Polarized Deep-Inelastic Lepton Scattering From The Polarized Two - and Three-Nucleon Bound States

PHYSICAL REVIEW C VOLUME 56, NUMBER 4 OCTOBER 1997
Polarized deep-inelastic lepton scattering from the polarized two- and three-nucleon bound states
R.-W. Schulze* and P. U. Sauer
Institute for Theoretical Physics, University Hannover, D-30167 Hannover, Germany
~Received 15 August 1996!
A relativistic description of polarized deep-inelastic lepton scattering from polarized nuclei is presented. It is
based on front form dynamics. Convolution formulas for the nuclear spin structure functions g A1 and g A2 are
derived. They require the front form spin distributions of nuclei and the nucleonic spin structure functions. The
description is applied to the deuteron and the trinucleon bound states. The numerical calculations show that
relativistic effects arising from the consistent quantum mechanical treatment of spin are small. Simple approxi-
mations to the convolution formulas for g A1 and g A2 are justified. The approximative convolution formulas allow
the subtraction of nuclear effects from measured spin structure functions and therefore the experimental
determination of the neutron spin structure functions. Polarized 3 He turns out to be a rather reliable effective
neutron spin target for deep-inelastic scattering of polarized leptons. The differences between descriptions
based on front form and on instant form dynamics are also discussed. @S0556-2813~97!03908-3#
PACS number~s!: 25.30.2c, 24.70.1s, 25.10.1s, 29.25.Pj
I. INTRODUCTION describe deep-inelastic lepton scattering from the deuteron

@12–14# and the three-nucleon bound state simultaneously.
When 3 He is described as a system of three nucleons Deep-inelastic lepton scattering from nuclei is described
interacting through instantaneous potentials, the total angular in front form dynamics. Front form dynamics allows the use
momentum of 3 He is carried to a large extent by the neutron. of impulse approximation consistently @15–17#. This is a for-
Employing realistic two-nucleon potentials the calculated mal quantum mechanical consistency. We believe that im-
neutron contribution to the nuclear spin ranges @1# from 0.85 pulse approximation is also physically reliable for the de-
to 0.90. This theoretical observation motivated experimental- scription of deep-inelastic lepton scattering. The form of
ists @2# to use polarized 3 He targets as substitutes for un- plane wave impulse approximation ~PWIA! used neglects the
available neutron spin targets. The tacid assumption is that interaction between the hadrons, produced on the struck
nucleon, and the spectator residual nucleus in the final state,
the impurities present in the substitute neutron spin targets,
and it neglects all many-hadron currents. The paper extends
i.e., the unwanted contributions to the 3 He total angular mo-
the successful description, given previously @18# for deep-
mentum arising from the proton spins and from the nucleonic inelastic scattering of unpolarized leptons from unpolarized
orbital motion will not blur the neutron properties one would 3
He, to the case with polarization. Section II recalls some
like to see. Reference @3# discusses an unfavorable case: In facts on front form dynamics needed for the actual calcula-
inclusive quasielastic scattering of polarized electrons from tion. Section III parametrizes the nuclear current tensor of a
polarized 3 He @4–9#, the spin-dependent response R TL 8 car- spin- 21 target; it introduces the spin-averaged and the spin-
ries some information on the poorly known neutron charge dependent structure functions; the generalization to spin-1
form factor. However, the neutron signal in R TL 8 remains targets, e.g., to the deuteron, is given in the Appendixes A
weak and gets overwhelmed by the proton one at small mo- and B. Section IV describes the cross section for longitudinal
mentum transfers. The neutron charge form factor is small lepton polarization in terms of the nuclear structure func-
compared to the proton one and therefore contributes little, tions. Section V derives convolution formulas for the spin-
despite the fact that the neutron makes up most of the 3 He dependent structure functions in PWIA. They require the
total angular momentum. spin distributions of the deuteron and the trinucleon bound
This paper discusses deep-inelastic scattering of polarized states in terms of front form variables; those front form spin
leptons from polarized 3 He. Experiments measure the beam distributions are determined in Sec. VI for the trinucleon
asymmetry, in particular @10,11# the 3 He spin structure func- bound state and in Appendix C for the deuteron. Our results
tion g A1 ; from g A1 the neutron contribution should get un- are discussed in Sec. VII. Conclusions are given in Sec. VIII.
folded and the neutron spin structure function g n1 should get
extracted. The conclusion of the paper will be that indeed the II. ELEMENTS OF FRONT FORM DYNAMICS
neutron spin structure function g n1 can successfully be ob-
tained in these experiments. In deep-inelastic lepton scatter- First, we give our notation. We exhibit the instant form
ing polarized 3 He is a reliable neutron spin target in contrast components of any four-vector A in round brackets as usual,
to the experience of Ref. @3#. However, the theoretical dis- i.e.,
cussion of the paper will often be kept general in order to
A5 ~ A 0 ,A 1 ,A 2 ,A 3 ! , ~2.1!
*Present address: Scandpower GmbH, Flotowstraße 41-43, D- and abbreviate the three spatial components (A 1 ,A 2 ,A 3 ) by
22083 Hamburg, Germany. AW . We use the metric g m n 5(1,21,21,21) and the four-
0556-2813/97/56~4!/2293~23!/$10.00 56 2293 © 1997 The American Physical Society

2294 R.-W. SCHULZE AND P. U. SAUER 56
dimensional totally antisymmetric tensor e m n ab with In the description of deep-inelastic lepton scattering the
e 012352 e 012351. For the description of front form dynam- front form null vectors n and l are chosen by
S A D
ics we choose the following basis vectors:
Q0 2Q 2 k̂ e 2 ~ k̂ e •Q̂ ! Q̂
n5 ~ 1,2n̂ ! , ~2.2a! n̂52 Q̂1 ,
A~ Q 0 ! 2 2Q 2 ~ Q 0 ! 2 2Q 2 A12 ~ k̂ e •Q̂ ! 2
l5 ~ 1,n̂ ! , ~2.2b! ~2.7!
k e being the initial four-momentum of the lepton and Q its
e 1 5 ~ 0,ê 1 ! , ~2.2c!
momentum transfer to the target nucleus. The choice ~2.7!
yields Q 1 5Q 0 1n̂•Q W 50. The basis vector e 1 is defined by
e 2 5 ~ 0,ê 2 ! . ~2.2d!
Q'
They satisfy the conditions n 2 5l 2 50, e i •e j 52 d i j , ê 1 5 , ~2.8!
u Q' u
2 n•l51 and n•e i 5l•e i 50. The vectors n and l are null
1
vectors. The three-dimensional vectors ê 1 , ê 2 , and n̂ form a the definition of e 2 follows then from the requirement ~2.3!.
positively oriented three-dimensional set of orthonormal ba- Thus, the momentum transfer Q has the representation
sis vectors such that
Q5 $ 2Q 0 , A2Q 2 ,0,0 % . ~2.9!
1
e e e m n ab n a l b 51, ~2.3!
2 1m 2n Second, we need the description of states for the nuclear
target and for the constituents of the nuclear target. For the
the carets over vector symbols indicate three-dimensional description we use eigenstates of the mass operator M , of the
unit vectors. Four-dimensional vectors are built up from front form components P of the four-momentum P and of
those basis vectors, i.e., the spin JW f . The spin operator JW f is defined in terms of the
1 1 Pauli-Lubanski vector W @16#, i.e.,
A5 A 2 n1 A 1 l1A 1 e 1 1A 2 e 2 ,
S D
~2.4a! m
2 2 1 21 P
~ 0,JW f ! m 5 L W n. ~2.10!
where the front form components M f M n
W,
A 2 5l•A5A 0 2n̂•A ~2.4b! The Lorentz transformation L f ( P/M ) boosts a particle of
mass M from rest to four-momentum P. In Eq. ~2.10! the
W, Lorentz transformation L f ( P/M ) is still operator valued in
A 1 5n•A5A 0 1n̂•A ~2.4c!
its dependence on mass M and momentum P. The corre-
sponding boosts L f ( P/M ) belong to the front form kine-
A 1 52e 1 •A, ~2.4d!
matic subgroup of the Lorentz group; thus, even for compos-
ite systems with interactions between constituents, the boosts
A 2 52e 2 •A ~2.4e!
L f ( P/M ) are interaction free; this is why the components
are introduced. We arrange the front form components of a P of the momentum are called kinematic. The eigenstates are
S D S D
vector in curly brackets, i.e., chosen in the form
A5 $ A 2 ,A 1 ,A 2 ,A 1 % , ~2.5a! M2 m2
P p
A5 $ A 2 ,A% , ~2.5b! u pl & 5 u pl & . ~2.11!
2W /M 2 2
j ~ j11 !
and abbreviate the three kinematic components 2 v •W/M l
$ A 1 ,A 2 ,A 1 % by A and the transverse components $ A 1 ,A 2 %
among them by A' . The three-dimensional basis vectors The defining spin operators are given covariantly, e.g.,
ê 1 , ê 2 , and n̂ of front form dynamics may be assumed to 2W 2 /M 2 5JW 2f . The spin projection 2 v •W/M is chosen
differ from those of instant form dynamics; thus, the compo- with respect to the direction of quantization v , v taken as a
nents $ A 1 ,A 2 % and (A 1 ,A 2 ) may be different; in this paper spacelike vector which satisfies v 2 521 and v • p50. The
we choose them to be identical. The four-dimensional vol- standard choice of v for front form dynamics is
ume element of integration is
p m
v5 2 n. ~2.12!
4 0
d A5dA dA dA dA , 1 2 3
~2.6a! m n•p
1 In this case the spin projection operator becomes

d 4 A5 dA 2 dA 1 dA 2 dA 1 , ~2.6b! 2 v •W/M 5n̂•JW f . For a spin- 21 particle the operator JW f is
2
realized by the vector of Pauli matrices sW /2. The notation of
W 5dA 1 dA 2 dA 3 and
we shall use the abbreviations d 3 A states does not keep the eigenvalues of mass and spin, m and
1 1
d A5dA dA dA 5dA' dA .
3 1 2
j; it only indicates the kinematic components of the momen-
56 POLARIZED DEEP-INELASTIC LEPTON SCATTERING . . . 2295
tum and the spin projection, p and l. The states ~2.11! are corresponds to different rest frame vectors JW c in instant and
orthonormal and complete, i.e., JW f in front form dynamics. They are related @16# by the
Melosh rotation R M (p), i.e.,
^ p8 l 8 u pl & 5 d ~ p8 2p! d l 8 l , ~2.13a!
(l E d 3 pu pl &^ pl u 5I. ~2.13b!

R M ~ p ! 5L 21
f SD SD
p
L
p
m c m
, ~2.15a!
They are connected to the states u pR l & in the particles rest J ic 5 (j R M ~ p ! i j J jf . ~2.15b!
frame with p R 5 $ m,0,0,m % by the unitary transformation
U @ L f (p/m) # according to The Melosh rotation is an operator in spin space. For a spin-
A F S DG
1
2 particle it takes the explicit form
m p
u pl & 5 U Lf u pR l & , ~2.14!
p 1 m m1p 1 1in̂• ~ sW 3 pW !
R M~ p !5 . ~2.15c!
A~ m1p 1 ! 2 1p'2
the extra factor Am/p 1 ensuring orthogonality in the form
~2.13a!. The transformation does not change the eigenvalue The quantum mechanical states are related to each other by
l.
Need will arise to also use basis states u pW s & in the quan-
tum mechanical framework of instant form dynamics. The
latter states are defined in the same way as the corresponding
u pW s & 5 (l u pl & AU U ]p
] pW
^luR1
M ~ p !u s & . ~2.16!
front form states u pl & are in Eq. ~2.11!. Differences are due In Eq. ~2.16! u ] p/ ] pW u 5 p 1 /p 0 is the Jacobian for the trans-
to the different choice of kinematic components of the mo- formation of momentum variables. The instant form states
mentum. Though the components pW are called kinematic, the u pW s & are also chosen to be orthogonal and complete without
instant form boosts L c (p/m) are not interactionfree for com- extra weight factors in their phase space as the front form
posite systems, in contrast to the properties of the front form states are chosen according to Eqs. ~2.13a! and ~2.13b!.
boosts L f (p/m); the indices c and f differentiate the form of
dynamics, c standing for canonical represents the instant III. NUCLEAR CURRENT TENSOR
form, f standing for front represents the front form of dy-
namics. In addition, the instant form boosts L c (p/m) of a The current tensor ^ n A u W Am n (Q, P A ) u n A & of the target A
particle with mass m from rest to the momentum p are rota- with mass m A , four-momentum P A , and polarization n A is
tionless, whereas the corresponding front form boosts required for the description of inclusive processes; it is de-
L f (p/m) contain a rotation. Thus, the same four-vector W fined by
^ n A u W Am n ~ Q, P A ! u n A & 5 ~ 2 p ! 6 (
b
X
E A 3
d PX
P1
A
mA
^ PA n A u J Am ~ 0 ! u PX b X & 2 d 3 ~ PX 2Q2PA ! d ~ P 2 2 2
X 2Q 2 P A !
3^ PX b X u J An ~ 0 ! u PA n A & A P1
A
mA
~3.1!
in terms of the nuclear current J Am (0) at time-space point 0; assumed to be in pure quantum mechanical states u PA n A & .
the proton charge e p is split off from the current. General They are defined as eigenstates of the mass M , the momen-
final states u PX b X & of c.m. momentum PX can be reached in tum P, the spin 2W 2 /M 2 , and of the spin projection
the scattering process, b X describing discrete quantum num- 2n A •W/M in the direction of polarization n A with eigen-
bers and also the modes of internal excitation; they are on value j, i.e., 1 and 21 , respectively. Unless stated otherwise,
mass shell, they are not observed in inclusive processes. The the plane-wave d -function producing part of the states
momentum transfer to the target nucleus is Q. The current u PA n A & will be considered to be split off in the following.
tensor ^ n A u W Am n (Q, P A ) u n A & is obtained from its spin- Though the target states are assumed to be quantum me-
operator form W Am n (Q, P A ) and from the density operator chanically pure for the purpose of calculation, polarization
r A (n A P A ) by will nevertheless be described by the density operator
r A (n A P A ). It is an operator in nuclear spin space. It is a
^ n A u W Am n ~ Q, P A ! u n A & 5Tr@ W Am n ~ Q, P A ! r A ~ n A P A !# . Lorentz scalar and normalized by Tr@ r A (n A P A ) # 51. When
~3.2!
referring to pure states as assumed, it has the property
3
Deuteron and He are the considered target nuclei. They r 2A (n A P A )5 r A (n A P A ). The polarization n A chosen by ex-
have spin 1 and 21 , respectively; for calculations they are perimentalists does in general not coincide with the standard
direction v of quantization in Eq. ~2.12!. Thus, the target Its Lorentz structure is built from the three four-vectors Q,
states u PA n A & are to be expanded in terms of the target basis P A , and W. The current tensor operator W Am n (Q, P A ) has the
states u PA l A & of Eq. ~2.11!, i.e., general form
u PA n A & 5 (
l A
u PA l A &^ l A u n A & , ~3.3! W Am n ~ Q, P A ! 5 F Q mQ n
Q 2 G
2g m n F A1 ~ x,Q 2 !
A
mA
the transformation parameters ^ l A u n A & depend on the mo- P̃ Am P̃ An 1

mentum P A ; the trace operation in Eq. ~3.2! is carried out 1 F A2 ~ x,Q 2 !
Q• P A mA
with respect to matrix elements of the basis states u PA l A & at
fixed target momentum PA .
1i
1 e m n ab
F
Q W g A ~ x,Q 2 !
j Q• P A a b 1
S D G
A. Trinucleon density operator
Q•W A
The full density operator r A (n A ) commutes with the mo- 1 W b2 P g A ~ x,Q 2 !
Q• P A A b 2 mA
mentum operator P. Its projection r A (n A P A ), i.e.,
~3.8!
^ PA l A8 u r A ~ n A ! u PA l A & 5 ^ l A8 u r A ~ n A P A ! u l A & , ~3.4!
with
onto the Hilbert sector of momentum eigenstates with eigen-
value PA is needed in Eq. ~3.2!. The density operator of a Q• P A
spin- 21 target is parametrized in the form P̃ Am :5 P Am 2 Q m, ~3.9a!
Q2
r A~ n A P A ! 5
1
2F12
2
n •W .
mA A G ~3.5!
x:52
AQ 2
2Q• P A
, ~3.9b!
The parametrization satisfies all required constraints. It is at

most linear in the Pauli-Lubanski vector W which is trace- x being the Bjorken scaling variable. The dependence of the
less, i.e., Tr@ W # 50. The vector n A is the polarization vector current tensor operator on the Pauli-Lubanski vector W is
according to which target polarization is defined. The param- linear for a spin- 21 target, j5 21 being the spin quantum num-
etrization ~3.5! satisfies all required constraints for a quan- ber. The current tensor operator is built from four basic ten-
tum mechanically pure target state, provided n 2A 521 and sor forms which satisfy all symmetry requirements. Those
n A • P A 50. basic tensor forms get augmented by structure functions
Matrix elements of the density operator of any momentum which are real-valued Lorentz scalars and therefore depend
P A are related to those of the rest frame, i.e., on the nontrivial scalars Q 2 and Q• P A or, equivalently, on
the Bjorken scaling variable x and Q 2 .
The current tensor ^ n A u W Am n (Q, P A ) u n A & follows from the
^ l A8 u r A ~ n A P A ! u l A & 5 ^ l A8 u 21 @ 11n̂ R • sW A # u l A & . ~3.6!
current tensor operator W Am n (Q, P A ) of Eq. ~3.8! and from the
density operator r A (n A P A ) of Eq. ~3.5! according to Eq.
For Eq. ~3.6! the identities
~3.2!. In fact, the current tensor ^ n A u W Am n (Q, P A ) u n A & is ob-
A F S D GF S D S D G
n tained from the operator ~3.8! by replacing the Pauli-
Wm mA PA PA m
sW A Lubanski vector W by the polarization vector n A according
uP l &5 U Lf Lf 0,
mA A A 1
PA m A mA n
2 to the relation
3 u PR l A & , ~3.7a!
n Am 5 ~ 1/ jm A ! Tr@ W m r A ~ n A , P A !# . ~3.10!
n Am 5L f S D PA
mA
m
n
~ 0,n̂ R ! n , ~3.7b! The two spin-averaged structure functions F A1 and F A2 and
the two spin-dependent structure functions g A1 and g A2 deter-
mine the current tensor in full; they are used in the notation
are used, P R 5 $ m A ,0,0,m A % denoting the target rest frame standard for deep-inelastic lepton scattering; they are related
momentum, n R 5(0,n̂ R ) its rest frame polarization and to the corresponding structure functions W A1 , W A2 , G A1 , and
sW A /2 its angular momentum operator. The constraints on G A2 of Ref. @3#, usually employed at lower energy and mo-
r A (n A P A ) are best proven in the rest frame. mentum transfers, by
The corresponding density operator of a spin-1 target is
given in Appendix A. mA A 2
F A1 ~ x,Q 2 ! 5 W ~ Q ,Q• P A /m A ! , ~3.11a!
A 1
B. Trinucleon current tensor operator
Q• P A A 2
The current tensor operator W Am n (Q, P A ) is Hermitian and F A2 ~ x,Q 2 ! 5 W 2 ~ Q ,Q• P A /m A ! , ~3.11b!
conserved, it preserves parity and time-reversal invariance. mA
Q• P A A 2 and n A 5 $ 2(n•n A ),n 1A ,n 2A ,(n•n A ) % 5(0,n̂ R ). The momen-

g A1 ~ x,Q 2 ! 5 G ~ Q ,Q• P A /m A ! , ~3.11c! tum transfer is used in the form ~2.9!. The kinematically
Am A 1
needed four-vectors can be given in terms of the basis vec-
~ Q• P A ! 2 tors n, l, e 1 , and e 2 , i.e.,
g A2 ~ x,Q 2 ! 5 G A2 ~ Q 2 ,Q• P A /m A ! . ~3.11d!
Am 3A 1
P A 5 m A ~ n1l ! , ~3.12a!
We are interested in describing asymmetry measurements in 2
deep-inelastic lepton scattering; they are carried out in order
to determine the spin-structure functions g A1 and g A2 . 1
Q5 Q 2 n1 A2Q 2 e 1 , ~3.12b!
All structure functions can be obtained from a given cur- 2
rent tensor by contractions with other tensors. The deuteron
current tensor is—as current tensor of a spin-1 target—quite 1
different from Eq. ~3.8! as Appendix B shows. Nevertheless, n A 5 ~ n•n A !~ n2l ! 2 ~ e 1 •n A ! e 1 2 ~ e 2 •n A ! e 2 ,
2
its dependence on the spin-structure functions g A1 and g A2 , ~3.12c!
being contained in the part proportional to e m n ab , is the same
also for the deuteron. Thus, the relations by which the spin
structure functions g A1 and g A2 are recovered from a computed
current tensor will hold for the deuteron in identical form.
n A2
Q•n A
Q• P A S
P A 52 ~ n•n A ! 1
A2Q 2
Q2
~ e 1 •n A ! n D
A2Q 2
C. Trinucleon current tensor in the nuclear c.m. system 2 ~ e 1 •n A ! l2 ~ e 1 •n A ! e 1
Q2
The nuclear current tensor W Am n (Q, P A ) is evaluated in the
nuclear c.m. system. We shall now use the notational identi- 2 ~ e 2 •n A ! e 2 . ~3.12d!
fications P A 5 P R for the four-momentum of the target and
n A 5n R for its polarization, i.e., P A 5 $ m A ,0,0,m A % 5(m A ,0W ) The nuclear current tensor takes the form
^ n A u W Am n ~ Q, P A ! u n A & 5 F Q mQ n
Q2
2g mn
G F A1 ~ x,Q 2 !
A
1
P̃ Am P̃ An A
m A Q• P A 2
1
F ~ x,Q ! 1
2
iA m n ab
mA mA
e n a1
2 A2Q 2
Q2
e 1a S D
H F
3 2n b
1
2 S
~ n•n A ! g A1 ~ x,Q 2 ! 1 ~ n•n A ! 1
A2Q 2
Q 2 D 1
G S
~ e 1 •n A ! g A2 ~ x,Q 2 ! 1l b ~ n•n A ! g A1 ~ x,Q 2 !
2
2
A2Q 2
Q 2 D
~ e 1 •n A ! g A2 ~ x,Q 2 ! 2e 1 b ~ e 1 •n A !@ g A1 ~ x,Q 2 ! 1g A2 ~ x,Q 2 !#
2e 2 b ~ e 2 •n A !@ g A1 ~ x,Q 2 ! 1g A2 ~ x,Q 2 !# . J ~3.13!
Equation ~3.13! rewrites Eq. ~3.8!—after the trace operation ~3.2! is carried out—with respect to the antisymmetric part of the
nuclear current tensor; only that part matters for the extraction of the spin-structure functions g A1 and g A2 ; it is the same for the
current tensors of the three-nucleon bound states and of the deuteron.
D. Extraction of spin-structure functions

The spin-structure functions g A1 and g A2 can be extracted from particular matrix elements or by contractions with four-
vectors and tensors available from the kinematic setup of the experiment as in Ref. @3#. We label the first extraction scheme
by I, the second extraction scheme by II.
The extraction scheme I reads the nuclear spin-structure functions g A1 and g A2 off from selected components of the nuclear
current tensor. The considered matrix element is written covariantly as a contraction with the basis vectors e 1 and e 2 , i.e.,
e 1 m e 2 n ^ n A u W Am n ~ Q, P A ! u n A & 52
iA
mA S
~ n•n A ! g A1 ~ x,Q 2 ! 12
A2Q 2
Q2
~ e 1 •n A ! g A2 ~ x,Q 2 ! , D ~3.14!
it refers to the kinematic components of the current. When specifying the nuclear polarization n A in the target rest frame, both
nuclear spin-structure functions can be extracted separately from that single relation, i.e.,
g A1 ~ x,Q 2 ! 52 ~ im A /A ! e 1 m e 2 n ^ n A u W Am n ~ P A ,Q ! u n A & u nW A i n̂ , ~3.15a!

Q2
g A2 ~ x,Q 2 ! 52 ~ im A /A ! e 1 m e 2 n ^ n A u W Am n ~ P A ,Q ! u n A & u nW A i ê 1 . ~3.15b!
2 A2Q 2
The extraction scheme II contracts the full current tensor with the tensors of other physical variables and extracts the
spin-structure functions from the resulting scalars, i.e.,
2i Q• P A Q•n A P 2A Q a n Ab 2Q• P A Q a P Ab
g A1 ~ x,Q 2 ! 5 e ^ n u W m n ~ Q, P A ! u n A & , ~3.16a!
2A Q•n A P 2A @ Q 2 1 ~ Q•n A ! 2 # 2 ~ Q• P A ! 2 abm n A A
i ~ Q• P A ! 2 Q•n A Q• P A Q a n Ab 2 @ Q 2 1 ~ Q•n A ! 2 # Q a P Ab
g A2 ~ x,Q 2 ! 5 e abm n ^ n A u W Am n ~ Q, P A ! u n A & . ~3.16b!
2A Q 2 Q•n A P 2A @ Q 2 1 ~ Q•n A ! 2 # 2 ~ Q• P A ! 2
The extraction scheme II uses relations ~3.11c! and ~3.11d!

for the spin-structure functions G Ai of Ref. @3#. Both extrac-
tion schemes—and all possible variants of them—are
d 2 s ~ pol !
dV 8e dE 8
5 S
1 d 2 s ~ 11 ! d 2 s ~ 21 !
2 dV 8e dE 8
2
dV 8e dE 8
, D ~4.4b!
H
equivalent as long as the current tensor is exact, i.e., satisfies
Lorentz covariance and current conservation. However, they d 2 s ~ pol ! Qe
5 s Mott2tan2 @ Ecosu A 1E 8 ~ cosQ e cosu A
are inequivalent for approximate current tensors. And the dV e8 dE 8 2
nuclear current tensor of this paper will only be calculated
approximately in PWIA and with wave functions which do g A1 ~ x,Q 2 !
not exhibit all required symmetry properties as Sec. VI will 1sinQ e sinu A cosf A !#
m AQ 0
explain in more detail.
22EE 8 @ cosu A 2 ~ cosQ e cosu A
J
IV. CROSS SECTION
g A2 ~ x,Q 2 !
The derivation of the inclusive cross section for inelastic 1sinQ e sinu A cosf A !# . ~4.4c!
m A~ Q 0 ! 2
lepton scattering is recalled. Let k e 5(E,kW e ) and
k e8 5(E 8 ,V W e8 E 8 ) be the initial and final momenta of the lep- It is the same for spin- 21 and the spin-1 targets; Q e is the
ton and Q5k e 2k 8e the four-momentum transfer. The cross scattering angle of the lepton, s Mott the Mott cross section. In
section is determined by the nuclear current tensor both cases the target polarization n A is described in the rest
^ n A u W Am n (Q, P A ) u n A & of Sec. III and by the corresponding frame by the polarization angles u A and f A , i.e.,
lepton current tensor ^ n e u h m n u n e & with n A 5(0,sinuAcosfA ,sinuAsinfA ,cosuA). The angles are mea-
sured with respect to the incoming lepton beam.
^ n e u h m n u n e & 52 ~ k 8e m k ne 1k 8e n k me 2g m n k e •k 8e
1im e e m n ab Q a n e b ! . ~4.1! V. PLANE WAVE IMPULSE APPROXIMATION
The calculations of this paper are carried out in PWIA.

When the lepton is longitudinally polarized before scattering,
They extend the work of Ref. @18# by the inclusion of beam
its polarization vector n e gets
and target polarization. Reference @18# also uses PWIA. Su-
ke perficially, impulse approximation assumes that only one-
m
n e 5h e ~4.2! nucleon currents j N(i) (0) contribute to the complete nuclear
me m
current J A (0), i.e.,
with h e 561 being the helicity and m e the mass of the lep- A
ton. In the rest frame of the target the cross section becomes
J Am ~ 0 ! 5 (
i51
j Nm ~ i ! . ~5.1a!
d s~ he!
2
a E8
2
5 ^ n e u h m n u n e &^ n A u W Am n ~ Q, P A ! u n A & . Stated in this manner, the assumption is clearly inconsistent
dV 8e dE 8 Q4 E
~4.3! because, for interacting systems, the current on the right-
hand side does not transform correctly as a four-vector. In
It is split into a helicity-independent and a helicity-dependent contrast, the restricted assumption
part, i.e.,
A
d 2s~ h e ! d 2 s ~ unpol! d 2 s ~ pol! JA ~ 0 ! 5 ( jN ~ i ! ~5.1b!

5 1h e . ~4.4a! i51
dV e8 dE 8 dV e8 dE 8 dV e8 dE 8
is consistent: Both sides of the latter equation transform cor-
Its polarization part has the form rectly under the front form kinematic subgroup of the Lor-
entz group; the full nuclear current can then be generated by mation is believed to be particularly reliable for deep-
translations and general Lorentz transformations. Further- inelastic processes; it yields the convolution formula for the
more, inclusive cross sections require knowledge of the nuclear current tensor.
nuclear current tensor which is quadratic in the current op-
erator and which therefore contains interference terms of the
m n A. Convolution formula for the nuclear current tensor
form j N(i) j N(k) . PWIA neglects those interference terms.
Reference @3# discusses this approximation of incoherence as Using the nuclear density operator r A (n A P A ), the nuclear
an approximation on the accessible final states; this approxi- current tensor ~3.1! takes the convolution form
^ n A u W Am n ~ Q, P A ! u n A & 5 (
tN
E d 3 pN
mN
p1
N
E `
2m A 1m A21
dM Tr@ W Nm ~nt N ! ~ Q N ~ M ! ,p N ! S ~ pN M t N ! r A ~ n A P A !# ~5.2a!
in PWIA. Its spin matrix elements are
^ l A8 u W Am n ~ Q, P A ! u l A & 5 (
tN
E d 3 pN
mN
p1
N
E `
2m A 1m A21
dM (
8
lNlN
^ l N8 u W Nm ~nt N ! @ Q N ~ M ! ,p N # u l N &^ l N l A8 u S ~ pN M t N ! u l N8 l A & .
~5.2b!
Because of the discussion on consistency at the beginning of this section, only the kinematic components m , n 51,2,1 will be
taken from the convolution formula ~5.2a!. In Eq. ~5.2a!, Q N is the momentum transfer to the struck single nucleon; the latter
momentum transfer
Q N ~ M ! :5 $ Q 2 1 P 2 2 2
A 2 P A21 ~ M 1m A ! 2 p N ,Q% ~5.2c!
2
is different from the momentum transfer Q to the whole nucleus. The momentum component P A21 (M 1m A ) of the spectator
2 1
(A21) nucleus is defined by P A21 (M 1m A )5 @ (M 1m A ) 1P(A21)' # / P A21 , M 1m A being the effective mass of the (A21)
2 2
mn
nucleus, m A21 is its minimum value. In Eq. ~5.2! the nucleonic current tensor W N(t N)
@ Q N (M ),p N # and the front form spectral
function S(pN M t N ) are introduced; their definitions will be discussed next. This section uses the capital M in a notational
meaning distinct from Sec. II.
B. Nucleonic current tensor

mn
The nucleonic current tensor W N(t N)
(Q N , p N ) is defined by
^ l N8 u W Nm ~nt N !~ Q N , p N ! u l N & 5 ~ 2 p ! 6
p1
N
(
m N bx
1
E d 3 px 1 ^ pN l N8 t N u j Nm ~ 0 ! u px 1 b x 1 & 2 d 3 ~ px 1 2Q2pN ! d @ p 2
1
2 2
x 2Q N ~ M ! 2p N #
3^ px 1 b x 1 u j Nn ~ 0 ! u pN l N t N & . ~5.3!
It is an operator in nucleonic spin space and depends on isospin t N . In Ref. @3# the corresponding instant form matrix elements
are given. After transformation to front form spin states according to
^ l N8 u W Nm ~nt N !~ Q N , p N ! u l N & 5 ( ^ l N8 u R 1 8 &^ s N8 u W Nm ~nt N ! ~ Q N ,p N ! u s N &^ s N u R M ~ p N ! u l N & ,

M ~ p N !u s N ~5.4!
8 sN
sN
the nucleonic current tensor takes the form
^ l N8 u W Nm ~nt N !~ Q N ,p N ! u l N & 5 d l N8 l N FF Q Nm Q Nn
Q 2N
2g mn
G N~ tN ! 2
F 1 ~ x N ,Q N !
1
1
p̃ Nm p̃ Nn N ~ t !
m N Q N •p N 2
F N ~ x N ,Q 2N !
1
mN G
1i
e m n ab
Q N •p N
K UF N~ t !
Q N a l N8 s b ~ sW N ! g 1 N ~ x N ,Q 2N ! 1 s b ~ sW N ! 2 S
Q N •s ~ sW N !
Q N •p N N b 2
p
N~ t !
g N ~ x N ,Q 2N ! l N D GU L
~5.5!
with the nucleonic Bjorken variable x N :52Q 2N /2Q N •p N and with

p N •Q N
p̃ Nm :5 p Nm 2 Q Nm , ~5.6a!
Q 2N
s ~ sW N ! :5 H S
m N p2N'
p1 2
N mN
n̂• sW N 12
pN'
mN
W W W
• s N' 2n̂• s N , s N' 1DpN'
mN
p1
n̂• s N , n̂• sW N .
W N
mN J ~5.6b!
In fact, s( sW N ) acts like the Pauli-Lubanski vector W, except for factors, i.e., ^ l N8 u s( sW N )/2 u l N & 5 ^ l N8 u (1/m N )W u l N & .
C. Spectral function
The spectral function is defined as the probability of finding a nucleon of isospin t N with momentum pN and the spectator
(A21) nucleus with an excitation f A21 of mass m A21 ( f A21 ), i.e.,
^ l N l A8 u S ~ pN M t N ! u l N8 l A & 5A (
l A21 f A21
d @ M 1m A 2m A21 ~ f A21 !# ^ PA l A8 u pN l N8 t N ~ PA 2pN ! l A21 f A21 &
3 ^ pN l N t N ~ PA 2pN ! l A21 f A21 u PA l A & . ~5.7!
The spectral function is an operator in nucleonic and in nuclear spin space. Reference @18# defines it relativistically for the
spin-averaged case; the extension to spin dependence is minor. The spectral function is calculated here in the nuclear c.m.
system.
The spectral function is related to the front form momentum density r (pN t N ) according to the sum rule
E dM ^ l N l A8 u S ~ pN M t N ! u l N8 l A & 5 ^ l N l A8 u r ~ pN t N ! u l N8 l A & ~5.8a!
with
^ l N l A8 u r ~ pN t N ! u l N8 l A & 5A (
l A21 f A21
^ PA l A8 u pN l N8 t N ~ PA 2pN ! l A21 f A21 &^ pN l N t N ~ PA 2pN ! l A21 f A21 u PA l A & . ~5.8b!
The front form momentum density will be calculated in Sec. VI.
D. Spin structure functions in deep-inelastic lepton scattering

The results of the previous subsections are now adapted to deep-inelastic lepton scattering.
1. Bjorken limit
We consider the Bjorken limit (Q ) @2Q 0 2 2
@m 2N
under the condition that the Bjorken variable x52AQ 2 /2Q• P A remains
constant. In terms of front form components for the momentum transfer, (Q 2 ) 2 @2Q 2 . In scattering from a many-nucleon
system Q 2 2
N ÞQ , in principle; however, in the Bjorken limit limB j Q N 5Q for all components @15#. In that limit the convolu-
tion formula for the nuclear current tensor of Eq. ~5.2a! becomes
^ n A u W Am n ~ Q, P A ! u n A & 5 (
tN
E d 3 pN
mN
p1
N
Tr@ W Nm ~nt N ! ~ Q,p N ! r ~ pN t N ! r A ~ n A P A !# . ~5.9!
The troublesome problem, that the nuclear current tensor W Am n (Q, P A ) when calculated in PWIA does not respect current
conservation, disappears in that limit, i.e., Q m W Am n (Q, P A )5W Am n (Q, P A )Q n 50. In the Bjorken limit the spin-structure func-
tions scale; experimentally one should see them to become independent from Q 2 . The convolution formula ~5.9! yields that
scaling behavior; the nucleonic structure functions only depend on the nucleonic Bjorken variable x N and the resulting nuclear
structure functions will also turn out to depend on the nuclear Bjorken variable x only; the dependence of the spin-structure
N(t )
functions g i N and g Ai on the momentum transfers Q 2N and Q 2 , respectively, will therefore be scratched from now on. The two
Bjorken variables are related by
1 Q• P A P1A
x N 5x 5x , ~5.10!
A Q•p N Ap 1
N
the momentum fraction p 1 1

N / P A 5x/Ax N will often be needed.
2. Extraction from selected current tensor matrix elements

Extraction scheme I of Sec. III D only needs kinematic components of the nuclear current tensor; it can therefore be
employed consistently in PWIA; it is our favorite extraction scheme in this paper. Using the convolution formula ~5.9! for the
matrix element ~3.14! of the nuclear current tensor, i.e.,
e 1 m e 2 n ^ n A u W Am n ~ Q, P A ! u n A & 5iA (
t
N
E d 3 pN
mN 1
p1
N pN
1
Tr FS N~ t !
n•s ~ sW N ! g 1 N ~ x N ! 22
A2Q 2
Q2
N~ t !
e 1 •s ~ sW N ! g 2 N ~ x N ! D
3 r ~ pN t N ! r A ~ n A P A ! , G ~5.11!
convolution formulas result for g Ai in the form

( E d j f N ~ j t N ! 5A, ~5.14b!
g A1 ~ x ! 5 (
t
E g
Aj 1 S D
d j N~ tN ! x
nW •sW ~ j t n ! u W ,
A j A N N A n A i n̂
t N
E
N
~5.12a!
( d jj f N ~ j t N ! 51. ~5.14c!
E S D
t
d j 1 N~ tN ! x N
g A2 ~ x ! 5 (
t
N
g
Aj Aj 2
nW •sW ~ j t n ! u W
A j A N N A n A i ê 1
,
~5.12b! The front form spin distribution sW N ( j t N n A ) also satisfies a
sum rule, i.e.,
provided the front form spin distribution is defined by
sW N ~ j t N n A ! :5 E S D
d 3 pN d j 2
p1
N
P1
A
Tr@ sW N r ~ pN t N ! r A ~ n A P A !# . E d j nW A •sW N ~ j t N n A ! 5 ^ s N ~ t N ! & . ~5.15!
~5.13!
Thus, it is related to the spin expectation value ^ s N(t N ) & in
The assumption nW A i n̂ is essential for the derivation of g A1 ,
the polarized trinucleon bound state u PA n A & 5 u C B n A & , de-
the assumption nW A i ê 1 is essential for the derivation of g A2 . fined by
The front form spin distribution sW N ( j t N n A ) will be calculated
K U( S D U L
in Sec. VI for the three-nucleon bound state and in Appendix 3
C for the deuteron. It takes on the role which the spin- 1
^ s N ~ t N ! & :5 C B n A 1t N t N ~ i ! nW A • sW N ~ i ! C B n A ,
averaged front form momentum distribution f N ( j t N ), i.e., i51 2
S D
~5.16a!
f N ~ j t N ! :5 E d 3 pN d j 2
p1
N
P1
Tr @ r ~ pN t N ! r A ~ n A P A !# ,
A i.e., to the nucleonic contribution to the nuclear spin, sW N (i)
~5.14a!
and t N (i) being the nucleonic spin and isospin projection
played for the calculation of the spin-independent structure operators. The spin expectation value ^ s N(t N ) & is given by
function F A2 in Ref. @18#. Normalization and momentum con- the S-, S 8 -, and D-state probabilities P(S), P(S 8 ), and
servation implied P(D) of the trinucleon wave function according to
^ s N~ t N !& 5
H 2
3 @ P ~ S 8 ! 2 P ~ D !# ,
P ~ S ! 1 31 @ P ~ S 8 ! 2 P ~ D !# ,
1
t N 51 ~ proton! ,
2
1
t N 52 ~ neutron! .
2
~5.16b!
The sum rule is calculated in the nuclear c.m. system for which n A 5(0,n̂ R ); it does not depend on the direction of the
polarization n A .
3. Extraction from the full current tensor

Extraction scheme II of Sec. III D requires the full nuclear current tensor which cannot be calculated consistently in PWIA.
We only discuss the extraction according to Eqs. ~3.16a! and ~3.16b! for curiosity, since it was used in Ref. @3# for nonrela-
tivistic phenomena. The extraction scheme II yields
gA1 ~x!52
1
A tN( Ed pN p 1 P 2 @ Q 2 1 ~ Q•n
3
mN
N A
Q 2 P 2A
A ! # 2 ~ Q• P A !
2
Q• P A
2 Q•p
N
Tr S HF n A •s ~ sW N ! 2
Q• P A P A •s ~ sW N !
P 2A Q•n A
2 S ~ Q•n A ! 2
Q2
2 2 2
P AQ
D
Q•n A
W N! N t !
~ Q• P A ! 2 Q•s ~ s
G
g 1 N ~ x N ! 1 n A • s̃ ~ sW N ! 2 2
~
PA
F
Q• P A P A • s̃ ~ sW N ! N ~ t N !
Q•n A G
g 2 ~ x N ! r ~ pN t N ! r A ~ n A P A ! , J D
~5.17a!
g A2 ~ x ! 5
1
A (
tN
E d 3 pN
mN
p1
N
~ Q• P A ! 2 Q• P A
P A @ Q 2 1 ~ Q•n A ! 2 # 2 ~ Q• P A ! 2 Q•p N
2
Tr S HF n A •s ~ sW N ! 2
Q 2 1 ~ Q•n A ! 2 P A •s ~ sW N ! Q•s ~ sW N !
Q• P A Q•n A
1
Q•n A
G
3g 1
N~ tN !
F W N!2
~ x N ! 1 n A • s̃ ~ s
Q 2 1 ~ Q•n A ! 2 P A • s̃ ~ sW N ! N ~ t N !
Q• P A Q•n A
G J
g 2 ~ x N ! r ~ pN t N ! r A ~ n A P A ! . D ~5.17b!
The abbreviation
Q•s ~ sW N !
s̃ ~ sW N ! :5s ~ sW N ! 2 p ~5.18!
Q•p N N
is introduced in Eqs. ~5.17a! and ~5.17b!. The choice of polarization for n A and the approach of the Bjorken limit are
noncommuting mathematical procedures. It seems to make sense to take the Bjorken limit only after a definite polarization n A
is assumed. In case of the polarization nW A i n̂, Eq. ~5.17a! yields the convolution formula Eq. ~5.12a! for the spin-structure
function g A1 in the Bjorken limit; in case of the polarization nW A i ê 1 , Eq. ~5.17b! yields the convolution formula Eq. ~5.12b! for
the spin-structure function g A2 in the Bjorken limit. Thus, there appears to be complete consistency between the two employed
extraction schemes. We shall therefore use the convolution formulas ~5.12a! and ~5.12b! in all applications of this paper.
However, that consistency is not general.
Under the assumption nW A i ê 1 an alternative convolution formula for g A1 results from Eq. ~5.17a! in the Bjorken limit; under
the assumption nW A i n̂ an alternative convolution formula for g A2 results from Eq. ~5.17b! in the Bjorken limit, i.e.,
g A1 ~ x ! 5
1
A (E
tN
d 3 pN
mN P1
p1
N
A
p1
N
Tr FH ê 1 • sW N 1n̂• sW N
mN
J
ê 1 •pN' N ~ t N !
g 1 ~ x N ! r ~ pN t N ! r A ~ n A P A ! , G ~5.19a!
g A2 ~ x ! 5
1
A (
tN
S D FS H F S D
E d pN3
mN
p1
N
2
P1
A
p1
N
Tr n̂• sW N 12
mN
p1
N
2
1
p2N'
p1
N mN
G 1
p1
m 2N
W
N pN' • s N
J N~ tN !
g1 ~xN!
F S
1 n̂• sW N 11D G D ~ ê 1 •pN ! 2
m 2N
2
pN' • sW N
p1
N
N~ tN !
g2 ~ x N ! r ~ pN t N ! r A ~ n A P A ! . G ~5.19b!
The alternative convolution formulas Eqs. ~5.19a! and ~5.19b!, are structurely different from those of Eqs. ~5.12a! and ~5.12b!.
They contain terms proportional to pN' which cannot be expressed by the front form spin distribution Ws N ( j t N n A ). Though the
spin-structure functions g A1 and g A2 are Lorentz scalars, independent from the nuclear polarization, the ones approximately
calculated remain dependent. We notice that dependence with amazement, especially, since in the instant form description of
quasielastic scattering no such dependence arises. However, we have checked the quantitative differences of the convolution
formulas ~5.12a! and ~5.19a! for g A1 ; the differences are not resolvable on the plots shown later.
VI. FRONT FORM MOMENTUM DENSITY OF THE TRINUCLEON BOUND STATE AND SPIN DISTRIBUTION
In this section the front form momentum density r (pN t N ) of Eq. ~5.8b! is calculated for the trinucleon bound state, i.e., for
A53. It is calculated as an operator in nucleonic and nuclear spin space. Introducing the spin-dependent operator
^ l i8 u O i ~ l N8 l N ! u l i & :5 d l i8 l N8 d l i l N ~6.1!
for nucleon i by its matrix elements, the front form momentum density r (pN t N ) is related to the nuclear bound state u PA l A &
with spin projection l A , i.e.,
^ l N l A8 u r ~ pN t N ! u l N8 l A & 5 PA l A8 K U( i
U L
d 3 ~ pN 2pi ! O i ~ l N8 l N ! d t N t i PA l A , ~6.2a!
^ l N l A8 u r ~ pN t N ! u l N8 l A & 5
A
P1
A
K US
PA l A8 d
p1
N
1
PA
2
p1
1
P1
A
D U L
d 2 ~ pN' 2p1' ! O 1 ~ l N8 l N ! d t N t 1 PA l A . ~6.2b!
In the operator the front form momentum fraction p 1 1

N / P A is The internal single-nucleon momenta (ki' j i ) are ex-
introduced. tended to the standard three-dimensional form of on mass
The single-nucleon front form momenta pi are related to shell particles
the c.m. momentum PA and to the internal front form mo-
menta ki' and the front form momentum fractions j i by k1
i 5jiM 0 , ~6.4a!
m 2i 1k2i'
k2
PA' 5 (j pj' , ~6.3a! i 5
k1
i
, ~6.4b!
P1
A5 (j p 1j , ~6.3b! F
kW i 5 k 1i ,k 2i ,
1
2
S
j iM 02
m 2i 1k2i'
jiM 0
DG ~6.4c!
with
ki' 5pi' 2
p1
i
P1
A
PA' , ~6.3c! M 05 S( j
m 2j 1k2j'
jj D 1/2
~6.4d!
p1 and k 2i 5m 2i . That three-dimensional form kW i is also con-

i
j i5 . ~6.3d! strained by ( i kW i 50. The internal momenta kW i are therefore
P1
A
related to the Jacobi momenta ( pW qW ), in which the ground
state u C B l A & usually is given, by
The internal single-nucleon momenta (ki' j i ) satisfy the two
constraints ( i ki' 50 and ( i j i 51. For the calculation of the 1
front form momentum density the nuclear c.m. system is pW 5 ~ kW 2 2kW 3 ! , ~6.5a!
2
assumed, i.e., PA' 50 and P 1 A 5m A . The internal part of the
nuclear bound state u PA l A & is a mass eigenstate as in Eq. qW 52kW 1 , ~6.5b!
~2.11!, a proper state of relativistic quantum mechanics;
however, when the interactions are added to the square of the assuming equal masses m i 5m N for the three nucleons. Thus,
free mass operator, the resulting eigenvalue equation be- a complete transformation of the wave function to the inter-
comes @16# formally identical with the eigenvalue equation nal front form momenta (ki' j i ) and the front form spins is
for an internal nonrelativistic Hamiltonian. Thus, the relativ- possible. However, the front form momentum density
istic mass eigenstate can be identified with the nonrelativistic r (pN t N ) samples the front form properties of a single
ground state u C B l A & of internal motion; it was already used nucleon according to Eq. ~6.2b!. Thus, a mixed set of mo-
in Eq. ~5.16a!. The mass eigenstate is independent of any
menta ( pW k1' j 1 ) is preferable in terms of which the Jacobi
choice of the dynamics; it is represented here in terms of
momenta can be expressed, i.e.,
front form variables for momenta and spin. However, that
identification of u P A l A & with u C B l A & has one serious draw-
pW 5 pW , ~6.6a!
back: Rotational invariance is violated, since u C B l A & is an
eigenstate of an angular momentum operator, built up from
the operators of constituents without interactions. That vio-
lation @17# gives rise to the well-known ambiguities when
F
qW 5 2k 11 ,2k 21 ,2
1
2
Sj 1M 02
m 2N 1k21'
j1M 0
DG . ~6.6b!
elastic form factors are extracted from approximately com-

puted current matrix elements. The same ambiguities are In this representation the multicoordinate quantity M 0 be-
likely to reoccur in the present context when extracting struc- comes
ture functions from an approximately computed current ten-
sor. This paper does not explore those ambiguities in full, but
at least it compares two extraction schemes according to
M 05 F 4 ~ m 2N 1 p 2 ! 1k21'
12 j 1
1
m 2N 1k21'
j1
G 1/2
~6.6c!
Secs. III D, V D 2, and V D 3 which weight spin matrix el-

ements differently, but which yield spin-structure functions with the notation p 2 5 pW 2 in the remainder of this section.
numerically almost indistinguishable. In view of that short- Equation ~6.6c! holds only approximately; it neglects angle-
coming, the suggestion of Ref. @19# to construct relativistic dependent terms pW •k1' /m 2N which are, however, small for
mass eigenstates directly is an interesting alternative, but it the physically relevant momenta in a nuclear bound state.
has not yet matured such that it can be employed simulta- The relativistic trinucleon bound state wave function takes
neously in the two- and three-nucleon systems. the following form in that mixed representation:
^ pW k1' j 1 l 1 s 2 s 3 t 1 t 2 t 3 u C B l A & 5 ( ^ l 1 u R 1
s1
M ~ k1' j 1 p ! u s 1 &
2
AU ] ~ pW qW !
] ~ pW k1' j 1 !
U^ pW qW ~ k1' j 1 p 2 ! s 1 s 2 s 3 t 1 t 2 t 3 u C B l A & . ~6.7!
The transformation of momenta requires the Jacobi determinant
U ] ~ pW qW !
] ~ pW k1' j 1 !
5UU
]q3
]j 1
~ k1' j 1 p 2 ! 5 U
M 0 ~ k1' j 1 p 2 !
4 j 1 ~ 12 j 1 !
; ~6.8!
in Eq. ~6.8! the argument list of M 0 is made explicit. Only the spin of nucleon 1 is Melosh rotated. The rotation matrix R M
depends on k 1 , but through M 0 on all internal variables (ki' j i ), therefore also on p 2 ; this is the reason why its argument k 1
is changed to the argument list (k1' j 1 p 2 ), i.e.,
m N 1 j 1 M 0 1i ~ s 1N k 21 2 s 2N k 11 !
R M ~ k1' j 1 p 2 ! 5 , ~6.9!
A~ m N 1 j 1 M 0 ! 2 1k21'
in contrast to the notation in Eq. ~2.15c!. The spins of the nucleons 2 and 3 are still taken to be canonical, since the expectation
value ~6.2b! sums over that spin dependence.
The front form momentum density is calculated according to
A
( (
P1t t
A 2 3 s 81 s 1 s 2 s 3
E W U]]j
d3p
q3
U
~ pN' j p 2 ! ^ l N u R 1 2
8 u R M ~ pN' j p 2 ! u l N8 &
M ~ pN' j p ! u s 1 &^ s 1
3^ C B l A8 u pW qW ~ pN' j p 2 ! s 18 s 2 s 3 t N t 2 t 3 &^ pW qW ~ pN' j p 2 ! s 1 s 2 s 3 t N t 2 t 3 u C B l A & u j 5p 1 / P 1 . ~6.10!

N A
At this stage we go back to the corresponding instant form momentum density r (pW N t N ) which reads, in the nuclear c.m.
system,
K U(
^ s N s A8 u r ~ pW N t N ! u s N8 s A & 5 C B s A8
i
U L
d ~ pW N 2kW i ! O i ~ s N8 s N ! d t N t i C B s A
5A ( (
t t s s
2 3 2 3
E d 3 pW ^ C B s A8 u pW ~ 2 pW N ! s N8 s 2 s 3 t N t 2 t 3 &^ pW ~ 2 pW N ! s N s 2 s 3 t N t 2 t 3 u C B s A &
5
1
2 K U
s N s A8 r 0 ~ p N t N ! 1 r 1 ~ p N t N ! sW N • sW A 1 r 2 ~ p N t N !
pW 2N
S
sW N •pW N sW A •pW N 1
2 sW N • sW A
3 DU L
s N8 s A
5
1
2
E K U
p 2 dp s N s A8 r̄ 0 ~ p N pt N ! 1 r̄ 1 ~ p N pt N ! sW N • sW A 1 r̄ 2 ~ p N pt N ! S sW N •pW N sW A •pW N 1
pW 2N
2 sW N • sW A
3 DU L
s N8 s A .
~6.11!
The first step gives the definition. In the next two steps of Eq. ~6.11! the spin structure of the instant form momentum density
r (pW N t N ) is recalled; in the last step it is realized that that spin structure already holds after integration on p̂, i.e., prior to the
integration on the magnitude p of pW . The functions r̄ 0 , r̄ 1 , and r̄ 2 in the result ~6.11! are numerically available and will be
used for calculating the front form momentum density. Since the calculation is carried out in the nuclear c.m. system, the
canonical and front form spins of the nucleus are identical, thus, u C B l A & 5 u C B s A & and therefore
1 1
2 P1
A
E U p 2 dp
]q3
~p jp2!
]j N'
UK U
l N l A8 R 1 F W W
M ~ pN' j p ! r̄ 0 ~ q pt N ! 1 r̄ 1 ~ q pt N ! s N • s A 1 r̄ 2 ~ qpt N !
2
3 S sW N •qW sW A •qW 1
qW 2
2 sW N • sW A
3 D GU qW 5qW ~ pN' j p 2 !
U LU
R M ~ pN' j p 2 ! l N8 l A
j 5p 1 / P1
N A
. ~6.12!
The result ~6.12! requires the replacement of the Jacobi momentum qW as qW (pN' j p 2 ) by the front form momenta (pN' j )
according to the prescription of Eq. ~6.6b!, given there in terms of (k1' j 1 ).
The determination of the spin-structure functions g A1 and g A2 in deep-inelastic lepton scattering according to Eqs. ~5.19!
requires the full spin-dependent momentum density r (pN t N ); however, the determination according to Eqs. ~5.12! which this
paper prefers needs the front form spin distribution Ws N ( j t N n A ) only, i.e.,
sW N ~ j t N n A ! 5
1
2
E 2
d pN' E U 2
p dp
]q3
~p jp2!
]j N' U( 3
i51
Tr@ sW N R 1
M ~ pN' j p ! s N R M ~ pN' j p !#
2 i 2
F
3 r̄ 1 ~ qpt N ! n iA 1 r̄ 2 ~ p N pt N ! S q i nW A •qW 1 i
qW 2
2 nA
3 D GU qW 5qW ~ pN' j p 2 !
. ~6.13!
The front form spin distribution sW N ( j t N n A ) was already in- It is characterized by the S-, S 8 -, and D-state probabilities
troduced in Eq. ~5.13!. The calculations of this paper are P(S)590.36%, P(S 8 )51.38%, and P(D)58.25%. The ef-
based on it. fective nucleon polarization ^ s N(t N ) & is defined in Eq. ~5.16a!
of Sec. V; it is related to the wave function probabilities
VII. RESULTS AND DISCUSSION according to Eq. ~5.16b!; the employed wave function yields
The spin-structure functions g A1 and g A2 of deep-inelastic the specific values ^ s n & 50.881 and ^ s p & 520.046. The
neutron component is also in the front form spin distribution
lepton scattering are calculated according to Eqs. ~5.12!. We
of overwhelming importance as Fig. 2 shows; this fact holds
use the theoretical apparatus developed for the deuteron in
true for longitudinal and transverse polarizations. In the front
Appendices A–C as well and therefore discuss deep-inelastic
form spin distribution of 3 H the proton and neutron change
lepton scattering from the deuteron and the three-nucleon
bound states simultaneously. We shall abbreviate the label roles. Thus, the technical tools for calculating the 3 H spin-
structure functions are also available, but no such calculation
N(t N ) from now on by p and n for t N 51 21 and t N 52 21 ,
is done in this paper.
differentiating proton and neutron contributions, respec-
We expected that the Melosh rotation from canonical to
tively.
front form spin is a sizable relativistic effect, important for
A. Front form spin distribution the actual magnitude and shape of the front form spin distri-
butions. This expectation did not turn out to be true; leaving
The deuteron and 3 He front form spin distributions the Melosh rotations out, i.e., putting them to unity in Eq.
Ws N ( j t N n A ) are displayed as functions of the momentum frac- ~C7! for the deuteron and in Eq. ~6.13! for 3 He creates
tion j in Figs. 1 and 2. The components nW A •sW N ( j t N n A ) are minute changes of less than 0.1% in the distributions outside
given for longitudinal and transverse polarizations, i.e., for zeros; the changes are invisible in the plots. In both cases,
nW A 5n̂ and nW A 5ê 1 , respectively. They are split into proton deuteron and 3 He, the front form spin distributions are
and neutron contributions. The calculation is based on rela- peaked around j 51/A. The approximation
tivistic mass eigenstates which are obtained @16# by reinter-
preting nonrelativistic bound states. Their wave functions are
derived from the Paris potential @20# for the deuteron and for
the trinucleon bound state.
nW A •sW Napp~ j t N n A ! 5 d j 2 S D
1
A
^ s N~ t N !& ~7.1!
1. Deuteron is a simple one; it is independent from the direction nW A of the

The employed relativitic deuteron wave function is ob- nuclear polarization. When calculating the nuclear spin-
tained from a nonrelativistic one, characterized by the S- and structure functions g Ai later on, it will turn out to also be a
D-state probabilities, i.e., P(S)594.23% and reliable one for both nucleonic distributions in both nuclei. It
P(D)55.77%. The effective nucleon polarization ^ s N(t N ) & is poorest for the 3 He proton distribution in longitudinal po-
larization whose peaking around j 51/A is least pronounced.
is defined in Eq. ~5.16a!; it is related to the wave function
However, the proton contribution to the 3 He spin-structure
probabilities according to Eq. ~C8! of Appendix C; in the
functions will turn out to be small anyhow.
isoscalar deuteron it is the same for proton and neutron; the
employed wave function yields the specific values
^ s p & 5 ^ s n & 50.457. The proton and neutron components of B. Nucleonic spin-structure functions g i
N„tN…
the front form spin distribution sW N ( j t N n A ) are shown in Fig.

The nuclear spin-structure functions g Ai are calculated ac-
1. They are identical. Furthermore, their longitudinal and
cording to the convolution formulas ~5.12!. The convolution
transverse components are indistinguishable on the scale of
formulas require a parametrization of the nucleonic spin-
the plot. N(t )
structure functions g i N . We employ the parametrization of
2. Trinucleon bound state 3 He Ref. @23#; it is displayed in Figs. 3 and 4. The parametriza-
N(t )
The employed relativistic trinucleon wave function is ob- tion of g 1 N in Ref. @23# accounts for the experimental pro-
tained from a nonrelativistic one, computed in Refs. @21,22#. ton data @24# and satisfies the Bjorken sum rule @25#
N(t ) N(t )
FIG. 3. Nucleonic spin-structure functions g 1 N and g 2 N as
FIG. 1. Components nW A •sW N ( j t N n A ) of the deuteron front form
functions of the nucleonic Bjorken variable x N in the scaling limit.
spin distribution as a function of the momentum fraction j . Neutron
The neutron spin-structure functions are shown as dotted curves, the
and proton components are identical. The longitudinal and trans-
proton ones as dashed curves, all in the parametrization of Ref.
verse contributions, i.e., the contributions for nW A 5n̂ and nW A 5ê 1 , N(t )
@23#, g 2 N according to the twist-2 prescription of Ref. @26#. The
respectively, are indistinguishable on the chosen scale.
data for the proton spin-structure function g 1p are taken from Ref.
@24#; only the statistical errors are shown. Older SLAC data for
E dx N @ g 1p ~ x N ! 2g n1 ~ x N !# 5
1 gA
6 gV
, ~7.2!
g 1p referred to in Ref. @24# are not plotted.
N(t )
does not hold for the parametrization of g 1 N according to
g A and g V being the axial and vector coupling constants of Ref. @27#, which seems, however, to be invalidated by the
N(t )
the nucleon. The parametrization of g 2 N is based on the existing proton data @24#.
twist-2 prescription of Ref. @26# which relates both nucleonic
spin-structure functions by C. Nuclear spin-structure functions g Ai
The deuteron and 3 He spin-structure functions g A1 and
g2
N~ tN !
~ x N ! 52g 1
N~ tN !
~ xN!1 E
xN
1 dx N8
x N8
N~ tN !
g1 ~ x N8 ! . ~7.3!
g A2are calculated using the convolution formulas ~5.12!. The
predictions are displayed in Figs. 5 and 6, respectively. They
are broken up into proton and neutron contributions. The
neutron contribution dominates the 3 He spin-structure func-
The first data @14# for g 2p , referring to the momentum trans- tions; the proton contribution is insignificant for 3 He. In con-
fer range 1 GeV 2 <2Q 2 <10 GeV 2 , are consistent with that trast, the proton and neutron make contributions of compa-
prescription, though the experimental precision is still unsat- rable magnitude, but of opposite sign to the deuteron spin-
isfactorily poor. structure functions. Both results are due to the fact that
N(t )
The proton and neutron spin-structure functions g 1 N N(t )
proton and neutron spin structure functions g i N are of
N(t N )
and g 2 are of comparable magnitude, though of different comparable size. Experimental data on the deuteron and
sign, in the parametrization of Figs. 3 and 4. This character- 3
He spin-structure function g A1 and on the deuteron spin-
istic is decisive for the successful extraction of the neutron structure function g A2 have recently become available, i.e.,
spin-structure functions from 3 He data. This characteristic Refs. @12,13,10,11,14#, respectively.
FIG. 2. Component nW A •sW N ( j t N n A ) of the 3 He front form spin N(t )

FIG. 4. Weighted nucleonic spin-structure functions x N g 2 N as
distribution as function of the momentum fraction j . Neutron and functions of the nucleonic Bjorken variable x N in the scaling limit.
proton components are shown for longitudinal and transverse polar- The neutron spin-structure function is shown as dotted curve, the
izations, i.e., for nW A 5n̂ and nW A 5ê 1 , respectively. The neutron com- proton one as dashed curve, both in the parametrization of Ref. @23#
ponents are given by the dotted curve; on the chosen scale of the according to the twist-2 prescription of Ref. @26#. The data for the
plot, they coincide for both cases of polarization. The longitudinal proton spin-structure function g 2p are taken from Ref. @14#; they
~transverse! proton component is displayed as short-dashed ~long- refer to two distinct scattering angles of the lepton, Q e 54.5° and
dashed! curve; the small proton components are also shown in an Q e 57.0°, indicated by solid dots and solid rectangles, respectively;
enlarged scale on the right side of the figure. only the statistical errors are shown.
FIG. 5. Deuteron spin-structure functions g A1 and g A2 as func- FIG. 7. Weighted deuteron spin-structure functions xg A1 and
tions of the Bjorken variable x. They are derived from the convo- xg A2 as functions of the Bjorken variable x. The theoretical predic-
lution formulas ~5.12! using the parametrizations of Fig. 3 for the tions are derived from the convolution formulas ~5.12! using the
nucleonic structure functions. The full results are shown as solid parametrization of Fig. 3 for the nucleonic structure functions; they
curves; the neutron and proton contributions are given by the dotted are compared to the experimental data for xg A1 of @12#, indicated by
and dashed curves, respectively. small solid dots, and of @13#, indicated by solid rectangles, and to
the experimental data for xg A2 of @14#; the data for xg A2 refer to two
distinct scattering angles of the lepton Q e 54.5° and Q e 57.0°,
The deuteron data @12–14# refer to a momentum transfer indicated by solid dots and rectangles, respectively; only the statis-
range 1 GeV 2 <2Q 2 <15 GeV 2 ; we assume them to scale tical errors are shown.
and therefore consider the description in scaled approxima-
tion developed in this paper as appropriate. Figure 7 com- roughly account for trends in the existing experimental data
pares the data with the predictions of Fig. 5. The experimen- @11#. Figure 8 compares the theoretical prediction with data.
tal error bars for the quantity xg A1 in Ref. @13# have become
quite small indicating that the theoretical spin-structure func- D. Unfolding the neutron spin-structure functions
tion changes sign at too large a value for the scaling variable from the nuclear ones
x; furthermore its peak is also shifted to too large x values. The front form spin distributions Ws N ( j t N n A ) are approxi-
Thus, as a consequence, also the assumed parametrization of mated in Eq. ~7.1! by a d -function form. This approximation
the neutron spin-structure function g n1 appears to be slightly yields approximate nuclear spin-structure functions
inconsistent with the data. In contrast, due to sizable error
bars the first data @14# for xg A2 cannot be considered a strin-
N(t )
gent test of the assumed g 2 N .
g Aiapp~ x ! 5 (
tN
N~ t
gi N!
~ x ! ^ s N~ t N !& ~7.4!
The He data @10,11# refer to an average momentum

3
in turn, i.e., very simple relations between nuclear and nucle-
transfer of 2Q 2 52 GeV 2 , scaling is not evident; the de-
onic spin-structure functions. Approximation ~7.4! accounts
scription in scaled approximation developed in this paper is
for the full calculation quite reliably. For the deuteron and
not quite appropriate. Outside the domain of Bjorken scaling
for 3 He the differences in g A1 and g A2 between the approxi-
the general convolution formulas for the nuclear spin-
structure functions have to be based on the front form spec- mated forms ~7.4! and the full convolution formulas ~5.12!
tral function and not on the density which the particular re- are not distinguishable in plots and are therefore not
lations ~5.12! use; in that case Q N is also not equal to Q. The displayed—with one rather inconsequential exception: As
front form spectral function is not calculated in this paper. expected from the discussion of the 3 He proton distribution,
Nevertheless, the description assuming scaling is able to the approximated proton contributions to g Ai show some mi-
nor deviations; they are given in Fig. 9. The transverse com-
ponent of the proton spin distribution in 3 He is compara-
FIG. 6. 3 He spin-structure functions g A1 and g A2 as functions of

the Bjorken variable x. They are derived from the convolution for- FIG. 8. 3 He spin-structure function g A1 as function of the
mulas ~5.12! using the parametrizations of Fig. 3 for the nucleonic Bjorken variable x. The theoretical prediction is derived from the
structure functions. The full results are shown as solid curves; the convolution formula ~5.12a! using the parametrization of Fig. 3 for
neutron and proton contributions are given by the dotted and dashed the nucleonic structure functions; it is compared to the experimental
curves, respectively. data of Ref. @11#; for the data only the statistical errors are shown.
FIG. 9. 3 He spin-structure functions g A1 and g A2 as functions of FIG. 10. Weighted neutron spin-structure function x N g n1 as a
the Bjorken variable x. The validity of the approximation ~7.4! is function of the Bjorken variable x N . The neutron spin-structure
checked. Only the small proton contributions are shown. Compared function assumed in Fig. 3 is compared to the ones extracted from
with Fig. 6 the resolution of the plots is increased. The proton the 3 He data @10#, indicated by big solid dots, and from the deu-
contributions are ~i! derived in a proper calculation according to the teron data @13#, indicated by solid dots, according to Eq. ~7.5!. The
convolution formulae of Eqs. ~5.12! and ~ii! estimated according to model dependence of the extraction due to the nonuniqueness of the
the approximation ~7.4!; the dashed curves refer to the proper cal- deuteron wave function is not folded into the experimental error; it
culation, the dot-dashed ones to the approximation. The correspond- is small compared with the existing errors in the experimental data.
ing neutron contributions and the full spin-structure functions are
volution formula ~7.4!. When starting out from the param-
not shown, since both results are indistinguishable in plots with the
etrization of the nucleonic spin-structure functions in Fig. 3,
chosen resolution.
calculating the nuclear ones without approximation accord-
ing to Eq. ~5.12! and then recovering g ni according to the
tively better approximated by Eq. ~7.1! than the longitudinal approximate relation ~7.5!, differences arise which are not
one as the different pronounciation of peaks around j 51/A resolvable in plots and are therefore not displayed. This sys-
in Fig. 2 shows. According to Eqs. ~5.12! the spin-structure tematic error therefore appears to be minor.
function g A1 depends on the longitudinal and the spin- ~2! The experimental errors of g Ai and g ip , i.e., Dg Ai and
structure function g A2 on the transverse component of the Dg ip , pollute the extracted neutron spin-structure functions.
front form spin distribution sW N ( j t N n A ). The differences be- The errors on Dg Ai in the existing experiments are still large.
tween the approximated form ~7.4! and the full convolution For the deuteron the error in the existing proton data contrib-
formula ~5.12b! for g A2 are therefore smaller than the corre- utes to the error in the extraction quite significantly, since
sponding differences for g A1 . This fact is bourne out in Fig. 9 Dg ip u ^ s p & / ^ s n & u 5Dg ip ; in contrast the same error has a very
for the proton contributions. We now turn to unfolding the small weight for the extraction from 3 He data, since
neutron spin-structure functions from the nuclear ones. Dg ip u ^ s p & / ^ s n & u .0.05Dg ip .
First, we assume that the convolution formulas ~5.12! are ~3! The nucleonic contributions ^ s N(t N ) & in the extraction
exact. However, we would like to use its approximated form ~7.5! are model dependent, though they lie within rather
~7.4! in order to extract the neutron spin-structure functions small limits for realistic bound state wave functions, i.e., for
g ni from the experimental data the deuteron 0.45< ^ s p,n & <0.47 and for 3 He @1#
U
0.85< ^ s n & <0.90 and 20.06< ^ s p & <20.04. The resulting
1
g ni ~ x N ! 5 @ g Ai ~ x ! 2g 1p ~ x ! ^ s p & # . ~7.5! spread of theoretical results is displayed in Fig. 11 for 3 He.
^ s n& x5x N Given the errors in the present experimental data, the model
dependence is not significant yet.
An example is given in Fig. 10. Even if the convolution
formula ~5.12a! were exact and were well described by Eq.
~7.4!, the extraction ~7.5! of the neutron spin-structure func-
tions meets three distinct types of errors, i.e.,
S
Dg ni ~ x N ! 5 Dg Ai ~ x N !
1
^ s n&
1Dg ip ~ x N ! U UD
^ s p&
^ s n&
1D ^ s p & U g ip ~ x N !
^ s n&
U U
1D ^ s n & ~ g Ai ~ x N !
2g ip ~ x N ! ^ s p & !
^ s n& 2
1
U 1D sysg ni ~ x N ! , ~7.6!
FIG. 11. Neutron spin-structure function g n1 as a function of the
Bjorken variable x N . The neutron spin-structure function assumed
according to Fig. 3 is shown as the middle curve and compared with
the ones extracted from the calculated 3 He spin-structure function
which we add up linearly for convenience. g A1 according to Eq. ~7.5!. The upper and lower curves give the
~1! There is a systematic error D sysg ni (x N ) due to the fact bandwidth of the theoretical uncertainty due to the model depen-
that the extraction ~7.5! is derived from the approximate con- dence @1# of the trinucleon bound state wave function.
Second, the convolution formulas ~5.12! may be concep-

tually wrong. There may be a multitude of effects, e.g., a
medium dependence of the nucleonic structure functions, a
F A2 binding~ x ! 5 (
tN
E djF2
N~ tN !
S D
x
f ~jt !
A j Nbinding N
~7.8a!
binding correction, or other contributing non-nucleonic
nuclear constituents, which invalidate the convolution for-
with the spin-averaged front form momentum distribution
mulas ~5.12! and which also contribute to D sysg ni (x N ). Those
S D
effects were discussed in the context of the spin-independent
structure function F A2 @18#. In the next subsection we worry
about corresponding consequences for the spin-structure
f Nbinding~ j t N ! :5 E d 3 pN d j 2
p1
N 1e
P1
A
functions g Ai and their bearing on extracting neutron spin-
structure functions. Anticipating that also those corrections 3Tr@ r ~ pN t N ! r A ~ n A P A !# , ~7.8b!
will turn out to be small, we conclude:
The approximate relation ~7.5! can reliably be used to into which binding is built in, the EMC effect in the EMC
extract the neutron spin-stucture functions g ni of the neutron ratio R EMC can by and large be accounted for. Assuming the
from deuteron and 3 He data. Figure 10 uses the 3 He data of same momentum shift in the nucleonic plus momentum for
Ref. @10# and the deuteron data of Ref. @13#. When the data the present discussion of the 3 He spin-structure function
from all experiments, presently being performed or being in g A1 , the front form spin distribution sW N ( j t N n A ) gets the cor-
the planning stage, will be in, the parametrization of the respondingly changed modification and becomes
neutron spin structure functions g ni , shown in Fig. 3, may
S D
have to be revised.
E. Is there an EMC effect

Ws Nbinding~ j t N n A ! :5 E d 3 pN d j 2
p1
N 1e
P1
A
in the 3 He spin structure function g A1 ?

3Tr@ sW N r ~ pN t N ! r A ~ n A P A !# ,
In the case of unpolarized deep-inelastic lepton scattering
~7.9a!
a deviation between the nucleonic structure function of a free
N(t )
nucleon, i.e., F 2 N , and of a bound nucleon, i.e., F A2 /A, is
indeed observed and called the EMC effect. There is a diver-
sity of theoretical mechanisms @28# able to account for that
observation. For example, the decrease of the EMC ratio
sW Nbinding~ j t N n A ! 5sW N FS j2
e
D G
t n .
mA N A
~7.9b!
The shift in the nucleonic plus momentum is only introduced

F A2 ~ x ! in the d function which contains the front form momentum
R EMC~ x ! :5 ~7.7! fraction p 1 1
2F 2p ~ x ! 1F n2 ~ x ! N / PA . The front form spin distribution
Ws Nbinding( j t N n A ) still satisfies the sum rule ~5.15! almost un-
changed: Compared with the original definition ~5.13! its de-
at intermediate values of the Bjorken variable x, defined here pendence on j is shifted, but since its nonvanishing values
for 3 He, is often attributed to a medium correction of the are centered around 1/A according to Figs. 1 and 2, the inte-
nucleonic structure function, which may be realized in a the- gration of nW A •sW Nbinding( j t N n A ) over j from 0 to 1 still col-
oretical description either by modifying the nucleonic front lects the complete domain in which the front form spin dis-
form momentum distribution due to nuclear binding or by an tribution is nonzero. The changed front form spin
additional meson contribution to the nuclear structure func- distribution in Eq. ~5.12a! yields the modified 3 He spin-
tion, the pion being the most important non-nucleonic structure function g A1binding according to
nuclear constituent; Ref. @18#, which this paper extends, used
S D
the latter mechanism. Both mechanisms yield comparable
effects for F A2 at intermediate x; they affect, however, the
spin-structure function g A1 quite differently: Whereas a bind-
g A1binding~ x ! 5 (
t
N
E d j N~ tN ! x
g
Aj 1
nW •sW ~ j t n !u W .
A j A Nbinding N A n A i n̂
ing modification of the spin-averaged momentum distribu- ~7.10!
tion will also yield a corresponding modification of the front
form spin distribution sW N ( j t N n A ) and will directly change When using that modified 3 He spin-structure function
the predictions for the nuclear spin-structure function g A1 , we g A1 binding—as a theoretical model for experimental data — in
do not see any simple mechanism for the spinless pions to the extraction procedure ~7.5! for the neutron spin-structure
contribute to g A1 . function g n1 , no significantly increased deviation occurs be-
In one recipe attempting to account for nuclear binding tween the neutron spin-structure function extracted and the
@29#, the nucleonic plus momentum p 1 1
N is shifted to p N 1 e
one assumed for calculating g A1 binding according to Eq. ~7.10!.
due to binding, e standing for an averaged binding energy, In fact, compared with the errors Dg n1 discussed in the last
often taken to be about 230 MeV. Indeed, such a simple subsection the deviation is increased by less than 1%; the
shift yields the decrease in the ordinary EMC ratio deviation would not be discernable in a plot with the resolu-
R EMC(x) at intermediate Bjorken variables x. Using the con- tion of Fig. 10. Thus, a possible binding correction in the
volution formula 3
He spin-structure function g A1 , as introduced in this subsec-
ergy transfers to the struck nucleon Q N and to the nucleus Q

is relatively small for deep-inelastic scattering, i.e.,
D:5Q 0 2Q 0N , u D u !Q 0 . Nevertheless, its effect on Q 2N
Q 2N 5Q 2 22Q 0 D1D 2 ~7.12!
can be sizable. One finds for the ratio of the scaling variables
x and x N
x Q2 2 p N •Q N
5 A , ~7.13a!
FIG. 12. EMC ratios S EMC(x) and R EMC(x). The spin ratio x N 2 P A •Q Q 2N
S EMC(x) of Eq. ~7.11! is given as solid curve and compared with the
ordinary EMC ratio R EMC(x) of Eq. ~7.7! given as a dashed curve. x 1 Ap N •Q
In the right figure the nuclear structure functions are used in the . , ~7.13b!
x N 11D/m N x P A •Q
A
forms g 1binding A
(x) and F 2binding (x) according to Eqs. ~7.10! and
~7.8a!. The left figure shows the same ratios but without the inclu- with
sion of binding effects; there, the nuclear structure functions are
used in the forms g A1 (x) and F A2 (x) according to Eqs. ~5.12a! and p N •Q N . p N •Q. ~7.13c!
~7.8a! with e 50.
The ratio x/x N is only at moderate values of x close to
tion, does not invalidate the extraction of the neutron spin- Ap 1 1
N / P A for all relevant nucleon momenta p N at which the
structure function according to Eq. ~7.5!. However, an EMC spectral function gives significant contributions. Comparing
effect can arise in the ratio the instant form and the front form convolution formulas we
therefore have to expect conflicting results for small values
g A1 ~ x ! of the Bjorken variable x.
S EMC~ x ! :5 , ~7.11! The noncovariant result for g A1 (x,Q 2 )5(Q• P A /
g 1p ~ x ! ^ s p & 1g n1 ~ x ! ^ s n & Am A )G A1 (Q 2 , P A •Q/m A ) is obtained from Eq. ~3.11c! of
Ref. @3# as function of x and Q 2 . The only modification with
S EMC being a much more sensitive quantity. This fact is respect to the quasielastic results is the replacement of the
documented in Fig. 12 which shows this ratio with and with- nucleonic structure functions by the corresponding deep-
out the inclusion of a binding correction according to Eq. inelastic ones according to
~7.10! in the calculation of the 3 He spin-structure function
g A1 . The ratio S EMC exhibits the same decrease at intermedi- N~ tN ! m N N~ tN !
ate Bjorken variables x as the ordinary EMC ratio R EMC of G1 ~ Q 2N ,p N •Q N /m N ! 5lim g ~ x N ,Q 2N ! ,
Bj p N •Q N 1
Eq. ~7.7! does. The trend is clearly observable in S EMC ,
~7.14a!
though its smooth dependence on the Bjorken variable x is
disturbed by the zero in the denominator of the ratio in defi-
m 3N
nition ~7.11!; the definition may therefore be considered un- G2
N~ tN !
~ Q 2N ,p N •Q N /m N ! 5lim
N~ t !
g N ~ x N ,Q 2N ! .
2 2
fortunate. B j ~ p N •Q N !
~7.14b!
F. Relation to instant form dynamics N(t )
For practical calculations the parametrizations of g i N
Reference @3# describes polarized inelastic lepton scatter- given in Sec. VII B are used; they do not carry a dependence
ing from 3 He noncovariantly in PWIA. It assumes instant N(t )
on Q 2N ; in contrast, the nucleonic structure functions G i N
form dynamics. Though the formalism is employed there for
a description of quasielastic scattering, it is general and may will remain dependent on Q 2N .
formally be applied to the deep-inelastic regime of the The solid curve of Fig. 13 shows the spin-structure func-
Bjorken limit as well. In this subsection we explore a de- tion g A1 (x,Q 2 ) derived from Eq. ~3.11c! of Ref. @3#. The
scription of deep-inelastic scattering based on Eqs. ~3.11a!– four-momentum transfer has been chosen to be 2Q 2 510
~3.11d! of Ref. @3#. However, the exploration is only a matter GeV 2 . However, no significant change of the results is seen
of curiosity: The boosts of instant form dynamics are inter- for any value larger than 2Q 2 >1 GeV 2 . Thus, theoreti-
action dependent for a composite system. The assumption of cally, scaling is observed for 2Q 2 @1 GeV 2 provided it is
PWIA that the nuclear current is built from single-nucleon assumed for nucleons. This fact may be taken as one justifi-
contributions cannot be made consistently. Thus, front form cation for assuming scaling in the description of the data of
dynamics remains the superior form of description for deep- Refs. @10,11#, collected for an average momentum transfer
inelastic scattering. Q 2 52 GeV 2 . However, Fig. 13 also proves that the concep-
The description of Ref. @3# in terms of instant form dy- tually required difference between the momentum transfer Q
namics made the same conceptual distinction between the to the nucleus and Q N to the nucleon matters. The unjustified
momentum transfer to the nucleus Q and to a nucleon Q N , recipe Q N 5Q makes the results very close to the prediction
to a nucleon bound in the nucleus though assumed to be on based on front form dynamics. References @30–34# employ
mass shell, as this paper does, i.e., Q N ÞQ due to their dif- this recipe Q N 5Q. We claim that the approximated form of
ference in energy transfer. The difference between the en- Eq. ~7.4! for g Ai , used for the interpretation of ex-
N(t )
g 2 N , respectively. In deep-inelastic scattering the simple
approximation ~7.4! for the convolution formulas is moti-
vated and proven to be reliable. References @1,32# already
related the nuclear spin-structure functions to the nucleonic
spin-structure functions weighted by integral properties of
the nuclear wave function. But they were unable to justify
that relation rigorously. The present approach gives the miss-
ing proper justification. This finding is important for the in-
terpretation of the experimental data with respect to the
analysis of nuclear effects.
FIG. 13. 3 He spin-structure function g A1 as function of the ACKNOWLEDGMENTS

Bjorken variable x. The result derived from the convolution for-
mula ~3.11c! of Ref. @3# using a noncovariant approach under the The work was started together with F. Coester who ac-
assumption of instant form dynamics is given as a dashed curve. In companied the authors for a long time with his expertise and
a second calculation based on the same formalism we have explic- encouragement; the authors are indebted to him; the authors
itly set Q N 5Q; that approximation cannot be justified theoretically owe their slowly acquired insight into the intricacies of front
in instant form dynamics; the result is presented as dotted curve. form dynamics entirely to him, their still limited understand-
For comparison the full calculation of Eq. ~5.12a! is shown as a ing of it is their own fault. The authors are grateful to R.
solid curve. Milner for discussions on the HERMES project. The work
was supported in part by the Deutsche Forschungsgemein-
perimental data, is well supported by front form dynamics; in schaft ~DFG! under Contract No. Sa 247/11-2.
contrast, its derivation based on instant form dynamics is
only poorly justified. APPENDIX A: DEUTERON DENSITY OPERATOR
The full density operator r A (n A ) commutes with the mo-

VIII. CONCLUSION mentum operator P. Its projection r A (n A P A ), i.e.,
The convolution formulas which relate the structure func-
tions for deep-inelastic scattering to the properties of con- ^ PA l A8 u r A ~ n A ! u PA l A & 5 ^ l A8 u r A ~ n A P A ! u l A & , ~A1!
stituent particles of the target nucleus depend on the assump-
tion of one-body current operators and on the covariance of onto the Hilbert sector of momentum eigenstates with eigen-
the current tensor of the bound constituents. Consistency value PA is needed in Eq. ~3.2!. The density operator of a
problems and ambiguities are due to the fact that one-body spin-1 target is parametrized in the form
currents cannot be covariant under all Lorentz transforma-
F G
tions in a system of bound particles.
1 3 3
This paper presents a detailed derivation of nuclear spin- r A~ n A P A ! 5 12 n A •W1 2 t A m n T m n ~ W !
dependent structure functions for deep-inelastic lepton scat- 3 2m A mA
tering and applies them to the two- and three-nucleon bound ~A2!
states. Front form dynamics is used for which the covariance
under the kinematic subgroup is sufficient to derive convo- with the Lorentz tensor T m n (W) defined by
lution formulas: The restriction to one-nucleon currents is
possible without inconsistencies. This is why we like the
description in terms of front form dynamics. The previous
calculation of Ref. @30# is closest in spirit to this paper. In
T m n ~ W ! :5
W m W n 1W n W m 1 P Am P An
2
1
3 P 2A S
2g m n W 2 . D
contrast the noncovariant approach using instant form dy- ~A3!
namics as discussed in Refs. @3,31–34# cannot be consis-
tently extended to the Bjorken limit in this sense. Our inves- The density operator is at most quadratic in the Pauli-
tigation of relativistic effects in the spin-dependent structure Lubanski vector W. The vector n A and the tensor t A describe
functions for the deep-inelastic regime is novel; it extends the polarization of the target. The parametrization ~A2! sat-
the work of Ref. @18# to polarization. The numerical results isfies the required constraints. The Pauli-Lubanski vector W
show that relativistic effects due to the proper treatment of and the tensor operator T m n (W) are traceless, i.e., Tr@ W # 50
spin are small and negligible compared to larger effects aris- and Tr@ T m n (W) # 50. In general, the polarization vector n A
ing from the conceptual difference between front form and and the polarization tensor t A are independent. However,
instant form predictions and arising from the model depen- when considering a quantum mechanically pure state as we
dence of the ground state wave functions. These findings are do, then n A is the direction according to which its polariza-
equally valid for the two- and three-nucleon systems. tion is defined; in this case the polarization vector n A and the
Front form dynamics allows the formulation of convolu- polarization tensor t A are connected by
tion formulas with a comparatively simple structure, i.e., the
nuclear spin-structure functions g A1 and g A2 are separately de-
termined by the convolution of the front form spin distribu-
N(t )
tion with the nucleonic spin-structure functions g 1 N and
t Am n 5 F
3 m n 1 P Am P An
n n 2
2 A A 3 P 2A S2g m n DG , ~A4!
besides Eq. ~A4! the general constraints n 2A 521 and

n A • P A 50 then hold for the polarization vector n A .
Matrix elements of the density operator of any momentum
t Am n 5L f S D S D
PA
mA
m
i
Lf
PA
mA
n
j
t iRj ~A6d!
P A are related to those of the rest frame, i.e.,

are used. In the form ~A5! for the target rest frame it is
K UF
^ l A8 u r A ~ n A P A ! u l A & 5 l A8
1
3
3
11 n̂ R •SW A
2
obvious that the density operator has only nine independent
elements as appropriate for a spin-1 particle, one arising
from the identity operator, three from the vector operator SW A ,
13 (i j t Ri j T iAj ~ S A ! GU L
l A . ~A5!
and five from the tensor operator T A (S A ). The Cartesian rest
frame tensor T iAj (S A ),
For Eq. ~A5! the identities S iA S Aj 1S Aj S iA 1

T iAj ~ S A ! :5 2 d i j SW 2A , ~A7!
A F S DG
2 3
m
W mA PA
uP l &5 U Lf is symmetric, i.e., T iAj (S A )5T Aji (S A ), and with respect to the
mA A A P1
A
mA
tensor lables traceless, i.e., ( i T iiA (S A )50; thus, the tensor
FS D G
3 Lf
PA
mA
m
r
~ 0,SW A ! r u PR l A & , ~A6a!
part of the density operator can equivalently be written as a
five-component tensor T [2] A (S A ) of second rank—in standard
relation to its Cartesian representation ~A7!. The constraints
S D
n Am 5L f
PA
mA
m
r
~ 0,n̂ R ! r , ~A6b!
on the density operator r A (n A P A ), from which, for example,
condition ~A4! results, are best proven in the target rest
frame, and their validity then carries over to any system by
A F S DG
Lorentz covariance.
T mn~ W ! mA PA
u PA l A & 5 U Lf
m 2A P1
A
mA APPENDIX B: DEUTERON CURRENT TENSOR
FS D S D G
3 Lf
PA
mA
m
i
Lf
PA
m A
n
j
T iAj ~ S A ! u PR l A & ,
The current tensor W Am n (Q, P A ) is Hermitian and con-
served, it preserves parity and time-reversal invariance. Its
Lorentz structure is built from the three four-vectors Q, P A ,
~A6c! and W. The current tensor has the general form
W Am n ~ Q, P A ! 5 F Q mQ n
Q2
2g m n GS F A1 ~ x,Q 2 ! 1
Q•T ~ W ! •Q
Q 2 m 2A
b A1 ~ x,Q 2 ! D A
1
P̃ Am P̃ An
m A Q• P A SF A2 ~ x,Q 2 ! 1
Q•T ~ W ! •Q A
Q 2m A2
b 2 ~ x,Q 2
!
1
mA D
1 1 1 e m n ab
1G m n ~ W ! b A3 ~ x,Q 2 ! 1H m n ~ W ! b A4 ~ x,Q 2 ! 1i
mA mA j Q• P A
F
3Q a W b g A1 ~ x,Q 2 ! 1 W b 2 S Q•W
P D
g A ~ x,Q 2 !
Q• P A A b 2
A
mA G ~B1!
with were permissible. Furthermore, the structure functions which

augment the basic tensors depend on the Lorentz scalars
Q• P A Q 2 ,Q• P A and (Q•W) 2 , where the dependence on (Q•W) 2
P̃ Am :5 P Am 2 Q m, ~B2a! is at most linear and could therefore explicitly be split off.
Q2 This strategy would result in a perfectly admissible and ab-
solutely general parametrization of the current tensor. How-
AQ 2 ever, those additional terms, specific for a spin-1 target, are
x:52 , ~B2b!
2Q• P A not traceless and therefore would yield a nonstandard param-
etrization upon spin averaging. In contrast, the modified,
x being the Bjorken scaling variable. The dependence of the though clumsier basic tensor forms in Eq. ~B1!:
current tensor operator on the Pauli-Lubanski vector W is at
F G F G
most quadratic for a spin-1 target, j51 being the spin quan-
tum number. Besides the standard basic tensor forms used Q mQ m8 Q n8Q n
G mn~ W ! 5 g mm82 T m 8 n 8~ W ! g n 8 n 2 ,
for a spin- 21 case the two additional ones @ W̃ m W̃ n 1W̃ n W̃ m # /2 Q2 Q2
and @ W̃ m P̃ An 1W̃ n P̃ Am # (Q•W)/2 with W̃:5W2(Q•W/Q 2 )Q ~B3a!
F
H mn~ W ! 5 g mm82
Q mQ m8
Q2
G @ P A Q a T a n 8~ W !
space. As in Eq. ~6.2b! of Sec. VI it is related to the nuclear
bound state u PA l A & with spin projection l A by
1T m 8 a ~ W ! Q a P A #
1
F g n8n2
Q n8Q n
G ^ l N l A8 u r ~ pN t N ! u l N8 l A &
m 2A
K US D
2
Q p1 p1
A N 1
~B3b! 5 PA l A8 d 2 d 2 ~ pN' 2p1' !
P1
A P1
A P1
A
and the scalar Q•T(W)•Q5Q m T m n (W)Q n are traceless
with respect to spin summation. Thus, the structure functions
proportional to the Pauli-Lubanski vector W, i.e., g Ai , and the
U L
3O 1 ~ l N8 l N ! d t N t 1 PA l A . ~C1!
structure functions proportional to the basis tensors G m n (W)

and H m n (W) and to the scalar Q•T(W)•Q, i.e., b Ai , cannot
The nuclear c.m. system is assumed for the calculation.
contribute in the spin-averaged case. When spin averaging,
The internal part of the nuclear bound state u PA l A & is a mass
the current tensor ~B1! reduces to the well-known form with
eigenstate and it can therefore be identified with the nonrel-
the two structure functions F A1 and F A2 .
ativistic ground state u C B l A & of internal motion. It is calcu-
The current tensor ^ n A u W Am n (Q, P A ) u n A & follows from the
lated in terms of the three-dimensional relative momentum pW
current tensor operator W Am n (Q, P A ) of Eq. ~B1! and from the
and in terms of canonical spins, but has now to be trans-
density operator r A (n A P A ) of Eq. ~A2! according to Eq.
formed to the appropriate front form variables.
~3.2!. In fact, the current tensor ^ n A u W Am n (Q, P A ) u n A & is ob- Internal single-nucleon front form momenta (ki' j i ) are
tained from the operator ~B1! by replacing the Pauli- introduced as in Sec. VI and extended to three-dimensional
Lubanski vector W and the tensor operator T m n (W) accord- form there; the momenta are constrained, (k1' j 1 ) are the
ing to the relations
independent ones. The relative momentum pW is expressed by
n Am 5 ~ 1/ jm A ! Tr@ W m r A ~ n A , P A !# , ~B4a! them, i.e.,
t Am n 5 ~ 1/m 2A ! Tr@ T m n ~ W ! r A ~ n A P A !# . ~B4b! 1

pW 5 ~ kW 1 2kW 2 ! , ~C2a!
We note that Eq. ~B4a! differs from the corresponding Eq. 2
~3.10! for a spin- 21 target, since the quantum number j is
different; this difference is obvious, since polarization is de-
fined with respect to the maximum angular momentum pro-
jection j; despite that difference the current tensors in Eqs.
F
pW 5 k 11 ,k 21 ,
1
S
j M 2
2 1 0
m 2N 1k21'
j1M 0
DG , ~C2b!
~3.8! and ~B1! are defined such that their dependence on the
polarization vector and on the spin-structure functions g Ai is with the abbreviation
the same for spin-1 and spin- 21 targets.
APPENDIX C: FRONT FORM MOMENTUM DENSITY

OF THE DEUTERON AND SPIN DISTRIBUTION
M 05 F m 2N 1k21'
j 1 ~ 12 j 1 !
G 1/2
. ~C2c!
In this section the front form momentum density r (pN t N )

of Eq. ~5.8b! is calculated for the deuteron, i.e., for A52. It The two-nucleon bound state wave function takes the follow-
is calculated as an operator in nucleonic and nuclear spin ing form in that representation:
^ k1' j 1 l 1 s 2 t 1 t 2 u C B l A & 5 ( ^ l 1 u R 1
s1
M ~ k1' j 1 ! u s 1 & AU ] pW
] ~ k1' j 1 !
U
^ pW ~ k1' j 1 ! s 1 s 2 t 1 t 2 u C B l A & . ~C3a!
The transformation of momenta requires the Jacobi determinant
U ] pW
] ~ k1' j 1 !
5 UU
]p3
]j 1
~ k1' j 1 ! 5 U
M 0 ~ k1' j 1 !
4 j 1 ~ 12 j 1 !
, ~C3b!
the transformation of spin the Melosh rotation
m N 1 j 1 M 0 1i ~ s 1N k 21 2 s 2N k 11 !
R M ~ k1' j 1 ! 5 ; ~C3c!
A~ m N 1 j 1 M 0 ! 2 1k21'
in Eq. ~C3b! the argument list of M 0 is made explicit. The front form momentum density samples the front form properties of
a single nucleon. Thus, only the spin of nucleon 1 is Melosh rotated. In contrast, the spin of nucleon 2 remains in canonical
representation, since the expectation value in Eq. ~C1! sums over that spin dependence. The front form momentum density
takes the form
A
P1
A
(t (
2 s 81 s 1 s 2
U ] p 3 ~ pN' j !
]j
U
^ l Nu R 1
M ~ pN' j ! u s 1 &^ s 8 8 &^ C B l A8 u pW ~ pN' j ! s 81 s 2 t 1 t 2 &
1 u R M ~ pN' j ! u l N
3^ pW ~ pN' j ! s 1 s 2 t 1 t 2 u C B l A & u j 5p 1 / P 1 . ~C4!

N A
At this stage we go back to the corresponding instant form momentum density r (pW N t N ) which reads in the nuclear c.m. system
K U(
^ s N s A8 u r ~ pW N t N ! u s N8 s A & 5 C B s A8
i
U L
d ~ pW N 2kW i ! O i ~ s N8 s N ! d t N t i C B s A 5A (
t2
(
s
^ C B s A8 u pW N s N8 s 2 t N t 2 &^ pW N s N s 2 t N t 2 u C B s A &
2
5
1
2K U
s N s A8 r 0 ~ p N t N ! 1 r 1 ~ p N t N ! sW N •SW A 1 r 2 ~ p N t N !
pW N 2
S
sW N •pW N SW A •pW N 1
2 sW N •SW A
3 D
1 r 3~ p Nt N ! (
M
T [2]
AM ~ S A ! Y * 8 sA
2M ~ p̂ N ! s N U L . ~C5!
The first step gives the definition. In the next two steps of Eq. ~C5! the spin structure of the instant form momentum density
r (pW N t N ) is recalled. The spin operator T [2]
A (S A ) was already introduced in Appendix A; it is a five-component tensor of rank
2 and is related to its Cartesian form ~A7! in standard fashion. In addition to the three functions r 0 , r 1 , and r 2 , which are
already present in the description of the spin- 21 target, a fourth function r 3 occurs, which is responsible for the tensor
polarization. The functions r 0 , r 1 , r 2 , and r 3 of Eq. ~C5! are numerically available and are used for calculating the front form
momentum distribution. However, neither r 0 nor r 3 will affect the spin-structure functions g A1 and g A2 ; both determine the
front form momentum distribution f N ( j t N ) of Eq. ~5.14a!, required for the calculation of the spin-independent deuteron
structure function F A2 . Since the calculation is carried out in the nuclear c.m. system, the canonical and front form spins are
identical, thus, u C B l A & 5 u C B s A & and therefore
^ l N l A8 u r ~ pN t N ! u l N8 l A & 5 U 2 P1
A
U
1 1 ] p 3 ~ pN' j !
]j
UK Ul N l A8 R 1 F W W
M ~ pN' j ! r 0 ~ pt N ! 1 r 1 ~ pt N ! s N •S A 1 r 2 ~ pt N ! S sW N •pW SW A • pW
pW 2
1
2 sW N •SW A 1
3 D (
M
T [2] 2M ~ p̂ ! r 3 ~ pt N !
AM ~ S A ! Y * GU pW 5pW ~ pN' j !
U LU
R M ~ pN' j ! l N8 l A
j 5p 1 / P1
N A
. ~C6!
The result ~C6! requires the replacement of the Jacobi momentum pW as pW (pN' j ) by the front form momenta (pN' j ) according
to the prescription of Eq. ~C2b!, given there in terms of (k1' j 1 ).
The determination of the spin-structure functions g A1 and g A2 in deep-inelastic lepton scattering according to Eqs. ~5.19!
requires the full spin-dependent momentum density r (pN t N ); however, the determination according to Eqs. ~5.12! needs the
front form spin distribution sW N ( j t N n A ) only, i.e.,
sW N ~ j t N n A ! 5
1
2
E 2
d pN' U ]p3
~p j!
]j N' U( 3
i51
Tr@ sW N R 1 F
M ~ pN' j ! s N R M ~ pN' j !# r 1 ~ pt N ! n A 1 r 2 ~ pt N !
i i
S p i nW A •pW 1 i
pW 2
2 nA
3 D GU pW 5 pW ~ pN' j !
.
~C7!
The front form spin distribution sW N ( j t N n A ) was already introduced in Eq. ~5.13!. The calculations of this paper are based on
it. It satisfies the sum rule ~5.15! relating it to the spin expectation value ^ s N(t N ) & in the polarized deuteron state u C B n A & ,
defined by Eq. ~5.16a!. The spin expectation value ^ s N(t N ) & is given by the S- and D-state probabilities of the two-nucleon
bound state wave function according to
^ s N~ t N !& 5
1
2S 1
P~ S !2 P~ D ! .
2 D ~C8!
@1# J. Friar, B.F. Gibson, G.L. Payne, A.M. Bernstein, and T.E. Côté, P. Pirès, and R. de Tourreil, Phys. Rev. C 21, 861
Chupp, Phys. Rev. C 42, 2310 ~1990!. ~1980!.
@2# E.W. Otten, Prog. Part. Nucl. Phys. 24, 103 ~1990!. @21# Ch. Hajduk and P.U. Sauer, Nucl. Phys. A369, 321 ~1981!.
@3# R.-W. Schulze and P.U. Sauer, Phys. Rev. C 48, 38 ~1993!. @22# A. Stadler, P.U. Sauer, and W. Glöckle, Phys. Rev. C 44, 2319
@4# C.E. Woodward et al., Phys. Rev. Lett. 65, 698 ~1990!. ~1991!; A. Stadler ~private communication!.
@5# C.E. Woodward et al., Phys. Rev. C 44, R571 ~1991!. @23# A. Schaefer, Phys. Lett. B 208, 175 ~1988!.
@6# A.K. Thompson et al., Phys. Rev. Lett. 68, 2901 ~1992!. @24# J. Ashman et al., Phys. Lett. B 206, 364 ~1988!; Nucl. Phys.
@7# C.E. Jones et al., Phys. Rev. C 47, 110 ~1993!. B328, 1 ~1989!.
@8# H. Gao et al., Phys. Rev. C 50, R546 ~1994!. @25# J.D. Bjorken, Phys. Rev. 148, 1467 ~1966!; Phys. Rev. D 1,
@9# J.-O. Hansen et al., Phys. Rev. Lett. 74, 654 ~1995!.
1376 ~1970!.
@10# P.L. Anthony et al., Phys. Rev. Lett. 71, 959 ~1993!.
@26# S. Wandzura and F. Wilczek, Phys. Lett. 72B, 195 ~1977!.
@11# D.W. Kawall, Ph.D. thesis, Stanford University, 1995 ~unpub-
@27# J. Kaur, Nucl. Phys. B128, 219 ~1977!.
lished!.
@28# E.L. Berger and F. Coester, Annu. Rev. Nucl. Part. Sci. 37,
@12# B. Adeva et al., Phys. Lett. B 302, 533 ~1993!.
463 ~1987!.
@13# K. Abe et al., Phys. Rev. Lett. 75, 25 ~1995!.
@29# S.V. Akulinichev, S.A. Kulagin, and G.M. Vagradov, Phys.
@14# K. Abe et al., Phys. Rev. Lett. 76, 587 ~1996!.
@15# E.L. Berger and F. Coester, in Quarks and Gluons in Particles Rev. B 158, 485 ~1985!.
and Nuclei, edited by S. Brodsky and E. Moniz ~World Scien- @30# R.M. Woloshyn, Nucl. Phys. A496, 749 ~1989!.
tific, Singapore, 1986!, p. 255. @31# C. Ciofi degli Atti, E. Pace, and G. Salmè, Phys. Rev. C 46,
@16# B. Keister and W. Polyzou, Adv. Nucl. Phys. 20, 225 ~1991!. R1591 ~1992!.
@17# F. Coester, Prog. Part. Nucl. Phys. 29, 1 ~1992!. @32# C. Ciofi degli Atti, S. Scopetta, E. Pace, and G. Salmè, Phys.
@18# U. Oelfke, P.U. Sauer, and F. Coester, Nucl. Phys. A518, 583 Rev. C 48, R968 ~1993!.
~1990!. @33# L.P. Kaptare, A.Yu. Umnikov, C. Ciofi degli Atti, S. Scopetta,
@19# J. Carbonell and V.A. Karmanov, Nucl. Phys. A581, 625 and K.Yu. Kazakov, Phys. Rev. C 51, 52 ~1995!.
~1995!. @34# S.A. Kulagin, W. Melnitchouk, G. Pillar, and W. Weise, Phys.
@20# M. Lacombe, B. Loiseau, S.M. Richard, R. Vinh Mau, J. Rev. C 52, 932 ~1995!.

Polarized Deep-Inelastic Lepton Scattering From The Polarized Two - and Three-Nucleon Bound States

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Polarized Deep-Inelastic Lepton Scattering From The Polarized Two - and Three-Nucleon Bound States

Uploaded by

Copyright:

Available Formats

PHYSICAL REVIEW C VOLUME 56, NUMBER 4 OCTOBER 1997

PACS number~s!: 25.30.2c, 24.70.1s, 25.10.1s, 29.25.Pj

I. INTRODUCTION describe deep-inelastic lepton scattering from the deuteron

0556-2813/97/56~4!/2293~23!/$10.00 56 2293 © 1997 The American Physical Society

1 In this case the spin projection operator becomes

(l E d 3 pu pl &^ pl u 5I. ~2.13b!

the transformation parameters ^ l A u n A & depend on the mo- P̃ Am P̃ An 1

The parametrization satisfies all required constraints. It is at

Q• P A A 2 and n A 5 $ 2(n•n A ),n 1A ,n 2A ,(n•n A ) % 5(0,n̂ R ). The momen-

2e 2 b ~ e 2 •n A !@ g A1 ~ x,Q 2 ! 1g A2 ~ x,Q 2 !# . J ~3.13!

D. Extraction of spin-structure functions

g A1 ~ x,Q 2 ! 52 ~ im A /A ! e 1 m e 2 n ^ n A u W Am n ~ P A ,Q ! u n A & u nW A i n̂ , ~3.15a!

The extraction scheme II uses relations ~3.11c! and ~3.11d!

The calculations of this paper are carried out in PWIA.

d 2s~ h e ! d 2 s ~ unpol! d 2 s ~ pol! JA ~ 0 ! 5 ( jN ~ i ! ~5.1b!

in PWIA. Its spin matrix elements are

B. Nucleonic current tensor

^ l N8 u W Nm ~nt N !~ Q N , p N ! u l N & 5 ( ^ l N8 u R 1 8 &^ s N8 u W Nm ~nt N ! ~ Q N ,p N ! u s N &^ s N u R M ~ p N ! u l N & ,

the nucleonic current tensor takes the form

with the nucleonic Bjorken variable x N :52Q 2N /2Q N •p N and with

3 ^ pN l N t N ~ PA 2pN ! l A21 f A21 u PA l A & . ~5.7!

E dM ^ l N l A8 u S ~ pN M t N ! u l N8 l A & 5 ^ l N l A8 u r ~ pN t N ! u l N8 l A & ~5.8a!

The front form momentum density will be calculated in Sec. VI.

D. Spin structure functions in deep-inelastic lepton scattering

the momentum fraction p 1 1

2. Extraction from selected current tensor matrix elements

convolution formulas result for g Ai in the form

3. Extraction from the full current tensor

tivistic phenomena. The extraction scheme II yields

In the operator the front form momentum fraction p 1 1

p1 and k 2i 5m 2i . That three-dimensional form kW i is also con-

elastic form factors are extracted from approximately com-

Secs. III D, V D 2, and V D 3 which weight spin matrix el-

The transformation of momenta requires the Jacobi determinant

3^ C B l A8 u pW qW ~ pN' j p 2 ! s 18 s 2 s 3 t N t 2 t 3 &^ pW qW ~ pN' j p 2 ! s 1 s 2 s 3 t N t 2 t 3 u C B l A & u j 5p 1 / P 1 . ~6.10!

1. Deuteron is a simple one; it is independent from the direction nW A of the

the front form spin distribution sW N ( j t N n A ) are shown in Fig.

FIG. 2. Component nW A •sW N ( j t N n A ) of the 3 He front form spin N(t )

The He data @10,11# refer to an average momentum

FIG. 6. 3 He spin-structure functions g A1 and g A2 as functions of

Second, the convolution formulas ~5.12! may be concep-

E. Is there an EMC effect

in the 3 He spin structure function g A1 ?

The shift in the nucleonic plus momentum is only introduced

ergy transfers to the struck nucleon Q N and to the nucleus Q

Q 2N 5Q 2 22Q 0 D1D 2 ~7.12!

FIG. 13. 3 He spin-structure function g A1 as function of the ACKNOWLEDGMENTS

The full density operator r A (n A ) commutes with the mo-

besides Eq. ~A4! the general constraints n 2A 521 and

P A are related to those of the rest frame, i.e.,

For Eq. ~A5! the identities S iA S Aj 1S Aj S iA 1

with were permissible. Furthermore, the structure functions which

structure functions proportional to the basis tensors G m n (W)

t Am n 5 ~ 1/m 2A ! Tr@ T m n ~ W ! r A ~ n A P A !# . ~B4b! 1

APPENDIX C: FRONT FORM MOMENTUM DENSITY

In this section the front form momentum density r (pN t N )

The transformation of momenta requires the Jacobi determinant

the transformation of spin the Melosh rotation

3^ pW ~ pN' j ! s 1 s 2 t 1 t 2 u C B l A & u j 5p 1 / P 1 . ~C4!

You might also like