τ )/B(Υ(nS) → µ µ )(n = 1, 2 ranges 0 < MA0 < 8.97 GeV/c (on-shell process) 2, 3) are motivated as tests of the lepton fla- 2 or 0 < Mχ < 4.44 GeV/c (off-shell process). For vor universality (LFU). In the SM these ratios the on-shell case, the photon energy is scaned in are expected to be close to 1. Any significant 353 steps that correspond to half the photon en- deviations would violate LFU and could be in- ergy resolution, and step size in the range from 25 troduced by the coupling to a light pseudoscalar MeV/c2 to 4 MeV/c2. For the off-shell case, 45 Higgs boson in supersymmetry, or leptoquarks and 2 scan points with a fixed step size of 100 MeV/c compositeness models. The values of Rτµ(Υ(1S)), 2 are used. When MA0 = 2.946 GeV/c , the local Rτµ(Υ(2S)), and Rτµ(Υ(3S)) obtained by CLEO significance of 2.1σ is largest; see Fig. 17. No sta- are 1.02 ± 0.02 ± 0.05, 1.04 ± 0.04 ± 0.05, and tistically significant signal is observed, and the up- 1.05 ± 0.08 ± 0.05 [153], which are consistent with per limits (90% C.L.) on the branching fractions the expectations from the SM. Recently, BaBar re- of the on-shell and the off-shell signals are given ported a value of Rτµ(Υ(3S)) to be (0.996±0.008± 18 2 Figure 17: Two-dimensional fit result for the on-shell process mass scan point with MA0 = 2.946 GeV/c , which has the highest ∗ local signal significance of 2.1σ from Belle [144]. Left: Mrecoil distribution. Right: Eγ distribution. The fitted components are tau-pair and continuum backgrounds (cyan dashed curve), Υ(1S) leptonic and hadronic decay backgrounds (magenta dashed curve), and the on-shell signal (red dashed curve). The blue solid curve shows the sum of all fitted components, and the black dashed curve shows the sum of all fitted background components.

Figure 18: 90% C.L. upper limits on the branching fractions of the on-shell process Υ(1S) → γA0 with A0 → χχ (left) and the oﬀ-shell process Υ(1S) → γχχ (right). The orange solid curves are the Belle limits [144] and the blue dashed curves are the BaBar limits [145].

Table 4: The most conservative upper limits at 90% C.L. on the B(Υ(1S)/Υ(2S)/Υ(3S) → γA0)B(A0 → µ+µ−/τ +τ −/hadrons) from CLEO [146] and BaBar [147, 148, 149, 150, 152].

Mode Υ(1S) → γA0,A0 → µ+µ− Υ(2S) → γA0,A0 → µ+µ− Υ(3S) → γA0,A0 → µ+µ− Branching fraction 9 × 10−6 [146, 148] 8.3 × 10−6 [147] 5.5 × 10−6 [147] Mode Υ(1S) → γA0,A0 → τ +τ − Υ(2S) → γA0,A0 → τ +τ − Υ(3S) → γA0,A0 → τ +τ − Branching fraction 1.3 × 10−4 [146, 150] - 1.6 × 10−4 [149] Mode Υ(1S) → γA0,A0 → hadrons Υ(2S) → γA0,A0 → hadrons Υ(3S) → γA0,A0 → hadrons Branching fraction - 8 × 10−5 [152] 8 × 10−5 [152]

19 0.014) [154]. The uncertainty in this result is al- 10−4 [164]. The transition was later confirmed by most an order of magnitude smaller than that in BaBar with a more precise branching fraction of the CLEO result. BaBar also tested LFU in the de- (2.39±0.31±0.14)×10−4 [165]. Meanwhile, the ra- cays Υ(3S) → Υ(1S)π+π−, Υ(1S) → `+`−, where tio B(Υ(2S) → ηΥ(1S))/B(Υ(2S) → π+π−Υ(1S)) −3 ` = µ, τ [155]. The resulting Rτµ(Υ(1S)) is found was measured to be (1.35±0.17±0.08)×10 [165]. to be 1.005 ± 0.013 ± 0.022 with improved preci- With a larger Υ(2S) data sample, Belle reported sion. No significant deviation of Rτµ(Υ(1S)) from a larger branching fraction B(Υ(2S) → ηΥ(1S)) the SM expectation is observed. = (3.57 ± 0.25 ± 0.21) × 10−4 and a larger ratio B(Υ(2S) → ηΥ(1S))/B(Υ(2S) → π+π−Υ(1S)) = The lepton flavor violating (LFV) processes were −3 searched for by CLEO [156] and BaBar [157]. No (1.99 ± 0.14 ± 0.11) × 10 [166]. These branch- significant signal was observed. The 95% C.L. up- ing fractions obtained from CLEO, BaBar, and per limits on the branching fractions for Υ(1S) → Belle are all slightly below the values predicted by µ±τ ∓, Υ(2S) → µ±τ ∓, and Υ(3S) → µ±τ ∓ are QCDME in Ref. [160]. Up to now, the transition of 6.0 × 10−6, 14.4 × 10−6, and 20.3 × 10−6 from Υ(3S) → ηΥ(1S) has not yet been observed. The upper limits on B(Υ(3S) → ηΥ(1S)) at 90% C.L. CLEO [156]. The more stringent constraints were −4 given by BaBar [157]. The 90% C.L. upper lim- in units of 10 are 1.8 from CLEO [164] and 1.0 its on the branching fractions for Υ(2S) → e±τ ∓, from BaBar [165]. Υ(2S) → µ±τ ∓, Υ(3S) → e±τ ∓, and Υ(3S) → ± ∓ −6 −6 −6 µ τ are 3.2 × 10 , 3.3 × 10 , 4.2 × 10 , and • Υ(4S) → Υ(1S) + π+π−/η 3.1×10−6. Effective field theory allows one to relate the above fractions to the mass scale Λ`τ of LFV be- The hadronic transitions Υ(4S) → Υ(1S) + 2 4 + − yond the SM physics using α`τ /Λ`τ = (B(Υ(nS) → π π /η were studied by BaBar [167] and ± ∓ + − 2 2 4 + − ` τ )/B(Υ(nS) → ` ` ))(2qb α /(mΥ(nS)) )(n = Belle [168]. The signature of Υ(4S) → π π Υ(1S) 1, 2, 3) [158, 159], where ` denotes e or µ, αlτ is the events is an invariant mass difference ∆M = NP coupling constant, qb = −1/3 is the charge of b Mππll −Mll compatible with the difference between quark, and α ≡ α(mΥ(nS)) is the fine structure con- the masses of Υ(4S) and Υ(1S), where Mππll is + − + − + − stant at the Υ(nS) mass. Assuming αeτ = αµτ = 1, the π π ` ` invariant mass, and Mll is the ` ` the branching fractions above can be translated to invariant mass. For Υ(4S) → ηΥ(1S), the signal the 90% C.L. lower limits Λeτ > 1.6 TeV and Λµτ > events are identified by the invariant mass differ- 1.7 TeV from BaBar [157], and the 95% C.L. lower ence, ∆Mη = M3π`` − M`` − M3π compatible with limit Λµτ > 1.34 TeV from CLEO [156]. M(Υ(4S))−M(Υ(1S))−M(η), where M3π`` is the + − 0 + − + − 0 π π π ` ` invariant mass, and M3π is π π π invariant mass. The ∆M and ∆M distributions 2.3. Transitions between Υ(mS) and Υ(nS) (m > η from Belle are shown in Fig. 19. Clear signals n) are observed in both Υ(4S) → π+π−Υ(1S) and Υ(4S) → ηΥ(1S). The ratio of their branching fractions is determined to be 2.07 ± 0.30 ± 0.11. • Υ(2S, 3S) → Υ(1S) + η/π+π− From BaBar, the corresponding ratio is 2.41±0.40± The QCD multipole expansion (QCDME) model 0.12. These results strongly disfavor the prediction can be used to describe the hadronic transitions be- by QCDME in Ref. [160]. This implies additional tween heavy quarkonia [160]. It has succeeded in implementations for QCDME. explaining the hadronic transitions between char- 0 monia, e.g., the relative rate of ψ → ηJ/ψ to 0 ψ0 → π+π−J/ψ [161]. It also predicts that the • Υ(4S) → Υ(1S) + η transitions between bottomonium states involving 0 The kinematically allowed transition Υ(4S) → an η or π meson are highly suppressed, since they Υ(1S) + η0 is expected to be as strong as Υ(4S) → require a spin flip of the heavy quark [160, 162, 163]. Υ(1S)+η [169]. The relative strength of the η0 to η The first process involving a b-quark spin flip, transitions depends on the relative uu¯ + dd¯ content Υ(2S) → ηΥ(1S), was observed by CLEO with of the mesons, and is predicted to range between a statistical significance of 5.3σ [164]. Its branch- 20% and 60%. Belle reported the observation of +0.7 0 ing fraction was measured to be (2.1−0.6 ± 0.3) × Υ(4S) → Υ(1S) + η with a significance of 5.7σ 20 + − Figure 19: The ∆M (left plot) and ∆Mη (right plot) distributions for Υ(4S) → π π Υ(1S) and Υ(4S) → ηΥ(1S) from Belle [168]. The data, best fits to them, and background contributions are shown by the dots with error bars, solid blue lines, and dashed red lines, respectively.

[170]. The Υ(1S) meson is reconstructed via its distributions for Υ(5S) → Υ(1S)π+π−, Υ(5S) → decay to two muons, which is considerably cleaner Υ(2S)π+π−, Υ(5S) → Υ(3S)π+π−, and Υ(5S) → than the dielectron mode. The η0 meson is recon- Υ(1S)K+K− are shown in Fig. 21. The branch- structed via the ρ0(→ π+π−)γ and η(→ γγ)π+π− ing fractions of these four decays are measured to final states. be (0.53 ± 0.03 ± 0.05)%, (0.78 ± 0.06 ± 0.11)%, (0.48+0.18 ±0.07)%, and (0.061+0.016 ±0.010)%, re- The signal events are identified by the variable −0.15 −0.014 0 spectively. Notably, these values exceed their coun- ∆M 0 = M(Υ(4S)) − M(Υ(1S)) − M(η ), where η terparts between lower Υ resonances by more than M(η0) = M(π+π−γ) or M(γγπ+π−), M(Υ(1S)) = 2 orders of magnitude [167, 168]. The unexpectedly M(µ+µ−), and M(Υ(4S)) = M(µ+µ−π+π−γ) or large branching fractions disagree with the expec- M(µ+µ−γγπ+π−). The signal and background tation of a pure b¯b state for the Υ(5S) resonance, yields are determined by an unbinned maximum unless there is a new mechanism to enhance the de- likelihood fit to the ∆Mη0 distributions, as shown in cay rate. The later observations of Zb(10610) and Fig. 20. With a simultaneous fit to these ∆M 0 dis- η Z (10650) [172] decaying into πΥ(1S), πΥ(2S) and tributions, the statistical significance is estimated b πΥ(3S) are supposed to contribute to the enhance- to be 5.8σ. It is reduced to 5.7σ with the system- ments in the dipion transitions of Υ(5S). atic uncertainty included. The resulting branching fraction B(Υ(4S) → Υ(1S) + η0) is (3.43 ± 0.88 ± 0.21) × 10−5. In addition, the ratios B(Υ(4S) → Υ(1S)+η0)/B(Υ(4S) → Υ(1S)+η) and B(Υ(4S) → 3. Υ(1S, 2S, 3S) measurements at the LHC Υ(1S) + η0)/B(Υ(4S) → Υ(1S) + π+π−) are obtained to be 0.20 ± 0.06 and 0.42 ± 0.11, where the 3.1. The pp, pP b, and P bP b datasets at the LHC uncertainties include both systematic and statistical uncertainties. The former ratio, in particular, is The integrated luminosities of the pp colliding in agreement with the expected value in the picture data recorded by LHCb, CMS, and ATLAS in the of Υ(4S) as an admixture of a state containing light past decade are displayed in Fig. 22[173]. The total quarks in addition to the b¯b pair [169]. The mea- −1 0 integrated luminosities are up to hundreds of fb surements of the η transition may shed light on in CMS and ATLAS, and only a few fb−1 in LHCb. the puzzle of hadronic transitions between heavy The data samples were collected in a wide C.M. en- quarkonia. ergy range from a few to 13 TeV. With these samples, LHCb, CMS, and ATLAS are able to measure • Υ(5S) → Υ(1S, 2S, 3S) + π+π−/K+K− the Υ(1S, 2S, 3S) cross sections and polarizations precisely. In addition to the pp colliding data, AL- Besides the hadronic transitions between lower ICE and CMS collected both pP b and P bP b collid- Υ resonances, the π+π− and K+K− transitions ing data, and LHCb collected pP b colliding data, from Υ(5S) to Υ(1S, 2S, 3S) were also studied by to probe the QGP mechanism by comparing the Belle [171]. The ∆M = M(µµππ/µµKK)−M(µµ) Υ(1S, 2S, 3S) yields from P bP b and pP b collisions 21 0 0 0 Figure 20: A simultaneous fit to the ∆Mη0 distributions for (a) Υ(4S) → Υ(1S) + η (→ ρ γ) and (b) Υ(4S) → Υ(1S) + η (→ ηπ+π−) from Belle [170]. The data, best fit to them, and background contributions are shown by the dots with error bars, solid blue lines, and dashed red lines, respectively.

Figure 21: The ∆M distributions for (a) Υ(5S) → Υ(1S)π+π−, (b) Υ(5S) → Υ(2S)π+π−, (c) Υ(5S) → Υ(3S)π+π−, and (d) Υ(5S) → Υ(1S)K+K− from Belle [171]. The solid red lines show the best ﬁts to the data, while the dashed red lines show the backgrounds.

22 µ µ with those from pp collisions. The total luminosi- > 3.5 GeV/c for |η | < 1.6 and pT > 2.5 GeV/c µ µ ties recorded in pp, pP b, and P bP b collisions at the for 1.6 < |η | < 2.4 at CMS [180], and pT > 4 µ µ LHC are summarized in Table5. GeV/c for |η | < 2.5 at ATLAS [184], where pT and ηµ are transverse momentum and pseudora- The experiments at the LHC have different fea- pidity of the muon. The dimuon invariant mass tures in detecting muons. At LHCb, muons are distributions in ranges of p < 30 GeV/c and |y| identified by a system composed of alternating T < 2 at CMS, 2 < p < 4 GeV/c and |y| < 1.2 at layers of iron and multiwire proportional cham- T ATLAS, and p < 15 GeV/c and 2 < |y| < 4.5 bers [174, 175]. The trigger system of LHCb con- T at LHCb are shown in Fig. 23. Here, p is the sists of a hardware stage based on information from T dimuon transverse-momentum, and y = 1 ln( E+p|| ) the muon systems, followed by a software stage, 2 E−p|| which employs a full event reconstruction [176]. is the rapidity, where E is the energy and p|| is the With high trigger efficiencies and good resolutions, momentum parallel to the beam axis of the muon the measurements consisting muon final states at pair. The Υ(1S, 2S, 3S) resonances are clearly vis- LHCb are very promising. ATLAS [177] and ible and fully resolved at CMS and LHCb. Assum- CMS [178] own the muon spectrometers equipped ing unpolarized Υ(1S, 2S, 3S) production, the prod- with drift tubes, cathode strip chambers, and re- ucts of the Υ(1S, 2S, 3S) cross sections and dimuon sistive plate chambers (CMS only). Owing to a branching fractions in pp collisions from CMS [180] larger magnetic field intensity (2 T at ATLAS and and LHCb [182] are summarized in Table6. 3.8 T at CMS) and extra resistive plate chambers, The Υ(nS)(n = 1, 2, 3) differential cross section the performances on the efficiency and resolution is computed as [180] for muons at CMS, especially in the low momentum region, are better than those at ATLAS. The d2σ(pp → Υ(nS)X) B(Υ(nS) → µ+µ−) = muon identification at ALICE utilizes a front ab- dpT dy sorber and an iron wall located between the track- NΥ(nS)(A, ε) ing and trigger systems [179]. The performances , L∆ ∆ on the efficiency and resolution for muons at AL- pT y

ICE are not very good compared to LHCb, AT- where NΥ(nS) is the signal yield corrected by the LAS, and CMS. The features of the performances dimuon event weights given by the inverse prod- in Υ → µ+µ− measurements at the LHC are shown uct of the detector acceptance A and the recon- in Table5, which are summarized from the related struction efficiency ε, L is the integrated luminos- public webs at CERN [173], technical design re- ity, and ∆pT and ∆y are the widths of the bins in ports [174, 175, 176, 177, 178, 179], and published transverse momentum (pT) and rapidity (y). The papers [180, 181, 182, 183]. We note that, the rapid- symbol X is used to indicate that the measure- ity, transverse momentum, and resolution denote ments include the feed-down contributions originat- the related parameters of Υ, while the trigger effi- ing from higher-mass states, such as the χb family ciency applies to a single muon. and Υ(3S). The differential Υ(1S) cross sections as a function of rapidity y within y < 2 were given by CMS, as illustrated in the left plot of Fig. 24[180]. 3.2. Υ(1S, 2S, 3S) productions in pp collisions As a complement to the phase-space coverage of CMS, LHCb studied the Υ(1S) cross sections over 3.2.1. Υ(1S, 2S, 3S) cross sections the range√ of 2.0 < y < 4.5 in proton-proton colli- Using data samples of 3.1 pb−1, 1.13 pb−1, and sions at s = 7 TeV (see Fig. 24 (right)) [186]. The −1 thick lines in the right plot of Fig. 24 show the fit 25 pb , the Υ(1S, 2S, 3S√) production cross sections in pp collisions at s = 7 TeV were mea- results with NRQCD color-octet model predictions sured by CMS [180], ATLAS [184], and LHCb [182], from Refs. [187, 188] in the region of 2.0 < y < 4.0, respectively. The Υ(1S, 2S, 3S) states are typi- and dashed lines show the extrapolations to the full cally reconstructed via the dimuon decay channel: region of 2.0 < y < 4.5. From Fig. 24, the rapid- Υ(1S, 2S, 3S) → µ+µ−. To ensure that the trigger ity y dependence of the cross section shows a slight and muon reconstruction efficiencies are high and decline towards the higher side. not rapidly changing within the acceptance win- With the increase of pT, several theoretical mod- µ dow, muon candidates are required to satisfy: pT els with higher-order corrections become more sen- 23 Figure 22: The integrated luminosities of the pp data recorded by LHCb, CMS, and ATLAS in the past decade [173].

Table 5: The features of the performances in Υ → µ+µ− measurements at the LHC [173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183].

Experiments LHCb CMS ATLAS ALICE 9 fb−1 (pp) 192 fb−1 (pp) 24.8 nb−1 (pP b) Total luminosity 182 fb−1 (pp) 33.4 nb−1 (pP b) 216.2 nb−1 (pP b), 2.5 nb−1 (P bP b) 1.6 nb−1 (P bP b) Rapidity 1.5 < |y| < 5 |y| < 2.4 |y| < 2.4 2 < |y| < 4.5 Transverse momentum <30 GeV/c <100 GeV/c <70 GeV/c <15 GeV/c Trigger eﬃciency >95% 70%−100% 40%−90% ∼50% Υ mass resolution ∼50 MeV/c2 50−100 MeV/c2 >100 MeV/c2 >100 MeV/c2

24 Figure 23: The left: The dimuon invariant mass distribution with pT < 30 GeV/c and |y| < 2 from CMS [180]; the solid line shows the result of a ﬁt to the invariant mass distribution, with the dashed line denoting the background component. The middle: The dimuon invariant mass distribution with 2 < pT < 4 GeV/c and |y| < 1.2 from ATLAS [184]; the shaded histogram shows the background contribution, and three other histograms show the three Υ states components. The right: The dimuon invariant mass distribution with pT < 15 GeV/c and 2 < |y| < 4.5 from LHCb [182]; the blue curve is the total ﬁtted result, and the dashed curves show the Υ(1S), Υ(2S), and Υ(3S) components.

Figure 24: The diﬀerential Υ cross sections as a function of rapidity y varying from 0.0 to 2.0 from CMS [180] (left) and 2.0 to

4.5 from LHCb [186] (right) in the transverse momentum range of pT < 30 GeV/c. In the left plot, the blue dashed line shows the normalized pythia prediction [185]. In the right plot, the thick lines show the ﬁt results with NRQCD color-octet model predictions from Refs. [187, 188] in the region 2.0 < y < 4.0, and dashed lines show the extrapolations to the full region 2.0 < y < 4.5.

25 authors to cover the range pT < 100 GeV/c. A Table 6: The products of the Υ(1S, 2S, 3S) cross sections and similar measurement [203] was performed using a dimuon branching fractions in pp collisions at 7 TeV from √ data set of 2.7 fb−1, but at a higher energy of s = CMS [180] and LHCb [182]. For the results from CMS [180], the first uncertainty is statistical, the second is systematic, 13 TeV. In order to perform the measurement in a kinematical region where muon acceptance is high, and the third is associated with the estimation of the inte- µ grated luminosity of the data sample. For the results from the requirements are applied as pT > 4.5 GeV/c for µ µ µ LHCb [182], the first uncertainty is statistical, the second is |η | < 0.3 and pT > 4.0 GeV/c for 0.3 < |η | < 1.4. systematic, and the third is due to the unknown polariza- Figure 27 shows the differential cross sections times tions of the three Υ states. dimuon branching fractions for 13 TeV CMS data, together with a comparison of the Υ(1S, 2S, 3S) dif- |y| < 2 & pT < 30 GeV/c [180] +0.61 σ(pp → Υ(1S)X)B(Υ(1S) → µ+µ−) (7.37 ± 0.13 ± 0.81) nb −0.42 ferential cross sections for the 7 and 13 TeV data +0.20 σ(pp → Υ(2S)X)B(Υ(1S) → µ+µ−) (1.90 ± 0.08 ± 0.21) nb −0.14 sets. The differential cross sections times dimuon +0.11 σ(pp → Υ(3S)X)B(Υ(1S) → µ+µ−) (1.02 ± 0.07 ± 0.11) nb −0.08 branching fractions for 13 TeV are well described 2.0 < y < 4.5 & pT < 15 GeV/c [182] +0.19 σ(pp → Υ(1S)X)B(Υ(1S) → µ+µ−) (2.29 ± 0.01 ± 0.01 ) nb −0.37 by NLO NRQCD [204, 205]. The cross sections of +0.048 σ(pp → Υ(2S)X)B(Υ(1S) → µ+µ−) (0.562 ± 0.007 ± 0.023 ) nb −0.092 +0.025 all the three Υ states at 13 TeV are factors of 2 to σ(pp → Υ(3S)X)B(Υ(1S) → µ+µ−) (0.283 ± 0.005 ± 0.012 ) nb −0.048 3 larger than the corresponding cross sections at 7 TeV, changing slowly as a function of pT. sitive. Therefore, the cross section measurement The measurements of the Υ(1S, 2S, 3S) cross sec- in high pT is of great importance for distinguishing √tions in pp collisions at a lower C.M. energy, i.e., s = 2.76 TeV were investigated by LHCb us- among the models.√ Using a pp collision data sample of 1.8 fb−1 at s = 7 TeV, ATLAS measured the ing a data sample of 3.3 pb−1 [206]. The differ- production cross sections of Υ(1S, 2S, 3S) mesons ential cross sections times dimuon branching frac- in a larger momentum region of pT < 70 GeV/c and tions as functions of pT and y were presented in in the rapidity interval of y < 2.25 [181]. Figure 25 Ref. [206]. The theoretical predictions, based on shows the measured differential cross sections of the NLO NRQCD calculation [26], provide a good Υ(1S) as a function of pT, in comparison with theo- description of the data in the higher kinematic re- retical calculations. By varying the spin-alignment gion. The products of the Υ(1S, 2S, 3S) cross sec- assumption from the nominal isotropic assump- tions and dimuon branching fractions in the regions tion for muons, the derived difference is shown by of pT < 15 GeV/c and 2.0 < y < 4.5, assuming un- the cyan bands in Fig 25. The predictions from polarised production,√ are listed in Table7. The the next-to-next-to-leading-order (NNLO) QCD- results at √s = 2.76 TeV are approximately half based calculations using the color-singlet mecha- of those at s = 7 TeV [182]. These√ results, to- nism (CSM) [189, 190, 191, 192, 193, 194] and the gether√ with the comparison between s = 7 TeV color evaporation model (CEM) [195, 196, 197, 198, and s =13 TeV from CMS [203], provide useful 199, 200] are indicated by the orange bands and information for characterizing the s-dependence of cross hatched violet bands. Apparently, the dis- Υ(1S, 2S, 3S) cross sections in pp collisions. crepancy between the predictions from CSM and data points is large if one extends the CSM calcula- Table 7: The products of the Υ(1S, 2S, 3S) cross sections tions to the higher pT region. It is probably because and dimuon branching fractions in pp collisions at 2.76 TeV CSM does not take into account the Υ productions from LHCb [206]. that arise from higher Υ or χbJ (nP ) decays, which mostly occur at modest rates. σ(pp → Υ(1S)X)B(Υ(1S) → µ+µ−) (1.111 ± 0.043 ± 0.044) nb σ(pp → Υ(2S)X)B(Υ(1S) → µ+µ−) (0.264 ± 0.023 ± 0.011) nb σ(pp → Υ(3S)X)B(Υ(1S) → µ+µ−) (0.159 ± 0.020 ± 0.007) nb CMS extended the maximum of pT to 100 GeV/c −1 to study the pT dependence√ using a 4.9 fb pp collision data sample at s = 7 TeV [201]. The Υ(1S, 2S, 3S) differential cross sections times 3.2.2. Υ(1S, 2S, 3S) polarizations dimuon branching fractions for |y| < 1.2 are shown in Fig. 26. The results are consistent with the mea- Measurements of the Υ(1S, 2S, 3S) polarizations, surements within pT < 70 GeV/c by ATLAS [181]. complementing their cross section measurements, The solid lines are the NLO NRQCD color-octet provide important information on their production calculations from Ref. [202] and extended by the mechanisms. CMS carried out these measurements 26 Figure 25: Differential cross sections multiplied by the dimuon branching fractions for the Υ(1S) production in the ranges of |y| < 1.2 (left) and 1.2 < |y| < 2.25 (right) from ATLAS [181]. Theoretical predictions are shown as ratios to the data in the lower panes for CEM (middle) and CSM (bottom), along with the variations of the cross section measurement under the spin-alignment scenarios.

Figure 26: The diﬀerential Υ(1S, 2S, 3S) cross sections times √ dimuon branching fractions as a function of pT at s = 7 Figure 27: The diﬀerential Υ(1S, 2S, 3S) cross sections times TeV in the rapidity range of |y| < 1.2 from CMS [201]. The √ solid lines are the NLO NRQCD color-octet calculations from dimuon branching fractions as a function of pT at s = 13 TeV in the rapidity range of |y| < 1.2 from CMS [203]. Ref. [202] extended by the authors to cover the range of pT < 100 GeV/c. The bands show the predictions by NLO NRQCD [204, 205]. The middle panel shows the ratios of measurements to the theory, and the lower panel shows the ratios of cross sections √ measured at s = 13 TeV to those measured at 7 TeV [201].

27 √ by studying the angular distribution of the leptons 2.2 < y < 4.5, for data collected at s = 7 and 8 produced in the Υ → µ+µ− decay [207, 208]: TeV. Similar cases are found for Υ(2S) and Υ(3S) states. No large polarization is observed. W (θ, φ) ∝ The bottomonium states are heavier and approxi- 1 2 2 (1 + λθcos θ + λφsin θcos2φ + λθφsin2θcosφ), mately non-relativistic systems, allowing the appli- 3 + λ θ cation of theoretical tools that simplify and con- where θ and φ are the polar and azimuthal angles strain the analyses of nonperturbative effects [46, of the outgoing leptons with respect to the quan- 47]. However, the corresponding predictions [217] tization axis of the chosen polarization frame; λ is of strong transverse polarizations are in stark con- the set of polarization parameters; the parameter trast with the negligible polarizations observed by λθ is 0 (1) for fully longitudinal (transverse) polar- the experiments [212, 213, 214, 215]. Accordingly, ization. The polarization parameters depend on the the measurements of Υ polarization have led to new reference frame in which they are measured. The theoretical interpretations for the quarkonium puz- three most commonly used reference frames are the zle. For instance, in Ref. [218], global fits using helicity (HX) [209], Collins-Soper (CS) [210], and both cross section and polarization measurements perpendicular helicity (PX) [211] frames. In the are performed to determine the nonperturbative pa- HX frame, the z axis is defined as the direction of rameters of bound-state formation. This study re- the Υ momentum in the C.M. frame of the colliding veals unexpected hierarchies in the phenomenolog- protons. In the CS frame, the z axis is defined such ical long-distance parameters, which brings a new that it bisects the angle between ~p1 and −~p2 in the understanding of the bound-state formation mech- rest frame of the Υ meson, where ~p1 and ~p2 are the anism in QCD. three-momenta of the colliding protons in the rest frame of the Υ meson. The PX frame is orthogonal to the CS frame. The y axis is always taken along 3.3. Sequential Υ suppression in heavy ion colli- the direction of the vector product of the two beam sions directions in the Υ rest frame. The x axis is defined to complete a right-handed coordinate system. 3.3.1. Υ suppression in P bP b collisions In the previous measurements from CDF [212] The first indication of Υ suppression√ in heavy and D0 [213], only the λθ parameter in a single po- ion collisions was reported by CMS at s = 2.76 larization frame was extracted. CMS measured all TeV [219]. The integrated luminosity used in this the polarization parameters λθ, λφ, and λθφ, in all measurement corresponds to 7.28 µb−1 for P bP b the three polarization frames mentioned above, plus −1 ˜ and 225 nb for pp collisions, the latter corre- the frame-invariant quantity λ = (λθ + 3λφ)/(1 − −1 √ sponding approximately to the equivalent nucleon- λφ), using a pp data sample of 4.9 fb at s = 7 nucleon luminosity of the P bP b run. Thanks to the TeV [214]. As an example, Fig. 28 shows the λθ, good momentum resolution of the CMS detector, λφ, and λθφ measurements as functions of pT in the the three Υ resonances in the dimuon mass spec- rapidity range of |y| < 0.6 in the HX frame. All po- trum can be well resolved. The same reconstruc- larization parameters are compatible with zero or tion algorithm and analysis criteria are applied to have small values in the three polarization frames. both data sets. The results are shown in Fig. 30, The polarizations of the√ Υ(1S, 2S, 3S) mesons where the three Υ peaks are clearly observed in the produced in pp collisions at s = 7 and 8 TeV in pp case, but Υ(2S) and Υ(3S) are not significant in the complementary rapidity region of 2.2 < y < the P bP b case. The suppression effects for Υ(2S) 4.5 were measured by LHCb using 1 and 2 fb−1 and Υ(3S) in P bP b collisions are more obvious than data samples, respectively [215]. The measure- that for Υ(1S). The ratio on the production rates ments were performed in three polarization frames: of Υ(2S + 3S) and Υ(1S) in P bP b and pp collisions HX, CS, and Gottfried-Jackson (GJ) [216] frames. can be derived according to the Υ(2S + 3S) and In the GJ frame, the z axis is defined as the direc- Υ(1S) signal yields directly since the acceptance tion of ~p1 in the rest frame of the Υ meson. Fig- and efficiency differences among the reconstructed ure 29 shows the polarization parameters λθ, λθφ, resonances are cancelled. An extended unbinned and λφ measured in the HX frame for the Υ(1S) maximum likelihood fit to the two invariant mass state in different bins of pT and rapidity region of distributions of Fig. 30 gives the double ratio of the 28 Figure 28: Values of the λθ (top), λφ (middle), and λθφ (bottom) parameters for the Υ(1S) (left), Υ(2S) (middle), and Υ(3S) (right) resonances, in the HX frame, as a function of pT in the rapidity range |y| < 0.6 from CMS [214]. observed signal yields and N obs(Υ(2S + 3S))/N obs(Υ(1S))| N obs(Υ(3S))/N obs(Υ(1S))| P bP b P bP b = obs obs = obs obs N (Υ(2S + 3S))/N (Υ(1S))|pp N (Υ(3S))/N (Υ(1S))|pp +0.19 0.31−0.15 ± 0.03, 0.06 ± 0.06 ± 0.06 (< 0.17 at 95% C.L.). µ µ in the kinematic region of pT > 4 GeV and |η | The measured values are considerably smaller than < 2.4. The statistical significance of the effect was unity. The significance of the observed suppression evaluated to be 2.4σ using an ensemble of 1 × 106 exceeds 5σ. pseudoexperiments. In addition, the absolute suppressions of all three In 2011, CMS collected√ a number of pp and P bP b individual Υ states are also studied using the nu- data samples at s = 2.76 TeV to measure the clear modification factor, RAA, defined as the yield sequential Υ suppression [220]. The total integrated per nucleon-nucleon collision in P bP b relative to −1 luminosity of the data samples is 150 µb , which that in pp. The RAA observable, is approximately 20 times larger than those used obs in Ref. [219]. The resultant dimuon invariant mass Lpp N (Υ(nS))|P bP b εpp RAA = obs , spectra are shown in Fig. 31 for the pp and P bP b TAANMB N (Υ(nS))|pp εP bP b datasets. The three Υ peaks are clearly observed is evaluated from the ratio of total Υ(nS) yields in in the pp case; the Υ(3S) peak is not prominent in P bP b and pp collisions corrected for the difference the P bP b case. The simultaneous fit to the dimuon in efficiencies ε and ε , with the average nu- invariant mass spectra in the P bP b and pp datasets pp P bP b clear overlap function T , number of minimum- gives the double ratios of the observed signal yields AA bias events sampled by the event selection NMB, obs obs N (Υ(2S))/N (Υ(1S))|P bP b and integrated luminosity of the pp dataset Lpp ac- obs obs = 0.21±0.07±0.02 N (Υ(2S))/N (Υ(1S))|pp counting for the normalization. The values of the 29 Figure 29: The polarization parameters λ (top), λ (middle), and λ (bottom) measured in the HX, CS, and GJ frames for θ θφ φ √ the Υ(1S) state in different bins of pT and rapidity region of 2.2 < y < 4.5, for data collected at s = 7 (left) and 8 TeV (right) from LHCb [215].

Figure 30: Dimuon invariant mass distributions obtained with the (a) 225 nb−1 pp and (b) 7.28 µb−1 P bP b data samples at √ s = 2.76 TeV from CMS [219]. The solid lines show the best ﬁts to the data.

30 RAA are determined to be agreement with the Υ(2S) double ratio results presented in Fig. 33. In this model, the dynamical evo- RAA(Υ(1S)) = 0.56 ± 0.08 ± 0.07, lution is treated using anisotropic hydrodynamics, where the relevant initial conditions are changed by RAA(Υ(2S)) = 0.12 ± 0.04 ± 0.02, varying the viscosity to entropy ratio, η/s, and the initial momentum-space anisotropy. Another theo- and retical curve from Du et al. in Ref. [224], based on RAA(Υ(3S)) < 0.10 (95% C.L.). a kinetic-rate equation approach and containing a small component of regenerated bottomonia, shows Centrality is an important parameter for the a similar level of agreement with the data. No sig- QGP formation since it is directly related to the nificant variations with pT (middle) or |y| (right) overlap region of the colliding nuclei. The exten- in Fig. 33 are observed. Predictions of Υ suppres- sive forward calorimetry of the CMS apparatus pro- sion as functions of pT [223, 224] and |y| [223] are vides a good chance for the centrality determina- in overall agreement with the data. √ √ tion in P bP b collisions. The event centrality ob- The Υ suppressions at s = 2.76 TeV and s = servable corresponds to the fraction of the total in- 5.02 TeV were also explored by ALICE in a com- elastic cross section, starting at 0 for the most cen- plementary rapidity range of 2.5 < y < 4 [183, 225]. tral collisions. The centrality classes used in this The integrated√ luminosities of√ the data samples study are (50 − 100)%, (40 − 50)%, (30 − 40)%, used for s = 2.76 TeV and s = 5.02 TeV are (20−30)%, (10−20)%, (5−10)%, and (0−5)%, or- 68.8 µb−1 and 225 µb−1, respectively. The ob- dered from the lowest to the highest energy deposit. tained values of R for Υ(1S) as a function of Using a Glauber-model calculation as described in AA centrality, pT, and y are shown in Fig. 34. AL- Ref. [221], the centrality variable can be expressed ICE has no ability to measure the cross sections by the average number of nucleons participating in of Υ in pp collisions. Therefore, they applied an the collisions (Npart). The double ratio of the ob- obs obs interpolation method utilizing the LHCb measure- N (Υ(2S))/N (Υ(1S))|P bP b served signal yields obs obs and N (Υ(2S))/N (Υ(1S))|pp ments in Ref. [206]. From Fig. 34, RAA decreases RAA for Υ(1S) and Υ(2S) as a function of Npart with increasing centrality, and no significant de- are shown in Fig. 32[220]. From Fig. 32 (left), the pendencies on pT and y are seen. The suppres- double ratio dependence on centrality is not pro- sions√ of the Υ(1S) state on the centrality and y nounced. While in Fig. 32 (right), the Υ(1S) and at s = 5.02√ TeV are slightly weaker than those Υ(2S) suppressions are observed to increase with measured at s = 2.76 TeV, which is not consis- collision centrality. These results indicate a signif- tent with our expectations since the bound state icant suppression of the Υ states in heavy-ion col- is more likely to dissolve at higher temperatures. lisions compared to pp collisions at the same per- Two transport models (TM1 and TM2) [224, 226] nucleon-pair energy. The Υ(1S) is the least sup- and one hydro-dynamical model [227] are compared pressed and the Υ(3S) is the most suppressed of with data points. Though TM1 and TM2 use dif- the three states, which supports the hypothesis of ferent rate equations, both of them take account of increased suppression of less strongly bound states. the feed-down contributions from higher-mass bottomonia to Υ(1S). Deviation between TM1 (TM2) The Υ suppression in P bP b collisions at √ and data is 2σ (1.4σ). The trends of R as a func- a higher C.M. energy, i.e. s = 5.02 TeV, AA tion of centrality, p , and y can be described by the was studied by CMS [222]. The double ra- T hydro-dynamical model to some extent. tio [Υ(nS)/Υ(1S)]P bP b/[Υ(nS)/Υ(1S)]pp was measured to be 0.308±0.055±0.019 for Υ(2S) and less The elliptic flow coefficient, denoted by ν2 [228], than 0.26 at 95% C.L. for Υ(3S). The double ra- reflects the response of the QGP medium to the ini- tio for Υ(2S) was studied as a function of collision tial anisotropy in non-central P bP b collisions. Us-√ centrality, as well as the Υ transverse momentum ing a P bP b collision data sample of 750 µb−1 at s and rapidity, as shown in Fig. 33. Predictions of Υ = 5.02 TeV, ALICE performed the first measure- suppression from Krouppa and Strickland [223], in- ment of the ν2 coefficient of Υ(1S) in the forward corporating color-screening effects on the bottomo- rapidity region of 2.5 < y < 4 [229]. The Υ mesons nium family and reflecting feed-down contributions are reconstructed via their µ+µ− decay channel. from decays of heavy quarkonia, are in overall The dimuon ν2 coefficient was determined using 31 √ Figure 31: Dimuon invariant mass distributions from the (a) 150 µb−1 P bP b and (b) 230 nb−1 pp data samples at s = 2.76 TeV from CMS [220]. The solid lines show the best fits to the data, while the dashed lines show the backgrounds.

Figure 32: Centrality dependence of the double ratio (left) and of RAA (right) for the Υ(1S) and Υ(2S) states from CMS [220].

Figure 33: The double ratios for Υ(2S) as a function of the collision centrality, transverse momentum, and rapidity from CMS [222].

32 √ √ Figure 34: The RAA factor of Υ(1S) as a function of centrality (top), pT (left) and y (right) at s = 2.76 TeV [183] and s = 5.02 TeV [225] from ALICE. the scalar product method [230, 231], correlating the reconstructed dimuons with the second-order harmonic event flow vector [228, 232]. Figure 35 shows the graph of ν2(Mµµ) as well as the fitted result, where a negative yield around Υ(1S) mass is observed. Here, only 5%−60% centrality interval is considered since the eccentricity of the initial collision geometry is small for the central collision (0%−5%), and the signal yield is low for the periph- eral collision (60%−100%). Figure 36 shows the ν2 coefficient of Υ(1S) as a function of transverse momentum. The results are compatible with 0 and consistent with the small positive values predicted by the available theoretical models within uncertainties. The values of ν2 for Υ(1S) are lower than those for J/ψ by 2.6 standard deviations, which indicates the limited dissociation for Υ(1S) in the early P bP b collision stage, and an additional regen- Figure 35: The graph of ν2 with respect to Mµµ in the ranges eration component needed for J/ψ. of 2.5 < y < 4 and 2 < pT < 15 GeV/c from ALICE [229]. The red solid line shows the fitted result, and the red dashed line shows the fitted backgrounds. 3.3.2. Υ suppression in pP b collisions In addition to hot-nuclear-matter (HNM) effects, regarding the QGP formation, the differences in quarkonium production between P bP b and pp 33 below 1, while the value of RAA for Υ(1S) in the backward region from LHCb is above 1. The blue band in Fig. 38 (left) shows the predicted results based on CSM at LO with EPS09 corrected parameter [239]. The calculation based on CEM at LO with EPS09 [240] is shown by the blue band in Fig. 38 (right). The parton energy loss calculations with and without EPS09 at NLO [241] are shown by green and red bands in Fig. 38 (right), respectively. All of the above theoretical predictions [239, 240, 241] favor the LHCb result that the value of RAA in the backward region is at least 1. The nuclear modification factors measured from LHCb also deliver information that the suppression for Υ(1S) is smaller than that for J/ψ.

Figure 36: The ν2 coefficient of Υ(1S) as a function of transverse momentum in the rapidity region of 2.5 < y < 4, compared to that of inclusive J/ψ from ALICE [229]. collisions can also arise from cold-nuclear-matter (CNM) effects [233]. Therefore, it is necessary to study Υ(1S, 2S, 3S) productions in pP b collisions in order to study the CNM effect separately. This knowledge can be extrapolated into P bP b collisions so that the fraction of the suppression due to HNM in P bP b collisions can be determined [234]. √ The Υ suppressions in pP b collisions at s = 5.02 TeV were investigated by CMS [235], LHCb [236], and ALICE [237], using data samples of 31 nb−1, −1 −1 Figure 37: Event activity integrated double ratios of the ex- 1.6 nb , and 10.82 nb , respectively. The studied cited states, Υ(2S) and Υ(3S), to the ground state, Υ(1S), √ rapidity intervals in CMS, LHCb, and ALICE are in pP b collisions at s = 5.02 TeV with respect to pp colli- √ |y| < 1.93, −5 < y < −2.5 (backward) && 1.5 < sions at s = 2.76 TeV from CMS [235], compared to their √ y < 4.0 (forward), and −4.46 < y < −2.96 (back- counterparts for P bP b collisions at s = 2.76 TeV from ward) && 2.03 < y < 3.53 (forward), respectively.√ CMS [220]. By utilizing the results in pp collisions at s = 2.76 TeV, CMS gave the pP b double ratios, which The Υ suppressions√ in pP b collisions at a higher are shown in Fig. 37, together√ with their counter- C.M. energy, i.e. s = 8.16 TeV were studied by parts in P bP b collisions at s = 2.76 TeV for com- LHCb [242] and ALICE [243]. Particularly, the parison. Clearly, the pP b ratios are much larger Υ suppression performances with increasing trans- than the corresponding P bP b ratios [220]. The nu- verse momentum pT were intensively investigated. clear modification RAA as a function of rapidity y Figure 39 shows the nuclear modification factor was determined by LHCb [236] and ALICE [237] in RAA for Υ(1S) and Υ(2S) as a function of pT in both of their forward and backward regions, where both the forward and backward regions. From the the input cross sections of Υ in pp collisions were figure, the values of RAA increase with pT. The obtained by a power-law interpolation from LHCb similar trends were also found in the RAA measure- measurements [182, 206, 238]. The resulting RAA ments from ALICE [243], as shown in Fig. 40. The distributions are shown in Fig. 38 with the calcu- predictions with EPPS16 [244, 245, 246, 247] and lations from several theoretical models. The values nCTEQ15 [248, 249, 250], as two different nuclear of RAA for Υ(1S) in the forward region from LHCb parton distribution functions, are used to compare and forward and backward regions from ALICE are the experimental results. The calculations based 34 Figure 38: The left: the nuclear modication factors for Υ(1S) (black dots), prompt J/ψ (red squares), and J/ψ from b-hadron √ decays (blue triangles) in pP b collisions at s = 5.02 TeV as a function of rapidity from LHCb [236]. The right: the nuclear √ modication factors for Υ(1S) in pP b collisions at s = 5.02 TeV as a function of rapidity from ALICE [237]. on EPS09 NLO [251, 252] are also used to compare light hadrons, pions, kaons, and protons in both the ALICE result, as shown by the red bands in Υ(1S) and Υ(2S) decays. The QΥ values measured Fig. 40. A good agreement between experimental by Belle are close to the “80% rule” line. However, data and theoretical predictions is found in the for- most of QΥ values from the measurements using ward region for both Υ(1S) and Υ(2S). While in CLEO data in Ref. [51] are below the “80% rule” the backward region, the values of RAA in data are line. The discrepancy is expected to be resolved by mostly less than 1, different from theoretical pre- using a large number of Υ(1S, 2S) events collected dictions. Overall, the suppression effects are seen at Belle II. in pP b collisions, but much smaller than those in P bP b collisions, which indicates the presence of the Abundant events of Υ(1S) → J/ψ + anything CNM effect. Such contributions can make small were seen by CLEO and Belle, and the distribu- corrections to the complete HNM suppression ef- tion of their momenta favors the expectation of the fects in P bP b collisions. color-singlet process. So far, no double charmonium process is evident except for Υ(1S) → J/ψ + χc1 with a significance of 4.6σ. This needs to be studied further at Belle II, in order to confirm the re- 4. Summary and prospects lated calculations with the NRQCD factorization approach. 4.1. Summary at e+e− colliders and prospects at The radiative decays of Υ(1S, 2S, 3S) to char- Belle II monia and bottomonia were systematically studied by CLEO, BaBar, and Belle. The first radia- Although the B-factories stopped taking data tive decay of the Υ(1S) into a charmonium state, more than ten years ago, the Belle II as a next- i.e., Υ(1S) → γχ was observed by Belle using a generation B-factory experiment came back in c1 Υ(1S) tagging technique. The branching fraction of March 2019. Datasets of Υ(1S, 2S, 3S) collected at Υ(1S) → γχ was obtained with large uncertain- Belle II are expected to be about 200 fb−1 for each c1 ties, and is slightly higher than the previous upper in the schedule [30], which offers a better chance to limit and much higher than the theoretical expecta- study Υ(1S, 2S, 3S) decays further. tion from NRQCD. At Belle II, the precise measure- The measurements of exclusive Υ(1S, 2S) decays ment of Υ(1S) → γχc1 and the searches for more into light hadrons were performed by Belle and au- Υ(1S, 2S, 3S) radiative decays to a charmonium, in- thors using CLEO data. From Belle, the evidences cluding Υ(1S, 2S, 3S) → γχc0/χc2/ηc(1S)/ηc(2S), ∗ 0 ¯ ∗ 0 + − were found in K (892) K2 (1430) , φK K , are expected to be done. Among the photon transi- ∗ 0 − + + − 0 0 0 + − K (892) K π , π π π π , and KSK π final tions of Υ(2S, 3S) to χbJ , the branching fractions of states in both Υ(1S) and Υ(2S) decays; from the Υ(3S) → γχbJ (1P ) decrease by one order of magni- measurements in Ref. [51], significance strengths tude compared to those of Υ(2S) → γχbJ (1P ) and were found in 17 decay modes consisting of 4 – 10 Υ(3S) → γχbJ (2P ). The ηb(1S) resonance was ob- 35 Figure 39: The nuclear modification factors for Υ(1S) (above) and Υ(2S) (below) in pP b collisions as a function of pT in the forward (left) and backward (right) regions from LHCb [242].

Figure 40: The nuclear modiﬁcation factors for Υ(1S) in pP b collisions as a function of pT in the backward (left) and forward (right) regions from ALICE [243].

36 served in the Υ(2S, 3S) radiative decays by Belle The transitions between Υ(mS) and Υ(nS) and CLEO in the inclusive photon energy spectra. (m > n) were intensively investigated in e+e− However, the observations of ηb(1S) and ηb(2S) in collision experiments. The ratio B(Υ(2S) → the exclusive hadronic final states of the Υ(1S) and ηΥ(1S))/B(Υ(2S) → π+π−Υ(1S)) is slightly below Υ(2S) radiative decays are much debated. the value predicted by QCDME, while B(Υ(4S) → ηΥ(1S))/B(Υ(4S) → π+π−Υ(1S)) strongly disfa- The charmonium-like states, exotic glueballs, vors the prediction from QCDME, which may im- light tetraquarks, and a stable six-quark state were ply additional implementations for QCDME are searched for in Υ(1S, 2S, 3S) inclusive or radia- needed. The B(Υ(4S) → η0Υ(1S))/B(Υ(4S) → tive decays. Unfortunately, no clear signals were ηΥ(1S)) was measured to be (0.20 ± 0.06), which is observed in all the studied modes, and only the in agreement with the expected value in the case of 90% C.L. upper limits on the production rates an admixture of a state containing light quarks in were determined. Nonetheless, searching for ex- addition to the b¯b pair for Υ(4S). The production otic states is still an important topic in hadronic rates of Υ(5S) → π+π−Υ(1S, 2S, 3S) exceed those physics. At Belle II, special attention should be of the dipion transitions between lower resonances paid to Υ(1S) → f1(1285)+G0−− to check whether by more than two orders of magnitude, which indi- 2 there is a peak around 3.92 GeV/c . The other cates a new mechanism to enhance the decay rates. exotic glueballs with J PC = 0+−, 1−+, and 2+− The latter observations of Zb(10610) and Zb(10650) predicted in Ref. [253] will also be searched for at decaying into πΥ(1S), πΥ(2S) and πΥ(3S) are sup- Belle II. Related studies with the radiative decays posed to contribute such enhancement in the dipion of Υ(1S, 2S, 3S) will be conducted further at Belle transitions of Υ(5S). II. The CLEO, Belle, and BaBar experiments studied the Υ(1S) → invisible decay to search for low 4.2. Summary in pp, pP b, and P bP b collisions and mass DM. No significant signal was observed in all outlook at the LHC these experiments. The most stringent upper limit was set by BaBar; it is B(Υ(1S) → invisible) < Measurements of the Υ(1S, 2S, 3S) cross sections 3.0 × 10−4 at 90% C.L. With a tagging method, in pp collisions at the unprecedented C.M. energies Belle and BaBar also set the upper limits at 90% of 2.76, 5.02, 7, 8, and 13 TeV have been under- C.L. on B(Υ(1S) → γA0)B(A0 → invisible) and taken, within the rapidity window of −2.0 < y < −5 B(Υ(1S) → γχχ) to a few times 10 and a few 4.5 and the dimuon momentum range of pT < 100 times 10−4, respectively. At Belle II, one can ex- GeV/c. In addition, the angular distribution of the pect a sensitivity of 1.3 × 10−5 at 90% C.L. for muons produced in the Υ(1S, 2S, 3S) decays has B(Υ(1S) → invisible) which is comparable to the been analyzed in different reference frames to deter- SM prediction B(Υ(1S) → νν¯) = 1.0 × 10−5 [21]. mine the polarization parameters. From the results As to the search for Υ(1S) → γ + invisible, Belle of CMS and LHCb, Υ(1S, 2S, 3S) polarization pa- II has the possibility to discover an excess of events rameters are all so close to zero that polarizations at 90% C.L. if B(Υ(1S) → γA0)B(A0 → invisible) can be neglected. The measurements of these cross > 5 × 10−7 and B(Υ(1S) → γχχ) > 5 × 10−6. sections and polarizations have shed light on the Υ(1S, 2S, 3S) production mechanisms in pp colli- Leptonic decays of Υ(1S, 2S, 3S) can be used sions. Analysis of new LHC data will extend the to test LFU. To date, the measured values of reach of the kinematics to test the NRQCD with B(Υ(1S, 2S) → `+`−)(` = e, µ, τ) are consis- higher-order corrections which becomes more sen- tent with each other, but with large uncertainties sitive with the increase of p . Additional studies of a few percent. The LFV processes have not been T of other charmonia, bottomonia, and hadron jets observed, and the 95% C.L. upper limits on the are expected to be conducted at the LHC to gain branching fractions are determined to be at a level a more complete understanding of the mechanisms of 10−7 by CLEO and BaBar. Belle II experiment is of hadron productions. expected to achieve better control of statistical and systematic effects, allowing improved determina- The QGP formation is a hot topic in modern par- tions of B(Υ(1S) → `+`−) and B(Υ(1S) → `+`0−), ticle physics. One of its most striking characteris- where `+ and `0− are different types of leptons. tics is the suppression of quarkonium states. With 37 the increased collision energy and detector capa- [7] S. Okubo, φ meson and unitary symmetry model, Phys. bility, CMS, LHCb, and ALICE have established Lett. 5, 165 (1963). the pattern of sequential Υ(1S, 2S, 3S) suppression [8] G. Zweig, CERN Report Nos. Th 401 and 412, 1964. in collisions involving heavy ions, where the ex- [9] K. Okada and O. Shito, Systematics and phenomenol- cited states are more suppressed than the ground ogy of boson mass levels. 3., Prog. Theor. Phys. 35, state. Besides, they proved that the Υ(1S, 2S, 3S) 1061 (1966). suppression increases with the collision centrality [10] J. Iizuka, Systematics and phenomenology of meson and energy, and does not change along with the family, Prog. Theor. Phys. Suppl. 37, 21 (1966). transverse momentum or rapidity. In addition to [11] E. Eichten et al., The Spectrum of Charmonium, Phys. P bP b collisions, the Υ(1S, 2S, 3S) suppression was Rev. Lett. 34, 369 (1975). [12] W. Buchmüllerand S. H. H. Tye, Quarkonia and Quan- also studied in pP b collisions by CMS, LHCb, and tum Chromodynamics, Phys. Rev. D 24, 132 (1981). ALICE to distinguish the non-HNM effects in the [13] S. N. Gupta, S. F. Radford, and W. W. Repko, Semirel- QGP formation. The slight Υ(1S, 2S, 3S) suppres- ativistic Potential Model for Heavy Quarkonia, Phys. sion observed in pP b collisions implies that both Rev. D 34, 201 (1986). HNM and CNM can influence the QGP formation. [14] T. Liu, Z. Chen, and T. Huang, A Study of a Possible In the future, the large datasets collected at the Unified Potential Model, Z. Phys. C 46, 133 (1990). LHC will be used to explore the sequential suppres- [15] T. Barnes, S. Godfrey, and E. S. Swanson, Higher char- sion of P -wave bottomonia. New observables, such monia, Phys. Rev. D 72, 054026 (2005). as azimuthal anisotropies and polarizations, will be [16] S. F. Radford and W. W. Repko, Potential model cal- also studied via angular analyses to figure out the culations and predictions for heavy quarkonium, Phys. properties of HNM in hadron collisions. Rev. D 75, 074031 (2007). [17] N. Brambilla et al., Effective Field Theories for Heavy Acknowledgements Quarkonium, Rev. Mod. Phys. 77, 1423 (2005). [18] G. T. Bodwin, E. Braaten, and G. P. Lepage, Rigorous We would like to thank Prof. Zhen Hu for fruit- QCD analysis of inclusive annihilation and production of heavy quarkonium, Phys. Rev. D 51, 1125 (1995). ful discussions. This work is supported by the [19] M. Okamoto et al. (CP-PACS Collaboration), Charmo- National Natural Science Foundation of China un- nium spectrum from quenched anisotropic lattice QCD, der Grants No. 11575017, No. 11761141009, No. Phys. Rev. D 65, 094508 (2002). 11975076, and No. 11661141008; National Key [20] W. M. Serenone, Heavy-quarkonium potential with in- R&D Program of China under the contract No. put from lattice gauge theory, arXiv:1408.3003 (2014). 2018YFA0403902; and the CAS Center for Excel- [21] L. N. Chang, O. Lebedev, and J. N. Ng, On the invisible lence in Particle Physics (CCEPP). decays of the Υ and J/ψ resonances, Phys. Lett. B 441, 419 (1998). [22] B. McElrath, Invisible quarkonium decays as a sensitive References probe of dark matter, Phys. Rev. D 72, 103508 (2005). [23] U. Ellwanger, C. Hugonie, and A. M. Teixeira, [1] S. W. Herb et al., Observation of a Dimuon Reso- The Next-to-Minimal Supersymmetric Standard Model, nance at 9.5 GeV in 400 GeV Proton-Nucleus Colli- Phys. Rep. 496, 1 (2010). sions, Phys. Rev. Lett. 39, 252 (1977). [24] P. Artoisenet et al.,Υ Production at Fermilab Teva- [2] W. R. Innes et al., Observation of Structure in the Υ tron and LHC Energies, Phys. Rev. Lett. 101, 152001 Region, Phys. Rev. Lett. 39, 1240 (1977). (2008). √ [3] C. Berger et al., Observation of a Narrow Resonance [25] J. P. Lansberg, J/ψ production at s = 1.96 and 7 + − Formed in e e Annihilation at 9.46 GeV, Phys. Lett. TeV: Color-Singlet Model, NNLO* and polarisation, J. B 76, 243 (1978). Phys. G 38, 124110 (2011). [4] C. W. Darden et al., Observation of a Narrow Reso- [26] K. Wang, Y. Q. Ma, and K. T. Chao, Υ(1S) prompt nance at 9.46 GeV in electron-Positron Annihilations, production at the Tevatron and LHC in nonrelativistic Phys. Lett. B 76, 246 (1978). QCD, Phys. Rev. D 85, 114003 (2012). [5] J. K. Bienlein et al., Observation of a Narrow Reso- [27] A. D. Frawley, T. Ullrich, and R. Vogt, Heavy flavor in + − nance at 10.02 GeV in e e Annihilations, Phys. Lett. heavy-ion collisions at RHIC and RHIC II, Phys. Rep. B 78, 360 (1978). 462, 125 (2008). [6] P. A. Zyla et al. (Particle Data Group), Review of parti- [28] T. Matsui and H. Satz, J/ψ Suppression by Quark- cle physics, to be published in Prog. Theor. Exp. Phys. Gluon Plasma Formation, Phys. Lett. B 178, 416 2020, 083C01 (2020). 38 (1986). Theories: Challenges and Perspectives, Eur. Phys. J. C [29] S. Digal, P. Petreczky, and H. Satz, Quarkonium feed 74, 2981 (2014). down and sequential suppression, Phys. Rev. D 64, [48] Y. F. Gu and X. H. Li, Ratio of Hadronic Decay Rates 094015 (2001). of J/ψ and ψ(2S) and ρπ puzzle, Phys. Rev. D 63, [30] E. Kou et al. (Belle II Collaboration), The Belle II 114019 (2001). Physics Book, Prog. Theor. Exp. Phys. 2019, 123C01 [49] C. P. Shen et al. (Belle Collaboration), First observa- (2019). tion of exclusive Υ(1S) and Υ(2S) decays into light [31] S. J. Brodsky, G. P. Lepage, and P. B. Mackenzie, hadrons, Phys. Rev. D 86, 031102 (2012). On the Elimination of Scale Ambiguities in Perturba- [50] C. P. Shen et al. (Belle Collaboration), Measurement tive Quantum Chromodynamics, Phys. Rev. D 28, 228 of exclusive Υ(1S) and Υ(2S) decays into Vector- (1983). Pseudoscalar final states, Phys. Rev. D 88, 011102 [32] S. E. Csorna et al. (CLEO Collaboration), A Measure- (2013). ment of the Direct Photon Spectrum From the Υ(1S), [51] S. Dobbs, Z. Metreveli, A. Tomaradze, T. Xiao, and Ka- Phys. Rev. Lett. 56, 1222 (1986). mal K. Seth, First Measurements of Exclusive Hadronic [33] B. Nemati et al. (CLEO Collaboration), Measurement Decays of Υ(1S) and Υ(2S), Phys. Rev. D 86, 052003 of the direct photon spectrum in Υ(1S) decays, Phys. (2012). Rev. D 55, 5273 (1997). [52] S. Li, Q. Xie, and Q. Wang, Contribution of color sin- [34] D. Besson et al. (CLEO Collaboration), Measurement glet process Υ → J/ψ + ccg¯ and Υ(1S) → J/ψ + X, of the direct photon momentum spectrum in Υ(1S), Phys. Lett. B 482, 65 (2000). Υ(2S), and Υ(3S) decays, Phys. Rev. D 74, 012003 [53] K. Cheung, W. Keung, and T. Yuan, Color octet J/ψ (2006). production in the Υ decay, Phys. Rev. D 54, 929 (1996). [35] H. Albrecht et al. (ARGUS Collaboration), Determina- [54] M. Napsuciale, Inclusive J/ψ production in Υ decay via tion of α − s From a Measurement of the Direct Pho- color octet mechanisms, Phys. Rev. D 57, 5711 (1998). ton Spectrum in Υ(1S) Decays, Phys. Lett. B 199, 291 [55] H. Trottier, Υ decay into charmonium and the color (1987). octet mechanism, Phys. Lett. B 320, 145 (1994). [36] A. Bizzeti et al. (Crystal Ball Collaboration), Measure- [56] R. A. Briere et al. (CLEO Collaboration), New mea- ment of the direct photon spectrum from Υ(1S) decays, surements of Υ(1S) decays to charmonium final states, Phys. Lett. B 267, 286 (1991). Phys. Rev. D 70, 072001 (2004). [37] TorbjörnSjöstrand, PYTHIA 5.7 and JETSET 7.4: [57] C. P. Shen et al. (Belle Collaboration), Search for XYZ Physics and manual, arXiv:hep-ph/9508391 (1994). states in Υ(1S) inclusive decays, Phys. Rev. D 93, [38] X. Garcia i Tormo and J. Soto, Soft, collinear and 112013 (2016). nonrelativistic modes in radiative decays of very heavy [58] S. Jia et al. (Belle Collaboration), Search for the 0−− quarkonium, Phys. Rev. D 69, 114006 (2004). Glueball in Υ(1S) and Υ(2S) decays, Phys. Rev. D 95, [39] X. Garcia i Tormo and J. Soto, Semi-inclusive radiative 012001 (2017). decays of Υ(1S), Phys. Rev. D 72, 054014 (2005). [59] K. Abe et al. (Belle Collaboration), Observation of dou- √ [40] X. Garcia i Tormo and J. Soto, Radiative decays and the ble cc¯ production in e+e− annihilation at s approxi- nature of heavy quarkonia, Phys. Rev. Lett. 96, 111801 mately 10.6 GeV, Phys. Rev. Lett. 89, 142001 (2002). (2006). [60] K. Abe et al. (Belle Collaboration), Study of double √ [41] R. D. Field, Radiative decays and the nature of heavy charmonium production in e+e− annihilation at s ∼ quarkonia, Phys. Lett. B 133, 248 (1983). 10.6 GeV, Phys. Rev. D 70, 071102 (2004). [42] T. Appelquist and H. D. Politzer, Orthocharmonium [61] B. Aubert et al. (BaBar Collaboration), Measurement and e+e− Annihilation, Phys. Rev. Lett. 34, 43 (1975). of double charmonium production in e+e− annihila- √ [43] A. De Rújulaand S. L. Glashow, Orthocharmonium and tions at s = 10.6 GeV, Phys. Rev. D 72, 031101 e+e− Annihilation, Phys. Rev. Lett. 34, 46 (1975). (2005). [44] M. E. B. Franklin et al., Measurement of ψ(3097) and [62] E. Braaten and J. Lee, Exclusive Double Charmonium ψ0(3686) Decays Into Selected Hadronic Modes, Phys. Production from e+e− Annihilation into a Virtual Pho- Rev. Lett. 51, 963 (1983). ton, Phys. Rev. D 67, 054007 (2003); Phys. Rev. D 72, [45] J. Z. Bai et al. (BES Collaboration), Measurements of 099901(E) (2005). ψ(2S) decays into vector tensor final states, Phys. Rev. [63] K. Y. Liu, Z. G. He, and K. T. Chao, Problems of double √ D 69, 072001 (2004). charm production in e+e− annihilation at s = 10.6 [46] N. Brambilla et al., Heavy Quarkonium: Progress, Puz- GeV, Phys. Lett. B 557, 45 (2003). zles, and Opportunities, Eur. Phys. J. C 71, 1534 [64] J. P. Ma and Z. G. Si, Predictions for e+e− →

(2011). J/ψηc with light-cone wave-functions, Phys. Rev. D 70, [47] N. Brambilla et al., QCD and Strongly Coupled Gauge 074007 (2004).

39 [65] K. Y. Liu, Z. G. He, and K. T. Chao, Inclusive char- heavy quarkonium magnetic dipole transitions in po- monium production via double cc¯ in e+e− annihilation, tential nonrelativistic QCD, Phys. Rev. D 87, 074024 Phys. Rev. D 69, 094027 (2004). (2013). [66] A. E. Bondar and V. L. Chernyak, Is the BELLE result [84] N. Brambilla, Y. Jia, and A. Vairo, Model-independent + − for the cross section σ(e e → J/ψηc) a real diﬃculty study of magnetic dipole transitions in quarkonium, for QCD?, Phys. Lett. B 612, 215 (2005). Phys. Rev. D 73, 054005 (2016). [67] K. Y. Liu, Z. G. He, and K. T. Chao, Search for excited [85] J. Segovia, S. Steinbeißer, and A. Vairo, Electric dipole √ charmonium states in e+e− annihilation at s = 10.6 transitions of 1P bottomonia, Phys. Rev. D 99, 074011 GeV, Phys. Rev. D 77, 014002 (2008). (2019). [68] Y. J. Zhang, Y. J. Gao, and K. T. Chao, Next-to-leading [86] M. Artuso et al. (CLEO Collaboration), Photon transi- + − √ order QCD correction to e e → J/ψηc at s = 10.6 tions in Υ(2S) and Υ(3S) decays, Phys. Rev. Lett. 94, GeV, Phys. Rev. Lett. 96, 092001 (2006). 032001 (2005). [69] Y. J. Zhang and K. T. Chao, Double charm produc- [87] K. Han et al. (CUSB Collaboration), Observation Of tion e+e− → J/ψcc¯ at B factories with next-to-leading P -Wave b¯b Bound States, Phys. Rev. Lett. 49, 1612 order QCD correction, Phys. Rev. Lett. 98, 092003 (1982). (2007). [88] C. Klopfenstein et al. (CUSB Collaboration), Observa- [70] Y. Jia, Exclusive Double Charmonium Production from tion of the Lowest P -Wave b¯b Bound States, Phys. Rev. Υ Decay, Phys. Rev. D 76, 074007 (2007). Lett. 51, 160 (1983). [71] J. Xu, H. R. Dong, F. Feng, Y. J. Gao, and Y. Jia, [89] U. Heintz et al. (CUSB Collaboration), b¯b spectroscopy

Exclusive decay of Υ into J/ψ + χc0,1,2, Phys. Rev. D from the Υ(3S) state, Phys. Rev. D 46, 1928 (1992). 87, 094004 (2013). [90] R. Nernst et al. (Crystal Ball Collaboration), Observa- [72] S. D. Yang et al. (Belle Collaboration), Evidence of tion of Three P States in the Radiative Decay of Υ(2S),

Υ(1S) → J/ψ +χc1 and search for double-charmonium Phys. Rev. Lett. 54, 2195 (1985). production in Υ(1S) and Υ(2S) decays, Phys. Rev. D [91] H. Albrecht et al. (ARGUS Collaboration), Radiative

90, 112008 (2014). Decays of the Υ(2S) Into the Three χb States, Phys. [73] C. P. Shen et al. (Belle Collaboration), Search for char- Lett. B 160, 331 (1985). monium and charmonium-like states in Υ(1S) radia- [92] P. Haas et al. (CLEO Collaboration), Observation of tive decays, Phys. Rev. D 82, 051504 (2010). Radiative Decays of the Υ(2S), Phys. Rev. Lett. 52, [74] X. L. Wang et al. (Belle Collaboration), Search for 799 (1984).

charmonium and charmonium-like states in Υ(2S) ra- [93] R. Morrison et al. (CLEO Collaboration), Inclusive χ2P diative decays, Phys. Rev. D 84, 071107 (2011). production in Υ(3S) decay, Phys. Rev. Lett. 67, 1696 [75] Y. J. Gao, Y. J. Zhang, and K. T. Chao, Radiative (1991). decays of bottomonia into charmonia and light mesons, [94] K. W. Edwards et al. (CLEO Collaboration), Measure- ¯ arXiv:hep-ph/0701009 (2007). ment of the mass splittings between the bb χbJ (1P ) [76] P. Katrenko et al. (Belle Collaboration), Observation of states, Phys. Rev. D 59, 032003 (1999).

the Radiative Decays of Υ(1S) to χc1, Phys. Rev. Lett. [95] M. Kornicer et al. (CLEO Collaboration), Measure- 124, 122001 (2020). ments of branching fractions for electromagnetic tran-

[77] S. Jia et al. (Belle Collaboration), Observation of sitions involving the χbJ (1P ) states, Phys. Rev. D 83, e+e− → γχ and search for e+e− → γχ , γχ , and 054003 (2011). √ c1 c0 c2 γηc at s near 10.6 GeV at Belle, Phys. Rev. D 98, [96] J. P. Lees et al. (BaBar Collaboration), Study of radia- 092015 (2018). tive bottomonium transitions using converted photons, [78] C. T. H. Davies et al., High precision lattice QCD con- Phys. Rev. D 84, 072002 (2011). fronts experiment, Phys. Rev. Lett. 92, 022001 (2004). [97] D. M. Asner et al. (CLEO Collaboration), Observation

[79] D. Besson and T. Skwarnicki, Υ spectroscopy, Annu. of χbJ (1P, 2P ) decays to light hadrons, Phys. Rev. D Rev. Nucl. Part. Sci. 43, 333 (1993). 78, 091103 (2008). [80] E. J. Eichten and C. Quigg, Mesons with beauty and [98] A. Abdesselam et al. (Belle Collaboration), Study of

charm: Spectroscopy, Phys. Rev. D 49, 5845 (1994). χbJ (1P ) Properties in the Radiative Υ(2S) Decays, [81] H. Mutuk, S-wave Heavy Quarkonium Spectra: Mass, arXiv:1606.01276 (2016). Decays and Transitions, Adv. High Energy Phys. 2018, [99] C. P. Shen et al. (Belle Collaboration), Search for dou- 5961031 (2018). ble charmonium decays of the P -wave spin-triplet bot- [82] N. Brambilla, P. Pietrulewicz, and A. Vairo, Model- tomonium states, Phys. Rev. D 85, 071102 (2012). independent Study of Electric Dipole Transitions in [100] V. V. Braguta, A. K. Likhoded, and A. V. Luchinsky, Quarkonium, Phys. Rev. D 85, 094005 (2012). Double charmonium production in exclusive bottomo- [83] A. Pineda and J. Segovia, Improved determination of nia decays, Phys. Rev. D 80, 094008 (2009).

40 [101] V. V. Braguta, A. K. Likhoded, and A. V. Luchinsky, and perspectives, arXiv:1907.07583 (2019). Double charmonium production in exclusive bottomo- [118] S. J. Brodsky, R. F. Lebed, and V. E. Lyubovit- nium decays, Phys. At. Nucl. 73, 1054 (2010). skij, QCD compositeness as revealed in exclusive vec- [102] V. V. Braguta, A. K. Likhoded, and A. V. Luchin- tor boson reactions through double-photon annihilation: + − ∗ 0 + − ∗ ∗ 0 0 sky, Observation potential for χbJ at the Tevatron and e e → γγ → γV and e e → γ γ → V V , CERN LHC, Phys. Rev. D 72, 094018 (2005). Phys. Lett. B 764, 174 (2017). [103] J. Zhang, H. R. Dong, and F. Feng, Exclusive decay [119] S. J. Brodsky, R. F. Lebed, and V. E. Lyubovitskij, of P -wave Bottomonium into double J/ψ, Phys. Rev. QCD dynamics of tetraquark production, Phys. Rev. D D 84, 094031 (2011). 91, 114025 (2015). [104] W. L. Sang, R. Rashidin, U. Kim, and J. Lee, Rela- [120] S. Jia et al. (Belle Collaboration), Search for + (0)− + − + (0)− √ tivistic Corrections to the Exclusive Decays of C-even Υ(1S, 2S) → Zc Zc and e e → Zc Zc at s Bottomonia into S-wave Charmonium Pairs, Phys. = 10.52, 10.58, and 10.867 GeV, Phys. Rev. D 97, Rev. D 84, 074026 (2011). 112004 (2018). [105] B. Aubert et al. (BaBar Collaboration), Observation [121] R. L. Jaﬀe and K. Johnson, Unconventional States of of the bottomonium ground state in the decay Υ(3S) → Conﬁned Quarks and Gluons, Phys. Lett. B 60, 201

γηb, Phys. Rev. Lett. 101, 071801 (2008). (1976). [106] B. Aubert et al. (BaBar Collaboration), Evidence for [122] C. F. Qiao and L. Tang, Finding the 0−− Glueball,

the ηb(1S) Meson in Radiative Υ(2S) Decay, Phys. Phys. Rev. Lett. 113, 221601 (2014). Rev. Lett. 103, 161801 (2009). [123] Y. Chen and M. Huang, Two-gluon and trigluon glue- [107] B. G. Fulsom et al. (Belle Collaboration), Observa- balls from dynamical holography QCD, Chin. Phys. C

tion of Υ(2S) → γηb(1S) decay, Phys. Rev. Lett. 121, 40, 123101 (2016). 232001 (2018). [124] S. Jia et al. (Belle Collaboration), Search for light [108] C. Hughes, R. J. Dowdall, C. T. H. Davies, R. R. Hor- tetraquark states in Υ(1S) and Υ(2S) decays, Phys. gan, G. V. Hippel, and M. Wingate (HPQCD Collab- Rev. D 96, 112002 (2017). oration), Hindered M1 Radiative Decay of Υ(2S) from [125] Z. R. Huang, W. Chen, T. G. Steele, Z. F. Zhang, and Lattice NRQCD, Phys. Rev. D 92, 094501 (2015). H. Y. Jin, Investigation of the light four-quark states [109] S. Dobbs, Z. Metreveli, A. Tomaradze, T. Xiao, and with exotic JPC = 0−−, Phys. Rev. D 95, 076017

Kamal K. Seth, Observation of ηb(2S) in Υ(2S) → (2017). γηb(2S), ηb(2S) → hadrons, and Conﬁrmation of [126] J. Preskill, Subgroup Alignment in Hypercolor Theo- ηb(1S), Phys. Rev. Lett. 109, 082001 (2013). ries, Nucl. Phys. B 177, 21 (1981). [110] S. Sandilya et al. (Belle Collaboration), Search for [127] R. L. Jaﬀe, Perhaps a Stable Dihyperon, Phys. Rev. Bottomonium States in Exclusive Radiative Υ(2S) De- Lett. 38, 195 (1977). cays, Phys. Rev. Lett. 111, 112001 (2013). [128] S. R. Beane et al. (NPLQCD Collaboration), Evidence [111] R. Mizuk et al. (Belle Collaboration), Evidence for for a Bound H-dibaryon from Lattice QCD, Phys. Rev.

the ηb(2S) and observation of hb(1P ) → ηb(1S)γ and Lett. 106, 162001 (2011). hb(2P ) → ηb(1S)γ), Phys. Rev. Lett. 109, 232002 [129] S. R. Beane et al. (NPLQCD Collaboration), Present (2012). Constraints on the H-dibaryon at the Physical Point ∗ [112] S. H. Lee, M. Nielsen, and U. Wiedner, DsD molecule from Lattice QCD, Mod. Phys. Lett. A 26, 2587 (2011). as an axial meson, J. Korean Phys. Soc. 55, 424 (2009). [130] T. Inoue, N. Ishii, S. Aoki, T. Doi, T. Hatsuda, [113] J. M. Dias, X. Liu, and M. Nielsen, Predicition for the Y. Ikeda, K. Murano, H. Nemura, and K. Sasaki ¯ ∗ ∗ ¯ decay width of a charged state near the DsD /Ds D (HALQCD Collaboration), Bound H-dibaryon in Fla- threshold, Phys. Rev. D 88, 096014 (2013). vor SU(3) Limit of Lattice QCD, Phys. Rev. Lett. 106, [114] A. Esposito, A. L. Guerrieri, F. Piccinini, A. Pilloni, 162002 (2011). and A. D. Polosa, Four-Quark Hadrons: an Updated [131] T. F. Carames and A. Valcarce, Examination of the H Review, Int. J. Mod. Phys. A 30, 1530002 (2015). dibaryon within a chiral constituent quark model, Phys. [115] H. X. Chen, W. Chen, X. Liu, and S. L. Zhu, The Rev. C 85, 045202 (2012). hidden-charm pentaquark and tetraquark states, Phys. [132] E. Braaten and H. W. Hammer, Universality in few- Rept. 639, 1 (2016). body systems with large scattering length, Phys. Rep. [116] F. K. Guo, C. Hanhart, U. G. Meißner, Q. Wang, Q. 428, 259 (2006). Zhao, and B. S. Zou, Hadronic molecules, Rev. Mod. [133] B. H. Kim et al. (Belle Collaboration), Search for an Phys. 90, 015004 (2018). H-dibaryon with mass near 2mΛ in Υ(1S) and Υ(2S) [117] N. Brambilla, S. Eidelman, C. Hanhart, A. Nefediev, decays, Phys. Rev. Lett. 110, 222002 (2013). C. P. Shen, C. E. Thomas, A. Vairo, and C. Z. Yuan, [134] G. R. Farrar, Stable Sexaquark, arXiv:1708.08951 The XYZ states: experimental and theoretical status (2017).

41 [135] J. P. Lees et al. (BaBar Collaboration), Search for a [152] J. P. Lees et al. (BaBar Collaboration), Search for a Stable Six-Quark State at BABAR, Phys. Rev. Lett. light Higgs boson decaying to two gluons or ss¯ in the 122, 072002 (2019). radiative decays of Υ(1S), Phys. Rev. D 88, 031701 [136] O. Tajima et al. (Belle Collaboration), Search for in- (2013). visible decay of the Υ(1S), Phys. Rev. Lett. 98, 132001 [153] D. Besson et al. (CLEO Collaboration), First Obser- (2007). vation of Υ(3S) → τ +τ − and Tests of Lepton Univer- [137] R. Rubin et al. (CLEO Collaboration), Search for In- sality in Υ Decays, Phys. Rev. Lett. 98, 052002 (2007). visible Decays of the Υ(1S) Resonance, Phys. Rev. D [154] J. P. Lees et al. (BaBar Collaboration), Precision 75, 031104 (2007). measurement of the B(Υ(3S) → τ +τ −)/B(Υ(3S) → [138] B. Aubert et al. (BaBar Collaboration), A Search for µ+µ−), arXiv:2005.01230 (2020). Invisible Decays of the Υ(1S), Phys. Rev. Lett. 103, [155] P. del Amo Sanchez et al. (BaBar Collaboration), Test 251801 (2009). of lepton universality in Υ(1S) decays at BaBar, Phys. [139] J. F. Gunion, D. Hooper, and B. McElrath, Light neu- Rev. Lett. 104, 191801 (2010). tralino dark matter in the NMSSM, Phys. Rev. D 73, [156] W. Love et al. (CLEO Collaboration), Search for Lep- 015011 (2006). ton Flavor Violation in Υ Decays, Phys. Rev. Lett. [140] K. Petraki and R.R. Volkas, Review of asymmetric 101, 201601 (2008). dark matter, Int. J. Mod. Phys. A 28, 1330028 (2013). [157] J. P. Lees et al. (BaBar Collaboration), Search for [141] R. Dermisek, J. F. Gunion, and B. McElrath, Prob- Charged Lepton Flavor Violation in Narrow Υ Decays, ing NMSSM Scenarios with Minimal Fine-Tuning by Phys. Rev. Lett. 104, 151802 (2010). Searching for Decays of the Υ to a Light CP -Odd Higgs [158] Z. K. Silagadze, Lepton ﬂavor violating decays as Boson, Phys. Rev. D 76, 051105 (2007). probes of quantum gravity?, Phys. Scr. 64, 128 (2001). [142] R. Dermisek and J. F. Gunion, New constraints on [159] D. Black, T. Han, H. J. He, and M. Sher, τ − µ ﬂavor a light CP -odd Higgs boson and related NMSSM Ideal violation as a probe of the scale of new physics, Phys. Higgs Scenarios, Phys. Rev. D 81, 075003 (2010). Rev. D 66, 053002 (2002). [143] G. K. Yeghiyan, Υ(1S) Decays into Light Scalar Dark [160] Y. P. Kuang, QCD multipole expansion and hadronic Matter, Phys. Rev. D 80, 115019 (2009). transitions in heavy quarkonium systems, Front. Phys. [144] I. S. Seong et al. (Belle Collaboration), Search for a China 1, 19 (2006) and references therein. light CP -odd Higgs boson and low-mass dark matter [161] H. Mendez et al. (CLEO Collaboration), Branching at the Belle experiment, Phys. Rev. Lett. 122, 011801 Fractions for Transitions of ψ(2S) to J/ψ, Phys. Rev. (2019). D 78, 011102 (2008). [145] P. del Amo Sanchez et al. (BaBar Collaboration), [162] M. B. Voloshin, Charmonium, Prog. Part. Nucl. Phys. Search for Production of Invisible Final States in 61, 455 (2008). Single-Photon Decays of Υ(1S), Phys. Rev. Lett. 107, [163] Y. A. Simonov and A. I. Veselov, Single η production 021804 (2011). in heavy quarkonia: Breakdown of multipole expansion, [146] W. Love et al. (CLEO Collaboration), Search for Very Phys. Lett. B 673, 211 (2009). Light CP -Odd Higgs Boson in Radiative Decays of [164] Q. He et al. (CLEO Collaboration), Observation of Υ(1S), Phys. Rev. Lett. 101, 151802 (2008). Υ(2S) → ηΥ(1S) and search for related transitions, [147] B. Aubert et al. (BaBar Collaboration), Search for Phys. Rev. Lett. 101, 192001 (2008). Dimuon Decays of a Light Scalar Boson in Radiative [165] J. P. Lees et al. (BaBar Collaboration), Study of Transitions Υ → γA0, Phys. Rev. Lett. 103, 081803 Υ(2S, 3S) → ηΥ(1S) and Υ(2S, 3S) → π+π−Υ(1S), (2009). Phys. Rev. D 84, 092003 (2011). [148] J. P. Lees et al. (BaBar Collaboration), Search for di- [166] U. Tamponi et al. (Belle Collaboration), Study of the muon decays of a low-mass Higgs boson in radiative hadronic transitions Υ(2S) → (η, π0)Υ(1S) at Belle, decays of the Υ(1S), Phys. Rev. D 87, 031102 (2013). Phys. Rev. D 87, 011104 (2013). [149] B. Aubert et al. (BaBar Collaboration), Search for a [167] B. Aubert et al. (BaBar Collaboration), Study of low-mass Higgs boson in Υ(3S) → γA0, A0 → τ +τ − hadronic transitions between Υ states and observation at BABAR, Phys. Rev. Lett. 103, 181801 (2009). of Υ(4S) → ηΥ(1S) decay, Phys. Rev. D 78, 112002 [150] J. P. Lees et al. (BaBar Collaboration), Search for a (2008). low-mass scalar Higgs boson decaying to a tau pair in [168] E. Guido et al. (Belle Collaboration), Study of η and single-photon decays of Υ(1S), Phys. Rev. D 88, 071102 dipion transitions in Υ(4S) decays to lower bottomonia, (2013). Phys. Rev. D 96, 052005 (2017). [151] J. P. Lees et al. (BaBar Collaboration), Search for [169] M. B. Voloshin, Hadronic transitions from Υ(4S) as a hadronic decays of a light Higgs boson in the radiative probe of four-quark admixture, Mod. Phys. Lett. A 26, decay Υ → γA0, Phys. Rev. Lett. 107, 221803 (2011). 773 (2011).

42 [170] E. Guido et al. (Belle Collaboration), Observation Collisions at 7.0 TeV, Mod. Phys. Lett. A 28, 1350120 of Υ(4S) → η0Υ(1S), Phys. Rev. Lett. 121, 062001 (2013). (2018). [189] J. F. Owens, E. Reya, and M. Gluck, Detailed Quan-

[171] K. F. Chen et al. (Belle Collaboration), Observation tum Chromodynamic Predictions for High pT Pro- of anomalous Υ(1S)π+π− and Υ(2S)π+π− produc- cesses, Phys. Rev. D 18, 1501 (1978). tion near the Υ(5S) resonance, Phys. Rev. Lett. 100, [190] V. G. Kartvelishvili, A. K. Likhoded, and S. R. 112001 (2008). Slabospitsky, D Meson and ψ Meson Production in [172] A. Bondar et al. (Belle Collaboration), Observation Hadronic Interactions, Yad. Fiz. 28, 1315 (1978) [Sov. of two charged bottomonium-like resonances in Υ(5S) J. Nucl. Phys. 28, 678 (1978)]. decays, Phys. Rev. Lett. 108, 122001 (2012). [191] C. H. Chang, Hadronic Production of J/ψ Associated [173] See public webs at CERN, http://twiki.cern.ch/ With a Gluon, Nucl. Phys. B 172, 425 (1980). twiki/bin/view/Main/PublicWebs. [192] E. L. Berger and D. L. Jones, Inelastic Photoproduc- [174] A. A. Alves Jr. et al. LHCb collaboration, The LHCb tion of J/ψ and Υ by Gluons, Phys. Rev. D 23, 1521 Detector at the LHC, JINST 3, S08005, (2008). (1981). [175] A. A. Alves Jr. et al. LHCb collaboration, Perfor- [193] R. Baier and R. Ruckl, On Inelastic Leptoproduction mance of the LHCb muon system, JINST 8, P02022 of Heavy Quarkonium States, Nucl. Phys. B 201, 1 (2013). (1982). [176] A. A. Alves Jr. et al. LHCb collaboration, The LHCb [194] R. Baier and R. Ruckl, Hadronic Collisions: A Trigger and its Performance in 2011, JINST 8, P04022 Quarkonium Factory, Z. Phys. C 19, 251 (1983). (2013). [195] H. Fritzsch, Producing Heavy Quark Flavors in [177] G. Aad et al. ATLAS Collaboration, The ATLAS Ex- Hadronic Collisions: A Test of Quantum Chromody- periment at the CERN Large Hadron Collider, JINST namics, Phys. Lett. 67B, 217 (1977). 3, S08003 (2008). [196] F. Halzen, Cvc for Gluons and Hadroproduction of [178] S. Chatrchyan et al. CMS Collaboration, The CMS Quark Flavors, Phys. Lett. 69B, 105 (1977). Experiment at the CERN LHC, JINST 3, S08004 [197] M. Gluck, J. F. Owens, and E. Reya, Gluon Contri- (2008). bution to Hadronic J/ψ Production, Phys. Rev. D 17, [179] K. Aamodt et al. ALICE collaboration, The AL- 2324 (1978). ICE experiment at the CERN LHC, JINST 3, S08002 [198] V. D. Barger, W. Y. Keung, and R. J. N. Phillips, On (2008). ψ and Υ Production via Gluons, Phys. Lett. 91B, 253 [180] V. Khachatryan et al. (CMS Collaboration), Υ Pro- (1980). √ duction Cross-Section in pp Collisions at s = 7 TeV, [199] J. F. Amundson, O. J. P. Eboli, E. M. Gregores, and F. Phys. Rev. D 83, 112004 (2011). Halzen, Colorless states in perturbative QCD: Charmo- [181] G. Aad et al. (ATLAS Collaboration), Measurement of nium and rapidity gaps, Phys. Lett. B 372, 127 (1996). Υ production in 7 TeV pp collisions at ATLAS, Phys. [200] J. F. Amundson, O. J. P. Eboli, E. M. Gregores, and F. Rev. D 87, 052004 (2013). Halzen, Quantitative tests of color evaporation: Char- [182] R. Aaij et al. (LHCb Collaboration), Measurement of monium production, Phys. Lett. B 390, 323 (1997). √ Υ production in pp collisions at s = 7 TeV, Eur. Phys. [201] V. Khachatryan et al. (CMS Collaboration), Mea- J. C 72, 2025 (2012). surements of the Υ(1S), Υ(2S), and Υ(3S) diﬀerential √ [183] S. Acharya et al. (ALICE Collaboration), Suppres- cross sections in pp collisions at s = 7 TeV, Phys. sion of Υ(1S) at forward rapidity in P bP b collisions Lett. B 749, 14 (2015). √ at sNN = 2.76 TeV, Phys. Lett. B 738, 361 (2014). [202] B. Gong et al., Complete next-to-leading-order study [184] G. Aad et al. (ATLAS Collaboration), Measurement on the yield and polarization of Υ(1S, 2S, 3S) at the of the Υ(1S) production cross-section in pp collisions Tevatron and LHC, Phys. Rev. Lett. 112, 032001 √ at s = 7 TeV in ATLAS, Phys. Lett. B 705, 9 (2011). (2014). [185] T. Sj¨ostrand,S. Mrenna, and P. Z. Skands, PYTHIA [203] A. M. Sirunyan et al. (CMS Collaboration), Measure- 6.4 Physics and Manual, J. High Energy Phys. 05, 026 ment of quarkonium production cross sections in pp col- √ (2006). lisions at s = 13 TeV, Phys. Lett. B 780, 251 (2018). [186] R. Aaij et al. (LHCb Collaboration), Forward produc- [204] Y. Q. Ma, K. Wang, and K. T. Chao, J/ψ(ψ0) produc- √ 4 4 tion of Υ mesons in pp collisions at s = 7 and 8 TeV, tion at the Tevatron and LHC at O(αsv ) in nonrela- J. High Energy Phys. 11, 103 (2015). tivistic QCD, Phys. Rev. Lett. 106, 042002 (2011). [187] L. S. Kisslinger, M. X. Liu, and P. McGaughey, Heavy [205] H. Han, Y. Q. Ma, C. Meng, H. S. Shao, Y. J. Zhang,

Quark State Production In pp Collisions, Phys. Rev. D and K. T. Chao, Υ(nS) and χb(nP ) production at 84, 114020 (2011). hadron colliders in nonrelativistic QCD, Phys. Rev. D [188] L. S. Kisslinger and D. Das, ψ and Υ Production In pp 94, 014028 (2016).

43 [206] R. Aaij et al. (LHCb Collaboration), Measurement of sion of Excited Υ States Relative to the Ground State √ √ Υ production in pp collisions at s = 2.76 TeV, Eur. in Pb-Pb Collisions at sNN=5.02 TeV, Phys. Rev. Phys. J. C 74, 2835 (2014). Lett. 120, 142301 (2018). [207] P. Faccioli et al., Towards the experimental clarifica- [223] B. Krouppa and M. Strickland, Predictions for bot- tion of quarkonium polarization, Eur. Phys. J. C 69, tomonia suppression in 5.023 TeV Pb-Pb collisions, 657 (2010). Universe 2, 16 (2016). [208] P. Faccioli et al., J/psi polarization from fixed-target to [224] X. Du, R. Rapp, and M. He, Color Screening and collider energies, Phys. Rev. Lett. 102, 151802 (2009). Regeneration of Bottomonia in High-Energy Heavy-Ion [209] M. Jacob and G. C. Wick, On the General Theory of Collisions, Phys. Rev. C 96, 054901 (2017). Collisions for Particles with Spin, Annals Phys. 7, 404 [225] S. Acharya et al. (ALICE Collaboration), Υ suppres- √ (1959). sion at forward rapidity in Pb-Pb collisions at sNN = [210] J. C. Collins and D. E. Soper, Angular Distribution 5.02 TeV, Phys. Lett. B 790, 89 (2019). of Dileptons in High-Energy Hadron Collisions, Phys. [226] K. Zhou, N. Xu, and P. Zhaung, Υ Production in Rev. D 16, 2219 (1977). Heavy Ion Collisions at LHC, Nucl. Phys. A 931, 654 [211] E. Braaten, D. Kang, J. Lee, and C. Yu, Optimal spin (2014). quantization axes for the polarization of dileptons with [227] B. Krouppa, A. Rothkopf, M. Strickland, Bottomo- large transverse momentum, Phys. Rev. D 79, 014025 nium suppression using a lattice QCD vetted potential, (2009). Phys. Rev. D 97, 016017 (2018). [212] T. Aaltonen et al. (CDF Collaboration), Measure- [228] S. Voloshin and Y. Zhang, Flow study in relativistic ments of Angular Distributions of Muons From Υ Me- nuclear collisions by Fourier expansion of Azimuthal √ son Decays in pp¯ Collisions at s = 1.96 TeV, Phys. particle distributions, Z. Phys. C 70, 665 (1996). Rev. Lett. 108, 151802 (2012). [229] S. Acharya et al. (ALICE Collaboration), Measure- [213] V. M. Abazov et al. (D0 Collaboration), Measurement ment of Υ(1S) elliptic flow at forward rapidity in Pb-Pb √ of the polarization of the υ and υ states in pp¯ colli- collisions at s = 5.02 TeV, Phys. Rev. Lett. 123, √ 1S 2S NN sions at s = 1.96-TeV, Phys. Rev. Lett. 101, 182004 192301 (2019). (2008). [230] C. Adler et al. (STAR Collaboration), Elliptic flow [214] S. Chatrchyan et al. (CMS Collaboration), Measure- from two and four particle correlations in Au+Au col- ment of the Y (1S),Y (2S) and Y (3S) Polarizations in lisions at s(NN)**(1/2) = 130-GeV, Phys. Rev. C 66, √ pp Collisions at s = 7 TeV, Phys. Rev. Lett. 110, 034904 (2002). 081802 (2013). [231] S. A. Voloshin, A. M. Poskanzer, and R. Snellings, [215] R. Aaij et al. (LHCb Collaboration), Measurement of Collective phenomena in non-central nuclear collisions, √ the Υ polarizations in pp collisions at s = 7 and 8 Landolt-Bornstein 23, 293 (2010). TeV, J. High Energy Phys. 1712, 110 (2017). [232] J. Barrette et al. (E877 Collaboration), Observation [216] K. Gottfried and J. D. Jackson, On the Connection of anisotropic event shapes and transverse flow in Au between production mechanism and decay of resonances + Au collisions at AGS energy, Phys. Rev. Lett. 73, at high-energies, Nuovo Cim. 33, 309 (1964). 2532 (1994). [217] B. Gong, J. X. Wang, and H. F. Zhang, QCD cor- [233] R. Vogt, Cold Nuclear Matter Effects on J/ψ and rections to Υ production via color-octet states at the Υ Production at the LHC, Phys. Rev. C 81, 044903 Tevatron and LHC, Phys. Rev. D 83, 114021 (2011). (2010). [218] P. Faccioli et al., Quarkonium production in the LHC [234] Z. Hu, N. T. Leonardo, T. Liu, and M. Haytmyradov, era: a polarized perspective, Phys. Lett. B 736, 98 Review of bottomonium measurements from CMS, Int. (2014). J. Mod. Phys. A 32, 1730015 (2017). [219] S. Chatrchyan et al. (CMS Collaboration), Indications [235] S. Chatrchyan et al. (CMS Collaboration), Event Ac- √ of suppression of excited Υ states in PbPb collisions tivity Dependence of Y(nS) Production in s =5.02 √ √ NN at SNN = 2.76 TeV, Phys. Rev. Lett. 107, 052302 TeV pPb and s=2.76 TeV pp Collisions, J. High En- (2011). ergy Phys. 1404, 103 (2014). [220] S. Chatrchyan et al. (CMS Collaboration), Observa- [236] R. Aaij et al. (LHCb Collaboration), Study of Υ pro- tion of Sequential Upsilon Suppression in PbPb Colli- duction and cold nuclear matter effects in pPb colli- √ sions, Phys. Rev. Lett. 109, 222301 (2012). sions at sNN =5 TeV, J. High Energy Phys. 1407, [221] S. Chatrchyan et al. (CMS Collaboration), Observa- 094 (2014). tion and studies of jet quenching in PbPb collisions [237] S. Acharya et al. (ALICE Collaboration), Production at nucleon-nucleon center-of-mass energy = 2.76 TeV, of inclusive Υ(1S) and Υ(2S) in p-Pb collisions at √ Phys. Rev. C 84, 024906 (2011). sNN = 5.02 TeV, Phys. Lett. B 740, 105 (2015). [222] A. M. Sirunyan et al. (CMS Collaboration), Suppres- [238] R. Aaij et al. (LHCb Collaboration), Production of

44 J/psi and Upsilon mesons in pp collisions at sqrt(s) = 8 TeV, J. High Energy Phys. 1306, 064 (2013). [239] E. G. Ferreiro, F. Fleuret, J. P. Lansberg, N. Matagne, and A. Rakotozafindrabe, Υ production in p(d)A collisions at RHIC and the LHC, Eur. Phys. J. C 73, 2427 (2011). [240] J. L. Albacete et al., Predictions for p+Pb Collisions at sqrt s NN = 5 TeV, Int. J. Mod. Phys. E 22, 1330007 (2013). [241] F. Arleo and S. Peigne, Heavy-quarkonium suppression in p-A collisions from parton energy loss in cold QCD matter, J. High Energy Phys. 1303, 122 (2013). [242] R. Aaij et al. (LHCb Collaboration), Study of Υ pro- √ duction in pPb collisions at sNN = 8.16 TeV, J. High Energy Phys. 1811, 194 (2018). [243] S. Acharya et al. (ALICE Collaboration), Υ pro- √ duction in p–Pb collisions at sNN =8.16 TeV, arXiv:1910.14405 (2019). [244] H. S. Shao, HELAC-Onia 2.0: an upgraded matrix- element and event generator for heavy quarkonium physics, Comput. Phys. Commun. 198, 238 (2016). [245] H. S. Shao, HELAC-Onia: An automatic matrix element generator for heavy quarkonium physics, Comput. Phys. Commun. 184, 2562 (2013). [246] K. J. Eskola, P. Paakkinen, H. Paukkunen, and C. A. Salgado, EPPS16 – Bringing nuclear PDFs to the LHC era, arXiv:1802.00713 (2019). [247] J. P. Lansberg and H. S. Shao, Towards an automated tool to evaluate the impact of the nuclear modification of the gluon density on quarkonium, D and B meson production in proton?Cnucleus collisions, Eur. Phys. J. C 77, 1 (2017). [248] K. Kovarik et al., nCTEQ15 – Global analysis of nuclear parton distributions with uncertainties in the CTEQ framework, Phys. Rev. D 93, 085037 (2016). [249] K. J. Eskola, P. Paakkinen, H. Paukkunen, and C. A. Salgado, EPPS16: Nuclear parton distributions with LHC data, Eur. Phys. J. C 77, 163 (2017). [250] A. Kusina, J. P. Lansberg, I. Schienbein, and H. S. Shao, Gluon Shadowing in Heavy-Flavor Production at the LHC, Phys. Rev. Lett. 121, 052004 (2018). [251] J. L. Albacete et al., Predictions for Cold Nuclear √ Matter Effects in p+Pb Collisions at sNN = 8.16 TeV, Nucl. Phys. A 972, 18 (2018). [252] R. Vogt, Shadowing effects on J/ψ and Υ production at energies available at the CERN Large Hadron Col- lider, Phys. Rev. C 92, 034909 (2015). [253] L. Tang and C. F. Qiao, Mass spectra of 0+−, 1−+, and 2+− exotic glueballs, Nucl. Phys. B 904, 282 (2016).