SPT-3G: A Next-Generation Cosmic Microwave Background
Polarization Experiment on the South Pole Telescope
B. A. Benson
a,b,c
, P. A. R. Ade
d
, Z. Ahmed
e,f,g
, S. W. Allen
e,f,g
, K. Arnold
h
,
J. E. Austermann
i
, A. N. Bender
j
, L. E. Bleem
b,k
, J. E. Carlstrom
b,l,m,k,c
, C. L. Chang
k,b,c
,
H. M. Cho
g
, J. F. Cliche
j
, T. M. Crawford
b,c
, A. Cukierman
n
, T. de Haan
j
, M. A. Dobbs
j,o
,
D. Dutcher
b,m
, W. Everett
i
, A. Gilbert
j
, N. W. Halverson
i,p
, D. Hanson
j
, N. L. Harrington
n
,
K. Hattori
q
, J. W. Henning
i
, G. C. Hilton
r
, G. P. Holder
j,o
, W. L. Holzapfel
n
, K. D. Irwin
e,f,g
,
R. Keisler
e,f
, L. Knox
s
, D. Kubik
a
, C. L. Kuo
e,f,g
, A. T. Lee
n,t
, E. M. Leitch
b,c
, D. Li
r
,
M. McDonald
u
, S. S. Meyer
b,l,m,c
, J. Montgomery
j
, M. Myers
n
, T. Natoli
b,m
, H. Nguyen
a
,
V. Novosad
v
, S. Padin
w
, Z. Pan
b,m
, J. Pearson
v
, C. L. Reichardt
x,n
, J. E. Ruhl
y
,
B. R. Saliwanchik
y
, G. Simard
j
, G. Smecher
z
, J. T. Sayre
y
, E. Shirokoff
b,c
, A. A. Stark
aa
,
K. Story
b,m
, A. Suzuki
n
, K. L. Thompson
e,f,g
, C. Tucker
d
, K. Vanderlinde
bb,cc
, J. D. Vieira
dd,ee
, A. Vikhlinin
aa
, G. Wang
k
, V. Yefremenko
k
, K. W. Yoon
e,f,g
a
Fermi National Accelerator Laboratory, MS209, P.O. Box 500, Batavia, IL 60510-0500
b
Kavli Institute for Cosmological Physics, University of Chicago, 5640 South Ellis Avenue, Chicago, IL 60637
c
Department of Astronomy and Astrophysics, University of Chicago, 5640 South Ellis Avenue, Chicago, IL 60637
d
School of Physics and Astronomy, Cardiff University, Cardiff CF24 3YB, United Kingdom
e
Kavli Institute for Particle Astrophysics and Cosmology, Stanford University, 452 Lomita Mall, Stanford, CA 94305
f
Department of Physics, Stanford University, 382 Via Pueblo Mall, Stanford, CA 94305
g
SLAC National Accelerator Laboratory, 2575 Sand Hill Road, Menlo Park, CA 94025
h
Department of Physics, University of California, San Diego, CA 92093
i
CASA, Department of Astrophysical and Planetary Sciences, University of Colorado, Boulder, Colorado 80309, USA
j
Department of Physics, McGill University, 3600 Rue University, Montreal, Quebec H3A 2T8, Canada
k
Argonne National Laboratory, High-Energy Physics Division, 9700 S. Cass Avenue, Argonne, IL, USA 60439
l
Enrico Fermi Institute, University of Chicago, 5640 South Ellis Avenue, Chicago, IL 60637
m
Department of Physics, University of Chicago, 5640 South Ellis Avenue, Chicago, IL 60637
n
Department of Physics, University of California, Berkeley, CA 94720
o
Canadian Institute for Advanced Research, CIFAR Program in Cosmology and Gravity, Toronto, ON, M5G 1Z8, Canada
p
Department of Physics, University of Colorado, Boulder, CO 80309
q
High Energy Accelerator Research Organization (KEK), Tsukuba, Ibaraki 305-0801, Japan
r
NIST Quantum Devices Group, 325 Broadway Mailcode 817.03, Boulder, CO, USA 80305
s
Department of Physics, University of California, One Shields Avenue, Davis, CA 95616
t
Physics Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720
u
Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, 77 Massachusetts Avenue,
Cambridge, MA 02139
v
Argonne National Laboratory, Material Science Division, 9700 S. Cass Avenue, Argonne, IL, USA 60439
w
California Institute of Technology, 1200 E. California Blvd., Pasadena, CA 91125
x
School of Physics, University of Melbourne, Parkville, 3010 VIC, Australia
y
Physics Department, Case Western Reserve University, Cleveland, OH 44106
z
Three-Speed Logic, Inc., Vancouver, B.C., V6A 2J8, Canada
aa
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138
bb
Dunlap Institute for Astronomy & Astrophysics, University of Toronto, 50 St George St, Toronto, ON, M5S 3H4, Canada
cc
Department of Astronomy & Astrophysics, University of Toronto, 50 St George St, Toronto, ON, M5S 3H4, Canada
dd
Astronomy Department, University of Illinois, 1002 W. Green Street, Urbana, IL 61801 USA
ee
Department of Physics,University of Illinois, 1110 W. Green Street, Urbana, IL 61801 USA
ABSTRACT
We describe the design of a new polarization sensitive receiver, spt-3g, for the 10-meter South Pole Telescope
(spt). The spt-3g receiver will deliver a factor of ∼20 improvement in mapping speed over the current receiver,
Further author information: (Send correspondence to B.A. Benson)
E-mail: bbenson@kicp.uchicago.edu, Telephone: 1 773 702 6452
Millimeter, Submillimeter, and Far-Infrared Detectors and Instrumentation for Astronomy VII,
edited by Wayne S. Holland, Jonas Zmuidzinas, Proc. of SPIE Vol. 9153, 91531P · © 2014
SPIE · CCC code: 0277-786X/14/$18 · doi: 10.1117/12.2057305
Proc. of SPIE Vol. 9153 91531P-1
spt-pol. The sensitivity of the spt-3g receiver will enable the advance from statistical detection of B-mode
polarization anisotropy power to high signal-to-noise measurements of the individual modes, i.e., maps. This
will lead to precise (∼0.06 eV) constraints on the sum of neutrino masses with the potential to directly address
the neutrino mass hierarchy. It will allow a separation of the lensing and inflationary B-mode power spectra,
improving constraints on the amplitude and shape of the primordial signal, either through spt-3g data alone
or in combination with bicep2/keck, which is observing the same area of sky. The measurement of small-scale
temperature anisotropy will provide new constraints on the epoch of reionization. Additional science from the
spt-3g survey will be significantly enhanced by the synergy with the ongoing optical Dark Energy Survey (des),
including: a 1% constraint on the bias of optical tracers of large-scale structure, a measurement of the differential
Doppler signal from pairs of galaxy clusters that will test General Relativity on ∼200 Mpc scales, and improved
cosmological constraints from the abundance of clusters of galaxies.
Keywords: B-modes, cosmic microwave background, cryogenics, inflation, gravitational lensing, neutrino mass,
optical design, polarization, transition-edge sensors
1. INTRODUCTION
The spt is a 10 meter telescope optimized for sensitive, high-resolution measurements of the CMB anisotropy
and mm-wave sky.
1, 2
The telescope is located at the NSF Amundsen-Scott South Pole station, one of the best
developed sites on Earth for mm-wave observations, with particularly low levels of atmospheric fluctuation power
on degree angular scales.
3, 4
The telescope is an off-axis, classical Gregorian design which gives a wide diffraction-
limited field of view, low scattering, and high efficiency with no blockage of the primary aperture. The current
telescope optics produce a 1
0
FWHM beamwidth at 150 GHz with a conservative illumination of the inner 8
meters of the telescope, and a ∼1 deg
2
diffraction-limited field of view.
5
The spt is designed to modulate the
beams on the sky by slewing the entire telescope at up to 4 deg s
−1
, eliminating the need for a chopping mirror.
The telescope operates largely remotely, with a high observing efficiency.
The spt has thus far been used for two surveys: 1) the completed 2500 deg
2
spt-sz survey
6
(2007-2011), and
2) the ongoing 500 deg
2
spt-pol survey
7
(2012-2015). The spt-sz survey observed 2500 deg
2
of sky with an
unprecedented combination of angular resolution (∼1 arcmin) and depth at mm-wavelengths, achieving a noise
level of approximately 36, 16, and 62 µK-arcmin
∗
at 95, 150, and 220 GHz, respectively .
8
The spt-pol survey
observes at 95 and 150 GHz, with added polarization sensitivity. By the end of 2015, the spt-pol survey is
expected to have observed 500 deg
2
of sky to a depth of 6 µK-arcmin at 150 GHz, a noise level approximately
seven times lower than the 143 GHz Planck first data release. In Figure 1, we show a 30 deg
2
cut-out of a spt-pol
map observed to full survey depth, which illustrates the improvements in resolution and depth of the spt-pol
data in comparison to wmap and Planck. The spt-sz and spt-pol observations have led to significant results
and new discoveries in three main areas: using the SZ effect to discover new galaxy clusters (particularly at high
redshift) ,
9–13
the systematic discovery of strongly lensed high-redshift star forming galaxies ,
14–16
measurements
of the fine-scale CMB temperature anisotropy ,
6, 17–21
and the first detection of the so-called “B modes” in the
polarization of the CMB.
22
The third-generation camera for the spt, spt-3g, will exploit the full power of ground-based CMB observa-
tions. The spt-3g camera will exploit two technological advances to achieve the necessary leap in sensitivity: 1)
an improved wide-field optical design that allows more than twice as many diffraction-limited optical elements
in the focal plane, and 2) multi-chroic pixels that are sensitive to multiple observing bands in a single detector
element. The combination of these two advances will deliver a factor-of-20 improvement in mapping speed over
the already impressive spt-pol camera. The spt-3g survey will observe for four years, from 2016-2019, and
cover 2500 deg
2
: an area equal to the original spt-sz survey but observed at a noise level 10× lower in tempera-
ture. In Section 2, we discuss the scientific motivation for spt-3g. In Section 3, we discuss the instrumentation
development necessary to achieve these goals.
∗
Here K
CMB
refers to equivalent fluctuation in the CMB temperature, i.e., the temperature fluctuation of a 2.73 K
blackbody.
Proc. of SPIE Vol. 9153 91531P-2
WMAP
W-band
30 deg
2
Planck
143 GHz
30 deg
2
SPTpol
150 GHz
30 deg
2
SPTpol
150 GHz
10 deg
2
SPT-CL J2337-5942
(z = 0.775)
SPT-CL J2341-5724
(z = 1.36)
SPT-CL J2329-5831
(z = 0.81)
Figure 1. A typical ∼30 deg
2
field from the spt-pol survey. Top: wmap W -band (Left) and Planck 143 GHz (Right)
data from the same region, high-pass filtered at `∼50, for comparison. The large-scale CMB features are measured with
high fidelity in both the wmap, Planck, and spt-pol maps. Bottom Left: Minimally filtered spt-pol data, showing
degree-scale and larger structure in the primordial CMB as well as small-scale features such as emissive sources and SZ
decrements from galaxy clusters. Bottom Right: Zoomed-in view of spatially filtered spt-pol data indicated by black
square in the bottom left panel. In this single 10 deg
2
region, we indicate three newly spt-discovered clusters at redshift
z > 0.75.
2. SCIENCE GOALS
The next frontier of CMB research is to extract the wealth of cosmological information available from its
polarization, in particular with regards to cosmic inflation and neutrinos .
23, 24
The current generation of CMB
polarization experiments use a variety of experimental approaches. For example, bicep2/keck,
25, 26
a pair of
ground-based instruments taking data at the South Pole, have ∼1-degree resolution and high raw sensitivity in
a single 150 GHz band; abs
27
is an instrument with similar design philosophy currently being deployed in Chile;
and the balloon-borne spider project,
28
with similar angular resolution, but with more observing bands and
drastically reduced atmospheric contamination compared to ground-based observatories, is expected to take its
first flight in the 2014-2015 Austral summer. These instruments are efficiently designed to focus on measuring
the tensor-to-scalar ratio r (and, hence, the energy scale of inflation), but they will have little or no sensitivity
to small-scale temperature and polarization. Planned upgrades and observations with the balloon-borne ebex
29
and ground-based act
30
and polarbear
31
experiments will, by contrast, have sufficient resolution to measure
smaller-scale signals—at the cost of added complexity in instrument and optics design.
The low-resolution, single-band approach was appropriate for a pathfinding mission in the era in which no
B-mode polarization was detected, and indeed this approach may have resulted in the first successful detection
of B modes from inflation .
32
However, if the goal is full characterization of the inflationary and lensing B-mode
signals (and the E-mode spectrum), there are several advantages afforded by a large telescope aperture and
Proc. of SPIE Vol. 9153 91531P-3
Experiment N
bolo
NET
T
Speed
T
NET
P
Speed
P
(µK
√
s) (µK
√
s)
spt-sz 960 22 1.0 - -
spt-pol 1,536 14 2.5 20 1.0
spt-3g 15,234 3.4 43 4.8 17
Table 1. The number of bolometers, sensitivity, and relative mapping speed of spt-sz, spt-pol, and spt-3g. The
sensitivity is quoted as noise-equivalent-temperature (NET) in CMB units for temperature (T) and polarization (P).
multi-band observation. First, the scope of science that can be targeted with a high-resolution, multifrequency
instrument is far broader: CMB lensing (only measurable on small angular scales) promises both significant
improvements in cosmological constraints and an opportunity to correlate tracers of structure with the underly-
ing matter field; fine-scale E-mode polarization can greatly increase science yield from the CMB damping tail;
small-scale temperature anisotropy measurements can provide information about the epoch of reionization, but
only if multiple bands are used to tease apart the SZ and foreground signals; and measurements of galaxy clusters
can inform models of dark energy and gravity (again, only if different signals can be distinguished spectrally).
Second, a high signal-to-noise map of the lensed B modes can be used with the measured E modes to reconstruct
the lensing potential and separate the lensing B-mode signal from the inflationary signal, thus improving the
constraints on r and the shape of the tensor spectrum —a process often referred to as “delensing”.
33, 34
Fi-
nally, many instrumental polarization systematics—particularly those having to do with beam mismatch—that
can contaminate low-` B modes for experiments with large beams, are drastically reduced in high-resolution
measurements .
35
Beyond resolution and frequency coverage, realizing the ambitious goals of exploring the neutrino mass scale,
delensing for inflationary B-mode searches, and exploiting the full scientific yield of small-scale CMB temperature
measurements requires a major leap in observing power. For example, the ability to delens is a strong function
of instrument resolution and map noise, and significant delensing is only possible with a
∼
<
10
0
beam and
∼
<
5
µK-arcmin noise in the B-mode map.
33
Similarly, data from neutrino oscillation experiments require that the
sum of the neutrino masses be Σm
ν
≥ 0.05 eV, although the true value of Σm
ν
will depend on the ordering of
the three masses—the so-called “mass hierarchy”. In the “inverted hierarchy” scenario, there are two neutrinos
with m
ν
≥ 0.05 eV and thus Σm
ν
≥ 0.1 eV. The limit on Σm
ν
from Planck alone is expected to be just above
this threshold; however, the combination of Planck and an experiment with
∼
<
5 µK-arcmin noise (in E and
B, and
√
2 lower in T ) over at least 1000 deg
2
would drive this limit closer to 0.05 eV, a regime in which the
mass hierarchy can be directly addressed. To map ≥1000 deg
2
to this noise level requires an order of magnitude
improvement in mapping speed over spt-pol.
The spt-3g instrument will deliver this with an unprecedented combination of sensitivity and resolution.
With ∼1
0
beams, a combined focal plane sensitivity of ∼4.8 µK
√
s in polarization, and 24-hour access to clean
patches of nearly foreground-free sky from one of the best available sites on Earth for mm-wavelength observa-
tions, spt-3g should achieve a B-mode noise level of ∼3.5µK-arcmin, enabling the production of high signal-
to-noise images of the lensing B modes. Combined with data from Planck and BOSS, spt-3g will achieve
σ(Σm
ν
)∼0.06 eV. Combined with spt-3g’s exquisite E-mode measurement, the lensing B mode measurement
will allow delensing of the primordial gravitational-wave B-mode signal with a factor of 4 reduction in power.
33
This delensing ability will enhance spt-3g’s own constraint on r and naturally complement the keck program,
which is observing the same region of sky. The measurement of small-scale temperature anisotropy from such a
survey will provide exciting new constraints on the epoch of reionization. Additional science from spt-3g will
be significantly enhanced by the synergy with the deep and wide optical survey, des, which will cover the full
spt-3g footprint: 1) The spt-3g CMB lensing data will be used to produce maps of the projected mass between
us and the CMB last scattering surface. These mass maps can be correlated with tracers of large-scale structure,
such as optically selected galaxies, effectively allowing us to “weigh” the galaxies. 2) The differential kinetic SZ
signal from pairs of galaxy clusters identified in des data will provide a unique test of General Relativity on ∼200
Mpc scales. 3) Both the temperature and polarization information will further improve constraints on cosmology
from the spt-3g and des galaxy cluster samples, particularly by sharpening the mass-observable calibration
with CMB-cluster lensing. Finally, as with spt-sz and spt-pol, the data from spt-3g will be released to the
community to enhance its scientific impact.
Proc. of SPIE Vol. 9153 91531P-4
10
2
10
3
L
10
-9
10
-8
10
-7
L
4
C
φφ
L
/
2
π
SPT-3G
Planck
0.0 0.1 0.2 0.3 0.4
2.9
3.0
3.1
3.2
3.3
Figure 2. Left: Projected error bars (per logarithmic bin size of d(ln`=0.1)) on the CMB lensing power spectrum for
spt-3g (red) and Planck (blue). Dashed vertical lines show the scale at which mapping of individual modes will be
possible for Planck (blue) and spt-3g (red). Right: Projected 1σ parameter constraints on Σm
ν
, the sum of the neutrino
masses and N
eff
, the effective number of neutrino species for spt-3g when combined with data from Planck and the
BOSS spectroscopic survey. The addition of the extremely deep CMB polarization and lensing maps from spt-3g will
substantially improve cosmological constraints on neutrino physics, including the sum of the masses and the number of
relativistic species.
In the following sections, we present projected results from spt-3g, including cosmological parameter con-
straints, in each of these areas of research. We assume a four-year survey over 2500 deg
2
and the focal plane
specifications outlined in Section 3.3 and Table 1. Assuming a 25% observing duty cycle (conservative compared
to the 60% efficiency achieved during winter months with spt-sz), this results in predicted map noise levels of
∼3.5µK-arcmin in E and B at 150 GHz (
√
2 lower in T ) and ∼6µK-arcmin in E and B at 95 and 220 GHz.
The choice of observing region size and location (2500 deg
2
roughly covering the footprint of the original spt-sz
survey) is motivated by achievable depth and predicted foreground levels, which drive us to smaller observing
area, and the constraint on Σm
ν
and the synergy with des, which drive us to larger area.
2.1 CMB Lensing
The spt-3g CMB lensing measurements will significantly improve the imaging of matter fluctuations between
us and the CMB surface of last scattering on small angular scales. This will lead to both a detailed measurement
of the lensing power spectrum and a high signal-to-noise map of the projected mass in the Universe. As shown
in Figure 2, the combination of improved map noise and resolution compared to Planck will result in an spt-
3g map of projected matter fluctuations with high signal-to-noise on scales larger than ∼15
0
(`
∼
<
750), and
spt-3g will measure the power spectrum of the lensing potential at high significance out to ` ≈ 5000. This
precise measurement of the growth of structure leads to strong constraints on cosmological parameters. Table 2
shows the combined parameter constraints from CMB lensing and primordial power spectrum constraints (see
Section 2.2), and constraints on one key parameter combination are shown in Figure 2. Significant improvements
over Planck alone are seen in many parameters, particularly in the neutrino sector. spt-3g+Planck will place
stringent constraints on the number of relativistic species at recombination and thus confirm or rule out the
hints of a fourth neutrino species from CMB and direct neutrino measurements.
34
The combined sensitivity to
Σm
ν
will be σ(Σm
ν
) ∼ 0.06 eV, which is roughly six times better than future beta decay experiments such as
KATRIN
36
and comparable to the largest neutrino mass splitting. At this level of precision, spt-3g will either
measure Σm
ν
and determine the mass scale for neutrinos, or will place strong pressure on an inverted neutrino
mass hierarchy. A yet tighter limit is achievable: if we assume the standard number of neutrinos and a perfect
measurement of H
0
, the constraint on Σm
ν
improves to σ(Σm
ν
) = 0.018 eV.
Additionally, the high signal-to-noise projected mass map can be cross-correlated with other probes of large
scale structure to measure the bias of these tracers to better than 1%, providing new clues on the link between
galaxies and dark matter halos. This large area mass map will be extremely useful for comparisons with cosmic
Proc. of SPIE Vol. 9153 91531P-5
