Status of rare exclusive B meson decays in 2018

# Status of rare exclusive B meson decays in 2018

Johannes Albrecht    Stefanie Reichert    Danny van DykaaaTUM-HEP-1146/18
###### Abstract

This review discusses the present experimental and theoretical status of rare flavour-changing neutral current -quark decays at the beginning of 2018. It includes a discussion of the experimental situation and details of the currently observed anomalies in measurements of flavour observables, including lepton flavour universality. Progress on the theory side, within and beyond the Standard Model theory is also discussed, together with potential New Physics interpretations of the present measurements.

Keywords: Flavour Physics, Rare Decays, Exclusive B Decays

Revised Day Month Year

PACS numbers: 13.20.He 13.30.Ce 12.15.Mm 12.60.Jv

## 1 Introduction

The year 2018 marks a special point for the field of particle physics as it is not only the end of second period of data-taking at the Large Hadron Collider(LHC) but also the start-up of the Belle II experiment. In the field of flavour physics, a number of anomalies have been observed when comparing measurements to the Standard Model ( Standard Model (SM)) expectation; from which a coherent and conclusive picture emerges. Those anomalies could, if they persist, point towards certain New Physics (NP) models such as leptoquarks or bosons. In this review, we will give a comprehensive overview of the current state-of-the-art of exclusive rare decays from both experimental and theoretical perspective. We will present the current experimental landscape with a focus of the observed anomalies and discuss potential new physics interpretations.

The LHCb experiment will continue to run for the next two decades and hence will be able to either confirm or rule out many of the present-day anomalies. With the imminent start-up of the data-taking at the Belle II experiment, these tensions will be independently cross-checked and further complementary measurements will be accessible.

As the progress in inclusive decays was limited after the end of the factories, this review focusses on exclusive measurements of rare decays of hadrons as these are excellent probes for many new physics scenarios. These rare decay processes are CKM-, GIM- and loop-suppressed, wherefore potential new physics effects can be large compared to the SM amplitude. Such indirect searches allow to probe NP models at much higher mass scales as are currently accessible through direct searches. In the past, exclusive decays with muons in the final state have been measured extensively by the LHC experiments, most notably by LHCb, as well as the factories. We review the status of these measurements, discuss their theoretical description in the SM and model-independently in the presence of new physics. A special focus is set on the anomalies seen in angular distributions of and in the branching fraction measurements of all exclusive decays as well as to the tensions observed in tests of lepton flavour universality in this class of decays. Measurements of the type are not covered in this review.

On the theoretical side, the current state of the SM predictions for the discussed observables is presented. The anatomy of the amplitudes, and their dependence on various local and non-local hadronic matrix elements is discussed. Particular focus is set on recent developments for the non-local matrix elements. The model-independent interpretation of the measurements is examined in the framework of the usual Effective Field Theory.

This review is structured as follows: Section 2 introduces the amplitudes of decays in an effective field theory approach. Section 3 discusses the experimental landscape with a focus on semileptonic decays, which is followed by a brief discussion of decays, purely leptonic and decays. The experimental part of the review closes with a brief discussion of decays.

The second part of the review, Section 4, examines the foundation for the SM predictions required in the first part of the review. Section 5 discusses then a combined interpretation of all presented measurements in the framework of global fits and possible interpretations of the observed patterns in NP frameworks.

The review closes with a discussion of the experimental outlook in Section 6.

## 2 Anatomy of the amplitudes

Within the SM and at the Born level there are no flavor-changing neutral currents (FCNC). As a consequence, any FCNC-mediated quark-flavor transitions, such as , emerge only from virtual loop corrections, in which a is first emitted and subsequently re-absorbed. The emergence at the loop level naturally suppresses the rate at which these processes occur. In the SM, the unitarity of the Cabibbo-Kobayashi-Maskawa (CKM) matrix introduces an additional suppression mechanism that renders transitions very rare . Moreover, descriptions of these rare decays in the SM are further complicated since they pose a multi-scale problem: as weak decays they are mediated through the exchange of and bosons, which are much heavier than the remaining particles (with the exception of the quark). The use of an Effective Field Theory (EFT) helps our understanding of such multi-scale dynamics. In the following, we will discuss the EFT used for predictions of rare decays below the electroweak breaking scale . Within the SM, this EFT captures the effects of the quark as well as the and bosons, which are no-longer dynamical degrees of freedom for scales .

The effective Lagrangian reads (see e.g. )

 (1)

where and denote the Lagrangians of the electromagnetic and strong interactions (with five quark flavours after integrating out the top quark), refers to the Fermi constant, and the last term captures the local effective operators with their effective couplings - or Wilson coefficients (WC)- at the renormalisation scale . For convenience, the product of CKM matrix elements has been extracted from the definition of the WC.

The matrix elements of the effective operators need to be evaluated at a low scale , which minimises logarithms in the perturbative expansion of the matrix elements. Consequently, one requires the WC evaluated at . Problems arising from large logarithms are resummed through Renormalisation Group (RG) improved running. Contemporary analyses use resummation of QCD-induced large logarithms up to Next-to-Next-to-Leading-Logarithm (NNLL). This requires knowledge of the relevant anomalous dimensions at the four-loop level, and the matching conditions at the three-loop level . The electroweak effects to NLL, which are numerically sub-leading, are known only for a subset of the WC .

For a consistent treatment at the leading order in the electromagnetic coupling , all operators of the following set - usually called the SM basis - are required (see e.g. ). The SM basis consists of the current-current operators ()

 O1q =[¯sγμTAPLq][¯qγμTAPLb], O2q =[¯sγμPLq][¯qγμPLb]; (2)

the QCD-penguin operators

 O3 =[¯sγμPLb]∑q[¯qγμq], O4 =[¯sγμTAPLb]∑q[¯qγμTAq], (3) O5 =[¯sγμνρPLb]∑q[¯qγμνρq], O6 =[¯sγμνρTAPLb]∑q[¯qγμνρTAq];

the electromagnetic and chromomagnetic dipole operators

 O7 =e16π2¯¯¯¯¯mb[¯sσμνPRb]Fμν, O8 =gs16π2¯¯¯¯¯mb[¯sσμνTAPRb]GAμν; (4)

and the semileptonic operators

 O9 =e216π2[¯sγμPLb][¯ℓγμℓ], O10 =e216π2[¯sγμPLb][¯ℓγμγ5ℓ]. (5)

In the above, we abbreviate , , , and sums over run over all active quark flavours , , , , and . Through using , universal QED corrections at NLO are taken care of .

Searches for NP effects at energies smaller than have so far not discovered either new interactions or new particles. Assuming that no such low-mass fields exist, the EFT framework, which is necessary for accurate theory predictions, can also be used to systematically describe NP effects. To this end, the effective Lagrangian Eq. (2) has to be modified in the following way :

1. []

2. the set of effective field operators is enlarged to include all operators allowed by field content and Lorentz symmetry up to a given mass dimension (typically mass dimension six),

3. their WC are assumed to be independent and uncorrelated parameters.

When limiting this enlarged set to only semileptonic operators, a basis of all operators up to and including mass dimension six can be formed by also including the chirality-flipped operators,

 O9′ =e216π2[¯sγμPRb][¯ℓγμℓ], O10′ =e216π2[¯sγμPRb][¯ℓγμγ5ℓ]; (6)

the (pseudo)scalar operators

 OS =e216π2[¯sγμPRb][¯ℓℓ], OP =e216π2[¯sγμPRb][¯ℓγ5ℓ], (7) OS′ =e216π2[¯sγμPLb][¯ℓℓ], OP′ =e216π2[¯sγμPLb][¯ℓγ5ℓ];

and the tensor operators

 OT =e216π2[¯sσμνb][¯ℓσμνℓ], OT5 =e216π2[¯sσμνb][¯ℓσμνγ5ℓ]. (8)

A complete and non-redundant set of dimension-six operators, including their one-loop anomalous dimensions in both QCD and QED is compiled in .

This bottom up approach probes model-independently for deviations from the SMthe WCs are sensitive to NP effects of new particles that are too heavy to be produced directly. Rare semileptonic decays therefore provide information that is complementary to the “direct” searches for new interactions and particles that are carried out at the Large Hadron Collider.

The above basis is further reduced when one assumes a manifest invariance of NP effects under the SM gauge group within a SM-like (i.e. linear ) Higgs model. For the operators, this was explicitly demonstrated in .bbb Note, however, that a nonlinear representation fully restores the basis of dimension six operators to the set introduced above . The effective Lagrangian in Eq. (2) can be matched onto the SM Effective Field Theory (SMEFT) (see e.g. ), whose operators are manifestly invariant under the SM gauge group. The required matching formulas are compiled in  to leading non-trivial loop level. Matching, basis transformations and Renormalization-Group-Equation (RGE) evolution can be conveniently achieved through a variety of computational tools, including but not limited to: the “DsixTools” Mathematica package ; the “Wilson Coefficient Exchange Format” and its reference Python implementation ; and the “Wilson” Python package .

Schematically, the matrix elements for all exclusive processes can now be expressed as

 A∼C10F10+[C9F9−2mbMBq2C7F7]−32π2M2Bq2H+O(αe), (9)

where refers to the invariant mass squared of the di-lepton pair. The refer to hadronic matrix elements of local currents as induced by the operators , while denotes matrix elements of time-ordered products involving four-quark operators and the chromomagnetic operator together with the electromagnetic current. In the presence of NP effects in the semileptonic and radiative operators, the hadronic matrix elements remain unchanged. However, the coefficients multiplying the latter are then modified. For the complete anatomy of the amplitudes of decays in the presence of NP operators of mass dimension six we refer to . A similar study for is presented in .

## 3 Experimental measurements

In this section, we draw a picture of the current experimental status of semi-leptonic (including a brief discussion of ) decays, purely leptonic and radiative transitions. Hereby, we will discuss measurements of the various hadronic final states, e.g. kaon and states, giving access to a broad range of observables such as branching fractions, and isospin asymmetries as well as angular observables. Within this section, the LHCb measurements refer to the full Run 1 dataset corresponding to 3 collected during the years 2011 and 2012, the BaBar results were obtained on their full dataset of 424 and the Belle publications exploit their complete dataset of 711 if not otherwise stated. Presented results integrated over the whole range have been obtained by vetoing the charmonium resonances and interpolating over this vetoed region. It should be noted that the exact range to veto the charmonium resonances depends on the respective analysis.

### 3.1 Semileptonic b→sℓ+ℓ− decays

Particular interest was raised throughout the past years by various measurements of semileptonic decays, in which several tensions between the SM predictions and experimental measurements have been observed. These deviations are mostly in the range of two to three standard deviations and show a consistent pattern, which can be explained by lowering the WC with respect to its SM value  as detailed further in Section 5.1.

#### 3.1.1 B→Kℓ+ℓ− decays

The differential branching fractions of the decays and versus the invariant mass squared of the dimuon pair, , were determined by the LHCb collaboration with the -averaged isospin asymmetry   defined as

 AI =Γ(B0→K(∗)0μ+μ−)−Γ(B+→K(∗)+μ+μ−)Γ(B0→K(∗)0μ+μ−)+Γ(B+→K(∗)+μ+μ−), (10) =B(B0→K(∗)0μ+μ−)−(τ0/τ+)⋅B(B+→K(∗)+μ+μ−)B(B0→K(∗)0μ+μ−)+(τ0/τ+)⋅B(B+→K(∗)+μ+μ−), (11)

with the partial widths and branching fractions of the corresponding decay channels and the ratio of to lifetimes . The analysis finds values individually compatible with the SM prediction. However, the entity of measurements lies systematically below the predictions as shown in Fig. 1. In the same analysis, the -averaged isospin asymmetry has been measured and is depicted in Fig. 1. In a similar analysis, the asymmetry

 ACP =Γ(¯B→¯K(∗)μ+μ−)−Γ(¯B→¯K(∗)μ+μ−)Γ(¯B→¯K(∗)μ+μ−)+Γ(¯B→¯K(∗)μ+μ−), (12)

was determined under the assumption of no direct violation in the control mode to be , where the uncertainties are statistical and systematic, respectively , and this measurement is consistent with the SM prediction .

The double differential decay rate of decays  is given by

 1ΓdΓdcosθl=34(1−FH)(1−cos2θl)+12FH+AFBcosθl, (13)

where the angle between the () and the oppositely charged kaon () of the () decay is denoted by , is the so-called flat term, and is the forward-backward asymmetry of the dimuon system.

As the flavour of the neutral meson cannot be determined for the self-conjugate final state , the double differential decay rate is extracted as a function of the absolute value of as

 1ΓdΓd|cosθl|=32(1−FH)(1−|cosθl|2)+FH, (14)

with the constraint of enforcing the expression to be positive-definite for all values of  . The results of and are compatible with the SM predictions  and are shown in Fig. 2.

Interference effects between the short- and long-distance contributions were studied in decays at low recoil . The LHCb collaboration reported the first observation of the decays and the subsequent decay with , which is determined under the assumption of lepton flavour universality and hence the second uncertainty stems from the known branching fraction. No significant signal was observed for the resonance, and an upper limit at C.L. is set. The mean and the width of the resonance were measured to be and , respectively, where the uncertainties comprise both statistical and systematic sources. The interference between the observed resonant and the non-resonant decay in the large region amounts to   leading to renewed interest in previous estimates of quark-hadron duality violation for these processes  as discussed in Section 4.2.

By assuming a model of hadronic resonances, the LHCb collaboration fits the differential branching fraction of to data. The branching fraction of the short-distance component is determined by setting the and resonance amplitudes to zero, and is found to be  , where the uncertainties are statistical (including the form factor uncertainties) and systematic. In addition, the phase difference between the short-distance and the narrow-resonance amplitudes in decays was determined by performing a fit to the mass of the dimuon pair; of which one possible solution is illustrated in Fig. 3. As the values of the phases are compatible with , the interference with the short-distance contributions far from the pole masses is small.

#### 3.1.2 B→K∗ℓ+ℓ− decays

In addition to the aforementioned studies of and states, channels with a pair of mesons in the final state have received particular attention over the past years as those decays show a rich phenomenology. Amongst these decays, events with a invariant mass close to the vector resonance (892) are the subject of numerous studies . Here and in the following, we discuss treating the (892) as a quasi-stable particle, and decays should be interpreted as the decay chain if not otherwise stated.

The asymmetry defined in Eq. (3.1.1) was measured by the LHCb collaboration to be , where the uncertainties are statistical and systematic, respectively , and the measurement is found to be consistent with the SM expectation of a small asymmetry in this decay.

The differential branching fraction of a charged meson into a final state, , has been measured by the LHCb collaboration; the results are shown in Fig. 4 in several bins of . A study of the neutral decay mode was performed, which resulted in the measurement of the isospin asymmetry as defined in Eq. (3.1.1) illustrated in Fig. 4. As the differential branching fractions of the decay had been previously reported in , the values were not updated in  until later in . The latter results are as well depicted in Fig. 5, and the analysis yields the most precise measurement to date of the -averaged branching ratio

 B(B0→K∗0μ+μ−)/δq2=(0.342+0.17−0.17±0.009±0.023)⋅10−7c4GeV2, (15)

with (in the bin ) where the uncertainties are statistical, systematic and from the uncertainty on the branching fraction of the normalisation channel . In the region and for the invariant mass range , the S-wave component is measured to be  , where the uncertainties are of statistical and systematic origin. Assuming the absence of high-order waves e.g. D- and F-waves, the pure P-wave component of the differential branching fraction has been determined for the first time. However, in previous analyses, the S-wave fraction was not taken into account as is the case in a measurement of the differential branching fraction published by the CMS collaboration , whose results are shown in Fig. 5.

The four-dimensional differential decay rate of () decays is expressed as a function of the invariant mass of the lepton pair , and the angles , and , where refers to the angle between the and the () flight directions in the di-lepton rest frame, and to the angle between the () and the () flight directions in the rest frame and corresponds to the angle between the planes defined by the di-muon and the kaon and pion in the rest frame . The fully differential decay rate is then given by

 d4¯Γ(B0→K∗0μ+μ−)dqdcosθℓdcosθK∗dcosϕ =932π∑¯Ii(q,θK∗)fi(θℓ,ϕ), (16) d4Γ(¯¯¯B0→¯¯¯¯K∗0μ+μ−)dqdcosθℓdcosθK∗dcosϕ =932π∑Ii(q,θK∗)fi(θℓ,ϕ), (17)

where the () terms depend on products of the spin amplitudes and the functions are the corresponding angular distribution functionscccThe differential decay rate taking the full basis including scalar and tensor operators has been derived in . assuming an on-shell meson . As the theoretical calculations define with respect to the negatively charged muon in both and decays, the angular distributions between theory and experiment differ. A translation scheme between the two conventions is given in . More common than expressing the differential decay rates as functions of and is using the -averaged and -asymmetric observables and as introduced in .

 Si =(Ii+¯Ii)/(dΓdq2+d¯Γdq2), (18) Ai =(Ii−¯Ii)/(dΓdq2+d¯Γdq2). (19)

One can construct a complete set of observables with a reduced form-factor dependence  in the large energy limit. They emerge from combinations of and , where is the most notable and hence relevant for further discussions in this review.

In the angular analysis of decays by the CMS collaboration on a dataset of recorded in 2012, both and were determined , by using a folded differential decay rate - an approach originally developed by the LHCb collaboration . After the folding, a multi-dimensional fit is performed, with the parameters of interest , and along with signal and background yields, whereas the values of the longitudinal polarisation of the meson, , and and the interference between S- and P-wave, , have been fixed to values determined in a previous analysis performed on the same dataset . The results of and are shown in Fig. 6 and are consistent with the SM predictions and previous measurements.

The aforementioned folding technique has also been employed by the Belle collaboration to extract and from the full dataset for both and decays, as well as the combination of both leptonic final states . The results on and are shown in Fig. 7. Tensions between measurement and SM prediction  are observed for for the muon final state in the region of , whereas the electron mode differs by in the very same region, leading to a combined tension of . In addition to the observables, the so-called observables are determined for the first time; any deviation of these observables from zero would be a clear sign for new physics . However in the Belle analysis, no deviation from zero is observed as can be seen in Fig. 7.

A more complete angular analysis was performed by the ATLAS collaboration on its 20.3 dataset from 2012 by exploiting four different folding schemes allowing to extract a set of four observables for each scheme . However neither nor can be measured with the chosen approach. The values for and , which are common to each folding scheme, have been compared between the four fits and have found to be consistent. The publication comprises results on , , and , from which and are shown in Fig. 8. The obtained results are consistent with the different SM predictions within less than three standard deviations.

In contrast to previous analyses, the LHCb collaboration has performed a full angular analysis, and has determined the set of angular observables by employing three different methods: by a maximum likelihood fit, by exploiting the angular principal moments , and in addition the zero-crossing points of and were determined from a fit to the decay amplitudes . The full angular analysis gives access to the correlation matrices, which provide crucial input for global fits to theoretical models. The results of the maximum likelihood and the method of angular principal moments are found to be compatible. As the maximum likelihood yields the most precise results, we restrict the discussion to this approach. However, we note in passing that the angular principal moments allow for smaller bin widths in and therefore provide more information of the shape of the angular observables compared to the maximum likelihood method. In this analysis performed by the LHCb collaboration, the fraction of the S-wave component was taken into account. Neglecting theoretical correlations, the observables appear to be compatible with the SM predictions. The theory correlations can be displayed best in the space of the observables with the most notable deviation of the SM in as can be seen from Fig. 9: in the regions and , deviations of the measured values from the SM prediction  of and can be observed, respectively. This confirms the tension seen in a previous LHCb analysis  in the region , which had a local significance of .

A fit to the complete set of -averaged observables, namely , and , as determined from the maximum likelihood fit is performed by the LHCb collaboration  using the EOS software package  for observables in the ranges below and . The fit result illustrated in Fig. 10 indicates a tension of between the measurements and the SM prediction of alone.

When the fully differential decay rate (see Eq. (3.1.2)) is integrated over and , the resulting expression depends on and , which is what the BaBar collaboration exploited to measure those parameters  on their full dataset and to extract . The studied decay channels include not only and decays but also their charged counterparts. The results are summarised in Fig. d and most results are compatible amongst each other and with the SM prediction. In the low region, a tension is observed in between and decays as well as the SM prediction. In the same region, there appears to be a small tension for .

Although most previous analyses focus on the muonic final states, several analyses measure e.g.. the branching fraction or angular observables for decays. On the 2011 LHCb dataset corresponding to 1, the branching fraction of the electron mode has been determined in the dielectron mass range to be  , where the uncertainties are statistical, systematic and from the normalisation mode. This result is found to agree with SM predictions. On the 3 dataset, an angular analysis has been performed  in the effective range yielding

 FL =+0.16±0.06±0.03, (20) P1=A(2)T =−0.23±0.23±0.05, (21) AImT =+0.14±0.22±0.05, (22) AReT =+0.10±0.18±0.05, (23)

where the uncertainties are statistical and systematic, respectively, and these results are in agreement with the SM expectations.

#### 3.1.3 B→K∗ℓ+ℓ− decays beyond the ground state

Apart from kaon and states, one can consider higher resonances above the mass range. The LHCb collaboration reported the first observation of the decay channels and , whose branching fractions are

 B(B+→K+π+π−μ+μ−) =(4.36+0.29−0.27±0.21±0.18)⋅10−7, (24) B(B+→ϕK+μ+μ−) =(0.82+0.19−0.17+0.10−0.04±0.27)⋅10−7, (25)

where the uncertainties are statistical, systematic and originating from the normalisation mode . For the channel, the differential branching fractions illustrated in Fig. 12 could be measured.

In the region above the mass around , the following resonances decaying to a final state contribute to the spectrum: the S-wave , the P-waves and and the D-wave . The LHCb collaboration has, for the first time, studied the differential branching fraction of the decay and performed a full angular analysis including S-, P-, and D-wave contributions in the region  . The former results are shown in Fig.