Slides - Harvard-Smithsonian Center for Astrophysics

IS THE LINE REAL?
NEW FORCES IN THE DARK SECTOR
Neal Weiner
Center for Cosmology and Particle Physics
New York University
May 21, 2014
debates on the nature of dark matter 2014
The Interactions of Dark Matter
e
k
i
l
P
M
I
W
0.5-1 GeV residual
20
10
15
0
10
5
-10
-20
20
10
0
-10
20
10-6 counts/cm2/s/sr
20
10
0
-10
0
-20
20
5
20
-20
2-5 GeV residual
20
Fully cosmological zoom of isolated Milky Way: Vmax~ 40 km/s
CDM
!
10
σ/m=0.5 cm2/g
σ/m=1 cm2/g
σ/m=10 cm2/g
3
0
2
Elbert et al., in prep
J. Bullock, UC Irvine
-10
1
-20
20
0
10
0
-10
-20
10-6 counts/cm2/s/sr
Can solve.Too.Big.To.Fail:.
.with.σ/m > 0.5 cm2/g
4
10
0
-10
-20
20
FIG. 6: Intensity maps (in galactic coordinates) after subtracting the
why not scattering for a
cosmic signal?
In general for weak scale cross sections,
at E ~ keV to MeV, rate is too low,
A dark force and an excited state:
allows excitation, followed by deexcitation into CR signal
allows a cross section as large as 1/q2 or 1/m2
look for signals in the keV-MeV range
Finkbeiner, NW ‘07
A LINE AT 3.55(ish) KeV
Submitted to ApJ, 2014 February 10
Preprint typeset using LATEX style emulateapj v. 04/17/13
DETECTION OF AN UNIDENTIFIED EMISSION LINE IN THE STACKED X-RAY SPECTRUM OF GALAXY
CLUSTERS
Esra Bulbul1,2 , Maxim Markevitch2 , Adam Foster1 , Randall K. Smith1 Michael Loewenstein2 , and
Scott W. Randall1
[astro-ph.CO] 10 Feb 2014
17 Feb 2014
1
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138.
2 NASA Goddard Space Flight Center, Greenbelt, MD, USA.
Submitted to ApJ, 2014 February 10
ABSTRACT
We detect a weak unidentified emission line at E = (3.55 3.57) ± 0.03 keV in a stacked XMM
spectrum of 73 galaxy clusters spanning a redshift range 0.01 0.35. MOS and PN observations
independently show the presence of the line at consistent energies. When the full sample is divided
into three subsamples (Perseus, Centaurus+Ophiuchus+Coma, and all others), the line is seen at
An unidentified line in X-ray spectra of the Andromeda galaxy and Perseus galaxy cluster
> 3 statistical significance in all three independent MOS spectra and the PN “all others” spectrum.
The line is also detected at the same
energy in the2Chandra ACIS-S
and ACIS-I spectra
of the Perseus
1
3,4
1,5
A.
Boyarsky
,
O.
Ruchayskiy
,
D.
Iakubovskyi
and
J.
Franse
cluster, with a flux consistent with XMM-Newton (however, it is not seen in the ACIS-I spectrum of
1
Instituut-Lorentz
for Theoretical
Physics,
Leiden,
Niels for
Bohrweg
2, Leiden,
Netherlands
Virgo). The
line is present
even if we
allowUniversiteit
maximum
freedom
all the
knownThe
thermal
emission
2
Ecole
Polytechnique
ed´eralean
deequivalent
Lausanne, FSB/ITP/LPPC,
CH-1015,ofLausanne,
lines. However,
it is
very weak F´
(with
width in the BSP,
full sample
only ⇠ 1Switzerland
eV) and located
within 50–110 3eV
of several
known
faint lines;Physics,
the detection
is atStr.
the14-b,
limit
of the
current
Bogolyubov
Institute
of Theoretical
Metrologichna
03680,
Kyiv,
Ukraineinstrument
4
capabilities and subject
significant
modelingAcademy”,
uncertainties.
On Str.
the 2,origin
this Ukraine
line, we argue that
NationaltoUniversity
“Kyiv-Mohyla
Skovorody
04070,ofKyiv,
there should be no5 Leiden
atomicObservatory,
transitions
in thermal
at this2,energy.
An Netherlands
intriguing possibility is
Leiden
University,plasma
Niels Bohrweg
Leiden, The
the decay of sterile neutrino, a long-sought dark matter particle candidate. Assuming that all dark
matter is inWe
sterile
neutrinos
= 2E
7.1 keV,
ourofdetection
in the
fulland
sample
corresponds
identify
a weak linewith
at E m
∼s 3.5
keV =
in X-ray
spectra
the Andromeda
galaxy
the Perseus
galaxy to
2
11
a neutrino
decay
mixing
angle sin (2✓)objects,
⇡ 7 ⇥for10which, there
below
the
previous
However,
based
cluster
– two
dark matter-dominated
exist
deep
exposuresupper
with thelimits.
XMM-Newton
X-ray
observatory.
Such
a line
was not previously
to be present
in the
spectra than
of galaxies
or galaxy
clusters.
on the cluster
masses
and
distances,
the line known
in Perseus
is much
brighter
expected
in this
model,
Although
the linefrom
is weak,
it hassubsamples.
a clear tendency
to become
stronger
towards
the centers
the objects; it bright
is
significantly
deviating
other
This
appears
to be
because
of an of
anomalously
the in
Perseus
clusterwhich
than for
the Andromeda
galaxy dielectronic
and is absent inrecombination
the spectrum of aline,
very although
deep
line at Estronger
= 3.62for
keV
Perseus,
could
be an Arxvii
“blankwould
sky” dataset.
individual
objects it isvalue
hard toand
exclude
the possibility
that the
is due In
its emissivity
have Although
to be 30for
times
the expected
physically
difficult
to feature
understand.
an instrumental
effect might
or an atomic
line of
anomalous
brightness,init other
is consistent
with the behavior
a line
principle,tosuch
an anomaly
explain
our
line detection
subsamples
as well,of though
it
originating
from
the
decay
of
dark
matter
particles.
Future
detections
or
non-detections
of
this
line
in
multiple
would stretch the line energy uncertainties. Another alternative is the above anomaly in the Ar line
targets may
helpkeV
to reveal
its nature.
combinedastrophysical
with the nearby
3.51
K line
also exceeding expectation by factor 10–20. Confirmation
Bulbul et al
73 Clusters, XMM, central, to z=0.35
incl Coma, Perseus !
Perseus Chandra, central
!
Virgo Chandra, central (not seen)
Boyarsky et al
M31 XMM central+non-central
!
Perseus XMM, non-central
0.8
-1
3.57 ± 0.02 (0.03)
0.7
XMM-MOS
Full Sample
6 Ms
1.5
XMM-PN
Full Sample
2 Ms
3.51 ± 0.03 (0.05)
-1
-1
Flux (cnts s keV )
-1
Flux (cnts s keV )
10
0.6
0.02
1
Bulbul et al
0.02
5.5
0
Residuals
2
Eff. Area (cm )
6.5
0.04
-1
-1
Flux Residuals
(cnts s keV )
2
Residuals
Exposure χ2 /d.o.f. Line position
Flux
∆χ2
0
0
[ksec]
[keV]
10−6 cts/sec/cm2
-0.01
-0.02
M31 ON - CENTER
978.9
97.8/74 3.53 ± 0.025
4.9+1.6
13.0
-0.02
−1.3
1020
315
M31 OFF - CENTER
1472.8 107.8/75 3.53 ± 0.03
< 1.8 (2σ)
...
310
+0.044
+2.6
P ERSEUS 1000
CLUSTER (MOS)
528.5
72.7/68 3.50−0.036
7.0−2.6
9.1
305
+3.1
300
P ERSEUS CLUSTER (PN)
215.5
62.6/62 3.46 ± 0.04
9.2−3.1
8.0
16
980
+0.019
+2.2
3
3.2
3.4
3.8
4
3
3.2
3.4
3.8
4
3.6
3.6
P ERSEUS
(MOS)
1507.4
191.5/142
3.518
8.6
(Perseus) 25.9
1.9 ⇥ 10 10 , consistentEnergy
with the
MOS detection. Figure
Gaussian line and re-fit
the(keV)
Perseus spectrum−0.022
removing −2.3
(keV)
Energy
+1.4
+ M31 ON - CENTER
(3 dof)
6 shows both XMM-Newton Perseus spectra.
the upper limits on the Ar xvii DR line. We obtained 4.6−1.4 (M31)
2
only
a
slightly
worse
fit
than
the
previous
case,
with
a
0.08
B LANK - SKY0.2
15700.2 33.1/33 3.53 ± 0.03
< 0.7 (2σ)
...
Dataset
Eff. Area (cm )
Residuals
0.04
0.01
3
of 598.8 (574 dof). The measured flux of the Ar xvii DR
+1.3
5
0.1
XMM - MOS
line
at 3.62 keV in this case was 4.8+0.7
0.8 ( 1.4 ) ⇥ 10
Perseus of combined observations
Basic properties
used in this paper. Second column denotes the sum of exposures of individual observa2
0
photons
s 1 , which is a factor of 30 above the pre(core cut)
2 cm
last column317
shows
changedicted
in ∆χmaximum
when 2 extra
d.o.f.
and flux
of the line)
flux of
the (position
Ar xvii DR
line based
on are added. The energies for Perseus are quoted
ks
TABLE I:
tions. The
in the rest frame of the object.
-0.1
the measured flux of the Ar xvii line at ⇠3.12 keV and
line rates. The predicted maximum flux of the
-0.04
0.2
Ar xvii DR line for the Perseus spectrum was 1.6 ⇥ 10 6
0.1
650
1
10.00
photons cm 2 s M31
(<
0.01 times the flux0.36
of the Ar xvii
ON-center
0
M31 ON-center
285
XMM-MOS
0.34
triplet
at
⇠3.12
keV).
No line at 3.5 keV
XMM-PN
-0.1
Centaurus +
inCentaurus
the Perseus
0.32
+
640 This test showed that the line detected
-0.2
Coma +
1.00
Coma
+
cluster
could
also
be
interpreted
as
an
abnormally
bright
0.30
Ophiuchus
280
Ophiuchus
305
Ar xvii DR line. We note that, however,0.28
that
obtaining
525.3 ks
630
168
ks
such a bright DR line relative to the He-like
triplet
at
300
0.10
0.26
3.12
keV
is
problematic.
The
emissivity
of
the
satellite
3
3.2
3.4
3.8
4
3.6
3
3.2
3.4
3.6
3.8
4
0.24
295
line peaks at kT=1.8 keV,
and
declines sharply
at lower
Energy
(keV)
Energy (keV)
0.22
temperatures, in addition to the change in
the ionization
3
3.2
3.4
3.8
4
3.6 0.01
-2
-2
1⋅10
+17
1⋅10
No line
at 3.5 the
keV Ar
No line at 3.5 keV
Energy (keV)
balance
which
reduces
content
of
the plasma.
0.008
Line at 3.5 keV
0.04
8⋅10-3
8⋅10-3
The
emissivity
ratio
for
the
DR/3.12
keV
has
its
max-3
Figure
7. 3 4 keV band of the core-excised
0.006
6⋅10 stacked MOS spec6⋅10-3
imum
value of 0.04 at kT=0.7 keV, but
the
emissivity
-3 the energy band,
trum of the Perseus cluster. The figures 4⋅10
show
0.02
-3
4⋅10
of
both
lines
is
weak
here,
so
any
hotter
component
will
where a new spectral feature at 3.57 keV 2⋅10
is detected.
The Gaus-3
0.002
-3
2⋅10observed.
sian lines with peak values of the flux normalizations
of K xviii
0
dominate
and
lead
to
a
lower
ratio
being
0⋅10
0
0
0
and Ar
xvii estimated using AtomDB were included
in the modTo avoid cool gas in the Perseus core0⋅10
contaminating
-2⋅10-3
els.
The red lines in the top panels show the model
and
the
excess
-0.002
-3
-4⋅10-3
the flux of the nearby Ar and K lines,-2⋅10
we also tried ex-0.02
emission in both spectra. The blue lines show
the total model after
-3
-3
0
-4⋅10
a Gaussian line is added, indicating that -6⋅10
the unidentified spectral
cising
the
central
1
region
of
the
cluster
and
performed
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
3.0
3.2
3.4
3.6
3.8
4.0
line315
can be modeled with a Gaussian.
the
fit
on
the
core-excised
co-added
MOS
spectrum.
We
1260
Energy [keV]
Energy [keV]
found that adding an extra Gaussian line at 3.57 keV has
310
XMM-PN
XMM-MOS
2
improved
the
fit
by
of
12.8
for
an
additional
Rest
of degree
the
Rest
of
the
1240
305
+0.7 +1.2of the central
Since
this is a single-cluster
we count
first
check
FIG. 1:spectrum,
Left: Folded
rate
(top)
and
residuals
(bottom)
for
the
MOS
spectrum
region of M31. Statistical Y-errorbars on the
of freedom with a best-fit flux of 2.1 0.6 ( 1.1
) ⇥ 10 5
Sample
Sample
whether
the Perseus signal
is
not
an
artifact
of
our
blue300
2
(69 the
Clusters)
(69 Clusters)
top plot are smaller than
the point size.
The line
is not
hence
theinnergroup of positive residuals. Right: zoom onto the line
photons
cm around
s 1 3.5
(seekeV
Figure
7).added,
Excising
1220
shifting
procedure. For this we fit the original, redshifted
1.8 Ms
4.9 Ms
0
295
region.
most 1 reduced the flux of the detected line by a factor
MOS spectrum with a line-free apec model. We obtained
290
of two, indicating that the most of the flux of this emis2 3.2
3
3.8
3.6 a Gaussian
a best-fit
463 for 385 3.4
dof. Adding
line at4
3.2
3.4
3.8
4
3.6
sion3 originates
from theEnergy
cool core.
2
Energy (keV)
(keV)The mixing angle that
3.57 keV (rest energy) improved
the fit by
of 16 for
corresponds to the line flux from the core-excised Perseus
an additional degree of freedom. The best-fit flux
was
5.3
Figure 5. Top panels: 53 4 keV band of2the stacked
MOS
(left
panel)
and
stacked PN
(right panel)
spectra1 of the
The for
figures
spectrum
is consistent
within
2 samples.
with those
the
± 1.2 (2.0) ⇥ 10 photons cm s 1 , is in agreement
[cts/sec/keV]
Normalized count rate
[cts/sec/keV]
Data - model
Residuals
Boyarsky et al
Eff. Area (cm )
[cts/sec/keV]
[cts/sec/keV]
Data - model
Normalized count rate
2
Eff. Area (cm )
-0.2
AtomDB
2
2
Eff. Area (cm )
Residuals
2
2
Eff. Area
(cm Area
)
Residuals
Eff.
(cm )
4.5
Passes the Toro test…
show the energy band where the new spectral feature is detected. The Gaussian lines with maximum values of the flux normalizations of K
BUT WHAT IS IT?
assuming it’s BSM physics, that is
Decaying
dark matter
DECAYING
DARK
MATTER
e±
• Sterile neutrino N → ν + γ
Ns
ν
ν
W∓
W∓
γ
γ
(a)
k
• R-parity violating gravitino
g˜ → ν + γ
ℓ
ℓ
ℓ˜
!
G
p
•
̸R
ν
p−k
• Also R-parity violating axino, . . .
• For bosonic DM axions (or axion-like particles) would decay a → γγ
Oleg Ruchayskiy
Xshamelessly
stolenDfrom talk by Ruchayskiy,
April 201411
ECAYING DARK MATTER IN
RAYS
22
Figure 12. Recent constraints on sterile neutrino production
models, assuming sterile neutrinos constitute dark matter (Abazajian et al. 2007). Straight lines in black show theoretical predictions
assuming sterile neutrinos constitute the dark matter with lepton
number L = 0, L = 0.003, L = 0.01, L = 0.1. Constraints from the
cosmic X-ray background are shown in the solid (blue and hatched
regions). The region is solid green is excluded based upon observations of the di↵use X-ray background (Abazajian et al. 2007).
Individual galaxy cluster constraints from XMM-Newton observations of the Coma and Virgo clusters are shown in light blue (Boyarsky et al. 2006). The horizontal pink band shows the mass scale
consistent with producing a 100 300 pc core in the Fornax dwarf
galaxy (Strigari et al. 2006), and limits from the Milky Way by
Boyarsky et al. (2006) is indicated with BMW. The orange region
at ms < 0.4 keV is ruled out by an application of the Tremaine-
10-11
Tremaine-Gunn / Lyman-α
2
Interaction strength Sin (2θ)
cases, such as the core of the Perseus cluster where many
neutral
filamentsDMare
known, it is possible that CX could
10-7
overproduction
be large-8 enough to create Excluded
a small fraction
the total
by X-ray of
observations
X-ray10emission, although it would not create or enhance
a line10
at-93.57 keV or the DR line at 3.62 keV. CX could
not dominate the overall emission, however, as it would
also create
10-10 Fe XVII and other lines that are not detected.
5.2. Sterile neutrino decay line?
An interesting
interpretation of the line is the decay
Not enough DM
10-12
signature of the sterile neutrino, a long-sought dark matter particle
candidate (Boyarsky et al. (e.g., 2009), see
10-13
1
10 would be dou- 50
our §1). The mass
of 2the sterile 5neutrino
matter
mass M
ble the decay photon Dark
energy,
ms =7.1
keV.
The line flux
DM [keV]
detected in our full sample corresponds to a mixing angle
for the decay sin2 (2✓) ⇠ 7 ⇥ 10 11 . This value is below
FIG.
4: Constraints
on sterile
neutrino
DM searches,
within νMSM
[4]. The
the
upper
limits placed
by the
previous
shown
would
corresponds
the stacked
best-fit value
from M31 if the
inblue
Fig.point
12. Our
detection
fromtothe
XMM-Newton
line comes
from DM
decay.clusters
Thick errorbars
are ±1σ
MOS
observations
galaxy
are shown
with limits
a staron the
Thin
errorbars
the uncertainty
in the DM
influx.
red in
that
figure.correspond
Figure 13toshows
the detections
anddistributionlimits
in thewe
center
of M31.
upper
obtained
from our various subsamples we
used in this work (based on the included cluster masses
and distances), as well as a comparison with previous upper limit placed using the Bullet cluster by Boyarsky et
al. (2008) at 3.57 keV, which is the most relevant earlier
constraint for us. Since the mixing angle is a universal
[1] A. Boyarsky, O. Ruchayskiy, and D. Iakubovskyi, JCAP 0903,
quantity, all the subsample measurements must agree.
The 005
line(2009).
in the subsample of fainter 69 clusters (full
[2]
J.
L.
Feng,
ARA&A
48, 495
(2010). and Centaurus)
sample sans Perseus,
Coma,
Ophiuchus
[3] S. Tremaine
J. E. Gunn,
Lett. 42, 407with
(1979).
corresponds
to a and
mixing
anglePhys.
thatRev.
is consistent
Boyarsky,the
D. same
Iakubovskyi,
O. Ruchayskiy,
the[4]
fullA.sample;
is seenand
(though
with a Phys.
mild Dark
Univ. 1, for
136the
(2012).
1.5 tension)
subsample of bright nearby clusters
[5] A. Boyarsky, O. Ruchayskiy, However,
and M. Shaposhnikov,
Ann. Rev.
Coma+Centaurus+Ophiuchus.
the brightness
Nucl.line
Part.inSci.
191 (2009). spectrum of Perseus
of the new
the59,
XMM-Newton
[6] A. Boyarsky,
J. Lesgourgues,
O. Ruchayskiy,
andthan
M. Viel,
corresponds
to a significantly
higher
mixing angle
Phys.
102, (by
201304
(2009).
that for
theRev.
full Lett.
sample
factor
8 for the MOS spec[7] E.
Bulbul,
M. Markevitch,
R. further
K. Smith,investiM. Loewentrum),
which
poses
a problemA.
inFoster,
need of
05
noitcudorprevo MD
snoitavresbo yar-X yb dedulcxE
2
5
01
]Vek[ MDM ssam rettam kraD
MD hguone toN
An intere
signature of
ter particle
our §1). Th
ble the deca
detected in
for the deca
Tremaine-Gunn / Lyman-α
the upper li
in Fig. 12. O
MOS observ
Interaction strength Sin (2θ)
in red in tha
upper limits
used in this
and distanc
per limit pl
al. (2008) a
12. Recent constraints on sterile neutrino production
constraint f
weFigure
should
compare
all
models
to
is
the
sterile
neutrino…
models, assuming sterile neutrinos constitute dark matter (Abazaquantity, al
jian et al. 2007). Straight lines in black show theoretical predictions
(cf Davoudiasl et al hep-ph/0405097)
The line
assuming sterile neutrinos constitute the dark matter with lepton
number L = 0, L = 0.003, L = 0.01, L = 0.1. Constraints from the
sample sans
cosmic X-ray background are shown in the solid (blue and hatched
corresponds
1
7-
01
01
01
8-
9-
01
01
01
011
0
11-
21-
31-
2
the model
neutral filam
be large en
X-ray emiss
a line at 3.5
not domina
also create F
CONSIDERING ALTERNATIVES
•
Important to have alternatives just to ask what to
test
•
Other observations may motivate other scales of
DM (e.g., the GeV excess in the GC)
•
?
VIRGO VS PERSEUS
Other Clusters MOS
Other Clusters PN
Perseus (Core-Cut) MOS
Perseus (Core-Cut)PN
Perseus ACIS-I
Perseus ACIS-S
Virgo ACIS-I
7.04
7.02
10
10
-11
sin
2
100
Figure 13. Sterile Neutrino Mass and Mixing angle measurements and upper limits obtained from the di↵erent samples used
in this study. The comparison of our stacking method with the
limits placed by the single well exposed Bullet cluster at 3.57 keV
10
7.06
Coma+Centaurus+Ophiuchus
MOS
Coma+Centaurus+Ophiuchus
PN
Ar XVII
5×10
7.08
Bullet Cluster (Boyarsky
et al. 2008)
Full Sample MOS
Full Sample PN
Flux (ph cm-2s-1keV-1)
ms (keV)
7.1
1.5×10
7.12
3
Figure 14.
The line wi
1300 km s
VIRGO VS PERSEUS
mvirgo~ few
14
10
M☉
dvirgo~ 15 Mpc
mperseus~ few
14
10
dvirgo~ 75 Mpc
M☉
II.
ForMODELS
δ > 2me , this will be followed by χ∗ → χe+ e− ,
The original XDM model was based on a simple U(1) dark force with the minim
2
ngian ,
INTEGRAL/SPI positron excess [3–5].
THE
MODEL
However,
for δ < 2m , it has been shown that - absen
grangian
,
The original XDM model was based on a simple U(1) dark
e
2
1
d
dµν
dµν
isFstable
onµν Fcosmological
An ob
L=χ
¯i D
̸ χi state
+ Fµν
+ ϵF
+ m2 φµ φµtimescales
+ Mi χ
¯i χi +[8,
δi χ9].
i χi .
4
1 d dµν
1 ∗ dµν
µν
2, which
µ
χ
σ
χF
the
inclusion
of
a
dipole
operator
L
=
χ
¯
D
̸
χ
+
F
F
+
ϵF
F
+
m
φ
φ
+
M
µν
i
µ
M µν the dark force,
µν
e kinetic mixing parameter ϵ gives SMi particles
ϵ
charge
under
all
4
lifetime
such
decay
uilibrium in the earlyThe
universe
via
χχ ↔aparameter
φφ,
and isφeϵ[10]
↔
γe.SM particles ϵ charge un
kineticfor
mixing
gives
The presence of theequilibrium
excited state
allows
possibility
χχ
→
in the
earlythe
universe
via χχof↔upscattering
φφ, and
↔
γe.
! φe "
2!
ke
M
δ > 2me , this will be
followed
by of
χ the
→ excited
χe e , state
potentially
[1, 6, o
τ = 0.5
secexplaining
×
The
presence
allows
the possibility
TeV
δ
∗
+ −
TEGRAL/SPI positron
[3–5].
For excess
δ > 2m
,
this
will
be
followed
by
χ
→
χe
e , potenti8
e
Thus, even for ∼ keV splittings, dipoles with M < 10
∗
+ −
However, for δ < 2mINTEGRAL/SPI
that -excess
absent[3–5].
any other interaction - the e
positron
e , it has been shown
timescales. This then motivates us to consider the imp
te is stable on cosmological
timescales
modification
to the
However,
for δ < [8,
2m9].
hasobvious
been shown
that - absent
anymo
o
e , it An
X-ray signals
beyond the 511 keV line. We dub this
varia
1 ∗ µν
∗
χ σ χFµν , whichtimescales
mediates [8,
the9].
decay
→ χγm
inclusion of a dipole
operator
state
is stable
An χobvious
M on cosmological
time for such a decay
[10]
theis inclusion
of a dipole operator M1 χ∗ σ µν χFµν , which mediat
III. X-RAY
OF XDM AND A FEATUR
lifetime
for such aSIGNALS
decay
is
[10]
"2 !
"3
!
−3.1
we hold the total mass inside
the virial
radius fixed
by varying
ρ0 .
+0.16
−5
−2
−1
+0.16
−5
−2
−1
[12]offind
rate ×
of10
0.49cm
sec . [11]
For find
Virgoan[11]
find limit
an up
−0.13 ×
31 [12] For
findM31
a rate
0.49a−0.13
sec10 . cm
For Virgo
upper
compare signals,−5we approximate
the Chandra ACIS-I FOV (which is a 2 × 2 array of
−2
−1
0.91
−2 × −1
10−5ofcm
sec10′. cm sec .
CCDs) by a 9 disc, which has nearly the same angular size. For cuspy profiles (as
A naive
estimate
of the total from
luminosity
from
can(assuming
be found (assuming
ve estimate
of the
total luminosity
Perseus
canPerseus
be found
an NFW
e necessary to explain the data), the (Perseus)
majority of the signal is in the central region,
profile) using the cluster parameters found in [13]
sing the cluster parameters found in [13]
he precise boundary is not
at the
leading order. For M31, whose flux has
"
#
! Rimportant
2
" 200#2 2 ρ(r)
! R200
ρ(r) 4πr
L =
⟨σscatt
v⟩ the inner 15’ as our region.
2 data,
ound L
by =
[12] from 4πr
XMM
we⟨σtakemv⟩
a radius
around
scatt χ
0
"
#"
#2
mχ
0
"
# scatt
" v⟩
#2 10GeV
will parametrize the predicted flux
⟨σ
49 as
× v⟩ −19 10GeV
.
⟨σscatt
49= 1.9 × 10 photons/sec
3
−1
10
cm
sec
m
= 1.9 × 10 photons/sec!×
.
(3)
"
!
"
χ
2
−19
3
−1
10 perseus
cm sec 10GeVmχ
⟨σv⟩
19,10
Φperseus
= Fperseus
.
(7)
With Perseus
78 Mpc
away,×this10yields
photon
flux
−19 cma3 local
−1
sec
mχ
seus 78 Mpc away, this yields a local
photon
flux
"
#"
#2
"
# ⟨σv⟩
" 19,10 #2 10GeV
−5
−5
−2
−1
inner slope profilesΦ
γ
=
(0.7,
1,
1.3)
we
find
F
=
(1.1,
2.0,
5.9)
×
10
cm
sec
.
=
2.6
×
10
photons/sec.
⟨σv⟩ 10−19 cm10GeV
3perseus
−1
−5
sec
mχ
Φ = 2.6 × 10 19,10 −19 3 −1
photons/sec.
(4)
19,10
−5
−2
−1
cm9.6,
sec62) × 10 mcm
for Virgo, we find Fvirgo10
= (2.7,
sec . Finally, for M31, FM 31 =
χ
Clearly, this cross section is well above the conventional thermal annihilation cros
y, this cross section is well above the conventional thermal annihilation cross section,
but since this is a scattering process, this cross section can be naturally large, as
5
this is a scattering process, this cross section can be naturally large, as we now
describe.
THE SIGNAL
The perturbative cross section for this scattering has a cross section
erturbative cross section for this scattering has a cross
2 2 section
4πMχ αd
,
2 σ2 =
calculation
As
abeen
more
careful
estimate,
we
now
take
value
inand
each
As
afrom
more
careful
estimate,
we now
takeestima
hassystem.
a fixed
value
in each
system.
Aswe
a more
careful
found
by
[12]
XMM
data,
take
a radius
aro
!
!
! and v is the
cross section
in the “moderately
relativistic” limit,
2parametrize the
2will
2
2−
predicted
flux
as
2
2cluster,
⟩n=assumed
σmrWevthat
− v⟨σ
,
(8)
⟨σ
is
independent
of
location
in
the
v⟩
=
σ
v
v
,
scatt
scatt
mr
⟨σ
v⟩
=
σ
v
−
v
thresh
thresh
scatt
mr
thresh ,
WIMPs. We take the (3D, single-particle) rms velocity dispersion
THE
SIGNAL
!
each system. As a more careful estimate, we now take
"
⟨σv⟩
relativistic”
limit,
andscale
vinis19,10
the“moderately
relative
e“moderately
section
in is
thevelocity
relativistic”
limit,
andradius
v is the
perseus
σmr
the
cross
the
relativistic”
vcross
the circular
atsection
the
radius.
At each
we tr
!“moderately
circ iswhere
Φperseus
=
F
×
perseus
2
−19 cm
3 sec−1
2−v
⟨σ
v⟩
=
σ
v
,
(8)
10
scatt
mr
thresh
(3D, velocity
single-particle)
rms
tovelocity
be
WIMPs.
We take
theWIMPs.
(3D,velocity
single-particle)
dispersion
of the
We dispersion
take the rms
(3D,
single-particle)
rm
ibution at the escape velocity, vesc (r).
"
−28
2the
19,10
locity
at
the
scale
radius.
At
each
radius
we
truncate
v
is
the
circular
velocity
at
the
scale
radius.
At
each
radius
we ×
tr
3/2v
,
where
v
is
circular
velocity
at
the
scale
radius
section
in
the
“moderately
relativistic”
limit,
and
v
is
the
relative
circ
circ
circ
ence value
of
σ
=
10
cm
,
we
get
F
=
(0.12,
0.29,
1.1)
For inner
1, 1.3) we find Fperseus
mrslope profiles γ = (0.7,
perseus
−5
We (0.7,1.0,1.3).
take
the
(3D,
single-particle)
rms
velocity
dispersion
to
be
vesc
(r).
the
velocity
distribution
at
the
escape
velocity,
v
(r).
ribution
at
the
escape
velocity,
v
(r).
19,10
esc
esc
svelocity,
for
F
=
(0.47,
2.0,
13.0)
×
10
in
a
9
arcmin
virgo
While for Virgo,
we find Fvirgo = (2.7, 9.6, 62) × 10−5cmra−
−28−5 we
2 truncate
−28
2
−28circular
2 Taking
the
velocity
at
the
scale
radius.
At
each
radius
−5
a
reference
value
of
σ
=
10
cm
,
we
get
F
rence
value
of
σ
=
10
cm
,
we
get
F
=
(0.12,
0.29,
1.1)
×
1
0
cm
,
we
get
F
=
(0.12,
0.29,
1.1)
×
10
in
mr
perseus
mr
perseus
perseus
=
(0.29,
1.3,
9.6)
×
10
in
a
15
arcmin
radius.
As
we
see,
the
va
31
−5(0.47, 2.0, 13.0) × 1
atfor
the
escape
velocity,
v
(r).
−5 (0.7,1.0,1.3). F
esc
a
9
arcmin
radius
for
=
s
(0.7,1.0,1.3).
F
=
(0.47,
2.0,
13.0)
×
10
inpicture
a
9
arcmin
rad
(0.47,
2.0,pronounced
13.0)virgo
× 10 forin this
a 9 arcmin
radius
for
rgo
is =
even
more
model. virgo
But the
is
quali
5
−28
2
−5
alue
of
σ
=
10
cm
,
we
get
F
=
(0.12,
0.29,
1.1)
×
10
in rad
−5
−5
mr
perseus
−5
(0.7,1.0,1.3).
F
=
(0.29,
1.3,
9.6)
×
10
in
a
15
arcmin
(0.29,
9.6) ×radius.
10
inAs
a 15
As we see, the va
M
31
10
a 151.3,
arcmin
we arcmin
see, theradius.
variation
31 = in
naive
model.
.7,1.0,1.3). Fvirgo = (0.47, 2.0, 13.0) × 10−5 in a 9 arcmin radius for
with
the
slope
γ
is
even
more
pronounced
for
this
model.
Bu
is
even
more
pronounced
for
this
model.
But
the
picture
is
qualit
nced
for
this
model.
But
the
picture
is
qualitatively
e see that for -5“pure”
NFW profiles, there
seems to be a-6conflict betw
-5
−5
(think
Perseus,
limit ofradius.
10 forAs
Virgo,
few the
x 10variation
for M31)
.29, 1.3,
9.6)10
× 10for in
a 15 arcmin
we see,
the same as the naive model.
naive
model.
inmore
Virgo
and
the
detection
in
Perseus.
However,
because
the
upsca
pronounced for this model. But the picture is qualitatively
result,
we see
that
“pure”
NFWtoprofiles,
there betwe
seems
eFW
see profiles,
thatAs
for
“pure”
NFW
there
seems
be a conflict
2 athere
seems
toprofiles,
be a for
conflict
between
the
on excess [3–5].
, it has been shown that - absent any other interaction - the
THE
SIGNAL
logical timescales [8, 9]. An obvious modification to the m
operator
1 ∗ µν
χ σ χFµν ,
M
which mediates the decay χ → χ
∗
is [10]
!
"2 !
"3
keV
M
.
τ = 0.5 sec ×
TeV
δ
8
splittings, dipoles with M < 10 TeV lead to decays on cosmo
could lookthe
like implications
2, could look like
motivates usSignal
to consider
of this XDM scena
511 keV line. We dub this variant of the XDM scenario “Xra
XDM MODELS FOR THE 3.55
LINE
Can be an internal symmetry (between states), or composite model
3.55 line as 21cm
Cline, Farzan, Liu, Moore and Xue
1404.3729
3.55 line as Lyman-
Kopp, Liu, NW in prep
ASSOCIATED SIGNALS
Data - model HctsêsecêkeVL
0.010
0.005
0.000
-0.005
3.0
3.5
Energy HkeVL
4.0
4.5
5.0
FIG. 1: The HD spectrum with gaussian fit for di↵erent energy levels. We fit the 3.5 keV line and
fix the flux for 2p ! 1s, while the other np flux are determined by the ratio in Table.I. We apply
Look for other “nearby” lines
a factor of 2 in solid line to represent the uncertainty in calculation as in Fig.7.
III.
A.
THE DARK HYDROGEN MODEL
The Model setup
My estimate is factor
of 5-10 conflict
unless dipole is
weak, in which case,
no conflict as
excited states
propagate out of the
center
Looking in the MW
center
!
S. Riemer-Sørensen in prep
5
-10
nts/cm2/s/sr
10
0
0
-10
A SIGNAL IN THE GC/IG
-20
20
0
10
0
-10
-20
20
-20
2-5 GeV residual
5
10
3
0
2
-10
1
-20
20
0
-10
-10
-20
20
10-6 counts/cm2/s/sr
4
0
0
5-20 GeV residual
20
10
10
-20
10
0
-10
-20
20
10
0
-10
-20
FIG. 6: Intensity maps (in galactic coordinates) after subtracting the best-fit Galactic di↵use model, Fer
isotropic templates. At energies between ⇠0.5-5 GeV (i.e. in the first three frames), the dark-matter-like e
visible around the Galactic Center.
analysis of Ref. [8], the cut on CTBCORE significantly
hardens the spectrum at energies below 1 GeV, rendering it more consistent with that extracted at higher latitudes (see Appendix A). Shown for comparison (as a solid
line) is the spectrum predicted from a 35.25 GeV dark
can vary non-negligibly with the choice of
Appendix C).
pulsar populatio
dark
matter
model.
tion
any new
signal is,using
in fact,
the result
dark signals
di↵useare
astrophysical
emission
pr
searches
cosmic
rays,of annihilating
where putative
tral
stellar
clust
we once
again note
the signal
described
nM
matter.
ther
the spectrum
or inthe morp
a↵ected by poorly Thirdly,
constrained
di↵usion
andthat
energy-loss
thematter
fact that glo
this
study
can
be
explained
by
a
very
simple
dark
signal.
In
particular,
the spheri
mu
There
are significant
reasons
to conclude,
however,
processes.
In other
words,
for the gamma-ray
signal at
numbers
of mill
candidate,
without
any baroque
orserved
otherwise
unexpected
emission
with
respect
to
⇠1
that the
gamma-ray
signal
described
in
this
paper
is
far
hand, there are relatively few “knobs to turn”, making
quence
of their
features.
Aftermatter
accounting
for uncertainties
in the
overall
not trace any
combination
ofmo
av
more likely
to
be
a
detection
of
dark
than
any
it less likely that one would be able to mistakenly
fit
a
central
stellar
c
(i.e. radiation,
gas, [17],
dust,
star
fo
mass
of the Firstly,
Milky Way’s
dark matter
halo profile
of the previously reported
anomalies.
this signal
pul
well-measured
astrophysical
signal
with
an
annihilating
follow the
square
of thepulsars
anticipa
lisecond
our results
favor and
darkhas
matter
with
an
annihiconsists of a very large number
of events,
been particles
of
dark
matter
model.
26 (with
3
detected with overwhelming
statistical
significance.
lation
cross section
of v The
= (0.7 The
3.9)astrophysical
⇥ 10 tion
cminterpretat
/s a num
to
4 again note that the¯signal described in
2.4
once
the excessThirdly,
consists we
of ⇠10
gamma
rays per square
(for
annihilations
to bb,meter,
see Fig. 15).
range
covers
nMSP
r signal
),
withinThis
the context
of/
this
tion
per year
1 GeV
(from
within
10
the
Galactic
thisabove
study
can be
explained
by aofvery
simple
matter
a large
of gam
unr
the
long
predicted
value
thatdark
is from
required
of apopulation
thermal
much
of the
con
Center).
Not only without
does this
large
number
of otherwise
events
sars. Thewith
millisecond
pulsarsthe
ob
candidate,
baroque
or
unexpected
relicany
that
freezes-out
in theenearly
universe
an abun⇠1
around
mo
able usfeatures.
to conclude
with
confidence
themeasured
signal is
are largely
located
either
dance
equal
touncertainties
the
dark
matter
After
accounting
forthat
incosmological
theWay
overall
pre
more
challengin
present, but it also allows
us to determine
its
26 spectrum
3
or in[17],
or around
the facGalacticCen
D
density
(2.2
⇥
10
cm
/s). No
substructure
boost
mass of the Milky Way’s dark matter halo
profile
pulsars could ac
and morphology in some tors,
detail.
And
as
shown,
the
meascale
height
of
zs ⇠ 1 kpc [11,
Sommerfeld
enhancements,
or
non-thermal
histories
bee
our
results
favor
dark
matter
particles
with
an
annihiof
this
excess,
w
sured spectrum, angular distribution, and normalization 26
lation
would
lead
to
a
di↵use
g
3
are required.
Furthermore,
it10
is not cm
difficult
to to
construct
cat
lation
cross
section
of
v
=
(0.7
3.9)
⇥
/s
at
least
⇠10
of this emission does indeed match well with that exhighly
elongated
along the disk,
simple
models
in
which
a
⇠30-40
GeV
particle
annihi¯
to bmatter
b, see particles.
Fig. 15). This range
covers with
pected(for
fromannihilations
annihilating dark
tions
for the sim
In
compatible
the constraints
latesvalue
to quarks
with
the required
cross section without
pul
the
long
predicted
that
is
required
of
a
thermal
consistent
with
Secondly, the gamma-ray signal from annihilating dark
example, the best-fit
model
of
violating
constraints
from direct
detection experiments,
stror
that
freezes-out
in the
early and
universe
with an
matterrelic
can be
calculated
straightforwardly,
generally
on abunthe population
of ifpresently
more,
the
requ
colliders,parameters.
or other indirect
searches
(for work
related
to a the
depends
on
only
a
few
unknown
The
morlisecond
pulsars,
predicts
mo
dance equal to the measured cosmological dark matter
present
⇠10
(⇠
particle
physics
models
capable
of
accommodating
this
of t
26
phology
of this (2.2
signal,
depends
only on the boost
gamma-ray
exhibiting
density
⇥ in
10particular,
cm3 /s).
No substructure
fac- emission
Center, a signifi
signal,
see
Refs.
[62–74]).
lise
distribution
of
dark
matter
in
the
Inner
Galaxy
(as
paWithin
10
of
the
Galactic
Cen
tors, Sommerfeld enhancements, or non-thermal histories
beensignal
resolvedsion
by
And
lastly,
dark matter
of this
rameterized in our study by
the
innerthe
slope,
). Theinterpretation
that millisecond
pulsars should
are required.
Furthermore,
it ismodels
not difficult
to construct
catalog
not
hard
to
make
≠
not
baroque
is
strengthened
by
the
absence
plausibledi↵use
or well
moti-(assumi
app
spectral shape of the signal depends only on the mass of of observed
emission,
and
simple
models
invated
which
a ⇠30-40
GeVModel
particlesignal
annihiluminosi
the dark
matter
particle
and
on
what
Standard
in this
paper.
alternatives.
There
is no reason
todescribed
expectsimilar
that
any
gam
lates
quarks inwith
the
required
cross
section without
(see
talk by
I. Yavin)
pulsars)
[11, how
44,
particles
areto
produced
its annihilations.
The
GalacTo evade this
conclusion,
siderations like vacuum stability or from the requirement that the vac
acceptable symmetry-breaking pattern. These are most simply iden
√
gauge, 2 H † = (h, 0) with real h, where the scalar potential takes th
"2
m20 2 λ 2 2 λS 4 λh ! 2
2
S + S h +
S +
h − vEW
V =
.
2
2
4
4
λh and vEW = 246 GeV are the usual parameters of the Standard Model
1. The Existence of a Vacuum: This potential is bounded from
that the quartic couplings satisfy the following three conditions:
-40
log σel (nucleon)
λS , λh ≥ 0
λS λh
!
cm2
"
LUX limits ~ 10-45cm2
-42
≥
and
λ2
for negative λ.
We shall assume that these relations are satisified and study the mini
potential.
-44
-46 Breaking Pattern: We demand the minimu
2. Desirable Symmetry
80
100requirement
20 to have
40 acceptable
60
80particle
100masses,
Burgess,
Pospelov
an
obvious
in
order
mS [GeV]
mS [GeV]
andis ter
Veldhuis
‘01 to
necessary
in order
-40ensure the longevity of S in a natural way. (S
60 is
"
40
cm2
20
mh = 120 GeV
⟨h⟩ =
̸ 0; and it must not break the symmetry S → −S, so ⟨S⟩ = 0. T
survive the age of the universe in order to play their proposed pres
!
mh = 100 GeV
the following two properties:
It must spontaneously break the electrowe
-48
aving a large enough For concreteness, we take the DM to be
ctive,
higher
dimensional opor
[12,
13].
1 2, coupled
a 54
2
fermion,
,
with
mass
m
to
a
real,
g
mmediately
ruled
out
V
=
V
+
m
a
+
a
+
V
,
(4)
onfronting electroweak sym2HDM
port
Ldark2 =a0y 0a0 ¯4i 0 .
allows
the
ets/photons.
diSM portion of The
Le↵ isglet,
not anpseudoscalar mediator,
a
,
through
†
0
Vport = iBa0 H1 H2 + h.c.
(5)
tor is
spin-dependent
ge
enough
The mediator couples to the SM via the Higgs porta
itruled
safe from
current
out
5parameter with
with
H
the
two
Higgs
doublets.
B
is
a
the
scalar
potential
which
is
¯
1,2
b
b
b
.
(2)
L R
R L
L
= y a ¯i
.
dark
0 L
dimensions
of
mass.
We
assume
that
dark and V are CPs.
The
digher
dimensional
conserving (i.e. B and y1 are both real, and there is no
lude the
Higgs fieldop(which
a 4
2 2
dependent
CP violation
V2HDM+) and
relaxing
V = Vin2HDM
mwe
awill
+comment
a0 +on
Vport
,
ng electroweak
syma
0
hich
then gets a vacuum
ex0
The mediator
couples
to
the
SM
via
the
Higgs
2
4
this assumption in Sec II B. In this case, a0 does not
plying
mediator
which
tioncurrent
ofa L
an can
m
e↵ is not
†the most general CP-conserving
develop
a
VEV.
We
write
the scalar potential
which
Vport = iBa
+ h.c.
0 H1 H2is
2HDM potential as
scalar-scalar interaction be-
gsional
the “Higgs
portal” operator with H1,2 the two
op✓ Higgs1doublets.
◆2
✓B is a parameter
◆2
w
2
2
bHiggs
.
(2)
L
v1 2
2
4v2
†
† a
doublet, since it is a dimensions
of2HDM
mass.
We
assume
V
are
V2HDM
=
+a0that
HL
H
V
=
V
+
m
+
V
1 H1 H
1
2+
2 a0and
dark
weak
sym2
port
a
tal has been well explored in
2 0
2
2
4
conserving (i.e.
B and y◆ are
both real,
and there i
✓
✓
◆
Higgs
field
(which
2
n is
its not
connection
to
DM
[14].
an
2
2
v
v
†
†
1 and we† will comment
2
in
V
)
on
relax
2HDM
Higgs sector
+
H
H
+
H
H
(6)
nexpand
gets athe
vacuum
ex- of CP violation
Vport3 = iBa
H
H
+
h.c.
1
2
1
0 21 2 2
2 case, a does
this
assumption
in
Sec
II
B.
In
this
dmediator
doublet, which
has
enough
h⇣
⌘⇣
⌘ ⇣
⌘⇣
⌘i0
which can
†
†
†
†
w forIpek,
a pseudoscalar
mix develop
We
write
the
most
general
CP-conserv
H
H
H
H
H
H
H
McKeen, toNelson
‘14 +a VEV.
4
1 1
2 2
1 2
2 H1
ator. In the presencewith
of CPH
two
param
2HDM
potential
as ⌘ doublets.
h ⇣ Higgs
i2
h B⇣ is a ⌘i
1,2 the
2
(2)
v
v
calar
interaction
be1 2
†
†
uce a pseudoscalar-scalar cou+ 5 Re H1 H2
+ 6 Im H1 H2
,
of mass.
that
L
and
V
2◆
iggsitportal”
operator
dark
ver
is puzzling
whydimensions
a new
✓ We assume
✓
◆
2 2
2 2
v1are both real,
v2 t
uplings
notitalso
have
†
†
oublet,would
since
isconserving
a a Vwith (i.e.
B
and
y
and
HWe
+ 2 annihilon)
all= i real.
also imposed
aHZ22Hsymmetry
ld
(which
2HDM
1 no
1
2
1 Hhave
fermion.
Including
two
Higgs
+
harder
hierarchy
problem
+
sannihilon
(scalar
been well exploredCP
in violation
22 !
2 o
under
which
H
!
H
and
H
H
to
suppress
flavorin
V
)
and
we
will
comment
1
1
2
an approximate
acuum
ex- symmetry of
✓ 2HDM ◆ ✓
◆
nnection to DM [14].
2
GC SIGNALS OF XDM
MODELS
Liu, NW, Xue in prep
Easy to write down
!
natural consequence of XDM
explanation of 3.5 keV line
!
determination of baroquocity left to
the reader
6
DarkPhoton Total
4⇡
4µ
4e
s
2.0
E2 J(E)[GeV cm
1.0
0.5
0.0
2.5
2.0
m = 9.0 GeV
m = 0.7 GeV
1.5
1.0
0.5
0.0
0.5
0.5
1.0
10
DarkPhoton Total
4⇡
4µ
4e
2
m = 7.0 GeV
m = 0.5 GeV
1.5
6
3.0
1
2.5
2
s
1
sr 1]
3.0
E2 J(E)[GeV cm
3.5 ⇥10
sr 1]
3.5 ⇥10
1
100
101
102
E [GeV]
(see talk by J. Shelton for more)
1.0
10
1
100
101
102
E [GeV]
Liu, NW, Xue in prep
CONCLUSIONS
•
Interesting new signal at 3.55 keV could be explained easily by excitations
•
XDM models make different predictions from decaying models so testable
•
Possible SIDM implications from “long range” forces
•
Simple (in words) models for the GC excess are often not simple in
practice
•
XDM setup provides simple explanation for the GC excess as natural
consequence of model -> consequences for many other searches for BSM
physics