Home Search Collections Journals About Contact us My IOPscience

On the properties of the Duffin-Kemmer-Petiau equation

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1993 J. Phys. G: Nucl. Part. Phys. 19 87

(http://iopscience.iop.org/0954-3899/19/1/006)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 140.117.111.1
This content was downloaded on 25/08/2014 at 07:56

Please note that terms and conditions apply.

1. Pbys. G: Nucl. Part. Phys. 19 (19s)87-98, Printed in the UK

On the properties of the Duffin-Kemmer-Petiau equation

Y Nedjadi and R C Barrett

Department of Physics, Universily of Surrey,Guildford GUZ SXH, UK

Received 5 March 1992

AbstraeL In view of recenl interest in the use of lhe Duffin-Kemmer-Peliau (DW

relatiVistic equation, we consider some of ils properlies which are needed for practical
calculalions. We also address the unresolved problem of the spinless DKP bmon in a
central field and derive lhe system of lint-order coupled differential radial equations
which enable the energy eigenvalues as well as the full wavefunctions to be evaluated.
As an example we calculate the free DKP spherical wayes and then solve the problem of
a pionic atom wilh a point Coulomb Interaction only.

1. Introduction

Recently, the first-order relativistic Duffin-Kemmer-Petiau equation has been used

to describe the interaction of S = 0 , l hadrons with nuclei [l-31. The conventional
relativistic approach, in contrast, involves solving the secondader Klein-Gordon (KG)
equation for S = 0 and the Proca equation for S = 1 [4].
The Dirac-like first-order relativistic DKP equation is not new and dates back to the
1930s [5-8]. Historically, the loss of interest in the DKP stems from the equivalence of
the DKP approach to the KG and the Proca descriptions in on-shell situations [9, lo],
in addition to the greater algebraic complexity of the DKP formulation.
As part of a larger programme to study the interactions of spinless mesons with
nuclei within the DKP approach, this paper is devoted to the study of some selected
properties of the DKP equation. Unlike earlier studies, the DKP is approached here
from the perspective of relativistic single-particle wave mechanics with an emphasis
on features needed for practical calculations.
After restating the DKP wave equation, we consider the DKP current for which
we introduce the Gordon decomposition so as to make its physical interpretation
transparent The free-particle solutions of the S = 0 DKP equation useful as a
preliminary to the central force problem are then briefly discussed.
It is of interest for scattering problems to construct the DKP propagator. We
derive in section 3.3 the Feynman propagator and the time-independent free DKP
Green's function in the asymptotic h i t .
In section 4, we investigate the solution of the S = 0 DKP equation for a
spherically symmeaic potential since previous authors considered only the DKP-
based Schrodioger-'(or KO-) equivalent equation to obtain eigenvalues and had no
procedure to construct and calculate the full DKP eigenfunction [l-31. We discuss
some general properties of the full DKP wavefunction and derive the set of first-order
coupled differential equations which determine all the DKP radial wavefunctions. This

0 1993 10P Publishing Ltd

0954-3899~~10087t12~7.SO 87
88 Y Nedjadi and R C Barrett

formalism is first applied to the free-particle case, to calculate the DKP spherical waves
needed in scattering problems, and then to the bound state problem of a charged
pion in the Coulomb field of an atomic nucleus.

2. Definition

The first-order relativistic DW equation for a free S = 0 or 1 particle of mass m is

(ip’8, - m)lCIDKp= 0 (1)

where the internal variables pJ’ (p = 0,1,2,3) satisfy the commutation relation

pJ’pP”pX+ pXP”pJ’= g’YpA +g“Xp”. (2)

The p’ matrices do not have an inverse and generate a ring consistently with integer-
spin algebra [ll]. The DKP algebra matrices p p have three irreducible representations:
a onedimcnsional representation which is trivial, a five-dimensional representation
associated with spinless particles and a ten-dimensional one relevant to vector bosons
[ll]. Explicitly, in the five-dimensional representation, the pJ’ are 5 x 5 matrices given
by

$=(o
0 0
0)
p i = (-Oi
PT
9) i=1,2,3

where
0 0
e = ( ; ;)
-1
.I=( 0 0 0)
0 -1 0 0 0 -1
p2=(o 0 0) p3=(o 0 0)

while pr designates the matrix transform of p.

The dynamical state QDKpof a spinless boson is a five-component spinor whereas
it has ten-component spinors for S = 1 particles.

3. Properties of the DKF’ equation

3.1. Adjoint equation and conserved current

The adjoint DW equation obtained from (1) and (2) takes the form

i8,,4pJ’+ mq5 = 0 (4)

with 4 = $too as the adjoint spinor for which qo = 2pDa- 1. The continuity
equation one gets from (1) and (4),

8J4P’Q) = 0 (5)
 O n the properties of the Duffin-Kemmer-Petiau equation 89

implies that &3’$ is the four-vector current density ( p = +pa$ , j = 4p$).

In order to exhibit the physical content of this D W current, one can perform a
Gordon decomposition. Writing the current as

j’ = q p v =i
2m (4ppwav+ - a , w p ~ + ) (6)

and using (l), (2) and (4) while separating the U = p terms from the U # p ones
yields

3.’ - ‘P
+jt;’ (7a)
where

= 3,’’
- i
.P
3,=, , - - (4(8’+)- (a’$)+)
2m

The j;=’ current does not have any p matrix and the corresponding expressions for p
and j are formally identical to those in the Schradinger theory. This is the convection
component of the DKP current &P+. The other component, (j& is
(pint,jint)),
associated to some internal motion since it involves average values of combinations
of the internal variables p’. We suggest in a study of a DKP particle in an
electromagnetic field [12] that 4[$,p]$ and G(p A p ) $ are associated with a
polarization density P and a magnetization density M, respectively, in such a way
.
that pint= -V P and jin, = i- V A M .
3.2. Solutwns of the DKP equation for a free S = 0 particle
If one seeks stationary solutions such as +(T) = u(k)e-i(Et-”.‘) = u(k)e-ik’s, the
free DKP equation can be expressed as

(E-m)u(k)=O (8)
where B = p’k and u(k) is a 5-component DW spinor independent of 2. In
Hamiltonian f o d , it can be shown from (1) and (2) to be

Hu = ( 6 - k + ~ a m ) u =E u (9)
90 Y Nedjadi and R C Bamett

with 6 = [po,p].Written in matrix form, the Hamiltonian reads

Nu=

(
O
m
0
0
0
m
0
-k,
-kz
-IC3
0
-k,
0
0
0
0
-k2
0
0
0
:
-k3

0
0
(10)

Its eigen-energies consist of the usual Duac values * d w as well as a threefold

degenerate 0 eigenvalue. The 0 eigenvalue solution is unexpected for this massive
spinless particle. An unusual feature also manifests itself in the spin states of the
particle. From the transformation properties of the DKP wavefunction under rotations
[lo],the spin operator can be expressed as
sk= i[p',p"] = i ~ ~ ~ , , , p ' p " ' ( k , l , m cyclic). (11)
More explicitly,

s,= (0 0)
sk

where s k are the usual 3 x 3 spin-one matrices. The eigenstates of Sz consist of

a doubly degenerate 0 eigenvalue together with a peculiar three-fold degenerate 2
eigenvalue. This spin-one solution is also present as an eigenstate of S3and of the
helicity operator S . k.
If one considers the commutation relation of the Hamiltonian with the orbital
angular momentum, one finds

= -i(6Ak)
[W,L] (13)
instead of 0. This implies that there is some missing angular momentum. Since
[H, S] = i(6 A k) (14)
it is only the total angular momentum (J = L + S ) which is conserved, i.e.
[ H , J ] = 0. (15)
Just as in the Dirac case, the Hamiltonian of this free spinless boson commutes with
the helicity operator, i.e.
- k
[H,S.b]=O (k==). (16)
It should be noted, however, that these unusual states are not measurable since
the expectation values of Sz, S, and S.b are zero in all the DKF' spin states.
Whereas the Dirac spinors are 4-component vectors reflecting the 2(2x 1) = 4 $+
degrees of freedom, the DKP spinor for spinless bosons has 5 components for only 2
degrees of freedom. The three lower components of the DKP spinor, which reflect the
presence of these extra degrees of freedom, were defined as redundant components
in earlier work [13,14]. In the context of pion-nucleus interactions, we propose a
physically more appealing interpretation of this massless spin-one object coupled to
the spinless DKP boson as the space part of the pionic axial current [15].
 On the properties of the Duffin-Kemmer-IPtiau equation 91

3.3. The DKP propagator

In order to calculate the scattering processes in the DKP theory one needs to specify
the DKP propagator. In analogy with the non-relativistic definition, the free relativistic
DKP propagator can be defined to satisfy a Green's function equation
(ipca', - m)GD,(zr,z) = J4(z'- z). (17)
G,(z,z
') accounts for the wave produced at z' = (t',r') by a unit source at
z ( t , r ) .Inserting the Fourier transform of GDKP(z',z ) into (17) gives

and therefore the free particle propagator is

1
GD,(k) = --- K-m kZ-m2
(E+$)
,G
, is a 5 x 5 matrix for S = 0 whereas it is a lox 10 matrix for S = 1. It is singular
at kZ = m2 (ko = &I-). Imposing the appropriate boundary conditions on
GDKp(z' - z), i.e. such that the positive energy DKP particle propagates forward in
time and the negative energy DKP particle propagates backward in time, one gets

with O(t' - t) being the unit step function. If one denotes by u(k) the normalized
particle spinor of (8) and by U( k) the spinor of the anti-particle, then

in which case the free propagator can be written

x (I\+e-ik'("'-z)O(t' - t ) + A-eik'(3'-")O(t - I()) (22)

where At = u6/2m (A- = vfi/2m) is the positive (negative) energy projection
operator.
For time-independent scattering problems, the free Green's function in coordinate
representation is

A lengthy but otherwise straightforward calculation yields, in the asymptotic limit,

Note that IC' N ki. G,,(r,r') appears in the form of the product of the usual free
Green's function with the positiveenergy DKP projection operator.
92 Y Nedjadi and R C Barrefr

4. The central force problem

The solution of the DKP equation for a particle in a central field needs consideration
since earlier work [l-31 involved solving a Schriidinger- or KG-equivalent equation
solely to determine eigenvalues with no procedure to construct and compute the full
DKP wavefunction. We discuss here some general properties of the DKF’ wavefunction
for a spinless boson bound in a spherically symmetric potential. The formulation
discussed below can be extended to vector-boson DKP problems.

4.1. Formalism
We assume the central interaction to be the sum of a Lorentz scalar (U;) and a
time-lie vector (U:) potential. The stationary states of the DKP particle in this case
are determined by solving

- + m c 2 t U, + PU;) +(r) = p0E+(v)

(p k c (25)
where the five component D W spinor

Eq.(25) cannot be expressed in the Hamiltonian form as in the freeparticle case. It

can be rewritten, in a suggestive way, as

which is a form somewhat analogous to the one often used for the Dirac equation.
Using the definition of [16], the second equation which couples and $qWcr
without involving the energy (or the time derivative in general) IS known as a
subsidiary condition of the first kind. More explicitly, equation (U)
is

+
(mc2 Us)4 = ( E - U:)v + hcV .A
h c V 4 = ( m e z + U,)A
+ US)V= ( E - U;)+
(4
where A is the vector (A,, A,, As).
In all eigenstates of (25), the total angular momentum J = L + S,that is,

J = ( ”0 L+s )
which commutes with p”, is a constant of the motion, i.e.
-
d (?,@”+) = 0.
dt
 On the properties of the Dufin-Kemmer-Petiau equation 93

Thus for a DW boson in a central potential, one can construct a simultaneous

eigenfunction of (B),J z and J3 with corresponding eigenvalues denoted E, J ( J t 1 )
and M.
Since the five-component wavefunction $ is simultaneously an eigenfunction of
J z and J3,

The full DKP spinor $ is not an eigenstate of Lz which is not a constant of motion.
However, L2 is equal to J z when it acts separately on so that any eigenfunction
of Jz is also an eigenfunction of Lz.If the eigenvalue of L2 acting on $upper
is tup,(lup,t 1) then tup.= J. On the other hand, this does not hold for $lowe,.
The most general eigensolution of (25) is therefore

where Y J M ( R )is the spherical harmonics of order J and Y J y , ( R ) stands for the
normalized vector spherical harmonics [17]

Inserting + J M ( ~ )as given in (32) into (28) while using the properties of vector
spherical harmonics [18] below

the spin-angular functions drop out and one gets the following set of first-order
coupled differential radial equations
94 Y Nedjadi and R C Barref1

1
=-
hC ((mC2t 'S)fmJ - U:)g.J).
-(E (354

with a~ = J ( J + 1)/(2J -t 1) and CJ = d m . If one redefines the radial

functions in such a way that

then the system of radial differential equations becomes

(E-U~)F(r)=(mc2+Us)G(r)

= -1( ( m c ' + U , ) F ( r ) - ( E - U , O ) G ( r ) ) .
tic (374
Whereas this system of radial equations is particular to the assumed Lorent?. scalar-
vector potentials, the method developed here to derive it can be used for interactions
of any Lorentz type provided they are spherically symmetric. The nature of the
coupling between the differential equations reflects the extent to which the assumed
interaction mixes the different components of the DKP spinor between themselves.
The radial wavefunetions in (37) are constrained in such a way that the full DKP
spinor must satisfy the normalization condition

/GJ,w(r)Pod'J,w(r)dT =1 (38)

(a non-covariant normalization is chosen here) in which case using the orthonormal-

ization properties of spherical harmonics yields

Jd- 1
Re[F'(r)G(r)]dr = -.
2 (39)

This constraint ditfers significantly from the traditional K G normalization condition.

Finally, to fully specify the DKP wavefunetion its parity needs to be defined. The
-
parity operator is II = # P ( O ) where q0 = 2p02 1, explicitly
/1 0 0 0 o\

\o 0 0 0 -1/
 On the properties of the Dufin-Kemmer-Petiau equation 95

and P(’)is the ‘orbital parity’ operator (i.e. P(.)+(T)= + ( - r ) ) . The parity of YJM
and Y?,-,,,is ( - l ) J t l so
is (-1), and the parity of the irrotational fields Y,~,,,,
that

Even though the two upper components and three lower ones have opposite parities,
$J M ( ~ has
) a well defined parity of (- 1) J .

4.2. Free DKP spherical waves

As a first application of the preceding formulation we assume U, = U: = 0, in which
case the stationary solutions with well defined angular momentum and parity are the
free DKP spherical waves. The corresponding system of differential equations to be
satisfied by the radial wavefunctions is
EF = mcZG
tic mc2Hl
dF J 1
tic -+-F
(dr ) =-mc2H-1
CJ

( m 2 F - EG) . (42d)
r
Eliminating G, HI and H-, in favor of F, one obtains the second-order
differential equation

This equation is the free radial KG wave equation and it would coincide with the
free radial Schrodinger equation if E2- m2c4were replaced by 2mc2e, E being the
non-relativistic energy. Equation (43) therefore verifies the consistency of the system
of coupled differential equations (37).
For any positive value of E* - m2c4, (43) has only one regular solution.
Introducing Icc = JEz - m2c4 (IEI 2 me2) and p = kr/ti, the regular solution
is

F(p)=Npjj(p) (44)
where j , is the Bessel function of order J and N is a normalization constant. Using
(42) together with known properties of the Bessel functions, the remaining radial
wavefunctions are found to be
96 Y Nedjadi and R C Barreft

4.3. The pionic atom

As another application of the formalism discussed in section 4.1, we consider the
bound states of a spinless charged pion (r-) in the Coulomb field of a nucleus.
Unlike pionic atoms found in nature, it is assumed here that there is no nuclear
interaction involved and the atomic nucleus is treated as a point charge Ze.
The eigensolution of the DKP wave equation

(P .kc + + 0%) $ ( r ) = PoEll(r) (6)

where V, is the Coulomb potential, can be expressed as in (32) whose radial
wavefunctions must satisfy the system of coupled equations (37) with U, = 0 and
U,.= v,.
In the particular case of only a point-like Coulomb potential, the coupling between
the differential radial equations is simple so that an exact analytical solution can be
found. If one introduces the parameters

the system of coupled equations becomes

QJ (dp- -
P
Hl

where A, = h / m c . Inserting (48a,6)into (48c) yields the second-order equation

with

Equation (49) appears as the radial KG equation for a point-charge Coulomb fieldt.
If one identifies J ( J + 1) - yZ with e,(e, +
l), equation (49) coincides with the
radial Schrodinger equation for a Coulomb interaction whose physically acceptable
solutions are expressed in terms of Laguene polynomials.

t This provides a further verification of the mnsislency of the set of coupled differential radial equations
(37) since, for a lime-like torentz vector interaction, the second order differential equation obtained for
the first component of the DW spinor from (28) is equivalent to the corresponding KG equation.
 On the properfies of fhe Dufin-Kemmer-Pefiau equafion 97

0.20 , I I

Figure 1. DKP radial wavefundions for the 36 Figure 2. Comparison of the Klein-Gordon and
pion-atomic state of Bi (2 = 83 ). the upper DW radial wavefunctions for the 3d pion-
atomic state of Bi.

Denoting by n the principal quantum number, the series terminates only if

X =n-J + e, (51)
so that the eigen-energy is
mc2
= J1+(.Ix,2
,
and the binding energy, B, = mc2 - E, ,, can be written
BnJ= m e -
2n2
The first term of (53) is the non-relativistic binding energy, namely, the Bohr formula.
The second term coincides with the Dirac relativistic correction if J is assumed to be
the total angular momentum of the corresponding charged fermion.
The radial wavefunction of (49) is of the form
F,,,(p) = Ne-$pf'LZfet1
n' (P) (54)
where n' = X -e, - 1, N is a normalization constant and L$t'(p) is the associated
Laguerre function [19]. The other radial wavefunctions needed to specify fully the
bound D W wavefunction are then
98 Y Nedjadi and R C Barrett
An illustration of the shape of the radial wavefunctions is shown in figure 1. As
a test of the computer code, which we have developed to solve numerically the set of
coupled differential radial equations for realistic descriptions of pionic atoms which
include nuclear interactions, one can seek a numerical solution of (37) for a point-
like Coulomb potential only. For instance, for the 3d pion-atomic bound state of Bi
(2= 83), whose binding energy we numerically compute to be 2.898299 MeV with
0.06% relative error compared to the exact analytical eigenvalue ((53)), the upper
components F ( r ) and G ( r ) are shown in figure 1. They are of the same magnitude
and both decay exponentially. The lower components H,(r) and H-,(r), which are
one order of magnitude smaller than the upper ones, also decay exponentially but
peak at a smaller radius.
In figure 2, the Kiein-Gordon wavefunction FKG(r)is compared with the upper
components FDu(r) and G(r). Since the KG probability density is

and the DKP one takes the form

r2eDw(r)= 2Re[F'(r)G(r)]= 2 ( E - K)IFDKp(r)12 (57)
mcz
while both are constrained by the normalization condition J;pZedr = 1, the
wavefunctions FKG and FDKpdiffer by a normalization constant even though they
satisfy the same wave equation (49). This also implies that the probability density
eKGcoincides with eDW Note that this is true only for interactions of time-like
Lorentz-vector type.

Acknowledgment
The financial support of SERC through GR/F/4105.1 is gratefully acknowledged.

References
[I] Glbermann G 1986 Pkys RN. C 34 2240
[2] Clark B C, Hama S, Kilbennann G, Mercer R L and Ray L 1985 Phys. Rm Lex 55 592
[3J Kozack R E, Clark B C, Hama S, Mishra V K, Mercer R t and Ray L 1989 Phys. Ro! C 40 2181
[4] Scadmn M D 1979 Advonced Q u a m m Theory (New York Springer)
[SI Petiau G 1936 Acod R. Bels CL Sci Mem CollccL 16 No 2
[6] Kemmer N 1938 Ror R SE 166 127; 1939 h R Soc. A 173 91
[7] Dumn R J 1938Pkys. Ro! 54 1114
[SI Krajcik R A and Nieto M M 1977 A m 3. Pkys. 45 818
[9] Fischbach E, Nieto M M and Scott C K 1972 Pros nteoz P& 48 574
[lo] Krajcik R A and Niela M M 1974 P/ys Reu D 10 4049
Ill] Fschbach E, Nieto M M and Scalt C K 1973 I. Math. Phys. 14 1760, 1974 Phys RN. IS 60
1121 Nedjadi Y and Barreit R C to be published
[13] Heitler W 1943 Proc R. Irish Acad 49 1
1141 Krajcik R A and Nieto M M 1975 Pkys Rn: D 11 1442
[IS] Nedjadi Y, Barrett R C and Cooper E D 10 be published
[16] Koutroulos C G 1991 Pkysics Essnys 4 228
1171
i18j Z
IW'E H 1954 ~ m . P@ 22 211
. ~ ,.
Dawdov A S 1976 Ounnavn Mechanics (New York Pereamon) D 337

. .
1191 Bethe H A and Salveter E E 1 9 n Q - u a n m Meckanics of One- and Two-EIecnon A l o m (New
York: Plenum) p i~