1: %\documentclass[prd,aps,preprint,tightenlines,nofootinbib,superscriptaddress]{revtex4}
2: %\documentclass[prd,aps,preprint,nofootinbib,superscriptaddress]{revtex4}
3: \documentclass[prl,aps,preprintnumbers,twocolumn]{revtex4}
4: %\voffset 1in
5: \usepackage{epsfig}
6: \usepackage{amsmath}
7: %\usepackage{axodraw}
8:
9: \begin{document}
10:
11: \preprint{BNL-NT-06/11} \preprint{RBRC-588}
12:
13: \title{A Unified Picture for Single Transverse-Spin Asymmetries in Hard Processes}
14:
15: \author{Xiangdong Ji$^{1,2}$, Jian-Wei Qiu$^3$, Werner Vogelsang$^{4,5}$, Feng Yuan$^5$}
16: \affiliation{${}^1$Physics Department, University of Maryland,
17: College Park, MD 20742\\
18: ${}^2$Physics Department, Peking University, Beijing 100871, P.R.
19: China\\
20: ${}^3$Department of Physics and Astronomy, Iowa State University,
21: Ames, IA 50011\\
22: ${}^4$Physics Department, Brookhaven National Laboratory, Upton,
23: NY 11973\\
24: ${}^5$RIKEN BNL Research Center, Building 510A, Brookhaven
25: National Laboratory, Upton, NY 11973}
26: \date{\today}
27:
28: \begin{abstract}
29: Using Drell-Yan pair production as an example, we explore the
30: relation between two well-known mechanisms for single
31: transverse-spin asymmetries in hard processes: twist-three
32: quark-gluon correlations when the pair's transverse momentum is
33: large, $q_\perp \gg \Lambda_{\rm QCD}$, and time-reversal-odd and
34: transverse-momentum-dependent parton distributions when $q_\perp$
35: is much less than the pair's mass. We find that although the two
36: mechanisms have their own domain of validity, they describe the
37: same physics in the kinematic region where they overlap. This
38: unifies the two mechanisms and imposes an important constraint on
39: phenomenological studies of single spin asymmetries.
40:
41: %\vspace{10cm}
42: \end{abstract}
43:
44: \maketitle
45:
46: \newcommand{\be}{\begin{equation}}
47: \newcommand{\ee}{\end{equation}}
48: \newcommand{\ben}{\[}
49: \newcommand{\een}{\]}
50: \newcommand{\beqn}{\begin{eqnarray}}
51: \newcommand{\eeqn}{\end{eqnarray}}
52: \newcommand{\Tr}{{\rm Tr} }
53:
54:
55: {\bf 1.} Single-transverse spin asymmetries (SSAs) have a long
56: history, starting from the 1970s and 1980s when large SSAs were first
57: observed in hadronic reactions at forward rapidities. The size of the
58: asymmetries came as a surprise and has posed a challenge for researchers
59: in Quantum Chromodynamics (QCD). SSAs
60: have again attracted much interest in recent years, both
61: experimentally and theoretically \cite{review}. Two mechanisms have been
62: proposed in QCD for explaining the large SSAs: effects due to
63: time-reversal-odd and transverse-momentum dependent parton
64: distributions (Sivers functions) \cite{Siv90}, or to
65: spin-dependent twist-three quark-gluon correlations
66: (Efremov-Teryaev-Qiu-Sterman (ETQS) mechanism) \cite{Efremov,qiu}.
67: Much progress has been made in understanding theoretical issues
68: involved in transverse-momentum-dependent (TMD) parton
69: distributions. Their gauge properties have been clarified
70: \cite{ColSop81,ColSopSte85,BroHwaSch02,Col02,BelJiYua02,BoeMulPij03},
71: and the relevant QCD factorization formulas have been established
72: \cite{ColSop81,ColSopSte85,JiMaYu04,ColMet04}. Phenomenological
73: studies have been made to explain the available data, using both
74: mechanisms\cite{qiu,siverscompare}.
75: An interesting relation between the two types of
76: distribution functions was discussed in \cite{BoeMulPij03}. However, a clear
77: connection between the two seemingly-different physical mechanisms,
78: particularly at the level of physical observables, has so far not been
79: established (see \cite {MaWa03}).
80:
81: In this letter we explore the relation between the two mechanisms
82: in Drell-Yan pair production at transverse momentum $q_\perp$. If
83: the pair's mass is $Q$, the quark-gluon correlation mechanism
84: works when $ q_\perp, Q \gg \Lambda_{\rm QCD}$, the strong
85: interaction scale. On the other hand, the TMD QCD factorization
86: formalism applies when $q_\perp\ll Q$. Therefore, one expects
87: there to be a common kinematic region, $\Lambda_{\rm QCD} \ll
88: q_\perp\ll Q$, where both mechanisms should apply and describe the
89: same physics. We perform calculations in this intermediate region
90: in both formalisms and find that they indeed yield the same
91: result. This conclusion imposes a rigorous constraint on
92: phenomenological studies of the single-spin asymmetry data.
93:
94: Consider scattering of a transversely polarized proton of spin
95: $S_\perp$ and momentum $P$ on an unpolarized hadron (another
96: proton, for example) of momentum $P'$, producing a virtual photon
97: that subsequently decays into a pair of leptons. The total
98: center-of-mass energy is denoted by $s=(P+P')^2$. At the lowest
99: order in QCD perturbation theory, there are two ways to produce
100: such a pair: either a quark-antiquark pair annihilates into a
101: virtual photon after emitting a gluon of transverse momentum
102: $q_\perp$, or a quark scatters with a gluon producing a quark and
103: a virtual photon. If the lepton pair has a large positive
104: rapidity, $y$ (in the forward direction of the polarized proton),
105: the dominant contribution comes from scattering involving a quark
106: from the polarized proton. Our interest is to calculate the {\it
107: spin-dependent} differential cross section
108: $d\Delta\sigma(S_\perp)/dQ^2dyd^2q_\perp$ at large $y$ with
109: $\Delta\sigma(S_\perp)=[\sigma(S_\perp)-\sigma(-S_\perp)]/2$.
110:
111: {\bf 2.} When $q_\perp$, $Q \gg \Lambda_{\rm QCD}$, the
112: spin-dependent cross section can be calculated in terms of a
113: twist-three quark-gluon correlation. The physics of this
114: correlation can be seen as follows: when a transversely-polarized
115: proton is traveling at nearly the speed of light, its internal color
116: electric and magnetic fields have preferred orientations in the
117: transverse plane. By parity invariance, the color electric
118: field must be orthogonal to the spin of the proton. If averaged
119: over the proton wave function, the field vanishes because the
120: proton is color-neutral (also because of time-reversal symmetry).
121: However, if one multiplies the color electric field with the quark color
122: current, the average may be non-zero. This average defines a
123: quark-gluon correlation function that characterizes a property
124: of a polarized proton.
125:
126: High-energy scattering probes the so-called light-cone correlations.
127: For these, quark and/or gluon fields are separated along the light-cone
128: direction $\xi^-$ (if $\xi^\mu$ denotes a space-time coordinate,
129: the light-cone variables
130: are defined as $\xi^\pm = (\xi^0\pm\xi^3)/\sqrt{2}, \xi_\perp =
131: (\xi^1, \xi^2)$). A typical diagram for scattering of an antiquark in the
132: polarized gluon background is shown in Fig.~1. The blob in the lower
133: part of the diagram represents the spin-dependent quark-gluon
134: correlation described above. One finds the following expression for the
135: correlation \cite{qiu}:
136: \begin{eqnarray}
137: &&T_F(x_1,x_2) =
138: \int\frac{d\zeta^-d\eta^-}{4\pi}e^{i(k_{q1}^+\eta^-+k_g^+\zeta^-)}
139: \, \epsilon_\perp^{\beta\alpha}S_{\perp\beta}
140: \label{eq1} \\
141: &&\left\langle PS|\overline\psi(0){\cal L}(0,\zeta^-)\gamma^+
142: gF_{\alpha}^{\ +}(\zeta^-){\cal L}(\zeta^-,\eta^-)
143: \psi(\eta^-)|PS\right\rangle \nonumber \ ,
144: \end{eqnarray}
145: where the sums over color and spin indices are implicit,
146: $|PS\rangle$ denotes the proton state, $\psi$ the quark field, and
147: $F_{\alpha}^{\ +}$ the gluon field tensor. In Eq.~(\ref{eq1}),
148: $x_1=k_{q1}^+/P^+$ and $x_2=k_{q2}^+/P^+$ are the fractions of the
149: polarized proton's light-cone momentum carried by the quark in
150: Fig.~1, while $x_g=k_g^+/P^+ = x_2-x_1$ is the fractional momentum
151: carried by the gluon; ${\cal L}$ is the light-cone gauge link,
152: ${\cal L}(\zeta_2,\zeta_1) = \exp\left(-ig\int^{\zeta_1}_{\zeta_2}
153: d\xi^- A^+(\xi^-)\right)$ that makes the correlation operator
154: gauge-invariant; and $\epsilon^{\alpha\beta}_\perp$ is the
155: 2-dimensional Levi-Civita tensor with $\epsilon^{12}_\perp=1$.
156:
157: \begin{figure}[t]
158: \begin{center}
159: \includegraphics[height=3.0cm]{general.eps}
160: \end{center}
161: \vskip -0.7cm \caption{\it A generic Feynman diagram contributing
162: to the single transverse-spin asymmetry for the Drell-Yan process
163: through the quark-antiquark scattering channel, containing both soft-
164: and hard-pole contributions. }
165: \end{figure}
166:
167: When the antiquark of the unpolarized proton scatters off a quark
168: of the polarized proton in the presence of a polarized color
169: electromagnetic field (the gluon) in Fig.~1, the lepton-pair yield
170: has a dependence on the transverse orientation of the spin. The
171: strong interaction phase necessary for SSAs arises from the
172: interference between an imaginary part of the partonic scattering
173: amplitude with the extra gluon and a real part of the scattering
174: amplitude without a gluon in Fig.~1. The imaginary part is from
175: the pole of the parton propagator associated with the integration
176: of the gluon momentum $x_g$. Depending on which propagator's pole
177: contributes, $\Delta\sigma(S_\perp)$ gets contributions from
178: $x_g=0$ (soft-pole) \cite{qiu} and $x_g\neq 0$ (hard-pole)
179: \cite{luo,xxx}.
180:
181: {\bf 3.} By considering all partonic diagrams in Fig.~1 and their
182: complex conjugates, one finds for the spin-dependent differential
183: cross section:
184: \begin{eqnarray}
185: \frac{d^3\Delta\sigma^{q\bar q\to\gamma^*g}(S_\perp)}
186: {dQ^2dyd^2q_\perp}\!\!&\!\!=\!\!&\!\!\int\frac{dx}{x}\frac{dx'}{x'}
187: \bar q(x')\delta(\hat s+\hat t+\hat
188: u-Q^2)\nonumber\\
189: &&\times \,C\cdot \big[H_q^s+H_q^h\big](x,\hat{s},\hat{t},\hat{u}) \ ,
190: \end{eqnarray}
191: where the sum over all quark flavors, weighted with their electric
192: charge squared, is implicit. We have defined the factor $C=
193: \sigma_0\epsilon^{\alpha\beta}{S}_{\perp\alpha}
194: q_{\perp\beta}\alpha_s/2\pi^2$ with $\sigma_0 = 4\pi\alpha_{\rm
195: em}^2/3N_c sQ^2$, $N_c=3$. $x'$ is the momentum fraction carried by
196: the parton from the unpolarized proton, and $x$ is the total parton
197: momentum fraction from the polarized proton while $x_g$
198: is fixed using the pole (on-shell) condition. The partonic
199: Mandelstam variables are defined as $\hat s=(xP+x'P')^2$, $\hat
200: t=(xP-q)^2$, $\hat u=(x'P'-q)^2$. $H_q^s$ and
201: $H_q^h$ are the soft- and hard-pole contributions,
202: respectively.
203:
204: The soft-pole contributions can be calculated in the same way as for
205: direct-photon and inclusive-hadron production \cite{qiu}.
206: A total of eight diagrams \cite{xxx} sum to
207: \begin{eqnarray}
208: H_q^s &=&\frac{1}{2N_c}\left[x\frac{\partial}{\partial
209: x}T_F(x,x)\frac{D^s_{q\bar q}}{-\hat u}+T_F(x,x) \frac{N^s_{q\bar
210: q}}{-\hat u}\right]\ , \label{hq}
211: \end{eqnarray}
212: where the hard coefficients $D^s_{q\bar q}$ and $N^s_{q\bar q}$ are
213: \begin{eqnarray} D^s_{q\bar q}&=&
214: \frac{\hat u}{\hat t}+\frac{\hat t}{\hat u}+\frac{2Q^2\hat s}{\hat
215: u\hat
216: t}\nonumber\\
217: N^s_{q\bar q}&=&\frac{1}{\hat t^2\hat u}\left[ Q^2(\hat u^2-\hat
218: t^2)+2Q^2\hat s(Q^2-2\hat
219: t)\nonumber\right.\\
220: &&\left. -(\hat u^2+\hat t^2)\hat t\right]\nonumber\ ,
221: \end{eqnarray}
222: with $1/2N_c$ the color factor. In the real photon limit,
223: $Q^2=0$, we obtain the direct-photon cross section.
224:
225: Kinematics allows only a $t$-channel propagator to
226: contribute to a hard pole, and there
227: are a total of 12 such diagrams \cite{xxx}.
228: The sum of their contributions is
229: \begin{eqnarray}
230: H_q^h=T_F(x,x-x_g) \frac{N^h_{q\bar q}}{-\hat
231: u}\left[\frac{1}{2N_c}+C_F\frac{\hat s}{\hat s+\hat u}\right] \ ,
232: \end{eqnarray}
233: where
234: $$N^h_{q\bar q}=\frac{(Q^2-\hat t)^3+Q^2\hat
235: s^2}{-\hat t^2\hat u}\ . $$ As mentioned above, for the hard-pole
236: contributions $x_g=-x\hat t/(Q^2-\hat t)$ differs from zero. We
237: find that the hard-pole contribution has no ``derivative-term'',
238: $\propto \partial T_F(x,x)/\partial x$, due to a cancellation
239: among the diagrams.
240:
241:
242: {\bf 4.} When $q_\perp\ll Q$, single spin asymmetries can be
243: generated from a spin-dependent TMD quark distribution introduced
244: by Sivers \cite{Siv90}. The physics of the Sivers function can be
245: understood as follows. Consider a transversely-polarized proton
246: traveling with large momentum $P$. The distribution of quarks with
247: longitudinal and transverse momentum $xP$ and $\vec{k}_\perp$ can
248: have a dependence on the orientation of $\vec{k}_\perp$ relative
249: to the polarization vector $\vec{S}_\perp$. The TMD quark
250: distributions are best defined through the following matrix:
251: \begin{eqnarray}
252: {\cal M}^{\alpha\beta} &=& p^+\int
253: \frac{d\xi^-}{2\pi}e^{-ix\xi^-P^+}\int
254: \frac{d^2b_\perp}{(2\pi)^2} e^{i\vec{b}_\perp\cdot
255: \vec{k}_\perp} \nonumber \\ && \times \left\langle
256: P\left|\overline{\mit \Psi}_v^\beta(\xi^-,0,\vec{b}_\perp) \gamma^+
257: {\mit \Psi}_v^\alpha(0)\right|P\right\rangle\ .
258: \end{eqnarray}
259: Here the vector $p=(p^+,0^-,0_\perp)$ is along the momentum
260: direction of the proton, ${\mit \Psi}_v(\xi) = {\cal
261: L}_{v}(-\infty;\xi)\psi(\xi)$, and the gauge link is $ {\cal
262: L}_{v}(-\infty;\xi) = \exp\left(-ig\int^{-\infty}_0 d\lambda
263: v\cdot A(\lambda v +\xi)\right)$. In a system where $P^+\gg P^-$,
264: the vector $v$ with $v^-\gg v^+, v_\perp=0$ is given a non-zero
265: $v^+$ component in order to regulate the light-cone singularities.
266: An expansion of the matrix in Dirac indices yields
267: \begin{equation}
268: {\cal M} = \frac{1}{2}\left[q(x,k_\perp)\not\! p + q_T(x,k_\perp)
269: \epsilon_{\mu\nu\alpha\beta}\gamma^\mu p^\nu k^\alpha S^\beta +
270: ...\right]
271: \end{equation}
272: where $q(x,k_\perp)$ is the spin-independent distribution and
273: $q_T(x, k_\perp)$ is the Sivers function. The dependence on $v$ is
274: of the form $\zeta^2=(2v\cdot P)^2/v^2$ and is implicit.
275:
276: \begin{figure}[t]
277: \begin{center}
278: \includegraphics[height=3.0cm]{factorization.eps}
279: \end{center}
280: \vskip -0.7cm \caption{\it Decomposition of the generic
281: diagram in Fig.~1 into different regions in the
282: transverse-momentum-dependent factorization approach. The large
283: transverse momentum of the Drell-Yan pair may come from (a) the
284: Sivers function, (b) the antiquark TMD distribution, and (c) the
285: soft factor. }
286: \end{figure}
287:
288: When $q_\perp\ll Q$, the Drell-Yan cross section in leading order
289: $q_\perp/Q$ can be factorized in terms of TMD quark distributions.
290: A generic Feynman diagram can be decomposed into various
291: contributions, each of which is represented by one of the reduced
292: diagrams shown in Fig.~2 \cite{ColSopSte85}. This factorization
293: property can be seen from the diagram in Fig. 1 by considering the
294: real gluon emitted either along the direction of $P$ or $P'$. The
295: single-spin-dependent part of the Drell-Yan cross section has the
296: following factorization \cite{JiMaYu04},
297: \begin{eqnarray}
298: \frac{d^3\Delta\sigma(S_\perp)}{dQ^2dyd^2q_\perp}\!\!&\!=\!&\!\!
299: \sigma_0\frac{\epsilon^{\alpha\beta}{
300: S}_{\perp\alpha} q_{\perp\beta}}{M_P} \int
301: d^2\vec{k}_{1\perp}d^2\vec{k}_{2\perp}d^2\vec{\lambda}_\perp H
302: \label{tmdfact} \\
303: &&\!\!\!\! \times \frac{\vec{k}_{1\perp}\cdot
304: \vec{q}_\perp}{\vec{q}_\perp^2}~\delta^{(2)}(\vec{k}_{1\perp}+
305: \vec{k}_{2\perp}+\vec{\lambda}_\perp-\vec{q}_\perp)
306: \nonumber\\
307: &&\!\!\!\! \times q_T(z_1,k_{1\perp},\zeta_1)~\bar
308: q(z_2,k_{2\perp},\zeta_2)~S(\lambda_\perp)^{-1} \ ,\nonumber
309: \end{eqnarray}
310: where $z_1=Q/\sqrt{s} e^y$ and $ z_2=Q/\sqrt{s} e^{-y}$ are the
311: momentum fractions of the colliding hadrons associated with the
312: observed Drell-Yan pair. $H$ is a hard factor and is entirely
313: perturbative. The soft-factor $S$ is a vacuum matrix element of
314: Wilson lines and captures the effect of soft gluon radiation.
315: Since the soft-gluon contribution in the TMD distributions has not
316: been subtracted, the soft-factor comes with inverse power. The
317: soft-gluon rapidity cut-off $\rho$ is defined as
318: $\rho=\sqrt{(2v_1\cdot v_2)^2/v_1^2v_2^2}$, where $v_1$ and $v_2$
319: are the directions of the gauge links in the Sivers and the
320: unpolarized antiquark distributions. In a special coordinate
321: frame, $z_1^2\zeta_1^2=z_2^2\zeta_2^2=\rho Q^2$ \cite{JiMaYu04}.
322:
323: {\bf 5.} From what we have discussed so far, the two mechanisms
324: for the single-spin asymmetries appear to be quite different. In
325: the case of the twist-three correlation, the nonperturbative
326: function itself is time-reversal even. On the other hand, in the
327: TMD factorization, the Sivers function is time-reversal odd and
328: inherently contains the final-state interaction through a Wilson
329: line. There is, however, a common kinematic region, $\Lambda_{\rm
330: QCD} \ll q_\perp \ll Q$, where both mechanisms should work. In
331: this region, $q_\perp$ is large enough so that the asymmetry is a
332: twist-three effect. But at the same time, $q_\perp \ll Q$, so the
333: TMD factorization formalism also applies in this region.
334:
335: In order to see the consistency of the two approaches, we need to
336: calculate the explicit $q_\perp$-dependence in the TMD
337: factorization. To do that, we let one of the transverse momenta
338: $\vec{k}_{i\perp}$ and $\vec{\lambda}_\perp$ be of the order of
339: $\vec{q}_\perp$ and the others small. When $\vec{\lambda}_\perp$
340: is large, for example, we neglect $\vec{k}_{i\perp}$ in the delta
341: function, and the integrations over these momenta yield either the
342: ordinary antiquark distribution or a moment of the Sivers
343: function. The latter is related to the twist-three correlation
344: \cite{BoeMulPij03}:
345: \begin{eqnarray}
346: \int d^2\vec{k}_\perp \bar q(x,k_\perp) = \bar q(x) \ ,
347: \int d^2\vec{k}_\perp \frac{\vec{k}_\perp^2}{M_P} q_T(k_\perp, x) = T_F(x,x)\ . \nonumber
348: \end{eqnarray}
349: When one of the $\vec{k}_{i\perp}$ is taken to be of order
350: $\vec{q}_\perp$ and $\vec{\lambda}_\perp$ is neglected in the
351: delta function, one also needs $\int d^2\vec{\lambda}_\perp
352: S(\lambda_\perp) = 1$.
353:
354:
355: \begin{figure}[t]
356: \begin{center}
357: %\includegraphics[width=7.5 cm,height=1.6cm]{f1.eps}
358: \includegraphics[height=3.0cm]{soft-hard-s.eps}
359: \end{center}
360: \vskip -0.7cm \caption{\it A typical Feynman diagram contributing to
361: the Sivers functions at a large transverse momentum, $k_\perp \gg
362: \Lambda_{\rm QCD}$, arising from the twist-three quark-gluon
363: correlation. The bars indicate the soft and hard poles.}
364: \end{figure}
365:
366: What remains is to calculate the large-$k_\perp$ behavior of the
367: TMD distributions and the soft factor. The latter has been
368: determined in Ref. \cite{JiMaYu04} and reads
369: \begin{equation}
370: S(\lambda_\perp)=\delta^{(2)}(\vec{\lambda}_\perp) +
371: \frac{\alpha_s}{2\pi^2}\frac{1}{\vec{\lambda}_\perp^2}C_F\left(\ln
372: \rho^2-2\right) \ ,
373: \end{equation}
374: which is infra-red finite. To the same order, the unpolarized
375: antiquark TMD distribution is given by
376: \begin{eqnarray}
377: && \bar q(z_2,k_\perp) = \bar q(z_2)\delta^{(2)}(\vec{k}_\perp)+
378: \frac{\alpha_s}{2\pi^2}\frac{1}{\vec{k}_\perp^2}C_F \\ && \times
379: \int\frac{dx'}{x'} \bar q(x')
380: \left[\frac{1+\xi_2^2}{(1-\xi_2)_+}+\delta(\xi_2-1)\left(\ln\frac{z_2^2\zeta_2^2}
381: {\vec{k}_\perp^2}-1\right)\right] \nonumber \ ,
382: \end{eqnarray}
383: where $\xi_2=z_2/x'$ and an additional contribution involving the gluon
384: distribution has been neglected \cite{xxx}.
385:
386: As discussed above, the Sivers function at large $k_\perp$ can be
387: calculated in terms of the twist-three quark-gluon correlation. There are
388: also soft and hard pole contributions in this case. We show one example
389: of the relevant Feynman diagrams in Fig.~3.
390: There are a total of eight diagrams
391: for the soft-pole contribution, and twelve for the hard-pole one. Adding all
392: contributions, we find:
393: \begin{eqnarray}
394: q_T^{(1)}(z_1,k_\perp)&=&\frac{\alpha_s}{4\pi^2}\frac{2M_P}
395: {(\vec{k}_\perp^2)^2}\int\frac{dx}{x}
396: \left\{A+C_FT_F(x,x)\right.\nonumber\\
397: &&\times
398: \left.\delta(\xi_1-1)\left(\ln\frac{z_1^2\zeta_1^2}{\vec{k}_\perp^2}-
399: 1\right)\right\} \ ,
400: \end{eqnarray}
401: where the $1/(k_\perp^2)^2$ behavior follows from a dimensional
402: analysis, and where
403: \begin{eqnarray} A\!\!&\!=\!&\!\! \frac{1}{2N_c}\left[
404: x\frac{\partial}{\partial
405: x}T_F(x,x)(1+\xi_1^2)+T_F(x,x-\tilde{x}_g)\frac{1+\xi_1}{(1-\xi_1)_+}\right.
406: \nonumber \\
407: &&\left.
408: +T_F(x,x)\frac{(1-\xi_1)^2(2\xi_1+1)-2}{(1-\xi_1)_+}\right]\nonumber\\
409: &&+C_FT_F(x,x-\tilde{x}_g)\frac{1+\xi_1}{(1-\xi_1)_+}
410: \end{eqnarray}
411: with $\tilde{x}_g=(1-\xi_1)x$ and $\xi_1=z_1/x$.
412:
413: Plugging the above results into the TMD factorization formula~(\ref{tmdfact}),
414: we obtain the spin-dependent cross section for the $q+\bar
415: q$ channel at large $q_\perp$:
416: \begin{eqnarray}
417: \frac{d^3\Delta\sigma^{q\bar q\to \gamma^*g}(S_\perp)
418: }{dQ^2dyd^2q_\perp}&=&\frac{C}{(q_\perp^2)^2}\int
419: \frac{dx}{x}\frac{dx'}{x'} \bar q(x') \label{tmd}\\
420: &&\times \left\{\delta(\xi_2-1)A+\delta(\xi_1-1)B\right\} \nonumber\
421: ,
422: \end{eqnarray}
423: where $A$ and $C$ have been given above and
424: \begin{eqnarray}
425: B\!\!&\!=\!&\!\! C_FT_F(x,x)\left[\frac{1+\xi_2^2}{(1-\xi_2)_+}
426: +2\delta(\xi_2-1)\ln\frac{Q^2}{\vec{q}_\perp^{\,2}}\right] \, \nonumber
427: \end{eqnarray}
428: where $\xi_2=z_2/x'$.
429: To compare this to the twist-three correlation calculation, we
430: expand Eq.~(2) for small $q_\perp/Q$. At leading order in $q_\perp/Q$,
431: the partonic Mandelstam variables become $\hat
432: s={\vec{q}_\perp^{\,2}}/{(1-\xi_1)(1-\xi_2)}$, $ \hat
433: t=-{\vec{q}_\perp^{\,2}}/{(1-\xi_2)}$, and $ \hat
434: u=-{\vec{q}_\perp^{\,2}}/{(1-\xi_1)}$. Using the above, we find that the
435: spin-dependent cross section to leading order in $q_\perp/Q$ is exactly
436: the same
437: as the one shown in (\ref{tmd}). The same conclusion applies for the
438: contribution to the single-spin asymmetry generated by gluon-quark scattering.
439:
440: To summarize, we have studied the single transverse-spin asymmetry
441: in Drell-Yan pair production at both large and small transverse
442: momenta $q_\perp$ of the lepton pair. At large $q_\perp$, one has
443: a result in terms of a twist-three quark-gluon correlation
444: function. At small $q_\perp$, a factorization formula in terms of
445: TMD parton distributions is valid. We have demonstrated that in
446: the intermediate region, both approaches give the same answer.
447: This establishes in an explicit calculation for a physical process
448: the connection between these two mechanisms for generating
449: single-spin asymmetries. It is this connection that unifies the
450: physical pictures for the underlying dynamics of single
451: transverse-spin asymmetries and imposes an important constraint on
452: phenomenological studies of the existing and future data.
453: Extension to the gluon initiated subprocess and other physical
454: processes, such as the semi-inclusive deep inelastic scattering,
455: will be presented elsewhere, along with the details of the
456: calculations described above.
457:
458: We thank George Sterman for helpful discussion. X.~J. is supported
459: by the U. S. Department of Energy via grant DE-FG02-93ER-40762 and
460: by a grant from the Chinese National Natural Science Foundation
461: (CNSF). J.Q. is supported in part by the U. S. Department of
462: Energy under grant No. DE-FG02-87ER-40371. W.~V. and F.~Y. are
463: grateful to RIKEN, Brookhaven National Laboratory and the U.S.
464: Department of Energy (contract number DE-AC02-98CH10886) for
465: providing the facilities essential for the completion of their
466: work.
467:
468: \begin{thebibliography}
469: \frenchspacing
470: %\cite{Anselmino:1994gn}
471: %\bibitem{Anselmino:1994gn}
472: \bibitem{review} for reviews, see:
473: M.~Anselmino, A.~Efremov and E.~Leader,
474: %``The theory and phenomenology of polarized deep inelastic
475: %scattering,''
476: Phys.\ Rept.\ {\bf 261}, 1 (1995) [Erratum-ibid.\ {\bf 281}, 399
477: (1997)];
478: %%CITATION = HEP-PH 9501369;%%
479: %\cite{Barone:2001sp}
480: %\bibitem{Barone:2001sp}
481: %\cite{Liang:2000gz}
482: %\bibitem{Liang:2000gz}
483: Z.~t.~Liang and C.~Boros,
484: %``Single spin asymmetries in inclusive high energy hadron hadron
485: % collision
486: %processes,''
487: Int.\ J.\ Mod.\ Phys.\ A {\bf 15}, 927 (2000);
488: % [arXiv:hep-ph/0001330];
489: %%CITATION = HEP-PH 0001330;%%
490: V.~Barone, A.~Drago and P.~G.~Ratcliffe,
491: %``Transverse polarisation of quarks in hadrons,''
492: Phys.\ Rept.\ {\bf 359}, 1 (2002).
493: %%CITATION = HEP-PH 0104283;%%
494: %\cite{Bass:2004xa}
495:
496: \bibitem{Siv90}
497: %\cite{Sivers:1989cc}
498: %\bibitem{Sivers:1989cc}
499: D.~W.~Sivers,
500: %``Single Spin Production Asymmetries From The Hard Scattering Of
501: % Point- Like
502: %Constituents,''
503: Phys.\ Rev.\ D {\bf 41}, 83 (1990);
504: %%CITATION = PHRVA,D41,83;%%
505: %\cite{Sivers:1990fh}
506: %\bibitem{Sivers:1990fh}
507: %D.~W.~Sivers,
508: %``Hard Scattering Scaling Laws For Single Spin Production
509: % Asymmetries,''
510: Phys.\ Rev.\ D {\bf 43}, 261 (1991).
511: %%CITATION = PHRVA,D43,261;%%
512:
513: %\cite{Efremov:1981sh}
514: \bibitem{Efremov}
515: A.~V.~Efremov and O.~V.~Teryaev,
516: %``On Spin Effects In Quantum Chromodynamics,''
517: Sov.\ J.\ Nucl.\ Phys.\ {\bf 36}, 140 (1982)
518: [Yad.\ Fiz.\ {\bf 36}, 242 (1982)];
519: %%CITATION = SJNCA,36,140;%%
520: %\cite{Efremov:1984ip}
521: %\bibitem{Efremov:1984ip}
522: % A.~V.~Efremov and O.~V.~Teryaev,
523: %``QCD Asymmetry And Polarized Hadron Structure Functions,''
524: Phys.\ Lett.\ B {\bf 150}, 383 (1985).
525: %%CITATION = PHLTA,B150,383;%%
526:
527: %\cite{Qiu:pp}
528: \bibitem{qiu}
529: J.~W.~Qiu and G.~Sterman,
530: %``Single Transverse Spin Asymmetries,''
531: Phys.\ Rev.\ Lett.\ {\bf 67}, 2264 (1991);
532: %%CITATION = PRLTA,67,2264;%%
533: %\cite{Qiu:1991wg}
534: %\bibitem{Qiu:1991wg}
535: % J.~w.~Qiu and G.~Sterman,
536: %``Single transverse spin asymmetries in direct photon production,''
537: Nucl.\ Phys.\ B {\bf 378}, 52 (1992);
538: %%CITATION = NUPHA,B378,52;%%
539: %\cite{Qiu:1998ia}
540: %\bibitem{Qiu:1998ia}
541: %J.~w.~Qiu and G.~Sterman,
542: %``Single transverse-spin asymmetries in hadronic pion production,''
543: Phys.\ Rev.\ D {\bf 59}, 014004 (1999).
544: %%CITATION = HEP-PH 9806356;%%
545:
546: \bibitem{ColSop81}
547: %\cite{Collins:1981uk}
548: %\bibitem{Collins:1981uk}
549: J.~C.~Collins and D.~E.~Soper,
550: %``Back-To-Back Jets In QCD,''
551: Nucl.\ Phys.\ B {\bf 193}, 381 (1981) [Erratum-ibid.\ B {\bf 213},
552: 545 (1983)];
553: %%CITATION = NUPHA,B193,381;%%
554: %\cite{Collins:va}
555: %\bibitem{Collins:va}
556: %J.~C.~Collins and D.~E.~Soper,
557: %``Back-To-Back Jets: Fourier Transform From B To K-Transverse,''
558: Nucl.\ Phys.\ B {\bf 197}, 446 (1982);
559: %%CITATION = NUPHA,B197,446;%%
560: %\cite{Collins:1981uw}
561: %\bibitem{ColSop81p}
562: %J.~C.~Collins and D.~E.~Soper,
563: %``Parton Distribution And Decay Functions,''
564: Nucl.\ Phys.\ B {\bf 194}, 445 (1982).
565: %%CITATION = NUPHA,B194,445;%%
566:
567: \bibitem{ColSopSte85}
568: %\cite{Collins:1984kg}
569: %\bibitem{Collins:1984kg}
570: J.~C.~Collins, D.~E.~Soper and G.~Sterman,
571: %``Transverse Momentum Distribution In Drell-Yan Pair And W And Z
572: % Boson
573: %Production,''
574: Nucl.\ Phys.\ B {\bf 250}, 199 (1985).
575: %%CITATION = NUPHA,B250,199;%%
576:
577: \bibitem{BroHwaSch02}
578: %\bibitem{Brodsky:2002cx}
579: S.~J.~Brodsky, D.~S.~Hwang and I.~Schmidt,
580: %``Final-state interactions and single-spin asymmetries in
581: % semi-inclusive deep inelastic scattering,''
582: Phys.\ Lett.\ B {\bf 530}, 99 (2002);
583: %%CITATION = HEP-PH 0201296;%%
584: %\cite{Brodsky:2002rv}
585: %\bibitem{Brodsky:2002rv}
586: %S.~J.~Brodsky, D.~S.~Hwang and I.~Schmidt,
587: %``Initial-state interactions and single-spin asymmetries in Drell-Yan
588: % processes,''
589: Nucl.\ Phys.\ B {\bf 642}, 344 (2002).
590: %%CITATION = HEP-PH 0206259;%%
591:
592: \bibitem{Col02}
593: %\cite{Collins:2002kn}
594: %\bibitem{Collins:2002kn}
595: J.~C.~Collins,
596: %``Leading-twist single-transverse-spin asymmetries: Drell-Yan and
597: % deep-inelastic scattering,''
598: Phys.\ Lett.\ B {\bf 536}, 43 (2002).
599: %%CITATION = HEP-PH 0204004;%%
600:
601: \bibitem{BelJiYua02}
602: %\cite{Ji:2002aa}
603: %\bibitem{Ji:2002aa}
604: X.~Ji and F.~Yuan,
605: %``Parton distributions in light-cone gauge: Where are the final-state
606: % interactions?,''
607: Phys.\ Lett.\ B {\bf 543}, 66 (2002);
608: %%CITATION = HEP-PH 0206057;%%
609: %\bibitem{ji2}
610: %\cite{Belitsky:2002sm}
611: %\bibitem{Belitsky:2002sm}
612: A.~V.~Belitsky, X.~Ji and F.~Yuan,
613: %``Final state interactions and gauge invariant parton
614: % distributions,''
615: Nucl.\ Phys.\ B {\bf 656}, 165 (2003).
616: %%CITATION = HEP-PH 0208038;%%
617:
618: \bibitem{BoeMulPij03}
619: %\cite{Boer:2003cm}
620: %\bibitem{Boer:2003cm}
621: D.~Boer, P.~J.~Mulders and F.~Pijlman,
622: %``Universality of T-odd effects in single spin and azimuthal
623: % asymmetries,''
624: Nucl.\ Phys.\ B {\bf 667}, 201 (2003).
625: %%CITATION = HEP-PH 0303034;%%
626:
627: \bibitem{JiMaYu04}
628: %\cite{Ji:2004wu}
629: %\bibitem{Ji:2004wu}
630: X.~Ji, J.~P.~Ma and F.~Yuan,
631: %``QCD factorization for semi-inclusive deep-inelastic scattering at
632: % low
633: %transverse momentum,''
634: Phys.\ Rev.\ D {\bf 71}, 034005 (2005);
635: % [arXiv:hep-ph/0404183];
636: %%CITATION = HEP-PH 0404183;%%
637: %\bibitem{JiMaYu04p}
638: %\cite{Ji:2004xq}
639: %\bibitem{Ji:2004xq}
640: %X.~Ji, J.~P.~Ma and F.~Yuan,
641: %``QCD factorization for spin-dependent cross sections in DIS and
642: % Drell-Yan
643: %processes at low transverse momentum,''
644: Phys.\ Lett.\ B {\bf 597}, 299 (2004);
645: % [arXiv:hep-ph/0405085].
646: %%CITATION = HEP-PH 0405085;%%
647: %\cite{Ji:2005nu}
648: %\bibitem{JiMaYu05}
649: % X.~Ji, J.~P.~Ma and F.~Yuan,
650: %``Transverse-momentum-dependent gluon distributions and
651: % semi-inclusive
652: %processes at hadron colliders,''
653: JHEP {\bf 0507}, 020 (2005).
654: % [arXiv:hep-ph/0503015].
655: %%CITATION = HEP-PH 0503015;%%
656:
657:
658: %\cite{Collins:2004nx}
659: %\bibitem{Collins:2004nx}
660: \bibitem{ColMet04}
661: J.~C.~Collins and A.~Metz,
662: %``Universality of soft and collinear factors in hard-scattering
663: %factorization,''
664: Phys.\ Rev.\ Lett.\ {\bf 93}, 252001 (2004).
665: %[arXiv:hep-ph/0408249].
666: %%CITATION = HEP-PH 0408249;%%
667:
668: %\cite{Anselmino:2005an}
669: \bibitem{siverscompare}
670: see example, M.~Anselmino {\it et al.},
671: %``Comparing extractions of Sivers functions,''
672: arXiv:hep-ph/0511017.
673: %%CITATION = HEP-PH 0511017;%%
674:
675: %\cite{Ma:2003ut}
676: %\bibitem{Ma:2003ut}
677: \bibitem{MaWa03}
678: J.~P.~Ma and Q.~Wang,
679: %``On unique predictions for single spin azimuthal asymmetry,''
680: Eur.\ Phys.\ J.\ C {\bf 37}, 293 (2004).
681: % [arXiv:hep-ph/0310245].
682: %%CITATION = HEP-PH 0310245;%%
683:
684: %\cite{Luo:1994np}
685: \bibitem{luo}
686: M.~Luo, J.~W.~Qiu and G.~Sterman,
687: %``Anomalous nuclear enhancement in deeply inelastic scattering and
688: %photoproduction,''
689: Phys.\ Rev.\ D {\bf 50}, 1951 (1994);
690: %%CITATION = PHRVA,D50,1951;%%
691: %%Cited 64 times in SPIRES-HEP
692: %\cite{Guo:1997it}
693: %\bibitem{guo}
694: X.~Guo,
695: %``Nuclear dependence in Drell-Yan transverse momentum
696: %distribution,''
697: Phys.\ Rev.\ D {\bf 58}, 036001 (1998);
698: % [arXiv:hep-ph/9711453];
699: %%CITATION = HEP-PH 9711453;%%
700: %%Cited 8 times in SPIRES-HEP
701: %\cite{Guo:1997mm}
702: %\bibitem{Guo:1997mm}
703: % X.~f.~Guo,
704: %``Nuclear dependence in transverse momentum distribution for
705: % Drell-Yan
706: %pair,''
707: Nucl.\ Phys.\ A {\bf 638}, 539C (1998).
708: % [arXiv:hep-ph/9801450].
709: %%CITATION = HEP-PH 9801450;%%
710: %%Cited 0 times in SPIRES-HEP
711:
712: \bibitem{xxx} X.\ Ji, J.~W.\ Qiu, W.\ Vogelsang, F.\ Yuan, {\it
713: in preparation}.
714:
715: \end{thebibliography}
716:
717: \end{document}
718: