B meson decays
© Stone et al 2009
Received: 20 February 2009
Accepted: 20 February 2009
Published: 20 February 2009
Skip to main content
© Stone et al 2009
Received: 20 February 2009
Accepted: 20 February 2009
Published: 20 February 2009
We discuss the most important Physics thus far extracted from studies of B meson decays. Measurements of the four CP violating angles accessible in B decay are reviewed as well as direct CP violation. A detailed discussion of the measurements of the CKM elements V cb and V ub from semileptonic decays is given, and the differences between resulting values using inclusive decays versus exclusive decays is discussed. Measurements of "rare" decays are also reviewed. We point out where CP violating and rare decays could lead to observations of physics beyond that of the Standard Model in future experiments. If such physics is found by directly observation of new particles, e.g. in LHC experiments, B decays can play a decisive role in interpreting the nature of these particles.
PACS Codes: 13.25.Hw, 14.40.Nd, 14.65.Fy
The forces of nature generally reveal their properties by how they act on matter. On the most fundamental material scale that we are aware of matter is formed from fermions. These take two forms, leptonic matter and quark matter. The former do not have any strong interactions, that is they lack a property called "color charge", which allow quarks to bind together either mesonically or baryonically. New and therefore as yet unknown forces could effect both leptons and quarks. Here, we concentrate on how such forces effect quarks, especially the b quark.
Light matter consists mostly of u, d and s quarks. In 1963 Cabibbo showed that weak interactions of mesons and baryons containing s quarks were suppressed with respect to those without s quarks by an amount tan θ C , where the "Cabibbo" angle θ C must be determined experimentally . The s was further shown to have an important and at that time a mystifying role, by the discovery of CP violation in decays in 1964 [2, 3]. (In CP, C stands for charge conservation invariance and P for parity invariance. The operation of C takes a particle to anti-particle state and the operation of P changes left to right. [4–6].) When the c quark was discovered in Nov. 1974 [7, 8] (though its existence was speculated earlier [9–11]), it became clear that θ C was the mixing angle between two quark doublets, (c, s) and (u, d). However, it is not possible to generate a CP violating phase with only two quark doublets.
This was recognized even before the discovery of the charm quark by Kobayashi and Maskawa, who postulated the existence of yet another quark doublet (b, t) , in work for which they were awarded the Nobel Prize in 2008. While the t quark is the heaviest, having a mass of 173 GeV, they are difficult to produce and decay before they can form a hadron, thus excluding many otherwise possible studies. Much interesting work has been done with the s and c quarks, but in this article we discuss the physics of the b quark, which turns out to be the most interesting of the six quarks to study.
First we will discuss how particles formed with b-quarks fit into current paradigm of particle physics, the "Standard Model" (SM) [13–15]. The SM has at its basis the gauge group SU(3)xSU(2)xU(1). The first term corresponds to the strong interaction and SU(3) describes the octet of colored gluons which are the strong force carriers of quantum chromodynamics. SU(2)xU(1) describes the weak interaction and is the product of weak isospin and hypercharge. We speak of the fundamental objects being spin-1/2 quarks and leptons and the force carriers generally spin-1 objects. The spin-0 Higgs boson, yet to be discovered, is necessary for generating mass (to view a cartoon of the mechanism of Higgs mass generation, see ).
Particles containing b quarks can be B 0, B -, B s , or B c mesons, depending on whether the light anti-quark that it pairs with is , , , or or a baryon containing two other quarks. Mesons containing b and quarks are also interesting especially for studies of quantum chromodynamics, but will not be discussed further in this article, as we will concentrate on weak decays and discuss strong interactions as an important and necessary complication that must be understood in many cases to extract information on fundamental b quark couplings.
The quarks come in three repetitions called generations, as do the leptons. The first generation is d u, the second s c and the third b t. In the second and third generations the charge +2/3 quark is heavier than the charge -1/3; the first generation has two very light quarks on the order of a few MeV with the d thought to be a bit heavier. (Isospin invariance is related to the equality of u and d quark masses. The PDG  gives the u quark mass between 1.5–3.3 MeV and the d mass between 3.5–6.0 MeV, where the large range indicates the considerable uncertainties.) Decays usually proceed within generations, so the c decays predominantly to the s quark via the quark level process c → W + s, though some decays do go to the first generation as c → W + d. The ratio of these amplitudes approximate the Cabibbo angle discussed earlier.
"Dark Matter" was first shown to exist by Zwicky studying rotation curves of galaxies . The motion could only be explained if there was massive cloud of matter that was not luminous. We still do not know what composes this dark matter, though hopes are it will be discovered at the LHC. An even more mysterious phenomena called "Dark Energy" may also have a connection to particle physics experiments , perhaps via "Extra Dimensions" .
When the Universe began with the Big Bang, there was an equal amount of matter and antimatter. Now we have mostly matter. How did it happen? Sakharov gave three necessary conditions: Baryon ( ) number violation, departure from thermal equilibrium, and C and CP violation . (The operation of Charge Conjugation (C) takes particle to anti-particle and Parity (P) takes a vector to - .)
These criteria are all satisfied by the Standard Model. is violated in Electroweak theory at high temperature, though baryon minus lepton number is conserved; in addition we need quantum tunneling, which is powerfully suppressed at the low temperatures that we now have. Non-thermal equilibrium is provided by the electroweak phase transition. C and CP are violated by weak interactions. However the violation is too small. The ratio of the number of baryons to the entropy in the observed part of the Universe needs to be ~5 × 10-11, while the SM provides many orders of magnitude less. Therefore, there must be new physics .
Our worry is why the Planck scale at ~1019 GeV is so much higher than the scale at which we expect to find the Higgs Boson, ~100 GeV. As Lisa Randall said  "The gist of it is that the universe seems to have two entirely different mass scales, and we don't understand why they are so different. There's what's called the Planck scale, which is associated with gravitational interactions. It's a huge mass scale, but because gravitational forces are proportional to one over the mass squared, that means gravity is a very weak interaction. In units of GeV, which is how we measure masses, the Planck scale is 1019 GeV. Then there's the electroweak scale, which sets the masses for the W and Z bosons. These are particles that are similar to the photons of electromagnetism and which we have observed and studied well. They have a mass of about 100 GeV. So the hierarchy problem, in its simplest manifestation, is how can you have these particles be so light when the other scale is so big." We expect the explanation lies in physics beyond the Standard Model .
When we make measurements on B decays we observe the contributions of SM processes as well as any other processes that may be due to physics beyond the SM or New Physics (NP). Other diagrams would appear adding to those in Figure 2 with new intermediate particles. Thus, when it is declared by those who say that there isn't any evidence of NP in B decays, we have to be very careful that we have not absorbed such new evidence into what we declare to be SM physics. There are several approaches that can be followed.
One approach is to simply predict the decay rate of a single process in the SM with known couplings and compare to the measurements. The classical case here is b → s γ and we will discuss this and other specific examples later. Another approach is make different measurements of the CKM parameters in different ways and see if they agree. This is normally done by measuring both angles and sides of the CKM triangle, but other quantities can also be used. This is the approach used by the CKM fitter  (see also ) and UT fit groups . In yet a third approach, the exact same quantity can be measured in several ways, even if cannot be predicted in the SM. An example here is measuring the CP violating angle β using B 0 → J/ψ K S decays that proceed through the diagram in Figure 2(b), at least in the SM, and another process that uses the "Penguin" diagram in Figure 2(f), e.g. B 0 → ϕ K S .
The punch line is that if new, more massive particles exist in a mass range accessible to the LHC then they MUST contribute to rare and CP violating B decays! Even if measurements are precise enough only to limit the size of these effects, the properties of these new particles will be much better understood. This is the raison d'être for the further study of B decays.
where H refers to the heavier and L the lighter of the two weak eigenstates, and ϕ = arg(-M 12/Γ12). We expect that ΔΓ is very small for B 0 mesons but should be significant for B s mesons.
Data from many experiments has been combined by "The Heavy Flavor Averaging Group," (HFAG) to obtain an average value Δm d = (0.507 ± 0.004) × 1012 ps-1 .
where B B is a parameter related to the probability of the d and quarks forming a hadron and must be estimated theoretically, F is a known function which increases approximately as , and η QCD is a QCD correction, with value about 0.8. By far the largest uncertainty arises from the decay constant, f B .
In principle f B can be measured. The decay rate of the annihilation process is proportional to the product of . Of course there is a substantial uncertainty associated with |V ub |. The experimental evidence for is discussed in Section 5.1, and substantial uncertainty exists in the branching ratio measurement. Thus, we need to rely on theory for a value of f B .
ignoring the higher order term in V ts . Solving the ratio of the two above equations for gives a circle centered at (1,0) in the ρ - η plane whose radius depends on x d /x s . The theoretical errors are now due to the difference between having a light quark versus a strange quark in the B meson, called SU(3) splitting.
For many years experiments at the Z 0 using e + e - colliders at both LEP and the SLC had set lower limits on mixing . In 2006 B s mixing was measured by the CDF collaboration [36, 37]. The D0 collaboration had previously presented both a lower and an upper limit .
where t is the proper time.
where the first error is statistical and the second systematic.
In order to translate the mixing measurements to constraints on the CKM parameters ρ and η, we need to use theoretical values for the ratios and in Eq. 8. It is interesting that in practice it is not possible to measure either of these numbers directly. It is usually assumed, however, that , which could in principle be measured via the process B - → τ - ν is the same as f B . There is no way to measure . The charm system, on the other hand, provides us with both D + → ℓ+ ν and , and both rates have been measured. They also have been calculated in unquenched lattice quantum electrodynamics (QCD).
The combined efforts of the HPQCD and UKQCD collaborations predict MeV , while the CLEO measurement is in astonishing agreement: (205.8 ± 8.5 ± 2.5) MeV . Furthermore, the measurements of are not in such good agreement with the Follana et al calculation of (241 ± 3) MeV. The average of CLEO and Belle results as determined by Rosner and Stone is (273 ± 10) MeV . (For updated results see [43, 44].) The discrepancy is at the 3 standard deviation level and not yet statistically significant, though it bears watching.
Unfortunately, the group that has calculated is not the same group that has determined . The theoretical calculations used for the B B terms and the f B terms are summarized by Tantalo . He suggests using values of , , and B B = 0.83 ± 0.01 ± 0.06. These numbers allow us to measure the length of one side of the CKM triangle using Eq. 8. Use of this measurement will be discussed in more detail in Section 6.
We have two quantum mechanical operators: Charge Conjugation, C, and Parity, P. When applied to a particle wavefunction C changes particle to antiparticle and vice-versa. Applying P to a wavefunction ψ (r) we have P ψ (r) = ψ (-r). The P operator can be thought of changing the natural coordinate system from right-handed to left-handed. If nature was blind to handedness, then P would be always conserved. By applying the P operator twice we end up with P 2 ψ (r) = ψ (r), so the eigenvalues of P are ± 1. Therefore wave-functions, or particles represented by such wave-functions, have either intrinsic positive parity +1 (right-handed) or -1 (left-handed).
Weak interactions, characterized by a combination of vector minus axial-vector currents, are known to be left-handed. Therefore, handedness matters and its well known that Parity is maximally violated in weak decays . Since C changes left-handed particles to right-handed anti-particles, the product CP symmetry could have been preserved, but nature decided otherwise. Different particle transitions involve the different CP violating angles shown in Figure 1. Measuring these independently allows comparisons with measurements of the sides of the triangle and any differences in constraints in ρ and η can be due to the presence of new physics.
Any two amplitudes will do, though its better that they be of approximately equal size. Thus charged B decays can exhibit CP violation as well as neutral B decays. In some cases, we will see that it is possible to guarantee that is unity, so we can get information on the weak phases. In the case of neutral B decays, mixing serves as the second amplitude.
CP is violated if ϵ B ≠ 0, which occurs if |q/p| ≠ 1.
where m = (m L + m H )/2, Δm = m H - m L and Γ = Γ L ≈ Γ H , and t is the decay time in the B 0 rest frame, the so-called "proper time". Note that the probability of a B 0 decay as a function of t is given by ⟨B 0 (t)|B 0(t)⟩*, and is a pure exponential, e -Γt/2, in the absence of CP violation.
which requires only that λ acquire a non-zero phase, i.e. |λ| could be unity and CP violation can occur.
for either B 0 or for B s mesons, separately.
A comment on neutral B production at e + e - colliders is in order. At the ϒ(4S) resonance there is coherent production of B 0 pairs. This puts the B's in a C = -1 state. In hadron colliders, or at e + e - machines operating at the Z 0, the B's are produced incoherently. For the rest of this article we will assume incoherent production except where explicitly noted.
This is quite lucky for the study of B d mesons, as the maximum size of the coefficient for any x is -0.5, close to the measured value of x d .
is real to order 1/λ 4.
which is real to order λ 4. It necessary to include this term, however, since there are other formulations of the CKM matrix than Wolfenstein, which have the phase in a different location. It is important that the physics predictions not depend on the CKM convention. (Here we do not include CP violation in the neutral kaon since it is much smaller than what is expected in the B decay.)
In summary, for the case of f = J/ψ K s , Imλ = - sin(2β). The angle β is the best measured CP violating angle measured in B meson decays. The process used is B 0 → J/ψ K S , although there is some data used with the ψ (2S) or with K L (where the phase is + sin(2β)).
(Note that the sign of the term changes depending on the CP of the final state, but not the term.)
The most precise measurements of sin(2β) have been made by the BaBar and Belle experiments [47, 48]. These measurements are made at the ϒ(4S) resonance using e + e - → ϒ(4S) → B 0 , and with one of the neutral B's decaying into J/ψ K 0. Because the ϒ(4S) is a definite quantum state with C = -1, the formulae given above have to modified somewhat. One important change is the definition of t. The time used here is the difference between the decay time of the J/ψ K 0 and the other B 0, also called the "tagging" B because we need to determine its flavor, whether B 0 or , in order to make the asymmetry measurement.
The average value as determined by the Heavy Flavor Averaging Group is sin(2β) = 0.671 ± 0.024, where the dominant part of the error is statistical. No evidence is found for a non-zero term with the measured average given as 0.005 ± 0.020.
Determining the sine of any angle gives a four-fold ambiguity in the angle. The decay mode B 0 → J/ψ K*0, K*0 → K s π 0 offers a way of measuring cos 2β and resolving the ambiguities.
The non-zero value of shows the presence of at least two amplitudes, presumably tree and penguin, in addition to the mixing amplitude, making extraction of α difficult. All is not lost, however. Gronau and London  showed that α could be extracted without theoretical error by doing a full isotopic spin analysis of the π π final state. The required measurements include: , , , , and . The last two items require a flavored tagged analysis that has not yet been done and whose prospects are bleak. Grossman and Quinn have showed, however, that an upper limit on Γ (B 0 → π 0 π 0)/Γ(B + → π + π 0) can be used to limit the penguin shift to α [57–59]. Unfortunately, current data show a relative large (B 0 → π 0 π 0) = (1.55 ± 0.19) × 10-6 rate, compared with , implying a large penguin contribution [17, 34], and the limit is very weak.
Use of the ρ + ρ - final state is in principle quite similar to π + π -, with some important caveats. First of all, it is a vector-vector final state and therefore could have both CP + and CP- components. It is however possible, doing a full angular analysis to measure the CP violating phase separately in each of these two amplitudes. The best method for this is in the "transversity" basis and will be discussed later  in Section 2.2.6. It is possible, however, for one polarization component to be dominant and then the angular analysis might not be necessary. In fact the longitudinal polarization is dominant in this case. BaBar measures the fraction as , and Belle measures it as . Thus we can treat this final state without worrying about the angular analysis and just apply a correction for the amount of transverse amplitude.
In addition, ρ mesons are wide, so non-B backgrounds could be a problem and even if the proper B is reconstructed, there are non-resonant and possible a 1 π contributions.
The task at hand is to do a time dependent analysis of the magnitude of the decay amplitudes and phases. The analyses of both collaborations allow for ρ(1450) and ρ(1700) contributions in addition to the ρ(770). There are a total of 26 free parameters in the fit. The statistics for such an analysis are not overwhelming: BaBar has about 2100 signal events , while Belle has about 1000 .
The LHCb experiment expects to be able to significantly improve on the determination of α using the ρ π mode. They expect 14,000 events in for an integrated luminosity of 2 fb-1, with a signal to background ratio greater than 1 . It is expected that this amount of data can be accumulated in one to two years of normal LHC operation.
where r B reflects the amplitude suppression for the b → u mode and δ B is the relative phase. We have not yet used identical final states for D 0 and decays, yet we can see that there will be an ambiguity in our result between δ B and γ that could be resolved by using different D 0 decay modes.
There are several suggestions as to different D 0 decay modes to use. In the original paper on this topic Gronau and Wyler [72, 73] propose using CP eigenstates for the D 0 decay, such as K + K -, π + π - etc.., combining with charge specific decays and comparing B - with B + decays. In the latter, the sign of the strong phase is flipped with respect to the weak phase. In fact modes such as D*0 or K*- can also be used. (When using D*0 there is a difference in δ B of π between γ D 0 and π 0 D 0 decay modes .)
Measurements have been made by the BaBar, Belle and CDF collaborations [75–79]. These data, however, are not statistically powerful enough by themselves to give a useful measurement of γ. Atwood, Dunietz and Soni suggested using double-Cabibbo suppressed decays as an alternative means of generating the interference between decay amplitudes . For example the final state B - → D 0 K - can proceed via the tree level diagram in Figure 2(a) when the . Then if we use the doubly-Cabibbo suppressed decay D 0 → K + π -, we get a final state that interferes with the final state with the , which is Cabibbo allowed. The advantage of this method, besides extending the number of useful final states, is that both amplitudes are closer to being equal than in the above method, which can generate larger interferences. Any doubly-Cabibbo suppressed decay can be used. Similar equations exist relating the measured parameters to γ and the strong phase shift.
Thus far, the best method for measuring γ uses the three-body final state K S π + π -. Since the final state is accessible by both D 0 and decays interference results. By its very nature a three-body state is complicated, and consists of several resonant and non-resonant parts. It is typically analyzed by examining the Dalitz plot where two of the possible three sub-masses (squared) are plotted versus one another . In this way, the phase space is described as a uniform density and thus any resonant structure is clearly visible. For our particular case a practical method was suggested by Giri et al , where the decay is analyzed in separate regions of the Dalitz plot. Results from this analysis have been reported by BaBar  (they also include the mode D 0 → K S K + K -) and Belle . BaBar finds γ = (76 ± 22 ± 5 ± 5)°, and Belle finds . In both cases the first error is statistical, the second systematic, and the third refers to uncertainties caused by the Dalitz plot model. (There is also a mirror solution at ± 180°.)
It has been shown [87, 88] that measurement of the amplitude magnitudes and phases found in the decays of ψ (3770) → (CP ± Tag)(K S π + π -) D provide useful information that help the narrow model error. The CLEO collaboration is working on such an analysis, and preliminary results have been reported . (CLEO also is working on incorporating other D 0 decay modes.)
Since there are usually two photons present in the η decay, experiments at hadron colliders, which can perform time-dependent studies of B s mesons, preferentially use the J/ψ ϕ final state. This, unfortunately, introduces another complexity into the problem; as the B s is spinless the total angular momentum of the final state particles must be zero. For a decay into a vector and scalar, such as J/ψ η, this forces the vector J/ψ to be fully longitudinally polarized with respect to the decay axis. For a vector-vector final state both angular momentum state vectors are either longitudinal (L), both are transverse with linear polarization vectors parallel (||) or they are perpendicular (⊥) to one another . Another way of viewing this is that a spin-0 B decay into two massive vector mesons can form CP even states with L = 0 or 2, and a CP odd state with L = 1. The relative populations in the two CP states are determined by strong interactions dynamics, but to study the weak phase here we are not particularly interested in the actual amount, unless of course one state dominated. We do not expect this to be the case however, since the SU(3) related decay B 0 → J/ψ K*0, K*0 → K + π - has a substantial components of both CP states; the PDG quotes gives the longitudinal fraction as (80 ± 8 ± 5)% .
The decay rate for can be found by replacing A ⊥ in the above expression with -A ⊥.
The quantities δ ⊥ and δ || are the strong phases of A ⊥ and A || relative to A 0, respectively . The expression for mesons can be found by substituting and .
The Standard Model allowed region is a very thin vertical band centered near zero at ϕ s of -0.036 ± 0.002 (shown in red). (Recall ϕ s ≡ -2χ ≡ -2β s .) The region labeled "all" (green) shows the allowed region at 68% confidence level. Although the fit uses several input components besides the CP asymmetry measurements in B s → J/ψ ϕ, including use of measured total widths introduced via the constraint equation , it is the measurement of ϕ s that dominates the fit result. We see that there is a discrepancy that may be as large as 2.7 standard deviations [95, 96]. While this is as not yet significant, it is very tantalizing. The LHCb experiment plans to vastly improve this measurement .
It has been pointed out, however, that there is likely an S-wave K + K - contribution in the region of the ϕ that contributes 5–10% of the event rate as estimated using decays . For example, analysis of the analogous channel B 0 → J/ψ K*0 reveals about 8% S-wave in the K π system under the K*0, and BaBar has used this to extract a value for cos 2β thus removing an ambiguity in β . The S-wave amplitude and phase needs to be added to Eq. 53. Note that the errors will increase due to the addition of another amplitude and phase. The S-wave can manifest itself as a π + π -, so it is suggested that the decay B s → J/ψ f 0 (980) be used to measure χ; here the f 0 → π + π - ; the estimate is that the useful f 0 rate would be about 20% of the ϕ rate, but the J/ψ f 0 is a CP eigenstate, so an an angular analysis is unnecessary, and these events may provide a determination of χ with an error comparable to that using J/ψ ϕ.
showing a large statistical significance.
The difference between and is not naively expected in the Standard Model and Belle suggests that this may be a sign of New Physics. Peskin commented on this possibility .
All CP violation in the quark sector is proportional to the parameter η in Eq. 1. In fact all CP asymmetries are proportional to the "Jarlskog Invariant", J = A 2 λ 6 η, which represents the equal area of all the CKM triangles . Since we know the value of these three numbers, we do know the amount of CP violation we can expect, even without making the measurements. We also can estimate the amount of CP violation necessary using cosmology. To reproduce the observed baryon to entropy ratio requires many orders of magnitude more CP violation than thus far found in heavy quark decays . Thus we believe there are new sources of CP violation that have not yet been found.
There are two experimental methods to determine |V cb |: the exclusive method, where |V cb | is extracted by studying the exclusive decay process; and the inclusive method, which uses the semileptonic decay width of b-hadron decays. In both methods, the extraction of |V cb | is systematics limited and the dominant errors are from theory. The inclusive and exclusive determinations of |V cb | rely on different theoretical calculations of the hadronic matrix element needed to extract it from measured quantities, and make use of different techniques which, to a large extent, have uncorrelated experimental uncertainties. Thus, the comparison between inclusive and exclusive decays allows us to test our understanding of hadronic effects in semileptonic decays. The latest determinations differ by more than 2σ, with the inclusive method having a stated error half of the size of the exclusive one.
Due to confinement and the non-perturbative aspect of the strong interaction, the concept of the quark masses cannot be tied to an intuitive picture of the rest mass of a particle, as for leptons. Rather, quark masses must be considered as couplings of the SM Lagrangian that have to be determined from processes that depend on them. As such the b-quark mass (m b ) is a scheme-dependent, renormalised quantity.
In principle, any renormalisation scheme or definition of quark masses is possible. In the framework of QCD perturbation theory the difference between two mass schemes can be determined as a series in powers of α s . Therefore, higher-order terms in the perturbative expansion of a quantity that depends on quark masses are affected by the particular scheme employed. There are schemes that are more appropriate and more convenient for some purposes than others. Here we examine the main quark mass definitions commonly used in the description of B decays.
Pole mass: The pole mass definition is gauge-invariant and infrared-safe [106–108] to all orders in perturbation theory and has been used as the standard mass definition of many perturbative computations in the past. By construction, it is directly related to the concept of the mass of a free quark. The presence of a renormalon ambiguity [109, 110] makes the numerical value of the pole mass an order-dependent quantity, leading to large perturbative corrections for Heavy Quark Effective Theory (HQET) parameters (see below for a discussion of HQET). These shortcomings are avoided by using quark mass definitions that reduce the infrared sensitivity by removing the ΛQCD renormalon of the pole mass. Such quark mass definitions are generically called "short-distance" masses.
mass: The most common short-distance mass definition is the mass [111–113], where the scale μ is typically chosen to be the order of the characteristic energy scale of the process. In the scheme the subtracted divergencies do not contain any infrared sensitive terms and the mass is only sensitive to scales larger than m b . The mass arises naturally in processes where the b-quark is far off-shell, but it is less adequate when the b-quark has non-relativistic energies.
where μ kin is the nominal kinetic mass renormalisation scale. For μ kin → 0 the kinetic mass reduces to the pole mass.
1S mass: The kinetic mass depends on an explicit subtraction scale to remove the universal infrared sensitive contributions associated with the non-relativistic b-quark dynamics. The 1S mass [117, 118] achieves the same task without a factorisation scale, since it is directly related to a physical quantity. The b-quark 1S mass is defined as half of the perturbative contribution to the mass of the ϒ(S 1) in the limit m b ≫ m b v ≫ m b v 2 ≫ ΛQCD.
The exclusive |V cb | determination is obtained by studying the decays B → D* ℓν and B → D ℓν, where ℓ denotes either an electron or a muon. The exclusive measurements of a single hadronic final state, e.g. the ground state D or D*, restrict the dynamics of the process. The remaining degrees of freedom, usually connected to different helicity states of the charmed hadron, can be expressed in terms of form factors, depending on the invariant mass of the lepton-ν pair, q 2. The shapes of those form factors are unknown but can be measured. However, the overall normalization of these functions needs to be determined from theoretical calculations.
Isgur and Wise formulated a theoretical breakthrough in the late 1980's. They found that in the limit of an infinitely heavy quark masses QCD possess additional flavor and spin symmetries. They showed that since the heavier b and c quarks have masses much heavier than the scale of the QCD coupling constant, they are heavy enough to posses this symmetry but that corrections for the fact that the quark mass was not infinite had to be made. They showed there was a systematic method of making these corrections by expanding a series in terms of the inverse quark mass . This theory is known as Heavy Quark Effective Theory (HQET).
Note that the first term in the non-perturbative expansion in powers of 1/m Q vanishes . QED corrections up to leading logarithmic order give η A = 0.960 ± 0.007. Different estimates of the corrections, involving terms proportional to , and 1/(m b m c ), have been performed in a quark model with QCD sum rules, and, more recently, with an HQET based lattice gauge calculation. The best estimate comes from lattice QCD, (1) = (1) = 0.921 ± 0.013 ± 0.020 . This result does not include a 0.7% QED correction.
Since the phase-space factor tends to zero as w → 1, the decay rate vanishes and the accuracy of the |V cb | value extracted with this method depends upon experimental and theoretical uncertainties in the extrapolation. Experiments determine the product by fitting the measured dΓ/dw distribution.
This decay has been analyzed by CLEO, BaBar and Belle using B mesons from the ϒ(4S) decay, and by ALEPH, DELPHI, and OPAL at the Z0 center of mass energy. Experiments that exploit the ϒ(4S) have the advantage that w resolution is quite good. However, they suffer from lower statistics near w = 1 in the decay B → D*+ ℓν due to the lower reconstruction efficiency of the slow π ±. On the other hand, the decay B → D*0 ℓν is not affected by this problem . In addition, kinematic constraints enable these experiments to identify the final state including D* without large contamination from the poorly known portion of semileptonic decays with a hadronic system recoiling against the lepton-ν pair with masses higher than the D and D*, commonly identified as 'D**'. B-factories and CLEO fit for the signal and background components in the distribution of the cosine of the angle between the direction of the B and the direction of the D*ℓ system. At LEP, B mesons are produced with a large variable momentum (about 30 GeV on average), giving a relatively poor w resolution and limited physics background rejection capabilities. By contrast, LEP experiments benefit from an efficiency that is only mildly dependent upon w.
LEP experiments extracted |V cb | by performing a two-parameter fit, for and the slope ρ 2. The first measurements of both ratio R 1 and R 2, and ρ 2, were made by the CLEO collaboration . Belle  and BaBar  improved upon these measurements using the → D*+ e - e decay. They determined R 1, R 2, and ρ 2 using an unbinned maximum likelihood fit to the full decay distribution. BaBar and CLEO results are combined to give: R 1 = 1.396 ± 0.060 ± 0.035 ± 0.027, R 2 = 0.885 ± 0.040 ± 0.022 ± 0.013, and ρ 2 = 1.145 ± 0.059 ± 0.030 ± 0.035. The stated uncertainties are the statistical from the data and systematic uncertainty, respectively.
Experimental results based on B → D*ℓν after the correction to common inputs and the world average as obtained by HFAG .
31.6 ± 1.8 ± 1.3
0.50 ± 0.20 ± 0.09
5.44 ± 0.25 ± 0.20
33.9 ± 0.3 ± 1.1
1.18 ± 0.05 ± 0.03
4.53 ± 0.04 ± 0.13
34.9 ± 0.8 ± 1.4
1.11 ± 0.06 ± 0.08
5.40 ± 0.16 ± 0.25
35.7 ± 0.2 ± 1.2
1.20 ± 0.02 ± 0.07
34.7 ± 0.2 ± 1.0
1.16 ± 0.04 ± 0.03
34.7 ± 0.2 ± 1.0
1.16 ± 0.04 ± 0.03
4.75 ± 0.25 ± 0.19
41.3 ± 1.3 ± 1.8
1.37 ± 0.08 ± 0.18
6.02 ± 0.19 ± 0.20
36.3 ± 1.8 ± 1.9
1.04 ± 0.14 ± 0.15
5.53 ± 0.19 ± 0.34
DELPHI(part rec) 
35.8 ± 1.4 ± 2.3
1.18 ± 0.13 ± 0.25
5.00 ± 0.15 ± 0.18
36.9 ± 1.6 ± 1.5
1.24 ± 0.20 ± 0.14
5.17 ± 0.20 ± 0.38
OPAL(par rec) 
37.6 ± 1.2 ± 2.4
1.14 ± 0.13 ± 0.27
5.63 ± 0.27 ± 0.43
35.4 ± 0.5
1.16 ± 0.05
5.05 ± 0.10
where the dominant error is theoretical, and it will be difficult to improve upon.
Experimental results from B → Dℓν using corrections to the common inputs and world averages.
38.1 ± 11.8 ± 6.1
0.91 ± 0.98 ± 0.36
2.25 ± 0.18 ± 0.36
42.3 ± 1.9 ± 1.0
1.20 ± 0.09 ± 0.04
43.8 ± 0.8 ± 2.1
1.22 ± 0.04 ± 0.06
2.20 ± 0.11 ± 0.12
40.7 ± 4.4 ± 5.1
1.12 ± 0.22 ± 0.14
2.09 ± 0.12 ± 0.39
44.5 ± 5.9 ± 3.4
1.27 ± 0.25 ± 0.14
2.10 ± 0.13 ± 0.15
42.4 ± 0.7 ± 1.4
1.19 ± 0.04 ± 0.04
2.16 ± 0.12
consistent with the value extracted from B → D⋆ℓν decay, but with an experimental uncertainty about twice as large.
BaBar recently has also studied the differential decay widths for the decays B - → D 0 ℓν and B - → D⋆0ℓν to extract the ratio , compatible with the lattice theory prediction of 1.16 ± 0.04.
It is important to understand the composition of the inclusive B semileptonic decay rate in terms of exclusive final states for use in semileptonic B decay analyses. The B → D (*)ℓν decays are well measured, but a sizeable fraction of semileptonic B decay are to D**ℓν. The D** resonances have larger masses than the D* and not well studied. (D** refers to the resonant states, , , and D 1.) The measurements of these branching fractions require a good understanding of the different D (*) nπ systems, which can be either resonant or non-resonant, each with characteristic decay properties.
There are four orbitally excited states with L = 1. They can be grouped in two pairs according to the value of the spin on the light system, j = L ± 1/2 (L = 1). States with j = 3/2 can have JP = 1+ and 2+. The 1+ state decays only through D*π, and the 2+ through D π or D*π. Parity and angular momentum conservation imply that in the 2+ the D* and π are in a D wave but allow both S and D waves in the 1+ state. However, if the heavy quark spin is assumed to decouple, conservation of j = 3/2 forbids S waves even in the 1+ state. A large D-wave component and the fact that the masses of these states are not far from threshold imply that the j = 3/2 states are narrow. These states have been observed with a typical width of 20 MeV/c 2. On the contrary, j = 1/2 states can have JP = 0+and 1+, so they are expected to decay mainly through an S wave and manifest as broad resonances, with typical widths of several hundred MeV/c 2.
The ALEPH , CLEO , DELPHI , and D0  experiments have reported evidence of the narrow resonant states (D 1 and ) in semileptonic decays, whereas more recent measurements by the BaBar  and Belle  experiments provide semi-inclusive measurements to D (*) π ℓν final states [137, 138].
Inclusive versus sum of exclusive measured B semileptonic branching fractions (%).
B → Dℓν
2.20 ± 0.16
2.30 ± 0.10
2.09 ± 0.16
2.16 ± 0.12
2.32 ± 0.09
B → D*ℓν
4.53 ± 0.14
5.37 ± 0.21
4.42 ± 0.25
5.05 ± 0.10
5.66 ± 0.18
B → D πℓν
0.42 ± 0.09
0.42 ± 0.07
0.43 ± 0.09
0.42 ± 0.06
0.43 ± 0.06
0.42 ± 0.05
B → D* πℓν
0.48 ± 0.09
0.59 ± 0.06
0.57 ± 0.22
0.68 ± 0.11
0.49 ± 0.08
0.61 ± 0.05
7.63 ± 0.25
8.68 ± 0.25
7.51 ± 0.73
8.13 ± 0.19
9.01 ± 0.21
10.14 ± 0.43
10.90 ± 0.47
10.46 ± 0.38
11.17 ± 0.38
10.33 ± 0.28
10.99 ± 0.28
Inc. – Σ (Exc.)
2.51 ± 0.50
2.22 ± 0.53
2.95 ± 0.82
2.20 ± 0.34
1.98 ± 0.35
All measured rates for the D** narrow states are in good agreement. Experimental results seem to point towards a larger rate for broader states. If it is due mainly to and decay channels, these results disagree with the prediction of QCD sum rules. However, Belle set an upper limit for the channel below the rate measured by BaBar and DELPHI. More measurements need to be performed to elucidate this puzzle.
Inclusive determinations of |V cb | are obtained using combined fits to inclusive B decay distributions [139, 140]. These determinations are based on calculations of the semileptonic decay rate in the frameworks of the Operator Product Expansion (OPE)  and HQET [139, 142]. They predict the semileptonic decay rate in terms of |V cb |, the b-quark mass m b , and non-perturbative matrix elements. The spectator model decay rate is the leading term in a well-defined expansion controlled by the parameter ΛQCD/m b [142–146] with non-perturbative corrections arising to order . The key issue in this approach is the ability to separate perturbative and non-perturbative corrections (expressed in powers of α s ). Thus the accuracy of an inclusive determination of |V cb | is limited by our knowledge of the heavy quark nonperturbative matrix elements and the b quark mass. It also relies upon the assumption of local quark hadron duality. This is the statement that a hadronic matrix element can be related pointwise to a theoretical expression formulated in terms of quark and gluon variables.
Perturbative and non-perturbative corrections depend on the m b definition, i.e. the non-perturbative expansion scheme, as well as the non-perturbative matrix elements that enter the expansion. In order to determine these parameters, Heavy Quark Expansions (HQE) [139, 146, 147] express the semileptonic decay width ΓSL, moments of the lepton energy and hadron mass spectra in B → X c ℓν decays in terms of the running kinetic quark masses and as well as the b-quark mass in the 1S expansion scheme. These schemes should ultimately yield consistent results for |V cb |. The precision of the b-quark mass is also important for |V ub |.
The shape of the lepton spectrum and of the hadronic mass spectrum provide constraints on the heavy quark expansion, which allows for the calculation of the properties of B → X c ℓν transitions. So far, measurements of the hadronic mass distribution and the leptonic spectrum have been made by BaBar , Belle , CLEO [150, 151], DELPHI . CDF  provides only the measurement of the hadronic mass spectrum with a lepton momentum cut of 0.6 GeV in the B rest frame.
corresponding to a distribution's width, kurtosis and so on.
The measurement of |V cb | from inclusive decays requires that these decays be adequately described by the OPE formalism. The motivation of the moment approach is to exploit the degree of experimental and theoretical understanding of each moment and for different , directly examining the dependence of the various coefficient functions f i on these terms. The possibility of deviations from the OPE predictions due to quark hadron duality violations have been raised . To compare the OPE predictions with data, one also has to define how uncertainties from corrections are estimated. These uncertainties are hard to quantify reliably, as the only information on the size of the matrix elements of the dimension six operators comes from dimensional analysis. The only method to check the reliability of |V cb | extraction from inclusive semileptonic decays is how well the OPE fits the data.
To compare with theoretical predictions, the moments are measured with a well defined cut on the lepton momentum in the B rest frame. The measured hadronic mass distribution and lepton energy spectrum are affected by a variety of experimental factors such as detector resolution, accessible phase space, radiative effects. It is particularly important to measure the largest fraction of the accessible phase space in order to reduce both theoretical and experimental uncertainties. Each experiment has focused on lowering the lepton energy cut.
The hadronic mass spectrum in B → X c ℓν decays can be split into three contributions corresponding to D, D*, and D**, where D** here stands for any neutral charmed state, resonant or not, other than D and D*. Belle , BaBar  and CLEO  explored the moments of the hadronic mass spectrum as a function of the lepton momentum cuts. CLEO performs a fit for the contributions of signal and background to the full three-dimensional differential decay rate distribution as a function of the reconstructed quantities q 2, , cos θ ℓ. Belle [155, 156] and BaBar  use a sample where one of the B mesons, produced in pairs from ϒ(4S) decays, is fully reconstructed and the signal side is tagged by a well identified lepton. The 4-momentum p X of the hadronic system X, recoiling against the lepton and neutrino, is determined by summing the 4-momenta of the remaining charged tracks and unmatched clusters. Belle reconstructs the full hadronic mass spectrum and measures the first, second central and second non-central moments of the unfolded spectrum in B → X c ℓν, for lepton energy thresholds, E min, varying from 0.7 to 1.9 GeV . BaBar extracts the moments from the measured distributions using a calibration curve derived from Monte Carlo data, with a minimum momentum for the electron in the B meson rest frame of 0.9 GeV. The latest BaBar analysis  measures the first, second central and second non-central moments of the spectrum for E min from 0.9 to 1.9 GeV. The main systematic errors originate from background estimation, unfolding and signal model dependence.
DELPHI follows a different approach in extracting the moments, measuring the invariant mass distribution of the D** component only and fixing the D and D* components. DELPHI measures the first moment with respect to the spin averaged mass of D and D*. At LEP b-quarks were created with an energy of approximately 30 GeV allowing the measurement of the hadronic mass moments without a cut on the lepton energy .
The shape of the lepton spectrum provides further constraints on the OPE. These measurements are sensitive to higher order OPE parameters and are considerably more precise experimentally. Moments of the lepton momentum with a cut p ℓ ≥ 1.0 GeV/c have been measured by the CLEO collaboration . BaBar  extract up to the third moment of this distribution, using a low momentum cut of p ℓ ≥ 0.6 GeV/c. Both BaBar and CLEO use dilepton samples. The most recent measurement of the electron energy spectrum is from Belle [155, 156]. Events are selected by fully reconstructing one of the B mesons, produced in pairs from ϒ(4S) decays and it determines the true electron energy spectrum by unfolding  the measured spectrum in the B meson rest frame. Belle measures B 0 and B + weighted average partial branching fractions and the first four moments of the electron energy spectrum in B → X c e ν, for E min from 0.4 to 2.0 GeV . All the lepton moment measurements are consistent with theory and with the moment of the hadronic and b → s γ photon energy spectrum. Hadronic and lepton energy measurements are consistent within their errors. When compared with theory there is no sign of inconsistencies. The B-factories provide the most precise HQE parameter estimates.
Using the moment measurements described above, it is possible to determine the CKM matrix element |V cb | and HQE parameters by performing global fits in the kinetic and 1S b-quark mass schemes . The photon energy moments in B → X s γ decays  are also included in order to constrain the b-quark mass more precisely. Measurements that are not matched by theoretical predictions and those with high cutoff energies are excluded (i.e. semileptonic moments with E min > 1.5 GeV and photon energy moments with E min > 2 GeV). The results are preliminary.
Experimental results using corrections to the common inputs and world averages for the 1S scheme.
|V cb |(×10-3)
X c ℓν + X s γ
4.700 ± 0.030
-0.315 ± 0.026
X c ℓν
4.656 ± 0.060
-0.343 ± 0.046
Experimental results using corrections to the common inputs and world averages for the kinetic scheme.
|V cb |(×10-3)
m b (GeV)
X c ℓν + X s γ
4.601 ± 0.034
0.440 ± 0.040
X c ℓν + X s γ †
4.605 ± 0.031
0.312 ± 0.060
X c ℓν
4.659 ± 0.049
0.428 ± 0.044
All the measured moments of inclusive distributions in B → X c ℓν and B → s γ decays are used in a fit to extract |V cb | and the b and c quark masses. The |V cb | and m b values obtained are listed in the Table 7. The default fit also gives m c = 1.16 ± 0.05 GeV, and . The errors are experimental and theoretical (HQE and ΓSL) respectively. In this fit the following variations were considered when in the extraction of the HQ parameters, ± 20 MeV for the b and c quark masses, ± 20% for and , ± 30% for the 3 rd order non perturbative terms and α 2 = 0.22 ± 0.04 for the perturbative corrections. The bias corrections uncertainties for B → s γ were varied by the full amount of their value.
There are open issues relating to the global fits. First of all, the χ 2/n.d.f. are very small, pointing to an underestimate in the theoretical correlations. In a recent study , the theoretical correlations used in the fit were scrutinized, and new correlation coefficients were derived from the theory expressions using a "toy Monte Carlo" approach, showing that the theoretical correlations were largely underestimated. The result of this new fit is shown in Table 7. The second issue is related the size of the theoretical error. Recently the NNLO full two-loop calculations become available [164–166]. In the Kinetic scheme NNLO calculations include an estimate of the non-BLM terms and lead to a roughly 0.6 % reduction of the |V cb | value -0.25 × 10-3. In the 1S scheme the shift on |V cb | is of about -0.14 × 10-3 . From the new power corrections at NLO we expect the chromo-magnetic corrections to be more important as the tree level corrections are more important, and a change of about 20–30% in the extracted value of in the pole expansion (may be less with the other schemes) .
HQE has been carried out up to and the effects are expected to be of the order δ (4)Γ/Γ ≈ 0.25% . All these newly calculated results can be used to scrutinize the earlier error estimates. In the kinetic scheme, the full NNLO value for A pert = 0.919, which is in good agreement with A pert = Γ (B → X c ℓν)/Γ (B → X c ℓν) tree = 0.908 ± 0.009. In the 1S scheme the estimated uncertainty from the non-BLM two-loop is half of the BLM part, equivalent to 1.5% of the tree-level rate, more than 3 times the actual correction. The new results are not implemented in the global fits as they are available mainly in the pole-mass scheme.
The error on the inclusive and exclusive determination of V cb is limited by theory. The two method have a 2σ disagreement, even when results in the same experiment are compared.
For exclusive determinations, the experimental determinations in B → D*ℓν from different experiments are not in agreement. Improving the statistical error for the determination from B → Dℓν will help in elucidating the origin of this discrepancy.
In the inclusive method, errors of less than 2% are quoted. However, the latest theoretical results and the introduction of better correlation between theoretical error show shifts in the central value larger than the quoted fit error. The situation needs to be reevaluated when all the new calculations and corrections are implemented in the fit. Another puzzling result of the global fit to B → X c ℓν moments only is the value of , which seems to be in disagreement with other determinations, such as the ones in Table 2. (For a detailed comparison between and different mass determinations see A. Hoang, talk at Joint Workshop on |V ub | and |V cb | at the B-Factories Heidelberg, December 14–16, 2007.)
The parameter |V ub | determines one of the sides of the unitarity triangle, and thus affects one of the crucial tests of the Yukawa sector of the Standard Model. Also in this case, there are two general methods to determine this parameter, using B meson semileptonic decays. The first approach relies on the determination of branching fractions and form factor determinations of exclusive semileptonic decays, such as . The relationship between experimental measurements and |V ub | requires a theoretical prediction of the hadronic form factors governing these decays. The complementary approach relies on measurements of inclusive properties of B meson semileptonic decays. In this case, the hadronic matrix element is evaluated in the context of the OPE.
Both methods pose challenges to both experimenters and theorists. The branching fractions are small, and a substantial background induced by the dominant needs to be suppressed or accounted for. The large data samples accumulated at the two b-factories, Belle and BaBar, have made possible the development of new experimental techniques that have reduced the experimental errors substantially. In parallel, theorists devoted considerable efforts to devise measurable quantities for which reliable predictions could be produced, and to improve the accuracy of the predictions through refinements in the calculation. Although the precision of the stated errors improved, so did the difference in central values between the inclusive and exclusive estimates of |V ub |, at least in most determinations. Possible interpretations of this discrepancy will be discussed.
where G F is the Fermi constant, , and f + (q 2) is the relevant form factor.
The first experimental challenge is to reconstruct this exclusive channel without significant background from the dominant charm semileptonic decays, and the additional background component from other transitions. The advantage of e + e - B-factories is that the B decaying semileptonically originates from the decay e + e - B . Thus, if the companion B is fully reconstructed, the ℓ 4-momentum can be inferred from the rest of the event.
CLEO pioneered this approach by reconstructing the ν from the missing energy (E miss ≡ 2E B - Σ i E i ), and momentum in the event; in these definitions the index i runs over the well reconstructed tracks and photons, and cuts are applied to enhance the probability that the only missing particle in the event is the . This approach allows the application of a low momentum cut of 1.5 GeV on the lepton for and 2 GeV for . Averaging the results from these two exclusive channels, they obtain . Using their relatively modest full data set (16 fb-1) at the center-of-mass energy of the ϒ(4S) and considering only the channel, they get the branching fraction (B → π + ℓ ℓ) = (1.37 ± 0.15 ± 0.11) × 10-4, and .
BaBar  uses a sample of 206 fb-1 to obtain (B → π + ℓ ℓ) = (1.46 ± 0.07 ± 0.08) × 10-4. Recently, the availability of very large data sets at Belle and BaBar have made possible tagged analyses, where semileptonic decays are studied in samples where the other B is fully reconstructed, thus defining the event kinematics even more precisely. The first implementation relies on the the partial reconstruction of exclusive B → D (⋆) ℓ+ ν ℓ decay to tag the presence of a B event [172, 173].
Partial and total branching fractions, in units of 10-4, for the decay B → πℓ+ ν ℓ.
q 2 > 16 GeV2
1.37 ± 0.15 ± 0.11
0.41 ± 0.08 ± 0.04
1.46 ± 0.07 ± 0.08
0.38 ± 0.04 ± 0.04
1.38 ± 0.19 ± 0.14 ± 0.03
0.36 ± 0.10 ± 0.04 ± 0.01
B→ D(*)ℓν Tag
1.49 ± 0.26 ± 0.06
0.31 ± 0.12 ± 0.01
B 0 → π -ℓ+ν ℓ (hadron tags)
1.53 ± 0.20 ± 0.06
0.39 ± 0.12 ± 0.02
Combined B 0 and B + tags
1.33 ± 0.17 ± 0.11
0.46 ± 0.10 ± 0.06
Combined B 0 and B + tags
1.34 ± 0.06 ± 0.05
0.37 ± 0.03 ± 0.02
In order to interpret these results, we need theoretical predictions for the form factor f + (q 2). This problem can be split into two parts: the determination of the form factor normalization, f + (0), and the functional form of the q 2 dependence. Form factor predictions have been produced with quark models  and QCD sum rule calculations . Lattice calculations provide evaluations of f + (q 2) at specific values of q 2 or, equivalently, pion momenta (p π ).
maps t + <t < ∞ onto |z| = 1 and -∞ <t < t+ onto the z interval [-1,1].
All are refinements of the old ansatz of a simple pole shape, now rarely used. The first lattice calculations were carried out with the quenched approximation that ignores vacuum polarization effects [178, 179]. In 2004, preliminary unquenched results were presented by the Fermilab/MILC  and HPQCD  collaborations. These calculations employed the MILC collaboration N f = 2 + 1 unquenched configurations, which attain the most realistic values of the quark masses so far. Using these calculations and the most recent value of the partial branching fraction for q 2 > 16 GeV2, shown in Table 8, we obtain |V ub | = (3.51 ± 0.096 exp ± 0.49 th ) × 10-3 with the HPQCD normalization, and |V ub | = (3.70 ± 0.10 exp ± 0.37 th ) × 10-3 with the Fermilab/MILC normalization. Fits to experimental data, combining lattice predictions, and dispersion relations reduce the theoretical errors. For example,  obtains |V ub | = (3.5 ± 0.17 ± 0.44) × 10-3, and, more recently,  obtains |V ub | = (3.47 ± 0.29) × 10-3. The most recent lattice calculation  performs a simultaneous fit of improved lattice numerical Monte Carlo results and the 12-bin BaBar experimental data on |V ub |f + (q 2)  and derives |V ub | = (3.38 ± 0.35) × 10-3. The ~10% error includes theoretical and experimental errors, not easily separable.
where μ ~ m b is the scale at which α S needs to be evaluated, while the scale μ ⋆ applies to the non-perturbative expansion parameters, namely the b quark mass m b , the chromo-magnetic operator, λ 2, and the kinetic operator, μ π . It is clear that, even restricting our attention to the total width, a precise knowledge of the b quark mass is critical, and considerable theoretical effort has been devoted to a reliable extraction of this parameter from experimental observables. Other uncertainties, such as the effects of weak annihilation or violations of quark-hadron duality, will be discussed later.
Summary of the experimental |V ub | determinations using the lepton endpoint; |V ub | is extracted with the "shape function" method .
Next a whole host of papers proposed alternative "model independent" approaches to measure |V ub | from inclusive decays [190–192], with the common goal of identifying a region of phase space where experimentalists can suppress the b → c background, and where the OPE works. The first proposal by Bigi, Dikeman, and Uraltsev proposed considering semileptonic events where M X ≤ 1.5 GeV . However, Bauer, Ligeti, and Luke pointed out  that the kinematic limit has the same properties of the lepton end point, and spoils the convergence of the OPE; the same authors proposed using dΓ/dq 2 up to and argue that this distribution is better behaved in the kinematic region of interest. This is the theoretical foundation of the so called "improved end point" method, which encompasses the simultaneous study of E ℓ and q 2. BaBar  used this technique, inferring the ν 4-momentum from the missing momentum ( ) in the event. There results give |V ub | central values between (3.88–4.93)×10-3, depending on the explicit model and with total errors of ≈ 10% on each value.
An alternative approach , incorporates hadronic structure functions to model the region of large hadronic energy and small invariant mass, not well modeled by the OPE, and applies the OPE to the kinematic region where the hadronic kinematic variables scale with M B , and smoothly interpolates between them. This approach is commonly referred to as "shape function" method, and uses the γ spectrum in inclusive B → X S γ to reduce the theoretical uncertainties. It combines the experimental data on high momentum leptons from B decays with the constraints from inclusive radiative decays, to produce a precise value of |V ub | inclusive.
The study of charmless inclusive semileptonic decays benefits from the use of hadronic tags. Belle used the information of the tag momentum to boost the electron into the B meson rest frame and to select a sample of high purity. They then reconstruct the ν energy and momentum from the measured 4-momentum vectors of the ϒ(4S), B tag, lepton, and the additional tracks not used to form the tag or the lepton. They then evaluate the invariant mass M X and the quantity , where . Both M X and P + are smaller for b → u transitions. They define different signal region for p ℓ > 1 GeV: P + < 0.66 GeV/c, M X < 1.7 GeV/c2, and M X < 1.7 GeV/c2 combined with q 2 > 8 GeV2/c2.
Perturbative corrections to the structure functions W 1–3 are now known up to order  and power corrections are included through . The separation between perturbative and non-perturbative physics is set by a cut-off scale μ = 1 GeV, in the "kinetic scheme" , which takes input parameters from fits to the b → c moments. Alternative choices of kinematical variable have been used, for example, in the SCET approach, the variables and are used. In particular, the "shape function" method  provides theoretical expression for the triple differential d 3Γ/dE ℓ dP + dP - and relates it to moments of the shape function extracted from B → X S γ . Finally, the approach originally proposed by Gardi  uses resummed perturbation theory in momentum space to provide a perturbative calculation of the on-shell decay spectrum in the entire phase space. The method used, dressed gluon exponentiation, is a general resummation formalism for inclusive distributions near a kinematic threshold. Finally, another model based on soft-gluon resummation and an analytical time-like QCD coupling has been proposed . Clearly a lot of work has gone into bringing to fruition the original promise that the inclusive |V ub | determination is a more precise method to determine this important CKM parameter.
An effect that can influence the inclusive |V ub | is the contribution due to topologies where the incoming pair annihilates into a W boson, with a concomitant soft gluon emission. These contributions are known as "weak annihilation" and they appear as a delta function at the end point of the lepton spectrum. This effect introduces a difference between B 0 and B +semileptonic decays.
at 90 % confidence level, where f W A (2.3 – 2.6) represents the fraction of weak annihilation rate contributing in the momentum interval Δp ℓ comprised between 2.3 and 2.6 GeV used in this analysis, believed to be close to unity.
where Γ0 (c → sℓν) is the bare parton semileptonic rate c → sℓν, f D is the D decay constant, and is the non-factorizable term that affects charmless semileptonic decays as well. Using this formula with the CLEO-c results for inclusive D 0 semileptonic decay , the sum of D s exclusive semileptonic decays , and m c (m c ) = 1.268 GeV , we obtain an estimate of non-factorizable effects equal to (9 ± 5) %. Note that annihilation effect induced uncertainties are estimated to contribute 1.6% in the BLNP approach, 1.5% in the DGE approach, and (+0.0, -3.1)% in the GGOU approach.
In summary, the values if |V ub | extracted with the inclusive method have quoted errors between 5.5% and 10%, with central values that change well outside these uncertainties and are all considerably larger than the |V ub | value obtained from the exclusive branching fraction .
In general, we define as "rare" B decays processes that are suppressed at tree level. They are interesting because they are typically mediated by loop diagrams which may be characterized by a matrix element whose strength is comparable to components with similar Feynman diagrams, where new particles appear in loops. Thus evidence for new physics may appear either through enhancements in branching fractions relative to the Standard Model expectation, or through interference effects.
Note that a quantity that is sensitive to the same coupling is the ratio . The present experimental value for this parameter is (41.6 ± 11.7 ± 5.2)% , which constrains g s to be ≤ 1.79. Thus these two measurements provide complementary constraints, which have very different systematic uncertainties. Thus an improvement in the knowledge of leptonic and semileptonic branching fractions involving a τ lepton in the final state is very important.
Radiative B decays involving b → s(d)γ transitions are very sensitive to new physics processes. These processes are ideals for indirect searches for physics beyond the Standard Model, such as two-Higgs doublet models, supersymmetric theories and extended technicolor scenarios . Hence, comparison of results from these theories with experimental measurements places constraints upon new physics. Moreover, B → X s γ is an ideal laboratory for studying the dynamics of the b-quark inside the B meson: since the motion of the b-quark inside the B meson is universal, information gained from a measurement of the energy spectrum of the emitted photon in this transition is applicable to other processes, for instance semileptonic decays.
where the Q i are local operators, the C i are Wilson coefficients, and μ is the QCD renormalization scale . The B → X s γ branching fraction directly probes the Wilson coefficient C 7. However, some new physics contributions may leave the B → X s γ rate unaltered, with changes to the C 7 amplitude. In this case the direct CP asymmetry is sensitive to new phases that may appear in the decay loop [214, 215] and the B → X s ℓ+ℓ- transition may provide information on the sign of the amplitude, since it also probes C 9 and C 10.
and the non-perturbative N(E 0) terms are . Note that the minimum value of E 0 for which this relation is valid is 1.6 GeV. The theoretical error from NLO perturbative calculations is about 10%, dominated by the renormalization scale dependence and a charm quark mass uncertainty. This uncertainty is mainly due to the change from the ratio in the pole scheme to the ratio in the scheme (see section 3.0.1). The large m c dependence is due to the fact that m c first enters at NLO matrix elements. Hence, the natural scale at which m c should be renormalized can only be determined by dedicated calculations at NNLO.
Measured branching fractions, minimum photon energy, and branching fractions for E min = 1.6 GeV photon energy for b → s γ .
at E min
E min = 1.6 GeV
E min = 1.6 GeV
306 ± 41 ± 26
336 ± 53 ± 42
367 ± 29 ± 34
366 ± 85 ± 60
331 ± 19 ± 37
324 ± 17 ± 24
312 ± 15 ± 16
294 ± 14 ± 12
Experimentally, two methods are used to extract the B → X s γ signal: the fully inclusive method and the semi-inclusive method. In the fully inclusive method, events containing a hard photon consistent with B → X s γ are selected. In this method the subtraction of a very large background, primarily from continuum events, is the main issue. There are several options to reduce this large background. Particularly effective in the suppression of continuum backgrounds is the requirement of a high momentum lepton, tagging the semileptonic decay of the accompanying B meson. Alternatively, a fully reconstructed B can be used to identify B decays. In this measurement the large background increases towards the lower photon energy, making it impossible to measure the full photon spectrum. Hence, all measurements require a minimum photon energy where the signal to noise ratio is still acceptable. The total branching fraction is then obtained by extrapolating the signal to the full phase space. This extrapolation is based on theoretical models and is an irreducible source of systematic error. It has been argued that the energy scale is significantly smaller than m b and therefore the above extrapolation error is underestimated. Therefore it is important to measure the b → s γ photon energy spectrum as precisely as possible. So far we have been able to lower the minimum photon energy to 1.7 GeV.
In the semi-inclusive method, the B → X s γ branching fraction is determined by summing up exclusive modes, with an extrapolation procedure to account for the unobserved modes. The semi-inclusive analysis provides a more precise photon energy in the B-meson rest frame and usually provides the flavor of the decay (i.e. b or ). The extrapolation to account for the unobserved modes is is the key issue in this approach; it is based on isospin symmetry and the B-meson Monte Carlo hadronization model, which is hardly precise and reliable. All the available measurements are summarized in Table 13. Note that the CLEO measurement on the branching fraction includes b → d γ events, which is expected to be quite small ~|V td /V ts |2 ≈ 5%.
At the parton level in the two-body decay b → s γ, the energy of the photon is E γ ≈ m b /2 in the b-quark rest frame. However, the B → X s γ is not mono-chromatic, due to several effects including the width of the X s mass distribution, gluon emission and the Fermi motion of the b-quark in the B meson. The non-perturbative Fermi motion effects will raise the photon energy above m b /2, while the gluon emission will give a long low-energy tail. The low-energy tail can be described by HQE apart non-perturbative effects that can be modeled by process-independent shape functions described by a few parameters, e.g. the b quark mass and the Fermi momentum ( ) that are considered to be universal in the following inclusive B decays: B → X s γ, B → X s ℓℓ, B → X c ℓν and B → X u ℓν. Recently a significant effort was made to combine all available data for B → X s γ and B → X c ℓν to determine the HQE parameters. Using these parameters, B → X s γ branching fraction results are combined together to provide a rather precise branching fraction. The method to combine these measurement has been provided by Limosani (A. Limosani, in a private communication, performed the average for this article) and it is different from the method used by HFAG, as explained below.
The extrapolated branching fractions do not include the published model uncertainties and the uncertainty on the extrapolation factor. These errors are included in the average by recalculation in the framework of a particular ansatz. Correlations between different measurements have been ignored. The parametric error on m b is evaluated by varying m b within its uncertainty. The world average has been calculated taking into account the correlations, when available (for the time being only Belle provides these correlations), between partial branching fraction measured at different photon energy thresholds for each single analysis. In the branching fraction average we use the energy threshold of each photon energy spectrum that corresponds to the optimal overall uncertainty on the full rate after extrapolation. This differs from the HFAG method, which uses the lowest energy thresholds for each spectrum measurement. The HFAG method penalizes analyses that quote measurements at low E min , which suffer from larger systematic uncertainties.
Two different calculations were used to extrapolate the measured partial branching fractions down to a photon energy lower threshold of 1.6 GeV. The extrapolation factors were determined using as input , or with and as input . The two world averages are listed in Table 13 and they agree within their theoretical errors.
The agreement between the measured B → X s γ branching fraction and the theoretical prediction constrains various new physics scenarios. One of the most popular examples is the lower bound on the type-II charged Higgs mass, since it always constructively interferes with the SM amplitude. The current limit is around 200 GeV for any tan β if no other destructive new physics contribution exists . This limit is significantly higher than the direct search limit.
CP Asymmetries for exclusive and inclusive b → s γ transitions for B 0, B ± and B 0/B ± admixture.
The B-factories also attempted to measure the polarization of the photon in the b → s γ transition, which can provide a test of the Standard Model, which predicts the photon to be mainly left-handed [214, 215]. The measurements rely on either the exploitation of the B 0 - interference [214, 215] or the interference between higher-Kaon resonances decaying into K π π 0 [221, 222]. However, both Belle and BaBar [223, 224] measurements are not precise enough to pin down the photon polarization. The list of established decay modes of the type B → K (*) X γ, where X is one or more flavorless mesons, is increasing. These channels are interesting as they can provide a measurement of the polarization of the photon. Many decay involving kaon resonances modes (e.g. K*(892), K 1(1270), ) are now well established and their branching ratios agree with the Standard Model predictions.
Summary of upper limits for B → ℓℓ decays.
The physics of B → X s ℓℓ is governed by the Wilson coefficients C 7, C 9 and C 10, which describe the strengths of the corresponding short-distance operators in the effective Hamiltonian, i.e. the electromagnetic operator O 7 and the semileptonic vector and axial-vector operators O 9 and O 10, respectively . The Wilson coefficients are experimental observables. Contributions from new physics appear in the experiment as deviations from the SM values, which have been calculated to next-to-next-to-leading order (NNLO). The experimental knowledge on the Wilson coefficient C 7 comes from the inclusive B → X s γ branching fraction, which determines its absolute value to about 20% accuracy, but not its sign. The partial b → X s ℓℓ decay rate in the lepton invariant mass range below the J/ψ resonance is sensitive to the sign of C 7.
Summary of inclusive and exclusive b → X s ℓℓ branching fractions.
CP Asymmetries for exclusive and inclusive b → X s ℓℓ transitions for B 0, B ± and B 0/B ± admixture.
-0.07 ± 0.22 ± 0.02
-0.04 ± 0.10 ± 0.02
-0.05 ± 0.09
-0.13 ± 0.17 ± 0.02
-0.13 ± 0.17
-0.08 ± 0.12 ± 0.02
-0.08 ± 0.12
-0.10 ± 0.10 ± 0.02
-0.10 ± 0.10
0.03 ± 0.23 ± 0.03
0.03 ± 0.23
-0.22 ± 0.26 ± 0.02
-0.22 ± 0.26
0.03 ± 0.23 ± 0.03
0.03 ± 0.23
The inclusive B → X s ℓℓ branching fraction, which constrains C 9 and C 10 , gives no information on the individual signs and magnitudes of these coefficients. To further pin down the values of these coefficients, it is necessary to exploit interference effects between the contributions from different operators. This is possible in B → X s ℓℓ decays by evaluating the differential inclusive decay rate as a function of the lepton invariant mass, q 2 = m(ℓℓ)2, or by measuring the forward-backward asymmetry in the exclusive decay B → K*ℓℓ.
where ξ is a function of q 2, and F 1,2 are functions of form factors. It is straightforward to determine C 10, Re(C 9) and the sign of C 7 from the A FB distribution as a function of q 2, using the value of |C 7| from B → X s γ and a few more assumptions: phases of C 10 and C 7 are neglected, and higher order corrections are known. These assumptions should be examined by comparing the results with the inclusive B → X s ℓℓ differential branching fraction as a function of q 2, since it is also sensitive to C 9 and C 10 in a different way. Most of the eight individual B → K(K*)ℓℓ modes have been established. Both experiments have searched for asymmetries with respect to lepton flavour. Currently available data clearly favour a negative sign for C 7, as predicted by the Standard Model.
In some SUSY scenarios the sign of the b → s γ amplitude (C 7) can be opposite to the SM prediction, while the transition rate may be the same. Within the SM there is a zero crossing point of the forward-backward asymmetry in the low q 2 region, while it disappears with the opposite sign C 7 if the sign of Re(C 9) is the same as in the SM. In another model with SU (2) singlet down-type quarks, tree-level Z flavor-changing-neutral-currents are induced. In this case, the larger effect is expected on the axial-vector coupling (C 10) to the dilepton than on the vector coupling (C 9). Because the forward-backward asymmetry is proportional to the axial-vector coupling, the sign of the asymmetry can be opposite to the SM. The same new physics effect is also a possibility for B → ϕ K S where anomalous mixing-induced CP violation can occur.
Belle attempted to extract C 9 and C 10 from A FB in B → K*ℓℓ with 357 fb-1 data . The Belle analysis constrains the sign of the product C 9 C 10 to be negative as in the SM. In this study higher order QCD correction terms are assumed to be the same as in the Standard Model. Only the leading order terms, C 9 and C 10, are allowed to float in the fit to the data.
Since B 0 → K 0*ℓ+ℓ- is an all charged particle final state, LHCb may be able to measure the zero crossing point with a better precision than a B factory. However, a model independent analysis requires measurement of the forward-backward asymmetry in B → X s ℓℓ, which is only possible at a super-B factory.
The precision of the B → K(K*)ℓℓ branching fractions is dominated by the theoretical errors, which have large model dependent irreducible uncertainties in the form-factors, which can be as large as a factor two. These uncertainties are much smaller if the ratio is measured. In the SM, the B → Ke + e - and B → K μ + μ - branching fraction are almost equal, apart from a small phase space difference due to the lepton masses. However, the B → K*e + e - branching fraction is expected to be larger than the B → K*μ + μ - branching fraction, due to a larger interference contribution from B → K*γ in B → K*e + e -. In new physics models with a different Higgs sector than that of the SM, scalar and pseudo-scalar types of interactions may arise in b → sℓℓ.
Therefore, is an observable that is sensitive to new physics. For example, a neutral SUSY Higgs contribution can significantly enhance only the B → K(K*)μ + μ - channel at large tan β. The current world average for the branching fractions gives: R K = 0.96 ± 0.26, dominated by the statistical error.
Finally, the transition is theoretically much cleaner than its charged counterpart, thanks to the absence of the photon penguin diagram and hadronic long-distance effects (charmonium resonances). From the experimental point of view these modes are very challenging due to the presence of the two neutrinos. Searches for such modes at the B factories use a fully reconstructed B sample or semileptonic tagged mode. Both Belle and BaBar report limits for these modes.
Values of Wolfenstein parameters from CKM fits assuming the Standard Model
0.2252 ± 0.0008
0.155 ± 0.022
0.342 ± 0.014
where q specifies the type of neutral B meson, and Δq gives the relative size of NP to SM physics. (The SM point has Re(Δq) = 1 and Im(Δq) = 0.)
To set limits it is necessary to make some simplifying, but very reasonable, assumptions (this type of analysis was done with somewhat different assumptions earlier by ). We assume that there is no NP in the tree level observables starting with the magnitudes |V ud |, |V us |, |V cb |, and |V ub |. Since since γ is measured using tree level B ∓ decays we include it in this category.
In both case quite large values of the real part of NP amplitudes are allowed. In the B s case a large negative value of the imaginary part is also preferred, but again the effect is only 2.7σ. We can conclude that we need to do a lot more work to constrain NP in these transitions.
The goals of our studies are to elucidate the features of the electroweak interactions and the fundamental constituents of nature. The b was discovered at fixed target experiment in a proton beam, by the first observation of narrow ϒ states in 1977. Experiments at several e + e -colliders have increased our knowledge enormously. CLEO provided the first fully reconstructed B mesons and measurements of their mass. The Mark II and MAC experiments at PEP gave the first measurements of the b lifetime, which were confirmed by PETRA experiments. ARGUS first measured B 0 mixing. CLEO, ARGUS and the LEP experiments competed on the first precision measurements of the CKM parameters V cb and V ub . The two most recent, BaBar and Belle, have provided measurements of CP violation.
The hadron collider experiments have measured B s mixing and are starting to probe CP violation in the B s system. The torch will soon be passed to first hadron collider experiment designed to measure b decays, LHCb. The ATLAS and CMS experiments will also contribute.
In the 31 years since the discovery of the b quark we have learned a great deal about the physics of b-flavored hadrons. We have seen that the Standard Model describes most decays well, yet studies continue as any new particles that we believe must exist have influence, especially on rare and CP violating B decays. We expect that future experiments will learn a great deal about these new particles either by observing their effects on B decays, or by seeing that they do not have observable effects. The latter case is one that experimentalists do not like, but it can be just as important in firming up our understanding of the basic constituents of matter and the forces that operate between them.
We thank Phillip Urquijo and Antonio Limosani for useful discussions concerning the CKM elements. M. Artuso and S. Stone thank the U. S. National Science Foundation for support.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.