
2
In both cases, the r2slopes are of the same order of
magnitude as predicted by theoretical calculations of the
CMW [8–12]. In particular, the STAR results exhibit
the expected centrality dependence. However, non-CMW
mechanisms could also contribute to the splitting of v±
2
as a function of Ach. A hydrodynamic study [20] claims
that the simple viscous transport of charges, combined
with certain initial conditions, will lead to a sizeable v2
splitting for charged pions. According to the analyti-
cal calculation of the anisotropic Gubser flow [19], the
∆v2for pions is proportional to both the shear viscosity
and the isospin chemical potential (µI) [20, 21]. On the
other hand, charge asymmetry Ach can also be linearly
related to µIwith the help of a statistical model, which
consequently connects ∆v2and Ach. This model further
predicts negative r2slopes for charged kaons and protons
with larger magnitudes than the pion slopes, because µI
as well as the strangeness chemical potential µSwill af-
fect these particles differently. These predictions warrant
the extension of our measurements to kaons and protons.
Local charge conservation (LCC) [18, 22–24] is also
able to qualitatively explain the finite r2slope observed
from data, when convoluted with the characteristic de-
pendence of v2on particle pseudorapidity (η) and trans-
verse momentum (pT). This is demonstrated with lo-
cally charge-conserved clusters, e.g., a pair of particles
with opposite charges, originating from a fluid element
or a resonance decay. Such a pair could contribute to a
non-zero Ach in an experiment, when one of the particles
escapes the limited detector acceptance. If this process
preferentially occurs in a phase space with smaller v2,
such as a lower-pTor higher-ηregion, then there would
be a positive r2slope, whether the escaping particle is
positive or negative. For example, the escape of a π+
with smaller v2effectively increases the v2of detected
π+’s, and decreases the observed Ach, causing a negative
slope for detected π+’s. Conversely, the escape of a π−
with smaller v2increases the v2of detected π−’s, and also
increases the observed Ach, causing a positive slope for
detected π−’s. A realistic estimate of such contributions,
however, appears to be smaller than that observed in the
STAR measurements [16]. Ref. [22] also proposes a test
with the r3measurements, defined as r3=d∆v3/dAch
with ∆v3=v−
3−v+
3, which should yield finite slopes ac-
cording to the LCC picture, while no slope is expected
from the CMW picture. Recently the CMS collaboration
at the LHC [25] has observed that normalized r2and r3
slopes are very similar to each other for charged hadrons
in Pb+Pb collisions at 5.02 TeV, supporting the LCC
picture. Such a test with the STAR data at 200 GeV is
reported in this paper.
The CMS measurements [25] also show, for charged
hadrons, a very similar Ach dependence of ∆v2in p+Pb
and Pb+Pb collisions at 5.02 TeV. In p+Pb collisions, the
magnetic field direction is presumably decoupled from
the event plane [26], and the r2slopes are dominated by
non-CMW contributions. The similar r2slopes in p+Pb
and Pb+Pb collisions [25] suggest that the r2slopes mea-
10−5−0 5 10
K
σ
n
0.5−
0
0.5
1
1.5
2
)
-2
(GeV c
2
m
1
10
2
10
3
10
4
10
5
10
6
10
7
10
Au+Au 200 GeV
< 1 GeV/c
T
|<1, 0.15 < pη|
)pp(
±
K
±
π
FIG. 1. Particle identification by the STAR TPC and TOF
detectors. nσdenotes the deviations from the theoretical
ln(dE/dx) curves measured by the TPC (here for kaons),
while m2denotes the mass information deduced from the
TOF.
sured in Pb+Pb are unlikely to originate from the CMW.
This disappearance of the CMW could arise from the
fact that the magnetic field strength drops in the vac-
uum much faster at the LHC energies than at RHIC [27],
and at the time of quark production, the magnetic field
could become too weak to initiate the CMW. The poten-
tial difference in the physics mechanisms between RHIC
and the LHC motivates us to present STAR measure-
ments of r2in small systems, i.e.,p+Au and d+Au at
200 GeV, and to compare them with results for Au+Au
and U+U collisions.
This paper is organized in the following way. The
STAR experiment and data collection are briefly intro-
duced in Sec. II. The analysis methods and systematic
uncertainties are described in Sec. III. The STAR results
of the Ach dependence of identified particle anisotropic
flow are presented and discussed in Sec. IV, where we
report: (A) the Ach dependence of mean pTand mean
|η|, and the ∆v2slope for charged pions selected using
different phase space requirements; (B) the r2slopes for
charged kaons and protons; (C) the r3slope for charged
pions; (D) the r2slopes for charged pions in p+Au, d+Au
and U+U. A summary is given in Sec. V.
II. EXPERIMENTAL SETUP AND DATA
SELECTION
The STAR detector complex consists of a series of sub-
systems located in both midrapidity and forward-rapidity
regions. The main detectors involved in this work are