The transverse structure of the nucleon
(resolving the quark motion inside a nucleon)
Mauro Anselmino, Torino University and INFN, JLab, December 15, 2006
the need for parton intrinsic motion
transverse momentum dependent distribution
and fragmentation functions (TMD)
spin and k┴: transverse Single Spin Asymmetries SSA in SIDIS
Spin effects in unpolarized e+e– → h1h2 X at BELLE what do we learn from TMD’s?
Plenty of theoretical and experimental evidence for transverse motion of partons within nucleons and of hadrons within fragmentation jets
GeV/c
0.2
fm
1
x p
uncertainty principle
gluon radiation
±1
± ±
k┴
qT distribution of lepton pairs in D-Y processes
Partonic intrinsic motion
p p
Q2 = M2
qT qL
l+
l–
*
pT distribution of hadrons in SIDIS
Hadron distribution in jets in e+e– processes
Large pT particle production in
pN hX
Transverse motion is usually integrated, but there might be important
spin-k┴ correlations
p xp
k┴
frame c.m.
* p hX
M. Arneodo et al (EMC): Z. Phys. C 34 (1987) 277
) ,
ˆ ( d )
, (
d f x Q
2 lq lqD
qhz Q
2q
q lhX
lp
Intrinsic motion in unpolarized SIDIS, [ O (
s0)]
q p x Q
2
2
2
2 q
Q
p l
q y p
in collinear parton model
2 2
2
ˆ 1 ( 1 )
ˆ
d ˆ
lqlq s u y
thus, no dependence on azimuthal angle Фh at zero-th order in pQCD
the experimental data reveal that
h h
h X
lh
lq
/ d Φ A B cos Φ C cos 2 Φ
d ˆ
Cahn: the observed azimuthal dependence is related to the intrinsic k┴ of quarks (at least for small PT values)
These modulations of the cross section with azimuthal angle are
assuming collinear fragmentation, φ = Φh
) 0 , sin
, cos
(
k k
k
2
1 cos
22ˆ 1
Q y k
Q sx k
s O
cos
221 1 2
) 1
ˆ (
Q k y
Q y k
x s
u O
h h
h lhX lq
Φ C
Φ B
A u
Φ s ˆ ˆ cos cos 2 d
d ˆ
2
2
SIDIS with intrinsic k
┴kinematics
according to Trento conventions (2004)
)
; ,
( )
; ,
ˆ ( d )
; ,
(
d
lplhX
qf
qx k
Q
2
lqlqy k
Q
2 D
qhz p
Q
2factorization holds at large Q2, and
P
T k
QCD Ji, Ma, YuanThe full kinematics is complicated as the produced hadron has also intrinsic transverse momentum with respect to the fragmenting parton
k
k
z
p
P
Tp
h k
2/ Q
2
neglecting terms of order one has
p k
P
Tz
assuming:
one finds:
with
clear dependence on and (assumed to be constant)
Find best values by fitting data on Φh and PT dependences
EMC data, µp and µd, E between 100 and 280 GeV
Large PT data explained by NLO QCD
corrections
dashed line: parton model with unintegrated distribution and fragmentation functions solid line: pQCD contributions at LO and a K factor (K = 1.5) to account for NLO effects
S p
p '
– p
PT
θ
sin θ
d d
d
d
TA
NS p P
– p '
5
;
4
;
3
;
2
;
1
;
M M M M M
5 independent helicity amplitudes
Im
5(
1 2 3 4)
AN
pp pp
z y
x
Example:
Transverse single spin asymmetries: elastic scattering
0
A
N needs helicity flip + relative phase + –+ +
Im
x+
+ + +
s q
N
E
A m
at quark level but large SSA observed at hadron level!
QED and QCD interactions conserve helicity, up to corrections
O ( m
q/ E )
Single spin asymmetries at partonic level. Example:
qq ' qq '
BNL-AGS √s = 6.6 GeV 0.6 < pT < 1.2
E704 √s = 20 GeV 0.7 < pT < 2.0
E704 √s = 20 GeV 0.7 < pT < 2.0
experimental data on SSA
X p
p
X p
p
observed transverse Single Spin Asymmetries
d d
d d
A
Nand AN stays at high energies ….
STAR-RHIC √s = 200 GeV 1.2 < pT < 2.8
) ˆ (
d )
( )
(
d
/ /, , , , ,
/
x f x D z
f
a b p b ab cd cg q q d c b a
p
a
PDF FF
pQCD elementary interactions a
b c
0D
X
X
ˆ
f
f
(collinear configurations)
X pp
0factorization theorem
p
p
X p
p
0RHIC data s 200 GeV
X p
p
Transverse Λ polarization in unpolarized p-Be scattering at Fermilab
X N
p
d d
d
P d
p p p
p
p
p
p p
d d
d d
AN
d d
d d
ANN
“Sivers moment”
X l
N
l
d d
d A d
) d d
( d d
) sin(
) d d
( d 2 d
) sin(
2 sin( )
S
S S
Φ Φ UT S
Φ Φ
Φ Φ
Φ Φ
A Φ
Φ S
“Collins moment”
d d
d A d
X l
N
l
) d d
( d d
) sin(
) d d
( d 2 d
) sin(
2 sin( )
S
S S
Φ Φ UT S
Φ Φ
Φ Φ
Φ Φ
A Φ
Φ S
and now ….?
Polarization data has often been the graveyard of fashionable theories.
If theorists had their way, they might
just ban such measurements altogether out of self-protection.
J.D. BjorkenSt. Croix, 1987
P sin (Φ Φ ) * p c.m. frame
A S p P
T
T π
S
needs k┴ dependent quark distribution in p↑ and/or p┴ dependent fragmentation of polarized quark
z y
Φ
SΦ
π x
X
p
S
PT
Transverse single spin asymmetries in SIDIS
in collinear configurations there cannot be (at LO) any PT
spin-k
┴correlations? orbiting quarks?
Transverse Momentum Dependent distribution functions Space dependent distribution functions (GPD)
) ,
( x k
q
) , ( b x q
k
?
b
q p T
q
N
f
M
f k
1/
2
p
k┴φ S q
p
q
φ Sp┴ q
ˆ ) ( ˆ
) ,
2 ( 1
) ,
( )
, (
/ / /
p p
S p
q q q
h N
q q h
h
p z D
p z D
z D
ˆ ) ( ˆ
) ,
2 ( 1
) , ( )
, (
/ / /
k p S
k
k x f
k x f
x f
p q N
p p q
q
Amsterdam group notations
q q h
h
N H
M z
D p
1
/ 2
spin-k
┴correlations
Sivers function Collins function
q p
q
N
h
M f k
1/
p
Sq k┴φ
q
p
q
φ S p┴ Λˆ ) ( ˆ
) ,
2 ( 1
) ,
2 ( ) 1 ,
(
/ / /
p p
S p
q q N
q q h
p z D
p z D
z D
ˆ ) ( ˆ
) ,
2 ( 1
) , 2 (
) 1 ,
(
/ / /
k p S
k
p q q N
p p q
q
k x f
k x f
x f
Amsterdam group notations
q q T
N D
M z
D p
1
/ 2
spin-k
┴correlations
Boer-Mulders function polarizing F.F.
8 leading-twist spin-k
┴dependent distribution functions
Sivers function
dependence on S
Tin cross section
The Sivers mechanism for SSA in SIDIS processes
) ,
ˆ ( ) ,
2 ( 1
/ 2 2
6
D z p
z J z dQ
x d f
d
d
qhh lq
lq p
q q
k k
Sivers angle
SSA:
after integration over k
┴one obtains:
) sin(
h SA
) 1
,
(
h
T z
J z zk
P p
] [
d
sin ] [
2 d
sin
d d
Φ
Φ d
d A
ΦΦ
data are presented for the sinΦ moment of the analyzing power
P k
p
Tz
Parameterization of the Sivers function
1 1
N
qq = u, d.
The Sivers function for sea quarks and antiquarks is assumed to be zero.
M.A., U.D’Alesio, M.Boglione, F. Murgia, A.Kotzinian, A Prokudin
from Sivers mechanism
) sin( S
A
UT Deuteron target
dh
hp u d N p
u N
UT f f D D
A h S
) / / 4
sin(
First p┴ moments of extracted Sivers functions, compared
with models
M.A, M. Boglione, U. D’Alesio, A. Kotzinian, F. Murgia, A. Prokudin
data from HERMES and COMPASS
) ,
4 (
/ 2
) 1 ( 1 )
1 (
d k m f x k
f f
p q N p
q T q
N
k
hep-ph/0511017
1
?
1
d T u
T
f
f
The first and 1/2-transverse moments of the Sivers quark distribution functions. The fits were constrained mainly (or solely) by the preliminary HERMES data in the indicated x-range. The
curves indicate the 1-σ regions of the various parameterizations.
) , ( )
(
2 1) 2 / 1 (
1
f x k
M d k
x
f
T qk
TqM. Anselmino, M. Boglione, J.C. Collins, U. D’Alesio, A.V. Efremov, K. Goeke, A. Kotzinian, S. Menze, A. Metz, F. Murgia, A. Prokudin, P. Schweitzer, W. Vogelsang, F. Yuan
) , 2
2 1(
2 2
) 1 (
1
f x k
M d k
f
T qk
Tqpredictions for JLab, proton target, 6 GeV
0.4 ≤ zh ≤ 0.7 0.02 ≤ PT ≤ 1 GeV/c 0.1 ≤ xB ≤ 0.6
0.4 ≤ y ≤ 0.85 Q2 ≥ 1 (GeV/c)2 W2 ≥ 4 GeV2 1 ≤ Eh ≤ 4 GeV
predictions for JLab, proton target, 12 GeV
0.4 ≤ zh ≤ 0.7 0.02 ≤ PT ≤ 1.4 GeV/c 0.05 ≤ xB ≤ 0.7
0.2 ≤ y ≤ 0.85 Q2 ≥ 1 (GeV/c)2 W2 ≥ 4 GeV2 1 ≤ Eh ≤ 7 GeV
ˆ ) ) (
, ( )
, (
ˆ ) ( ˆ
) ,
2 ( ) 1
, ( )
, (
1 /
/ / /
k M x
f k
x f
k x f
k x f
x f
a T p
a
p a N p
p a a
k p
S
k p
S k
Ta
p a
N f
M
f k 1
/
2
S
k
pˆ k
number density of partons with longitudinal momentum fraction x and
transverse momentum k┴, inside a proton with spin S
0 ) ,
/
(
2
adx d k
k
f
a px k
M. Burkardt, PR D69, 091501 (2004)
What do we learn from the Sivers distribution?
ˆ ( , )
2 cos ˆ
sin ˆ
ˆ ) ( ˆ
) ,
ˆ ( 2
) 1 ,
ˆ (
/ 2
/ / 2
k x f
k dk dx
k x f
k x f
d dx
p a N S
S
p a N p
a a
j i
k p
S k
k k
Then we can compute the total amount of intrinsic momentum carried by partons of flavour a
for a proton moving along the +z-axis and polarization vector
i j
S cos
Sˆ sin
Sˆ
S
a
) k sin(
ˆ )
( ˆ
p k
SS
sin ˆ cos ˆ GeV/c
13
. 0
GeV/c
cos ˆ sin ˆ
14
. 0
03 . 0
02 . 0 0.05 0.06 -
j i
k
j i
k
S S
d
S S
u
u
k
d
k
?
0
u k d
k
Sivers functions extracted from AN data in
X p
p give also opposite results, with
036 .
0
032 .
0
u
k
dk
Numerical estimates from SIDIS data
U. D’Alesio
Collins mechanism for SSA
q
q’ initial quark transverse spin is transmitted to the final
quark
q l
q
l 1 ( 1 )
2) 1
( 2
y y d
d
d d
q q
q q
q q
q q
(in collinear configuration, here ↑,↓ means perpendicular to the leptonic plane)
ˆ
d
ˆ ) ( ˆ
) ,
2 ( ) 1 ,
( )
,
(
/ //
Sq
pq
pq h N q
q h
h z p D z p D z p
D
C h
S q
q
(
pˆ
pˆ
) sin(
Φ
Φ) sin
Φ SThe polarization of the fragmenting quark q' can be computed in QED
depolarization factor
q
N
T
q d D
d
d ˆ
x y
CΦ
Φ
SΦ
hS
qS
ql l
S h
C
Φ Φ
Φ sin Φ
C sin( Φ
h Φ
S)
y q y
q
x q x
q
y S S y
y S S y
) ) (
1 ( 1
) 1
( ) 2
(
) ) (
1 ( 1
) 1
( ) 2
(
2 2
] [
) sin(
] d
2 [d
) sin(
d d
dΦ dΦ
Φ Φ
dΦ A dΦ
S h
S h
S Φ h
Φ UT h S
Collins effect in SIDIS
)
, ( )
, ˆ (
) sin(
) ,
ˆ (
) ,
(
/ 2 /
2
2 / 1
2 )
sin(
q q p q p
lq lq S
h
q h q h S
lq N lq q
S h
Φ Φ UT
p z D
k x dQ f
d d dΦ dΦ
Φ Φ
z dQ D
k d x h
d dΦ dΦ
A
h S
k
p k
:
1 or q
h q T transversity distribution
ˆ ) ( ˆ
) , ( )
,
( /
/
p Sq pq p
q h N q
h
ND z D z p
fit to HERMES data on sin(Φh ΦS )
A
UT 2 | |
h
1 q q
assuming
W. Vogelsang and F. Yuan Phys. Rev. D72, 054028 (2005)
A. V. Efremov, K. Goeke and P. Schweitzer
(h1 from quark-soliton model)
Collins function from e
+e
–processes
(spin effects without polarization)
e+e- CMS frame:
2 , 10 . 52 GeV
s
s z E
hX h h q
q e
e
BELLE @ KEK
e+
1
P
h 2P
h
e-e+
thrust-axis
z y
x
1
p
2
p
single quark or antiquark are not polarized, but there is a strong correlation between their spins
2 2 24 sin 3 cos
ˆ cos
ˆ
e
qs d
d d
d
) ,
( )
, cos (
ˆ
cos
, / 1 1 / 2 22 2 1 2 2
1 1 2
2 1
p p
p
p D z D z
d d d
d d
dz dz
d
q h q
spins h q
spins X
h h e e
cross section for detecting the final hadrons inside the jets
contains the product of two Collins functions
q q h q h q
q h q
N q
h N q
z D
z D
e
z D
z D
e
d dz dz
d d d
dz dz
d
A ( ) ( )
) ( )
) ( cos (cos
1 sin 8
1 1
cos 2
1
cos
2 /
1 / 2
/ 2 / 1
2 2 0
2
2 1
0 2
1 1
2 1
2
1
ND ( z ) d
2p
ND ( z , p ), etc.
0
1
2Cosine modulations clearly visible
0 2
1 0
0
) cos
(
P N P
N
2
1
0 0) ( N N
M. Grosse Perdekamp, A. Ogawa, R. Seidl
Talk at SPIN2006
BELLE data
P1
) , ( ) cos(
1 1 2 1 1 2
1
1 P z z
A A
L
U
U = unlike charged pionsL = like charged pions
) ( )
( 2
) ( )
( 5
) ( )
( 5
) ( ) ( 7 ) ( ) ( 5
) ( )
( 7
) ( )
( 5
cos 1
sin 8
) 1 , (
2 1
2 1
2 1
2 1
2 1
2 1
2 1
2 2 2
1 1
z D z
D z
D z
D z
D z
D
z D z D z
D z D
z D z
D z
D z
z D z P
U N U
N F
N U
N U
N F
N
U U
F F
U N U
N F
N F
N
etc.
,
, /
/ U
N u
N F
N u
ND D D D
fit of HERMES data on
A
UTsin(hS)Collins functions and transversity distributions from a global best
fit of HERMES, COMPASS and BELLE data
M.A., M. Boglione, U.D’Alesio, A.Kotzinian, F. Murgia, A Prokudin, C. Türk, in preparation
fit of COMPASS data on
A A
UTUTsin(sin(hhSS))fit of BELLE data on
P
1( z
1, z
2)
Extracted favoured and unfavoured Collins functions
positivity bound Efremov et al.
Vogelsang, Yuan
Extracted transversity distributions
Soffer bound
Predictions for Collins asymmetry
at JLab, proton target, 6 GeV
Predictions for Collins asymmetry
at JLab, proton target, 12 GeV
Sivers function and orbital angular momentum
Sivers mechanism originates from
S L
qthen it is related to the quark orbital angular momentum D. Sivers
For a proton moving along z and polarized along y
? ) 2
,
1
(
0 /
q y p
q
N
L
k x f
dx
Open issues and wild ideas …
Sivers function and proton anomalous magnetic momentum
M. Burkardt, S. Brodsky, Z. Lu, I. Schmidt
Both the Sivers function and the proton anomalous magnetic moment are related to correlations of proton wave functions with opposite helicities
? )
,
1
(
0 /
2
p q q
N
f x k C
d
dx
k
in qualitative agreement with large z data:
d u Φ
Φ UT
Φ Φ
UT
S S
A A
) sin(
) sin(