-
Hadronic transitions with
π+π− or η emittance have contributed significantly to discoveries of quarkonium(-like) states, such asY(4260) ine+e−→π+π−J/ψ via initial-state radiation (ISR) by the BaBar experiment [1]. While searching for ansˉs version ofY(4260) , BaBar discoveredY(2175) (now called 'ϕ(2170) ') ine+e−→π+π−ϕ via ISR [2], which was later confirmed by Belle [3]. While searching forϕ(2170) in the hadronic transition with η, BaBar studied thee+e−→ηϕ process via ISR using a232 fb−1 data sample and found an excess with a mass of(2125±22±10)MeV/c2 (tens ofMeV/c2 lower than the world average value ofϕ(2170) [4]) and a width of(61±50±13)MeV [5, 6]. Hereinafter, the first quoted uncertainties are statistical and the second ones are systematic. Belle measured this process with considerably larger statistics in a980 fb−1 data sample but did not find this excess, and the statistical significance ofϕ(2170) was only1.7σ [7].Interesting measurements have been obtained from the CMD-3 and BESIII experiments over the past years. The CMD-3 experiment measured the process
e+e−→K+K−η from 1.59 to2.007GeV and found it to be dominated by theηϕ contribution [8]. CMD-3 then calculated the contribution to the anomalous magnetic moment of the muon:aηϕμ(E<1.8GeV)=(0.321±0.015±0.016)×10−10 ,aηϕμ(E<2.0GeV)=(0.440±0.015±0.022)×10−10 . With a715 pb−1 data sample taken at 22 CM energy points in the range of 2.00 to3.08GeV , BESIII measured the Born cross sections ofe+e−→ηϕ [9] ande+e−→ϕη′ [10]. BESIII reported the observation ofϕ(2170) in theηϕ final state and determined its resonant parameters to bemϕ(2170)=(2163.5±6.2±3.0)MeV/c2 andΓϕ(2170)=(31.1+21.1−11.6±1.1)MeV [9], for which the width was considerably narrower than the world average value of approximately100 MeV [4]. BESIII also observed a resonance near2.17GeV/c2 in theϕη′ final state with a statistical significance exceeding10σ [10]. Assuming it isϕ(2170) , one can infer the ratioB[ϕ(2170)→ϕη]/B[ϕ(2170)→ϕη′]=(0.23±0.10±0.18) , which is smaller than the prediction ofsˉsg hybrid models by several orders of magnitude.It is puzzling that
ϕ(2170) is not significant in the η transition compared with theπ+π− transition, and the measurement ofϕ(2170) in theηϕ final state is still poor. The lineshape ofσ(e+e−→ηϕ) is considerably different from that ofσ(e+e−→π+π−ϕ) [2, 3], which may help us understand the difference betweene+e−→ηJ/ψ ande+e−→π+π−J/ψ . In thecˉc sector,σ(e+e−→ηJ/ψ)/σ(e+e−→π+π−J/ψ)≈1 at the peak ofY(4260) , whereasσ(e+e−→ηϕ)/σ(e+e−→π+π−ϕ)≫1 at the peak ofϕ(1680) and≪1 at the peak ofϕ(2170) in thesˉs sector. In a recent lattice quantum chromodynamics (QCD) calculation [11], the properties of the two lowest states were found to comply with those of ϕ andϕ(1680) but had no obvious correspondence toϕ(2170) .Besides
ϕ(1680) , there is one more state, known as 'X(1750) ,' that is a candidate of thesˉs quarkonium. The observation ofϕ(1680) inK+K− andKˉK∗(892) is sometimes cited as evidence that this state is ansˉs quarkonium, as the radial excitation of ϕ. However, it was argued that the one true evidence for ϕ as ansˉs state should be the large branching fractions to hidden strangeness modes such asηϕ [12]. The FOCUS experiment reported a high-statistics study of the diffraction photo-production ofK+K− and observedX(1750) with a mass of(1753.5±1.5±2.3)MeV/c2 and a width of(122.3±6.2±0.8) MeV [13]. Meanwhile, FOCUS observed a slight enhancement below theϕ(1680) region but no obviousX(1750) signal in theKˉK∗(892) final state.If
ϕ(1680) andX(1750) are the same state, the mass measured ine+e− collisions and photoproduction experiments typically has a difference of50−100MeV/c2 , withKˉK∗(892) dominance ine+e− collisions andK+K− dominance in photo-production. This may constitute evidence for two distinct states, although interference withqˉq(q=u,d) vectors may complicate a comparison of these two processes. This issue can be addressed by studying channels in which interference withqˉq(q=u,d) vectors is expected to be unimportant, notably,ηϕ . With a sample of 4.48 millionψ(2S) events, BESIII performed the first partial wave analysis ofψ(2S)→K+K−η and simultaneously observedϕ(1680) andX(1750) in theK+K− mass spectrum [14], which indicates thatX(1750) is distinct fromϕ(1680) . Meanwhile, BESIII determinedX(1750) to be a1−− resonance.Because the cross sections of
e+e−→ηϕ have been effectively measured by the BaBar, Belle, BESIII, and CMD-3 experiments, it is helpful to consider all of them to achieve a better understanding ofϕ(1680) ,X(1750) , andϕ(2170) . In this study, we combine the measuredσ(e+e−→ηϕ) from the BaBar, Belle, BESIII, and CMD-3 experiments to gain greater lineshape precision, which is helpful for the study of the anomalous magnetic moment of the muon. Then, we perform combined fits to these measured cross sections for the resonant parameters ofϕ(1680) andϕ(2170) and estimate the production ofX(1750) in theηϕ final state. -
The measurements of
σ(e+e−→ηϕ) from the BaBar, Belle, BESIII, and CMD-3 experiments are shown in Fig. 1, where (a) to (f) are the Belle measurement with theη→γγ mode [5], Belle measurement with theη→π+π−π0 mode [6], BaBar measurement with theη→γγ mode, BaBar measurement with theη→π+π−π0 mode [7], CMD-3 measurement [8], and BESIII measurement [9], respectively.Figure 1.
σ(e+e−→ηϕ) measured in (a) theη→γγ mode at Belle, (b) theη→π+π−π0 mode at Belle, (c) theη→γγ mode at BaBar, (d) theη→π+π−π0 mode at BaBar, (e) theη→γγ mode at CMD-3, and (f) theη→γγ mode at BESIII.● We show comparisons of the latest results from Belle and the previous measurements from BaBar, BESIII, and CMD-3 in Fig. 2. The comparisons reveal good agreement between the four experiments.
Figure 2. (color online) Measurements of
σ(e+e−→ηϕ) from the BaBar, BESIII, and CMD-3 experiments compared with the latest measurements from the Belle experiment. Plots (a) and (b) show the comparison between BaBar and Belle in theη→γγ mode andη→π+π−π0 mode, respectively; plots (c) and (d) show the comparison between CMD-3 and Belle and between BESIII and Belle, respectively, where the Belle measurement has theη→γγ andη→π+π−π0 modes combined.● In the BaBar measurements,
σ(e+e−→ηϕ) measured in theη→π+π−π0 mode is slightly lower than that measured in theη→γγ mode; however, both have similar lineshapes, including a small bump around2.13GeV . The expectedJ/ψ signal according to the world average value ofB(J/ψ→ηϕ) [4] is not clear in the BaBar measurements.● With a considerably larger data sample, the Belle measurements are approximately twice as accurate as those measured in the BaBar experiment. There are clear
J/ψ signals in both theη→γγ andπ+π−π0 modes, whereas enhancement is not observed around2.13GeV and2.17GeV .● BESIII reported the Born cross section of
e+e−→ηϕ . We calculate the dressed cross section ofe+e−→ηϕ with the vacuum polarization and Born cross sections from Ref. [9], as shown in Fig. 1(f).● The
50MeV interval in the BESIII data sample is a disadvantage in determining the lineshape of a structure with a width of tens ofMeV . When determiningϕ(2170) , BESIII relied onσ(e+e−→ηϕ) below2GeV measured by the BaBar experiment, which is dominated by the large contribution from theϕ(1680) signal.●
σ(e+e−→ηϕ) measured by CMD-3 are below2GeV , with a precision similar to that of the Belle measurement.● Clear
ϕ(1680) signals are observed in the BaBar, Belle, and CMD-3 measurements.The measurements of the dressed cross section of
e+e−→ηϕ from the four experiments are consistent with each other. Therefore, we combine these measurements to obtain the best precision ofσ(e+e−→ηϕ) . A preciseσ(e+e−→ηϕ) is helpful for studying the anomalous magnetic moment of the muon [8] and may provide hints ofϕ(2170) orX(1750) . The calculation for the combination usesˉx=∑ixi/Δx2i∑i1/Δx2i,
(1) ˉσ=∑iσi/Δσ2i∑i1/Δσ2i,
(2) (Δσ)2=1∑i1/Δσ2i,
(3) where
σi is the value of the ith (i=1,2,3,4,5,6 ) experimental measurement of the cross section at the energy pointxi (√si ) illustrated in Fig. 1, andΔσi andΔxi are their related uncertainties. The average ofxi takes into account the difference in√s in the data taking of the BESIII or CMD-3 experiment and the average√s reported by BaBar and Belle using ISR technology. The uncertainties of√s in the BESIII and CMD-3 experiments are of the1MeV level, and the two experiments have no overlap in the√s region. We take half of the√s bin width in the BaBar and Belle measurements as the uncertainty (Δxi ). However, there are correlations between the measurements, such as the branching fraction of ϕ or η decay. We revisit the estimation of Eq. (3) according to Ref. [15] and construct the matrices of the statistical uncertaintiesCstat and uncorrelated systematic uncertaintiesCuncor_syst asCstat/uncor_syst=(S1⋅σ2100⋯00S2⋅σ220⋯000S3⋅σ23⋯0⋮⋮⋮Si⋅σ2i⋮000⋯S6⋅σ26)
(4) with
Si=(δstati)2 or(δuncor_systi)2 , whereδstati andδuncor_systi are the statistical and uncorrelated systematic relative uncertainties ofσi . We then construct the matrix of the correlated systematic uncertaintiesCcor_syst asCcor_syst=(a11a12⋯a1ja12a22⋯a2j⋮⋮⋱⋮ai1ai2⋯aij)
(5) where
aij≡δcor_systi⋅δcor_systj⋅σi⋅σj . We obtain the effective global covariance matrixC=Cstat+Cuncor_sys+Ccor_syst.
(6) According to Ref. [15], we calculate the error of
ˉσ using(Δσ)2=(∑ij(C−1)ij)−1.
(7) We show the results of the combination in Fig. 3 and Table 1.
Figure 3. Cross section of
e+e−→ηϕ from the combination of the measurements by the BaBar, Belle, BESIII, and CMD-3 experiments.√s σ(e+e−→ηϕ) √s σ(e+e−→ηϕ) √s σ(e+e−→ηϕ) √s σ(e+e−→ηϕ) 1.565 3±30±1 2.165 376±65±22 2.76 104±38±5 3.36 31±18±2 1.585 143±44±4 2.179 217±17±10 2.78 116±31±7 3.388 21±15±2 1.596 476±60±11 2.205 211±15±8 2.80 103±36±5 3.40 −13±8±1 1.624 1068±112±36 2.225 218±53±8 2.82 42±23±1 3.42 44±29±3 1.644 1644±117±53 2.234 15±13±7 2.84 65±30±3 3.448 31±12±1 1.67 2341±157±90 2.265 206±56±11 2.86 65±22±3 3.46 41±19±2 1.68 2600±144±76 2.285 173±48±7 2.88 38±26±2 3.48 62±22±3 1.695 2226±139±62 2.31 160±11±6 2.90 35±2±2 3.50 55±20±3 1.722 2151±121±59 2.325 108±37±5 2.915 38±22±3 3.52 74±24±4 1.742 1761±124±48 2.345 176±48±9 2.948 27±5±1 3.54 8±12±7 1.759 1508±102±40 2.365 230±54±10 2.965 40±16±4 3.56 29±14±1 1.779 1221±100±34 2.388 139±10±5 2.98 26±5±1 3.58 18±13±3 1.799 1182±86±29 2.399 119±5±5 3.00 26±5±1 3.60 32±14±1 1.828 870±94±24 2.42 94±33±7 3.028 21±4±2 3.62 32±19±1 1.85 851±76±24 2.44 162±49±8 3.045 21±4±1 3.64 25±18±1 1.865 1168±120±35 2.46 62±26±8 3.06 29±5±1 3.66 32±21±2 1.877 764±32±12 2.48 113±44±6 3.085 29±2±1 3.68 38±23±2 1.902 864±73±22 2.50 91±33±2 3.10 475±64±24 3.70 31±18±2 1.923 738±63±19 2.52 125±39±4 3.12 224±50±10 3.72 7±9±1 1.945 768±61±21 2.54 106±29±6 3.148 70±22±3 3.74 24±12±1 1.963 743±60±19 2.56 167±49±9 3.16 82±28±4 3.76 −4±6±1 1.985 600±60±16 2.58 113±32±5 3.18 81±28±4 3.78 35±21±3 2.007 509±26±12 2.60 110±36±5 3.208 30±13±2 3.80 13±10±1 2.025 515±86±26 2.62 72±26±2 3.22 28±16±2 3.82 6±9±1 2.045 421±41±15 2.644 67±5±3 3.24 15±16±3 3.84 63±20±3 2.065 388±83±29 2.66 105±34±5 3.268 26±12±1 3.86 33±20±2 2.085 410±71±20 2.68 65±30±3 3.28 16±18±1 3.88 20±11±1 2.105 402±24±14 2.70 67±27±3 3.30 44±20±2 3.90 22±11±1 2.127 478±57±12 2.72 171±44±8 3.328 27±15±2 3.92 17±10±1 2.145 384±42±12 2.74 121±32±7 3.34 66±24±3 3.94 8±13±1 Table 1. Cross section of
e+e−→ηϕ versus√s calculated with the measurements from the BaBar, Belle, BESIII, and CMD-3 experiments. The first errors are statistical and the second ones are systematic. -
ϕ(1680) ,X(1750) , andϕ(2170) may exist in thee+e−→ηϕ process. We perform combined fits toσ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII and CMD-3 experiments and shown in Fig. 1. The fit range is from the threshold to2.85GeV/c2 . Assuming there areϕ(1680) ,X(1750) , andϕ(2170) components and a non-resonant contribution in theηϕ final state, we take the parameterization ofσ(e+e−→ηϕ) similar to that used in the BaBar analysis [5]:σ(e+e−→ηϕ)(√s)=12πPηϕ(√s)|An.r.ηϕ(√s)+Aϕ(1680)ηϕ(√s)+AX(1750)ηϕ(√s)+Aϕ(2170)ηϕ(√s)|2,
(8) where
Pηϕ is the phase space of theηϕ final state, the non-resonant amplitude takes the formAn.r.ηϕ(√s)=a0/sa1 , andAϕ(1680)ηϕ ,AX(1750)ηϕ , andAϕ(2170)ηϕ are the amplitudes ofϕ(1680) ,X(1750) , andϕ(2170) , respectively.For
Aϕ(1680)ηϕ andAX(1750)ηϕ , we describe the form with a Breit-Wigner (BW) function,AηϕX(√s)=√BηϕXΓe+e−X⋅√ΓX/Pηϕ(MX)⋅eiθXM2X−s−i√sΓX(√s),
(9) where X is
ϕ(1680) orX(1750) , the resonant parametersMX ,ΓX , andΓe+e−X are the mass, total width, and partial width toe+e− , respectively,BηϕX is the branching fraction ofX→ηϕ decay, andθX is the relative phase.The BaBar measurement [5] shows that
KˉK∗(892) andηϕ are two major decays ofϕ(1680) andBKˉK∗(892)ϕ(1680)≈3×Bηϕϕ(1680) , whereBKˉK∗(892)ϕ(1680) is the branching fraction of theϕ(1680)→KˉK∗(892) decay. We also take the same form as in Ref. [5]:Γϕ(1680)(√s)=Γϕ(1680)⋅[PKK∗(892)(√s)PKˉK∗(892)(Mϕ(1680))BKˉK∗(892)ϕ(1680)+Pηϕ(√s)Pηϕ(Mϕ(1680))Bηϕϕ(1680)+(1−Bηϕϕ(1680)−BKˉK∗(892)ϕ(1680))].
(10) Here,
PKˉK∗(892) is the phase space of theϕ(1680)→KˉK∗(892) decay. The other decays ofϕ(1680) are neglected, and their phase space dependence are correspondingly ignored. Because both theKˉK∗(892) andηϕ final states contain a vector meson (V) and pseudoscalar meson (P), the phase space takes the formPVP(√s)=[(s+M2V−M2P)2−4sM2Vs]3/2.
(11) We take the same form of
X(1750) as in the BESIII measurement [14]:ΓX(1750)(√s)=ΓX(1750)⋅M2X(1750)s⋅[p(√s)p(MX(1750))]2l+1,
(12) where
p(√s) [p(MX(1750)) ] is the momentum of a daughter particle in the rest frame of the resonance with energy√s (massMX(1750) ), and l is the orbital angular momentum of the daughter particle.We describe the amplitude of the
ϕ(2170)→ηϕ decay as in Ref. [7]:Aϕ(2170)ηϕ(s)=√Bηϕϕ(2170)Γe+e−ϕ(2170)⋅√Γϕ(2170)/Pηϕ[Mϕ(2170)]⋅eiθϕ(2170)M2ϕ(2170)−s−i√sΓϕ(2170)⋅B(p)B(p′),
(13) where
Mϕ(2170) andΓϕ(2170) are the mass and width ofϕ(2170) ,B(p) is the P-wave Blatt-Weisskopf form factor, and p (p′ ) is the breakup momentum corresponding to√s (massMϕ(2170) ). -
We perform several combined fits to
σ(e+e−→ηϕ) measured by BaBar, Belle, BESIII, and CMD-3. These are fits with 1) onlyϕ(1680) , 2)ϕ(1680) and the non-resonant component, 3)ϕ(1680) ,ϕ(2170) , and the non-resonant component, and 4)ϕ(1680) ,X(1750) ,ϕ(2170) , and the non-resonant component. The input data areσ(e+e−→ηϕ) and the related uncertainties shown in Fig. 1. Based on the fit results, which are described below, we obtain the nominal fit results from the third case.The input data of the combined fits are the values of
σ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII, and CMD-3 experiments, and a leastχ2 method with MINUIT [16] is used. According to Ref. [15], we defineχ2k of the kth energy point asχ2k=∑i,j(Δσik)(C−1)ij(Δσjk),
(14) where
Δσik (Δσjk ) is the difference between the measured value from the ith (jth) data sample and the fitted value ofσ(e+e−→ηϕ) , and the effective global covariance matrix C is described in Eq. (6). The totalχ2 is the sum ofχ2k over all energy points.In the measurement of one experiment, there may be a correlation between the two modes of η decays in one
√s bin, or a correlation between all√s bins. For the first correlation, we calculateχ′2i=(C′ii)⋅(σγγ,imeasured−σifit)⋅(σπ+π−π0,imeasured−σifit)
(15) for the ith
√s bin. Here, we also use the relative uncertainty(δcor_systi)2 between the two modes of η decays to calculate the elementsC′ii of the correlation matrix. We obtainχ′2=∑iχ′2i for the sum of all the√s bins in one experiment. Similarly, we calculateχ′′2 for the second correlation asχ′′2=∑i,j(C′′ij)⋅(σimeasured−σifit)⋅(σjmeasured−σjfit),i≠j.
(16) Here, the matrix element
C′′ij=δcor_systi⋅δcor_systj is for the correlation between the ith and jth√s bins. Note thatδcor_systi refers to different correlated systematic uncertainties in the calculations ofχ′2 andχ′′2 .We add
χ′2 andχ′′2 to the totalχ2 for the constraints owing to the two types of correlations in the combined fits. -
Fitting to
σ(e+e−→ηϕ) measured by the four experiments with only theAϕ(1680)ηϕ(√s) component in Eq. (8), we obtain reasonably good results with a quality ofχ2/ndf=381/254 , as illustrated in Fig. 4. Here,ndf is the number of all fitted data points minus the number of free parameters. We obtain the resonant parameters ofϕ(1680) asMϕ(1680)=(1723±6)MeV/c2 ,Γϕ(1680)=(376±9)MeV , andBηϕϕ(1680)Γe+e−ϕ(1680)=(197±5)eV . The world average values of the mass and width ofϕ(1680) are(1680±20)MeV/c2 and(150±50)MeV [4], respectively. We can see that the mass and width from this fit are considerably different from the world average values, which is due to the absence of several components in our fit, such as the non-resonant contribution andϕ(2170) . We also notice that the world average value of the width has a large uncertainty. -
Fitting to
σ(e+e−→ηϕ) with onlyAϕ(1680)ηϕ(√s) and the non-resonant contribution in Eq. (8), we obtain two solutions of equivalent quality withχ2/ndf=347/251 , as illustrated in Fig. 5 and Table 2. Hereinafter, we use all the measured data from the four experiments as input for the combined fits but show only the combinedσ(e+e−→ηϕ) from Fig. 3 to represent the data in the plots. We obtain the same resonant parametersMϕ(1680)=(1676±3)MeV/c2 andΓϕ(1680)=(161+5−4)MeV whileBηϕϕ(1680)Γe+e−ϕ(1680)=(88±3) eV or(162+5−3)eV from the two solutions. The two resonant parameters have good agreement with the world average values, and the precision is effectively improved. Meanwhile, the branching fraction ofϕ(1680)→ηϕ isBηϕϕ(1680)=(20+4−3)% or(24±3)% , which is close to the value∼17% that can be calculated according to the BaBar measurement [6]. We can see that most of the measuredσ(e+e−→ηϕ) from the four experiments are above the fit curve around2.17GeV in Fig. 5, which indicates the requirement ofϕ(2170) .Figure 5. (color online) Fitting to
σ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII, and CMD-3 experiments, includingϕ(1680) and the non-resonant contribution. The blue solid lines show the fit results, and the red, green, and gray dashed lines show theϕ(1680) and non-resonant components. The interference between the non-resonant component andϕ(1680) is not shown.Parameters with ϕ(2170) without ϕ(2170) Solution I Solution II Solution III Solution IV Solution I Solution II χ2/ndf 284/247 347/251 a0 −0.12±0.02 4.4±0.2 1.1±0.2 −5.0±0.2 −0.9+0.1−0.3 −5.0±0.3 a1 −4.8+0.2−0.1 2.8±0.1 −3.1±0.2 2.6±0.3 1.0+0.4−0.3 3.0±0.1 Bηϕϕ(1680))Γe+e−ϕ(1680)/eV 79±4 127±5 65+5−4 215+8−5 88±3 162+5−3 Mϕ(1680)/(MeV/c2) 1678+5−3 1676±3 Γϕ(1680)/MeV 156±5 161+5−4 Bηϕϕ(1680)(%) 19+2−1 22±2 24+4−3 19+2−1 20+4−3 24±3 Bηϕϕ(2170)Γe+e−ϕ(2170)/eV 0.56+0.03−0.02 0.36+0.05−0.03 38±1 41±2 − Mϕ(2170)/(MeV/c2) 2169±5 − Γϕ(2170)/MeV 96+17−14 − θϕ(1680)/(∘) −63±12 −95±9 −88±6 −122±7 102+7−6 −94+11−6 θϕ(2170)/(∘) 81+14−9 −77+10−5 −159+19−15 133+16−13 − Table 2. Results of fitting to
σ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII, and CMD-3 experiments with the non-resonant component,ϕ(1680) , andϕ(2170) , or withoutϕ(2170) . -
With
Aϕ(1680)ηϕ andAϕ(2170)ηϕ but noAX(1750)ηϕ in Eq. (8), we obtain four solutions of equivalent quality withχ2/ndf= 284/247 from the nominal combined fit, as illustrated in Fig. 6 and Table 2. The four solutions have the same resonant parameters,Mϕ(1680) ,Γϕ(1680) ,Mϕ(2170) , andΓϕ(2170) :Mϕ(1680)=(1678+5−3±7)MeV/c2 ,Γϕ(1680)=(156±5±9)MeV ,Mϕ(2170)=(2169±5±6)MeV/c2 , andΓϕ(2170)=(96+17−14±9)MeV . We can see that the mass and width ofϕ(2170) are close to the world average values:mϕ(2170)=(2162±7)MeV/c2 andΓϕ(2170)=(100+31−23)MeV [4]. The four solutions show thatBηϕϕ(1680)Γe+e−ϕ(1680)=(79±4±16) ,(127±5±12) ,(65+5−4±13 or(215+8−5±11)eV , andBηϕϕ(2170)Γe+e−ϕ(2170)=(0.56+0.03−0.02±0.07) ,(0.36±0.04±0.07) ,(38±1±5) or(41±2±6)eV . The branching fraction ofϕ(1680)→ηϕ isBηϕϕ(1680)≈20% , with uncertainties of less than 5%. Comparing the change ofΔχ2=63 andΔndf=4 between the fit with and withoutϕ(2170) , we obtain the statistical significance of theϕ(2170) resonance to be7.2σ . By fixing the mass and width ofϕ(2170) at the world average values [4], we obtain the fit results listed in Table 3, with curves similar to those in Fig. 6. We then estimate the statistical significance ofϕ(2170) to be7.4σ . We describe the systematic uncertainties in the fit results in Sec. V.Figure 6. (color online) Nominal fit to
σ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII, and CMD-3 experiments, including theϕ(1680) ,ϕ(2170) , and non-resonant components. The blue solid lines show the fit results, and the red, green, and gray dashed lines show theϕ(1680) ,ϕ(2170) , and non-resonant components, respectively. The interference among theϕ(1680) ,ϕ(2170) , and non-resonant components is not shown.Parameters with ϕ(2170) Solution I Solution II Solution III Solution IV χ2/ndf 288/249 a0 −0.45±0.05 0.19±0.02 −4.1±0.2 −4.4±0.5 a1 −0.27+0.05−0.04 2.8±0.2 −0.43+0.08−0.05 2.6+0.5−0.3 Bηϕϕ(1680)Γe+e−ϕ(1680)/eV 85±3 123+6−4 53±6 193+6−5 Mϕ(1680)/(MeV/c2) 1677+5−4 Γϕ(1680)/MeV 158+5−4 Bηϕϕ(1680)(%) 18+5−3 22±5 23+5−3 20±5 Bηϕϕ(2170)Γe+e−ϕ(2170)/eV 0.48+0.04−0.02 0.37±0.03 38±1 37+2−1 Mϕ(2170)/(MeV/c2) 2162(fixed) Γϕ(2170)/MeV 100(fixed) θϕ(1680)/(∘) 80±12 −109±11 −88+12−7 −54±6 θϕ(2170)/(∘) −61+14−10 −48+10−8 −173+16−12 −165±9 Table 3. Results of fitting to
σ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII, and CMD-3 experiments with theϕ(1680) ,ϕ(2170) , and non-resonant components. The mass and width ofϕ(2170) are fixed at the world average values [4].As discussed in Ref. [17], there are
2n−1 solutions in a fit with n components in the amplitude. They have the same goodness of fit and the same mass and width of a resonance. Unfortunately, we cannot find a proliferation of the solutions as in several previous measurements [3, 18−20]. -
To investigate the production of
X(1750) in thee+e−→ηϕ process, we perform the combined fit with Eq. (8). We fix the mass and width ofX(1750) at the world average values [4] and obtain eight solutions of an equivalent quality ofχ2/ndf=290/245 , with the same masses and widths ofϕ(1680) andϕ(2170) . The fit results are listed in Table 4, and the first two solutions are shown in Fig. 7. ComparingΔχ2=6 andΔndf=2 in the fits with and withoutX(1750) , the statistical significance ofX(1750) is2.0σ . BecauseX(1750) is not significant here, we determine the upper limit (UL) of its production (BηϕX(1750)Γe+e−X(1750) ) in the eight solutions at the 90% C.L. by integrating the likelihood versus theX(1750) yield, as listed in Table 4.Parameters Solution I Solution II Solution III Solution IV Solution V Solution VI Solution VII Solution VIII χ2/ndf 290/245 a0 2.2+0.4−0.2 −1.2±0.2 −0.15±0.04 −0.12±0.02 −0.62+0.04−0.03 0.14±0.04 1.4+0.3−0.2 −1.2±0.4 a1 2.9+0.3−0.2 2.7±0.3 0.61+0.04−0.02 −1.2+0.2−0.1 1.6±0.3 0.79±0.04 1.6±0.2 2.5±0.2 Bηϕϕ(1680)Γe+e−ϕ(1680) /eV247+9−7 93±6 107±9 159+11−7 244+9−5 170+11−7 114+9−5 280±12 Mϕ(1680) /(MeV/c2) 1680±4 Γϕ(1680) /MeV147±8 Bηϕϕ(1680)(%) 19±3 18±3 19+4−3 21±3 22±3 19±3 21+4−2 22±4 BηϕX(1750)Γe+e−X(1750) /eV210+18−14 102+15−12 167+22−16 172±19 227+20−16 250+23−17 289±22 102+18−14 UL of BηϕX(1750)Γe+e−X(1750) /eV249 136 197 214 269 287 322 142 MX(1750) /(MeV/c2) 1754 (fixed) ΓX(1750) /MeV120 (fixed) Bηϕϕ(2170)Γe+e−ϕ(2170) /eV0.34±0.02 38+2−1 0.57±0.04 39+2−1 0.42±0.02 37+2−1 0.44+0.04−0.02 41±2 Mϕ(2170) /(MeV/c2) 2169+8−6 Γϕ(2170) /MeV95+22−14 θϕ(1680)/(∘) 98+11−9 109±17 88±16 −97+11−9 −134±17 119±15 −125±19 −109+20−14 θX(1750)/(∘) −55±7 −68+18−12 −74±14 105+20−14 63±12 −59±14 108±15 131+18−13 θϕ(2170))/(∘) −118±15 −94+16−13 −108±14 132±21 −83±17 −69+15−11 −127±24 −111±17 Table 4. Results of fitting to
σ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII, and CMD-3 experiments with theϕ(1680) ,X(1750) ,ϕ(2170) , and non-resonant components. The mass and width ofX(1750) are fixed at the world average values [4].Figure 7. (color online) Two solutions from the fitting to
σ(e+e−→ηϕ) measured by the BaBar, Belle, BESIII, and CMD-3 experiments withϕ(1680) ,X(1750) , andϕ(2170) . The blue solid lines show the best fit results, and the red, orange, green, and gray dashed lines show theϕ(1680) ,X(1750) ,ϕ(2170) , and non-resonant components, respectively. The interference among theϕ(1680) ,X(1750) ,ϕ(2170) , and non-resonant components is not shown. -
We characterize the following systematic uncertainties for the nominal fit results and estimate the uncertainty of the parameterization in Eq. (9) with two different parameterization methods, which have the forms
A(M)∝√Γϕ(1680)(√s)Γe+e−s−M2+iMΓϕ(1680)
(17) and
A(M)∝M√s⋅√Γϕ(1680)(√s)Γe+e−s−M2+iMΓϕ(1680).
(18) By changing the fit range to
[1.6,2.9]GeV/c2 , we find that the systematic uncertainty due to the fit range is negligible. We useAn.r.ηϕ(s)=a0/s to estimate the model dependence of the non-resonant contribution. We obtain the uncertainty inBKˉK∗(892)ϕ(1680)/Bηϕϕ(1680) by varying1σ according to the previous measurement [5]. To estimate the uncertainty due to the possible contribution fromϕ(1680)→ϕππ , we takeB(ϕ(1680)→ϕππ)=B(ϕ(1680)→ϕη)/2 [3] and modify Eq. (10) toΓϕ(1680)(√s)=Γϕ(1680)⋅[PKK∗(892)(√s)PKˉK∗(892)(Mϕ(1680))BKˉK∗(892)ϕ(1680)
+Pηϕ(√s)Pηϕ(Mϕ(1680))Bηϕϕ(1680)+Pϕππ(√s)Pϕππ(Mϕ(1680))Bϕππϕ(1680)+(1−Bηϕϕ(1680)−Bϕππϕ(1680)−BKˉK∗(892)ϕ(1680))]
(19) in the combined fits. Here,
Pϕππ is the phase space of theϕ(1680)→ϕππ decay.Assuming all these sources are independent and summing them in quadrature, the total systematic uncertainties are listed in Table 5.
Parameter Source An.r.ηϕ(s) BKˉK∗(892)ϕ(1680)/Bηϕϕ(1680) Parameterization ϕ(1680)→ϕππ Sum Mϕ(1680) 2 3 3 5 7 Γϕ(1680) 3 5 5 4 9 Γe+e−ϕ(1680)Bηϕϕ(1680) Sol. I 8 10 7 6 16 Sol. II 4 9 6 2 12 Sol. III 7 5 7 6 13 Sol. IV 6 5 4 7 11 Bηϕϕ(1680) Sol. I 0.7 0.9 1.3 1 2 Sol. II 0.5 0.9 1.2 2 4 Sol. III 0.9 1.5 1.8 2 3 Sol. IV 0.7 0.4 1.8 1 3 Mϕ(2170) 3 5 2 2 6 Γϕ(2170) 6 2 6 3 9 Γe+e−ϕ(2170)Bηϕϕ(2170) Sol. I 0.02 0.03 0.05 0.03 0.07 Sol. II 0.02 0.04 0.04 0.03 0.07 Sol. III 2 2 3 2 5 Sol. IV 3 3 4 1 6 Table 5. Systematic uncertainties of the resonance parameters for
ϕ(1680) andϕ(2170) . M, Γ,Γe+e−B , andB are the mass with units ofMeV/c2 , total width with units ofMeV , the production of the branching fraction and the partial width toe+e− with units ofeV , and the branching fraction (%), respectively. -
Combining the measurements by the BaBar, Belle, BESIII, and CMD-3 experiments, we calculate
σ(e+e−→ηϕ) from the threshold to3.95GeV with improved precision. There are clearϕ(1680) andJ/ψ signals and lineshape changes around the mass ofϕ(2170) in theηϕ final state. We perform combined fits toσ(e+e−→ηϕ) measured by the four experiments and obtain the nominal fit results with theϕ(1680) ,ϕ(2170) , and non-resonant components. The statistical significance ofϕ(2170) is7.2σ . The mass and width ofϕ(2170) areMϕ(2170)=(2169±5±6)MeV/c2 andΓϕ(2170)=(96+17−14±9)MeV , which are consistent with the world average values [4]. The mass and width ofϕ(1680) areMϕ(1680)=(1678+5−3±7)MeV/c2 andΓϕ(1680)=(156±5±9)MeV , with a good precision compared with the world average values [4]. The branching fraction of theϕ(1680)→ηϕ decay is approximately 20%, with uncertainties of less than 6%. We also determineBηϕϕ(1680)Γe+e−ϕ(1680) andBηϕϕ(2170)Γe+e−ϕ(2170) from the fits. Assuming its existence in thee+e−→ηϕ process, the statistical significance ofX(1750) is only2.0σ . We determine the UL ofX(1750) ine+e−→ηϕ at the 90% C.L. -
We thank Prof. Changzheng Yuan for very helpful discussions.
Determination of the resonant parameters of excited vector strangenia with e+e−→ηϕ data
- Received Date: 2023-05-23
- Available Online: 2023-11-15
Abstract: We determine the resonant parameters of the vector states