Nuclear Physics B (Proc. Suppl.) 13 (1990) 227-232 227 North-Holland
SEMILEPTONIC DECAYS OF CHARMED MESONS
M.S. Witherell
University of California, Santa Barbara, CA 93106
I review the experimental status of the semileptonic decays of charm. There are new results on decay rates, polarization, and Cabibbo-suppressed decay.
1. INTRODUCTION
SemUeptonic decays of heavy quarks are of par-
ticular interest because they are the easiest decays
to interpret. For this reason, they are used to de-
termine the elements of the Kobayashi-Maskawa
matrix. For the case of charm decay, the elements
are constrained in the three-generation model to be
Vcs = 0.9743 ± 0.0006 and Vcd = 0.220 :I: 0.0002.1
At the present level of theoretical and experimental
knowledge, we can only check these values at a fairly
crude level. For comparison, the effect of the third
generation on Vcs is only 0.002, which is almost two
orders of magnitude better than the direct experi-
mental measurement.
Because the K-M matrix elements are so precisely
known, in the three-generation scheme, we can use
the measurements to tune up the form factor models.
These models should then be even more accurate in
B decays, and can be used for precise determinations
of Vcb and Vub, eventually. At the same time, the
form factors are of interest themselves as a source
of information about the wave functions of mesons
containing heavy quarks.
Exclusive semileptonic decays of charmed mesons
had been more difficult to measure than nonleptonic
decays, because no narrow mass peaks are formed.
Different techniques have been developed by E691
and Mark IH, however, to achieve good background
rejection and clean signals for these decays. The ba-
sic measurements are getting to be quite good, and
the attention is turning to more detailed measure-
ments, such as angular distributions.
The description of semileptonic decay is straight-
forward. It is assumed that it p r ~ e e ~ "..by specta-
tor decay only. In the weak interaction the charmed
quark decays c --* s(d)e +re and the light antiquark
q has no effect. All strong interaction effects are con-
tained in the form factor, which describes the prob-
ability for the sq system to be bound as a K, K*,
or other hadronic state. A simple isospin argument
shows that F(D + --* X0e+v) = F(D 0 --, X-e+~,) for
each final state, in spite of the very different lifetimes.
I will thus use partial rates rather than branching ra-
tios in comparing results with theory.
2. INCLUSIVE DECAYS
We have had for some time good measurement
from Mark HI on the inclusive semileptonlc decays. 2
They measure B(D o --* e+X) = (7.5 -I- 1.1 -I- 0.4)~
and B(D + ---, e+X) = (17.0 4- 1.9 ± 0.?)~. These can
be .combined with the lifetime measurements from
E6~13 to calculate the decay rate, r(D ° --, e+X) =
(17.4 ± 2.5 ± o.9}. m I0 s -1, and 1"(D + -, e+X} =
(lS.Z ± 1.8 ± s.e) • lO l° s -1. As expected, r(D ° --,
e+X) = r(D + -, e+X), and the average value 1"(D --~
e+X) = (16.3 ± 1.5). 10 I0 s -1.
In a simple quark decay picture,
G s _5 f ( ~ ) ,
where the function f corrects for the finite mass of the
strange quark and is about 0.5. This formula gives
the right answer for mc = 1.6 GeV, which is a rea-
sonable value. Unfortunately, the dependence on the
fifth power of the charmed quark mass means that
the rate varies by a factor of 2 for a change of 0.2
GeV in the quark mass. To do better, we must use
0920-5632/90/$03.50 © Elsevier Science Publishers B.V. (North-Holland)
228 M.S. Witheteli/Semileptonic decays of charmed mesons
exclusive decays.
3. EXCLUSIVE DECAYS
The amplitude for Cabibbo-favored semileptonic
decays can be written
ACF = ~22 Vcs L ~ H~
where L~ and H/z are the leptonic and hadronic cur-
rents, respectively. For decay into a pseudesc~lar X,
D --* Xe~,, Hp = (PD + Px)p fx( t ) , in the limit of
small lepton mass. The momentum transfer variable
t = (PD -- PX) 2 is just equal to the square of the
virtual W mass, M2v. There is jdst one vector form
factor, fx( t ) , needed to describe the decay. Often one
assumes the t-dependence with a single pole form, f+X(t)=fXt0~[ 1 1 + ~ / [1-t-:~yJ' where My is the mass of the
lowest c~ resonance with JP ----- 1 - , MD~ = 2.1 GeV.
More generally, this serves as a simple parameteri-
zation of the t-dependence. Assuming a mass of 2.1
GeV, the decay rate can be expressed
r (D ° -* K-e+v) -- Iv=12lf+X(0)l 2 15.3-1010S -1 ,
where we use the value of the form-factor at t = 0 as
the normalization parameter.
A measurement for D O --* K-e+~, was recently
published from the Fermilab charm photoproduc-
tion experiment E6914 We looked at the decay chain
D*+ k_~ ~r+D 0, D O .~ K-e+~,e. We require a K - e +
pair from the secondary vertex, in association with a
~r + from the primary. The electron identification ef-
ficiency is 72%, and the mlsidentification probability
is 0.5%. Assuming that the Kx comes from the de-
cay D O --. K - e + v , one can calculate the D~r mass.
There is a clear excess in the right-sign spectrum at
the D* mass, with a total of 250 events. The result-
ing branching ratio and rates are shown in Table 1.
The t-distribution is also measured, and a fit to the
single-pole form gives a good fit with My = 2.1 +0.4 -0.2
GeV/c ~.
The Max~ [ ] collaboration has also measured
D O -* K - e + v in double-tagged events, in which
Table 1. Results for D O --~ K - e + v
Branching Ratio Decay Rate (~) (1010s -1)
E691 3.8 4- 0.5 4- 0.6 8.8 -I- 1.2 -I- 1.4 Mark IH 3.4 4- 0.5 -t- 0.4 7.8 4- 1.2 4- 0.4 Average 3.5 4- 0.5 8.2 -I- 1.2
m
the D O is observed in one of a number of hadronic
decays. 5 The electron identifcation efficiency is 75%,
and the misidentification probability is 5%. The vari-
able U = Emiss - [Pmiss[ should be 0 for a single
missing neutrino. Figure I shows the U distribution
and a Monte Carlo fit. The shaded region shows the
shape expected for D O -* K-~r0e+v, but at 100 times
the expected rate. The very clean signal of 56 K - e + v
events corresponds to the results shown in Table 1.
~ 1 0 -
O d ~ " 5 - C
uJ 0 =0.2
12-88 6218A3
' I , I
0
U
K-e*v e
0.2
(GeV)
FIGURE 1
Distribution of U calculated for D O --~ K-e+ve can- didates from Mark III (histogram) for Monte Carlo events (curve), and for the background (xl00) from D O --* K-Tr0e+ve (shaded).
The results from the two experiments agree and
a weighted average gives B(D 0 --, K-e+~,) = (3.5 4-
0.5)~ and a decay rate of B(D 0 - , K - e + v ) = (8.2 +
1.3) • 101°s-1. Plugging this into the decay rate for-
mula above gives the result [fK(0)[2[Vcs[ -~ = 0.54 +
0.08. Assuming [Vcsl = 0.975, the best measure-
ment of fK(0) = 0.75 4- 0.05. This agrees well with
predictions of Wirbel, Stech, and Bauer 0 (0.76) and
Dominguez and Paver 7 (0.75). There are also early
results from lattice gauge calculations, some of which
were discussed at the conference. 8 The basic result
M.S. Witherell / Semileptonic decays of charmed mesons 229
is that there is good agreement of the experimen-
tal result with the calculated form factor, assuming
[vcs[ = cos 0c.
The other dominant Cabibbo-favored dec~y is
D -4 K~rev. The first goal is to measure the decay
rate for D --+ K*ev, and compare it to the calcu-
lations. The second is to measure the size of non-
resonant Klreu, and thus get a first measure of the
importance of hadronic final states other than the K
and K*. The general picture from form factor models
is that 6,9 the K* to K ratio should be greater than 1.
The D + --* K-~r+e+Ve rate has been measured in
E691.10 Although the D* cut available for D°decay is
not available, the vertex cuts are particularly effective
because of the long D + lifetime. There is a signal
of 250 events over 62 background as measured with
the wrong sign (K-lr+e - ) sample. With tighter cuts
on electron-identification and vertex isolation, the
numbers are 155 signal and 14 background.
The K~r mass spectra for the two samples are
shown in Figure 2. There is a clean K* peak, which
clearly dominates the signal. The background con-
tribution, averaged over the K* width, is about 10~
with the standard cuts and 4 ~ with the tight cuts.
The results of the fit are shown in Table 2. Less than
20~ of the decays are nonresonant, which corresponds
to a very small fraction of the inclusive semileptonic
decay rate. The decay rate for D + --* K*0e+v com-
bined with the E691 result for D O ~ K - e + v leads to
the ratio F(D --* K*e+v) / r (D --* Ke+v) = 0.45+0.12,
which is significantly lower than expected. This calls
for a re-examination of the form factors. Ther~ is
sor~e discussion of this problem by Wirbel in his pa-
per to this conference. 11
Table 2. E691 Results for D + --* K-~r+e+v
Mode Branching Decay Rate Ratio (~ ) (101°s -~)
D + --, K*0e+v 4.5 -I- 0.7 4- 0.5 4.2 :E 0.6:1:0.5 D + --. (K-~r+)NRe+v 0.3 • 0.2 4- 0.2 < 0.7
A
f j
60
40
(.9 v
i'M 0 (5 or) I-.. Z t.~ > h i
0
4(3
20
| | | g g
fl (o)
(b)
0 ~ ' ~ - ~ M m 0.6 1.0 1.4 !.8
K'Tr MASS (GeV/c 2)
FIGURE 2
The K~r mass spectra for D + -4 K-~r+eve candi- dates from E691 w'th right-sign (solid) and wrong- sign (dashed) electrons: (a) loose cuts, (b) tight c.uts. The curves are fits used to extract the K* compo- nent.
There are preliminary results on D --* K~ev from
Mark HI. They see D O decaying into three modes,
K-~rOe+v, K-6~r-e+v, and K0----~-/z+v, for a total of
23{) M.S. Witherell / $emileptonic decays of charmed mesons
16 events signal with 4 background. They also see 11
D + decays with 4 background, in the modes K-w+e+v
and K6~r0e+t,. The corresponding decay rates are
I"(D 0 --. [X~rl-e+v) _- (14_~3 4- 2) . 1ol0s - I and
p(D + [~rl0e+v) = (~ o+1.6~ 1Ol0s-1. "-~ " ~ - - 1 . 3 ; "
The D + decay rate agrees with E691, although the
D O rate is higher by a factor of 3. Although the statis-
tics are low, the discrepancy between the D O decay
rate and the D + rate from either experiment appears
significant. A violation of the isospin relation would
be most surprising and since the results are prelim-
inary, ] will not use them in the comparisons to fol-
low.
Do the exclusive decays to pseudoe~alar and vec-
tor mesons saturate the measured inclusive rate? In
Table 3 1 list the known components of the semilep-
tonic rate. For the nonrssonant Klrev decay, I mul-
tiply the K-g+e+ve rate from E691 by the isospin
factor of 1.5. The Cabibb~suppressed decay can be
estimated by taking the Cabibbo-favored decays and
multiplying by, for example, the calculation for the
ratio ~e~/Kev from reference 5. This takes into ac-
count the Cabibbo suppression, as well as differences
in phase space and form factors, and should be quite
accurate. The sum is (14.1 -I- 1.5). 1010s - I , compared
to (16.3 4- 1.5)- 1010s - I for the inclusive average. The
missing rate is (2.2-I- 2.2). 1010s - I , which is consistent
with zero. There is not much room for D --* K1rlre~,,
K~/ev, etc.
Table 3. Semileptonic Decay Rate S-mmary
Mode Source Rate (1010s -1)
D O --, K-e+~, Mark ]]I/E691 8.2 ± 1.2 D + --* K*-e+t , E691 4.2 4- 0.8
D + --, (K~)NRev E691 0.4 -I- 0.4 D --* (~r, p)e~, tan 2 8c 1.3 -t- 0.2
Total 14.1 4- 1.5
Inclusive Mark IH 16.3 -I- 1.5
4. ANGULAR DISTRIBUTIONS IN D --+ K*et,
Returning to the problem of the low D --* K*ev
rate, the next step is to determine which of the form
factors causes the discrepancy. There are three form
factors in K*ezs decay, one vector form factor V(t)
and two axial vector form factors A1 (t) and A2(t).
Fortunately, the extra information necessary to ex-
tract the form factors is available in the angular dis-
trlbutions. There are three angles which define *,he
distribution. The K* decay angle 6 v is the angle be-
tween the lr and the D momenta in the K* center of
mass. The leptonic decay angle ee is the angle be-
tween the electron and the D momenta in the rest
frame of the virtual W. The angle between the de-
cay planes is X- The entire specification of an event is
given by ee, 0V, X, and t.
The angular distribution for the general case has six terms: 12
[Ai 2 = G2lVcsl2tx
{[(I "}" cos 0e)2[H+l 2 -I- (I ~ COS 8e)~ IS~ 12] sin2 8V
+ [ 2 ( 1 - cos 2 ee)[H0[2]~ cos 2 e v
- [2(1 - cos 2 0e)Re(H+a*_)]~ sin 2 8 v cos 2X
- ~ S i n e e [ ( 1 + cos 0e)Re(R+H~)
+ (1 - cos ee)Re(H- H~)]2 sin 0V~os 8 v cos X }.
The functions H+, ~I0, and H_ correspond to helicity
amplitudes of the virtual W +. They are related to
the form factors by the relations
H~(t) = (M D "I" h/IK,)Al(t ) -4- 2 ~ V ( t )
1 r(S2 - M 2, - t) ( S v -I- MK.)AI(t ) H0(t) = 2Mx, v~t D 2 2
Integrating over the azimuthal angle X, there are two
terms: a transverse component with sin 2 0V, and lon-
gitudinal with cos 2 0 V.
To get a first look at the relative amount of trans-
verse and longitudinal decays, E691 has analyzed
the dependence on 0 V. The distribution of cos 6 V is
M.S. Withersll/Semileptonic decays of charmed mesons 231
CJ (3
Z bJ > bJ
20
! 0 , n 1 - I .0 0 ID
COS
FIGURE 3
The cos 8 v distribution for the D + --+ K*0e+ve can- didate from E691. The curve is from a fit to back- ground plus signal and corresponds to F~/FT = 2.4.
shown in Figure 3. A fit to the form dF/d cos 0 V =
1 + ~ cos 2 8 v gives the result a = 3 8+. 3"4 This cor-
responds to a ratio of longitudinal to transverse of
r L / r T = (1 + a ) / 2 = 2.4+1:9 ? -I-0.2. This ratio would
be 0.5 for unpolarized K*'s, and was expected to be
1.0 in the BSW model. The K* is longitudinally po-
larized, somewhat more so than expected. The prob-
lem now is to find a convincing explanation for both
the low K* rate and the longitudinal polarization.
This fit uses only cosSv, and varies the relative
size of FL and F T. We will next fit the same data
sample using cos 0v, cosSe, X and t, varying the size
of AI, A2, and V. It may be possible to l ~ r n the
source of the discrepancy by measuring the inc~vid-
ual form factors directly.
5. CABIBBO-SUPPRESSED DECAY
The K-M element Vcd can be measured in two
ways: in neutrino production of charm, and in Cahibl~
suppressed charm decay. The first of two measm~
ments from neutrino experiments is the result from
CDHS 13 of [Vcd[ 2 = 0.044 =I= 0,012. In his paper
submitted to this conference, Shaevit~ 14 ~ves a new
result from the CCFR experhnent corrssponding to
[Vcd[ 2 - 0.047 -I- 0.009. These both agree with
the expected result of 0.048. The results depend on
semlmuonic branching ratios, the ratio of cross sec-
tions for D + and D 0, and the fragmentation function.
Mark HI has observed the Cabibbo-suppressed de-
cay D O -* ~r-e+v. 5 The analysis parallels very closely
that used for the Kev decay discussed e~]ic ~ The
~r/K discrimination is provided by thne-of-flight mea-
surement, and by the U variable. The b ~ u n d
is 0.5 events, or less than 10~, and the resulting
branching ratio is B(D 0 --, ~-e+~,) = (0 ~o+0.23 ~_ • "-~'-0.11 -~ 0.04)~. By taking into account the difference in
phase space, and assuming equal values of ft(0), the = n 05~+0-038 -I- 0.00 , .% derived result is ]VcjVcs] 2 v. "-0.015
which is consistent with the expected result of 0.049.
The systematics on the measurement of the ratio are
quite good, because errors in the electron efficiency or
form factor normaliz.tion cancel. The only limitation
is the small number of events in the signal.
6. CONCLUSIONS
The decay rates for the Cabibbo-favored semilep-
tonic decays of charmed mesons are now reasonably
well measured. The average rate for D O --* K - e + o is
(8.2-1-1.2).10 l0 see -1, which agrees with expectations.
The decay rate for D + --* K,(Je+~ is (4.2 ~ 0.6 -I- 0.5).
1010 sec -1, which is significantly smaller than the
predicted value, and the K* is longitudinally polar-
ized, more than had been expected. The nonresone~t
decay D --* K1rev is less than about 7% of the in-