Studies of TMDs at HERMES Luciano Pappalardo
[email protected] (for the HERMES Collaboration)
JLAB, Newport News, Virginia USA , May 18-21 2010
Quantum phase-space tomography of the nucleon
σ
Longitudinal momentum structure of the nucleon 2
Quantum phase-space tomography of the nucleon Join the real 3D experience!!
σ
TM
Ds
by [fm]
G
s D P
bx [fm]
Longitudinal momentum structure of the nucleon
3D picture in coordinate space
3D picture in momentum space 3
The nucleon spin structure at leading twist momentum
helicity
transversity
+ +
-
Legenda (courtesy of A. Bachetta): Proton comes out of the screen photon goes into the screen
• functions in black survive integration over transverse momentum
4
The nucleon spin structure at leading twist momentum
helicity Boer-Mulders transversity
pretzelosity Sivers
worm-gear
Legenda (courtesy of A. Bachetta): Proton comes out of the screen photon goes into the screen
• functions in black survive integration over transverse momentum • functions in red are naive T-odd • functions in green box are chirally odd 5
TMDs can be studied by measuring azimuthal asymmetries in SIDIS
Distribution Functions (DF)
FF σ
DF
functions in green box are chirally odd
6
TMDs can be studied by measuring azimuthal asymmetries in SIDIS
Distribution Functions (DF)
FF σ
DF
DF The SIDIS cross section exhibits asymmetries in the azimuthal angles φ and φ S
Fragmentation Functions (FF) h D1 a U Unpol. FF d.
functions in green box are chirally odd
H1⊥ Collins FF 7
The SIDIS cross section up to twist-3 0 1 dσ = dσ UU + cos 2φ dσ UU +
d
1 1 2 3 cos φ dσ UU + λe sin φ dσ LU Q Q
6 1 1 4 5 7 + SL sin 2φ dσ UL + sin φ dσ UL + λe dσ LL + cos φ dσ LL Q Q
{
8 9 10 + ST sin(φ − φS ) dσ UT + sin(φ + φS ) dσ UT + sin(3φ − φS ) dσ UT
1 1 11 12 + sin( 2φ − φS ) dσ UT + sin φS dσ UT Q Q 1 1 13 14 15 + λe cos(φ − φS ) dσ LT + cos φS dσ LT + cos(2φ − φS )dσ LT Q Q
How can we disentangle all these contributions ? EXPERIMENT: setting the proper beam and target polarization states (U, L, T) ANALYSIS: e.g. fitting the cross section asymmetry for opposite spin states and extracting the relevant Fourier amplitudes based on their peculiar azimuthal dependences.
8
0 1 dσ = dσ UU + cos 2φ dσ UU +
1 1 2 3 cos φ dσ UU + λe sin φ dσ LU Q Q
6 1 1 4 5 7 + SLsin 2φ dσ UL + sin φ dσ UL +λ dσ LL + cos φ dσ LL e Q Q
Boer-Mulders effect •
{ ( x, p
⊥+ 1
∝h
8 9 10 ST sin( 2 φ − φ S⊥) dσ UT2 + sin(φ + φ S ) dσ UT + sin(3φ − φ S ) dσ UT
T
) ⊗ H 1 ( z , kT )
• correlation between parton transverse 1 momentum and parton transverse + sin( 2φ Q polarization in an unpolarized nucleon
− φ S ) dσ
11 UT
1 12 + sin φS dσ UT Q
1 1 13 14 15 +λe cos(φ − φS ) dσ LT + cos φS dσ LT + cos(2φ − φS )dσ LT Q Q 9
The Boer-Mulders effect Twist-2:
Cos 2φ dσ UU
ˆ ˆ ⋅ ⋅ − 2 ( P k )( P p ) k T ⋅ pT ∝ cos 2φ ⋅ ∑ eq2 I h ⊥ T h ⊥ T h1⊥ H1⊥ q MM h q
Boer-Mulders effect Twist-3:
Cosφ dσ UU
Cahn effect
ˆ ˆ ( Ph ⊥ ⋅ pT ) ( Ph ⊥ ⋅ kT ) 2 2M ⊥ ⊥q ∝ cos φ ⋅ ∑ eq I − x h1 H1 − x f1 D1 + ... Q Mh M q
Accessed through azimuthal modulations in SIDIS with unpol. H and D targets
The Boer-Mulders effect Twist-2:
Cos 2φ dσ UU
ˆ ˆ ⋅ ⋅ − 2 ( P k )( P p ) k T ⋅ pT ∝ cos 2φ ⋅ ∑ eq2 I h ⊥ T h ⊥ T h1⊥ H1⊥ q MM h q
Boer-Mulders effect Twist-3:
Cosφ dσ UU
Cahn effect
ˆ ˆ ( Ph ⊥ ⋅ pT ) ( Ph ⊥ ⋅ kT ) 2 2M ⊥ ⊥q ∝ cos φ ⋅ ∑ eq I − x h1 H1 − x f1 D1 + ... Q Mh M q
Accessed through azimuthal modulations in SIDIS with unpol. H and D targets analysis based on a multidimensional unfolding of data to correct for acceptance, detector smearing and higher order QED effects
The Boer-Mulders effect (Hydrogen target) cos( φ )
UU
∝ Ι[ − h1⊥ H 1⊥ − f1 D1 ]
negative cos(φ φ) amplitudes for both π+ and πcos( w φ )
UU
∝ Ι[ − h1⊥ H 1⊥ ]
cos(2φ φ) ampl. positive for πand slightly negative for π+ Similar results for D target
The Boer-Mulders effect (Hydrogen target) cos( φ )
UU
∝ Ι[ − h1⊥ H 1⊥ − f1 D1 ]
negative cos(φ φ) amplitudes for both π+ and πcos( w φ )
UU
∝ Ι[ − h1⊥ H 1⊥ ]
cos(2φ φ) ampl. positive for πand slightly negative for π+ Similar results for D target
Accessing the polarized cross section through SSAs Full HERMES transverse data (02-05 data with
PT ≈ 73% )
The Fourier amplitudes of the yields for opposite transverse target spin states were extracted through a ML fit alternately binned in x, z, and Ph⊥ but unbinned in φ and φS:
This is equivalent to perform a Fourier decomposition of the cross section asymmetry:
1 dσ h (φ , φS ) − dσ h (φ , φS + π ) A (φ , φS ) = | PT | dσ h (φ , φS ) + dσ h (φ , φS + π ) h UT
+ ... in the limit of very small φ and φS bins. I [...]: convolution integral over initial ( pT ) and final ( kT ) quark transverse momenta
14
0 1 dσ = dσ UU + cos 2φ dσ UU +
1 1 2 3 cos φ dσ UU + λe sin φ dσ LU Q Q
Collins effect •
∝ h1 ( x,
6 1 1 4 5 7 + SLsin 2φ dσ UL + sin φ dσ UL +λ dσ LL + cos φ dσ LL e Q Q 2 ⊥ 2 p ) ⊗ H ( z, k ) 1
T
T
{
• correlation between parton 8 + sin( φ − φ ) d σ S transverse polarization in a S UT T transversely polarized nucleon and transverse momentum of the produced hadron Ph⊥
9 10 + sin(φ + φS ) dσ UT + sin(3φ − φS ) dσ UT
1 1 11 12 + sin( 2φ − φS ) dσ UT + sin φS dσ UT Q Q
h
1 1 13 14 15 +λe cos(φ − φPS ) dσ LT + cos φS dσ LT + cos(2φ − φS )dσ LT Q Q h h⊥
15
Collins pions amplitudes
positive for π+ consistent with zero for π0 negative for π16
Collins pions amplitudes Non-zero Collins effect observed Both transversity and Collins function sizeable! Ampl. increase with x, i.e. towards the valence region Isospin symmetry fulfilled −
the large negative π amplitude suggests disfavored Collins FF with opposite sign:
positive for
measurement at e+e-
π+
consistent with zero for
π0
collider machines
negative for π17
pu bl is he d
so on !
Collins pions amplitudes
be
Both transversity and Collins function sizeable!
re su l
ts
to
Ampl. increase with x, i.e. towards the valence region
li n s
Isospin symmetry fulfilled −
Co l
the large negative π amplitude suggests disfavored Collins FF with opposite sign:
Fi na l
positive for
Non-zero Collins effect observed
measurement at e+e-
π+
consistent with zero for
π0
collider machines
negative for π18
2-D Collins pions amplitudes Kinematic dependencies often don’t factorize → correlations among variables bin in as many independent variables as possibles (multidim. analysis)
X vs. Z
X vs. Ph⊥⊥
19
0 1 dσ = dσ UU + cos 2φ dσ UU +
1 1 2 3 Sivers cos φ dσ UU +effect λe sin φ dσ LU Q Q ⊥
{
D1 ( z , kT )7 •1 ∝ f1T (5x, pT ) ⊗ 4 6 + SLsin 2φ dσ UL + sin φ dσ UL +λ dσ LL + cos φ dσ LL e Q transverse •Qcorrelation between parton momentum and nucleon transverse polarization 9 8 10 2
2
+ ST sin(φ − φS ) dσ UT + sin(φ + φS ) dσ UT + sin(3φ − φS ) dσ UT • requires orbital angular momentum
1 1 11h 12 + sin( 2φ q− φS ) dσ UT + sin φS dσ UT T Q Q qT
h 1 1 13 14 15 +λe cos(φ − φS ) dσ LT + cos φS dσ LT + cos(2φ − φS )dσ LT Q Q 20
Ph ys .
Re
v. Le tt. 10 3
(20
09
)1
52 0
02 ]
Sivers amplitudes
Main features confirmed by new high-statistics results
an r ap eti [Ai
U-quark dominance suggests sizeable u-quark orbital motion!
et al. ,
First observation of T-odd Sivers effects in SIDIS!
21
Sivers pions amplitudes Slignificantly positive clear rise with z rise at low Ph⊥⊥, plateau at high Ph⊥⊥
Sligthly positive
Consistent with zero Isospin symmetry fulfilled
22
Sivers pions amplitudes Slignificantly positive clear rise with z rise at low Ph⊥⊥, plateau at high Ph⊥⊥
Sligthly positive
Consistent with zero Isospin symmetry fulfilled Large positive π+ signal is dominated by scattering off u-quarks: +
2 sin(φ − φS )
π+ UT
∝−
f1T⊥,u ( x, pT2 ) ⊗W D1u →π ( z , kT2 ) u →π + 1
f ( x, p ) ⊗ D u 1
2 T
2 T
( z, k )
→ u-quark Sivers DF < 0
null signal for π- indicates that d-quark Sivers DF > 0 (cancellation) Confirmed by phenomenological fits (Torino group) and theoretical predictions (Gamberg)!
23
2-D Sivers pions amplitudes Kinematic dependencies often don’t factorize → correlations among variables bin in as many independent variables as possibles (multidim. analysis)
X vs. Z
X vs. Ph⊥⊥
24
The pion-difference asymmetry π+
Contribution by decay of exclusively produced vector mesons is not negligible
a new observable
π−
+
+
π −π AUT
−
−
+
−
π π π π 1 (σ U ↑ − σ U ↑ ) − (σ U ↓ − σ U ↓ ) (φ , φ S ) ≡ + − + − S T (σ Uπ ↑ − σ Uπ ↑ ) + (σ Uπ ↓ − σ Uπ ↓ )
Contribution from exclusive ρ0 largely cancels out!
• significantly positive Sivers and Collins amplitudes are obtained • measured amplitudes are not generated by exclusive VM contribution
25
The pion-difference asymmetry π+
Contribution by decay of exclusively produced vector mesons is not negligible
a new observable
π−
+
+
π −π AUT
−
−
+
−
π π π π 1 (σ U ↑ − σ U ↑ ) − (σ U ↓ − σ U ↓ ) (φ , φ S ) ≡ + − + − S T (σ Uπ ↑ − σ Uπ ↑ ) + (σ Uπ ↓ − σ Uπ ↓ )
Contribution from exclusive ρ0 largely cancels out
(cancellation of FFs assuming chargeconjugation and isospin symmetry)
provides access to Sivers u-valence distribution! 26
Sivers kaons amplitudes Slignificantly positive clear rise with z rise at low Ph⊥⊥, plateau at high Ph⊥⊥
Sligthly positive
27
Sivers kaons amplitudes Slignificantly positive clear rise with z rise at low Ph⊥⊥, plateau at high Ph⊥⊥
Sligthly positive
test presence of 1/Q2-suppressed contributions separate each x-bin in two Q2 bins hint of higher-twist contributions to the K+ amplitude 28
The Sivers π+/K+ riddle π+/K+ production dominated by scattering off u-quarks: ∝−
⊥ ,u 1T u 1
f
u →π + / K + W 1 u →π + / K + 1
( x, p ) ⊗ D 2 T 2 T
f ( x, p ) ⊗ D
( z , kT2 )
( z , kT2 )
?
π + ≡ ud , K + ≡ us → non trivial role of sea quarks
?
impact of different kT dependence of FFs in the convolution int.
⊗W
29
The Sivers π+/K+ riddle π+/K+ production dominated by scattering off u-quarks: ∝−
⊥ ,u 1T u 1
f
u →π + / K + W 1 u →π + / K + 1
( x, p ) ⊗ D 2 T 2 T
f ( x, p ) ⊗ D
( z , kT2 )
( z , kT2 )
?
π + ≡ ud , K + ≡ us → non trivial role of sea quarks
?
impact of different kT dependence of FFs in the convolution int.
⊗W
• Difference of π+ and K+ amplitudes • Separate each x-bin in two Q2 bins • only in low-Q2 region significant (90% c.l.) deviation is observed
?
Higher-twist contrib. for Kaons 30
0 1 dσ = dσ UU + cos 2φ dσ UU +
pretzelosity 1
•
+
∝
1 1 2 3 cos φ dσ UU + λe sin φ dσ LU Q Q
6 1 4 5 7 + SLsin 2φ dσ UL + sin φ dσ UL +λ dσ LL + cos φ dσ LL e ⊥ 2 ⊥ Q 2 Q h ( x, p ) ⊗ H ( z , k ) 1T
1
T
T
• correlation between parton transverse 8 9 parton transverse sin(φ − φand ) d σ + sin( φ + φ ) d σ ST momentum S UT S UT polarization in a transversely polarized nucleon 1
{
+
10 + sin(3φ − φS ) dσ UT
sin( 2φ − φS ) dσ
• can be linked to the shape Qof the nucleon (deviation from a sphere)
11 UT
1 12 + sin φS dσ UT Q
1of Ph⊥ with14 1 •λ suppressed by two powers 13 15 + e cos(φ − φS ) dσ LT + cos φS dσ LT + cos(2φ − φS )dσ LT respect to Collins and Sivers Q amplitudes Q 31
The sin(3φ-φS) Fourier component • Sensitive to pretzelosity • suppressed by two powers of Ph⊥ w.r.t. Collins and Sivers amplitudes • no significant non-zero signals oberseved
0 1 dσ = dσ UU + cos 2φ dσ UU +
1 1 2 3 cos φ dσ UU + λe sin φ dσ LU Worm-gear (UL) Q Q ⊥
{
+ ST sin(φ
⊥
H1 ( z , k )7 1• ∝ h1L (5 x, pT ) ⊗ 4 6 + SLsin 2φ dσ UL + sin φ dσ UL +λ dσ LL + 1 cos φ dTσ LL e Q Q • correlation between parton transverse momentum and parton transverse polarization9 in a longitudinally polarized 8 10 − φS ) dσ UT + sin(φ nucleon + φS ) dσ UT + sin(3φ − φS ) dσ UT 2
2
measurements 1 • accessible in UT 1 11 12 + sin( 2φ − φsin(2φ+φ ) d σ + sin φ d σ through ) Fourier component S UT S UT S Q Q
1 1 13 14 15 +λe cos(φ − φS ) dσ LT + cos φS dσ LT + cos(2φ − φS )dσ LT Q Q 33
The sin(2φ+φS) Fourier component • arises solely from longitudinal (w.r.t. virtual-photon direction) component of the target spin
• related to UL Fourier comp • sensitive to worm-gear
h1⊥L
• suppressed by one power of Ph⊥ w.r.t. Collins and Sivers amplitudes • no significant non-zero signal observed (except maybe for K+)
0 1 dσ = dσ UU + cos 2φ dσ UU +
{
+ ST sin(φ
1 1 2 3 cos φ d σ + sin φ d σ λ Worm-gear (LT) UU LU e Q Q
• ∝ g ⊥ ( x, p 2 ) ⊗ D ( z , k 2 ) 1T5 T 7 T 6 1 11 4 + SLsin 2φ dσ UL + sin φ dσ UL + d σ + cos φ d σ LL λe between LL • correlation parton Q Q transverse momentum and parton longitudinal polarization in a 8 9 10 polarized nucleon − φS ) dσ UT + sin(φ +transversely φS ) dσ UT + sin(3φ − φS ) dσ UT
1 1 11 12 + sin( 2φ − φS ) dσ UT + sin φS dσ UT Q Q
• accessible in UT measurements through sub-leading sin(2φ-φS) Fourier comp.
1 1 13 14 15 +λe cos(φ − φS ) dσ LT + cos φS dσ LT + cos(2φ − φS )dσ LT Q Q 35
The subleading-twist sin(2φ-φS) Fourier component • sensitive to pretzelosity, worm⊥ gear g1T and Sivers function:
• suppressed by one power of Ph⊥ w.r.t. Collins and Sivers amplitudes • no significant non-zero signal observed
The subleading-twist sin(φS) Fourier component ⊥
• sensitive to worm-gear g1T , Sivers function, Transversity, etc • significant non-zero signal observed for π- and K- !
The subleading-twist sin(φS) Fourier component ⊥
• sensitive to worm-gear g1T , Sivers function, Transversity, etc • significant non-zero signal observed for π- and K- ! Q2 dependence observed in π- signal:
Conclusions The existence of an intrinsic quark transverse motion gives origin to azimuthal asymmetries in the hadron production direction
• Non-zero Boer-Mulders effect observed for identified charged pions → clear evidence of non-zero Boer-Mulders function and Collins FF
• significant Collins amplitudes observed for charged π-mesons → enabled first extraction of transversity and Collins FF
• significant Sivers amplitudes observed for π+ and K+ → clear evidence of non-zero Sivers function → (indirect) evidence for non-zero quark orbital angular momentum → hint of non-trivial role of sea quarks and of higher-twist contrib. for positive kaons
• additional Fourier components recently extracted → no evidence of non-zero pretzelosity (though amplitude kinematically suppressed) ⊥ → first glimpse on worm-gears h1⊥L and g1T related observables h → significant non-zero sin(φS ) UT amplitudes for negatively charged mesons 39
Back-up slides
40
The HERMES experiment at HERA
TRD, Calorimeter, preshower, RICH: lepton-hadron > 98%
hadron separation
Aerogel n=1.03
C4F10 n=1.0014
π ~ 98%, K ~ 88% , P ~ 85% 41
The Boer-Mulders effect analysis based on a multidimensional unfolding of data to correct for acceptance, detector smearing and higher order QED effects = Kinematic range of integration
= signal not expected and not observed
x
= signal expected and observed
= signal expected but not observed
The Boer-Mulders effect for π+ Hydrogen vs. Deuteron target data The two samples are compatible
The Boer-Mulders effect for πHydrogen vs. Deuteron target data The two samples are compatible
Standard cuts
0.1 < y < 0.95
y < 0.95
0 .2 < z < 0 .7
Collins moments: Pion-kaon comparison •
K+
and
π + amplitudes
consistent (u-quark dominance) •
− K − and π amplitudes with
opposite sign (but
K − (u s )
originates from
fragmentation of sea quarks)
46
Siver samplitudes: additional studies
No systematic shifts observed between high and low Q2 amplitudes for both π+ and K+ No indication of important contributions from exclusive VM
47
2-D moments for Collins
π±: z vs. Ph⊥ Sivers
48
An alternative channel to access transversity Interference FF
ep→ →e’h1h2X
(does not depend on quark transv. momentum)
Chiral-odd
T- odd
Correlation between transverse spin of the fragmenting quark and the relative orbital angular momentum of the hadron pair.
Describes Spin-orbit correlation in fragmentation
h1
azimuthal asymmetries in the direction of the outgoing hadron pairs.
• Independent way to access transversity • No complications due to convolution integral → interpretation more transparent • ...but limited statistical power (v.r.t. single-hadron SSAs) • published on JHEP 06 (2008) 017
The extraction of the Distribution Functions dφS d Ph ⊥ sin(φ +φ S ) dσ UT ∫ = ∝I 2 ∫ dφS d Ph⊥ dσ UU 2
sin(φ + φS )
h UT
kT ⋅ Pˆh ⊥ 2 ⊥q 2 h1 ( x, pT ) H1 ( z , kT ) Mh
Convolution integral on transverse momenta
pT and kT
dφS d Ph ⊥ sin(φ −φ S ) dσ UT pT ⋅ Pˆh ⊥ ⊥ q ∫ 2 q 2 = ∝I f1T ( x, pT ) D1 ( z , kT ) 2 M ∫ dφS d Ph⊥ dσ UU 2
sin(φ − φS )
h UT
Experiment: only partial coverage of the full Ph ⊥ range (acceptance effects) Theory: difficult to solve h1 ( x) h1 ( x, p ) ≈ e 2 π pT ( x) 2 T
Gaussian ansatz −
pT2 pT2 ( x )
H
⊥q 1
H1⊥ q ( z ) ( z, k ) ≈ e 2 π kT ( z ) 2 T
(extraction assumption-dependent)
−
kT2 kT2 ( z )
50
Extraction of transversity and Sivers function form global analyses sin (φ +φ S ) AUT ∝ h1 ( x) ⊗ H1⊥ q ( z )
sin (φ −φ S ) ∝ f1,⊥T ( x) ⊗ D1q ( z ) AUT
Known
lp → l ' hX
ld → l ' hX
e + e − → h1h2 X lp → l ' hX
First extraction of Transversity!
ld → l ' hX
Anselmino et al. Phys. Rev. D 75 (2007)
Anselmino et al. Eur.Phys.J.A39 (2009)