The LSST project at BNL 04-17... · 2007. 4. 16. · DOE/Int’l. Data Management LSSTC NSF/DOE Brookhaven National Laboratory* ... – 2015: LSST first light – Survey 2015 –

The LSST project at BNL

P OrsquoConnorHEP Review

April 17 2007

2

LSST mission and goals

bull LSST is a dedicated 10-year all-sky survey using an 84m-class ground based telescope with three novel featuresndash extremely wide field of view (10 sq deg)ndash fast cadence (new image every 30s)ndash active optics for tight control of image quality (angular resolution limited

by atmosphere)

bull The main science mission of LSST is the physics of dark energy using a suite of techniques primarily gravitational weak lensing

bull The publicly-available LSST database will also enable a variety of astrophysical investigationsndash Milky Way mappingndash transient object discoveryndash solar system inventory

3

Gravitational lensing

Point mass

Background galaxies

Observed image

4

Weak lensing tomography

Observed galaxy shapes are subtly distorted (sheared tangentially) by the inhomogeneous gravitational field of mass concentrations along the line-of-sight between the galaxy and our telescope

5

Figure of merit for a survey telescope

Eacutetendue = AΩtEacutetendue = AΩt

humongous mirror enormous focal plane

SERIAL REGISTER

SERIAL REGISTER

50

0 X

20

00

50

0 X

20

00

40

00

X 2

00

04

00

0 X

20

00

BC

highly parallel readout segmented sensors ASIC electronics

6

Comparison with Keck 10m

Primary mirror effective area Field of view

KeckTelescope

003 sq deg

67 m2

LSST

10 sq deg

32 m2

10m

84m

X = Eacutetendue

2 m2-deg2

320 m2-deg2

7

Comparison with precursor surveys

Stage I and II

Stage III

Stage IV

LSST Dataset will have

4 billion galaxies low statistical uncertainty on cosmic shear

2000 exposuresfield suppress spurious correlations due to atmosphere + optics

six filters accurate redshift determination to z=3

8

The LSST 32 Gpixel camera

PACKAGEDCCD

RAFT TOWER

CRYOSTAT

FILTERSSHUTTERLENSESCCD

TOWERbull 3x3 sub-mosaic of CCDs bull front end electronics bull thermal management components

bull Tower is an autonomous fully-testable 144 Mpixelcamerabull A BNL deliverable

9

LSST project organization

LSST Corp

TelescopeSiteNOAO

NSFPrivate

CameraSLAC

DOEIntrsquol

DataManagement

LSSTCNSFDOE

Brookhaven National Laboratory California Institute of TechnologyColumbia UniversityGoogle IncHarvard UniversityHarvard-Smithsonian Center for AstrophysicsJohns Hopkins UniversityIN2P3Lawrence Livermore National LaboratoryNational Center for Supercomputing ApplicationsOak Ridge National Laboratory

Pennsylvania State UniversityPrinceton UniversityResearch Corporation Stanford Linear Accelerator CenterUniversity of Arizona University of California at DavisUniversity of Illinois at Urbana-ChampaignUniversity of PennsylvaniaUniversity of California at Santa CruzUniversity of TennesseeUniversity of Washington

Collaborating institutions ( = part of Camera Team)

BNL is responsible for sensor and front-end electronics development

BNLRACF participates in simulation and analysisWeak lensing science

10

We have made a detailed model of the dependence of Quantum Efficiency and Point Spread Function on

ndash thicknessndash wavelengthndash temperaturendash resistivityndash bias voltagendash flatness

Experimental verification is in progress on study devices provided by vendors

50 100 150 200

20

25

30

35

40

45

50

Thickness μm

σPS

F μ

m

477 nm870 nm1015 nm

Wavelength

LSST target

LSST acceptable

LSST target

LSST acceptable

at λ

=1μm

PS

F μ

m

P OrsquoC et al Proc SPIE 6276-75 (April 2006)

Sensor thickness optimization study

bull A thick high-resistivity CCD with high internal electric field is critical to achieve LSST goals for high near-IR QE and small PSFbull This technology beyond the present commercial state of the artbull LSST has contracted with several vendors to develop a custom CCD

11

lamp

mono

CCD controller

picoammeters

vacuum gauges

LN2 storage cylinder

Dewar

dark box

CCD characterization lab

12

Results on first LSST prototype

2K x 4K CCD2K x 4K CCD Cosmic tracks in dark imageCosmic tracks in dark image

Surface profileSurface profileResolution test targetResolution test target

13

Precision mechanical design of rafttower

bull Raft assembly requirementsndash All 9 CCDs coplanar to 65μm (peak-valley)ndash Minimum dead area ndash 180K operating temperature

bull 24 rafttowers will be integrated and cold-tested at BNL before installation in camera

Integration toolingIntegration tooling Thermal FEAThermal FEA

S Plate BNL

14

Current and planned level of effort

bull Instrumentation (detector development) ndash 3 scientists 15 FTE + 1FTE engineeringtech supportndash $392K cumulative total funding since FY04

bull Physics (science and detector development) ndash 2 scientists one postdoc 3 FTE (14 from core)ndash recruiting one senior scientist + junior position to lead Astro-cosmo groupndash $100K since FY05

bull Scenariondash $23M RampD proposal submitted to DOE-OHEP 22707 in response to

ldquoDiscovery of the Nature of Dark Energyrdquo announcement ($200K to BNL)ndash Summer rsquo07 issue RFP for prototype sensorsndash CD-1 in FY08 CD-2 in FY09 construction start in FY10ndash FY10-FY12 sensor procurement commission tower assembly facility

ramp to 8 FTE for tower integration and testndash 2015 LSST first lightndash Survey 2015 ndash 2025

BACKUPS

16

LSST Science

bull Computing effort in Physics Department centered around a small but significant cluster of 16 machines and about 20 TB of disk managed by RACF

bull Computing projects underway arendash Operation of prototype LSST pipelines by collaborators at University

of Arizonandash Calibration simulation in collaboration with SLAC and Harvardndash Image processing and weak lensing analysis of ESSENCE data in

collaboration with Stubbs group at Harvardndash Simulation projects

bull Collaboration with Harvard group has given us access to data andexpertise in image processing and analysis research associate resident at Harvard for a semester has provided a direct connection so that we have moved some of the data to our local cluster and are beginning an attempt to carry out a weak lensinganalysis of a nearby cluster in the ESSENCE survey (Abell 168) as a test bed for LSST This was a supernovae survey it presentsmany of the same challenges as LSSTmdashmany images have to be co-added without biasing the PSF the exposures are taken over a relatively long period of time over which the quality of the data varies considerably

Abell 168 in ESSENCE (P Challis CfA)

17

LSST Science

bull Development of hardware systems and analysis software to augment Instrumentation Division tests of CCD sensors being developed for the LSST camera

bull Brookhaven-Columbia collaborationndash How well will LSST determine cosmology S Wang et al Constraining the

evolution of dark energy with a combination of galaxy cluster observablesPhys Rev D 70 123008 (2004) and S Wang et alWeighing Neutrinos with Galaxy Cluster Surveys PRL 95 011302 (2005)

ndash In preparationbull Is Modified Gravity Required by Observation ndash An Empirical Consistency Test of Dark

Energy Models bull High Shear Regions in Weak Lensing Surveys Determine Cosmology

Foreground masses change the apparent shape of background galaxies through gravitational lensing (shear) The fractional area of sky with high shear values can place strong constraints on cosmological parameters such as the evolution of dark energy

bull Journal Club and lectures in cosmology by participants and interested local physicists members of LSST Weak lensing Science Collaboration

bull There is a search for a senior scientist with demonstrated expertise in observational astrophysics

18

Publications ndash Instrumentation

1 J Geary D Figer D K Gilmore P OConnor J Oliver V Radeka C Stubbs P Takas J A Tyson The LSST sensor technologies studies Proc SPIE Vol 6276 627601 High Energy Optical and Infrared Detectors for Astronomy II David A Dorn Andrew D Holland Eds June 2006

2 P OConnor V Radeka D Figer J G Geary D K Gilmore J Oliver C W Stubbs P Z Takacs J A Tyson Study of silicon thickness optimization for LSST Proc SPIE Vol 6276 62761W High Energy Optical and Infrared Detectors for Astronomy II David A Dorn Andrew D Holland Eds June 2006

3 PZ Takacs P OConnor V Radeka G Mahler J Frank J Geary LSST detector module and raft assembly metrology concepts Proc SPIE Vol 6273Optomechanical Technologies for Astronomy Eli Atad-Ettedgui Joseph Antebi Dietrich Lemke Editors 62733Q (Jul 2006)

4 K Gilmore S Kahn M Nordby D Burke P OConnor J Oliver V Radeka T Schalk R Schindler The LSST Camera System Overview Proc SPIE 6269 62690C Ground-based and Airborne Instrumentation for Astronomy Ian S McLean Masanori Iye Eds June 2006

5 OConnor P Figer D Geary J Gilmore K Oliver J Radeka V Stubbs C Takacs P Tyson A amp 2004 LSST Focal Plane and Detector Development AAS 10807

6 P OrsquoConnor J Geary K Gilmore J Oliver V Radeka P Takacs Technology of the LSST Focal Plane submitted to NIM-A

7 P OrsquoConnor V Radeka JG Geary DK Gilmore PZ Takacs Sensor Development for the Large Synoptic Survey Telescope to be published in Proc 7th Intrsquol Image Sensor Workshop

8 V Radeka ldquoCCD and PIN-CMOS Developments for Large Optical Telescopesrdquo BNL 76772-2006CP Proc SNIC Symp SLAC Stanford CA 3-6 April 2006httpwwwslacstanfordedueconfC0604032proceedingshtmtwo

9 V Radeka Z Li P OrsquoConnor PZ Takacs Charge Diffusion PSF in Thick Over-depleted Silicon Sensors presented at 6th Intrsquol Conference on Scientific Optical Imaging Cozumel Mexico Dec 2 2006

19

P5 and DETF recommendations

P5 report to HEPAP 22207

Dark Energy Task Force report to AAAC and HEPAP 906

20

A proposal for the construction of the LSST has just been submitted to the NSF

Unique mission WideFastDeepsurvey

Location Northern Chile(Andean front range 9200ft elev)

Completion date Sept 2015

Total cost $390M (2006USD)

Largest mirror 84m diam

21

bull The largest digital camera ever constructedndash 32 billion pixelsndash 16 m x 3 mndash 2800 kilograms

bull Sensor requirements push the frontiers of current technologyndash High QE across the whole visible bandndash 4-side buttable to efficiently fill focal planendash Tight flatness tolerances to maintain focusndash Highly parallel to enable fast readout

22

Sensors Key RequirementsSensors Key Requirementsndash High QE to 1000nm

bull Thick silicon (gt 75 microm)ndash PSF ltlt 07 arcseconds

bull High internal field in the sensorbull High resistivity silicon substrate (gt 5 kohmmiddotcm)bull High applied voltages (40 - 50 Volts)bull Small pixel size (02 arcseconds = 10 microm)

ndash Fast f12 focal ratiobull Sensor flatness lt 5microm p-vbull Package with piston tip tilt adj to ~1microm

ndash Wide Field of Viewbull ~ 3200 square cm focal planebull gt 200-sensor mosaic (~16 square cm each)bull Industrialized production processes

ndash High throughputbull gt 90 fill factor 4-side buttable package sub-mm gaps

ndash Fast readout (1 - 2 s) bull Segmented sensors (~3200 or 6400 total output ports ) bull 150 connections per sensor

ndash Low read noisebull lt only a few electrons

23

IR sensitivityIR sensitivity

Qua

ntum

Effi

cien

cy

Wavelength nm900 1000 1100

100

80

60

40

20

0

Thickness 150 250μm1007550

ndashPresent-generation CCDs are 15 ndash 40μm thickndashThicker silicon needed for high sensitivity in the IR

24

Expected range

Thic

knes

s μ

m

Temperature K

300

250

200

150

100

50100 140 180 200 260 300

QE at λ=1000nm

25

Sensor thickness comparison

LSSTf12

1rdquo

+ge 30V

conventional CCDf4

1rdquo

+10V

SNAPf11

thickness of Si needed for QE of

1rdquo

+ge100V

10

25

45

(T=173K)

26

Diffusion problem in thick CCDsDiffusion problem in thick CCDs

star image star image

entrancewindow

undepleted Si

depletion edgeE=0

Vsub gt Vdepl

Image of point source broadened by diffusion

Small point spread function

Partially depleted silicon Fully depleted silicon

High electric field in silicon is critical ndash needs high resistivity substrate and high voltage applied to entrance window

27

high-field modifications to PSF estimate

1 2

2e op

kTd q V

σ ⎛ ⎞= ⎜ ⎟⎜ ⎟⎝ ⎠

for low electric field

( )1 2 1 2

0( ) 2 1s

e op s

T EkTvd q V v

μσ ⎛ ⎞ ⎡ ⎤= +⎜ ⎟ ⎢ ⎥⎜ ⎟ ⎣ ⎦⎝ ⎠

velocity saturation for high fields gives

also the diffusion coefficient is field dependentwhich counteracts velocity saturation somewhat

-20 -10 0 10 20

23

45

67

Displacement μm

PSF σ

μ m

E=2kVcm

477nm [g]870 [z]1015 [Y]

-20 -10 0 10 202

34

56

7

Displacement μm

E=5kVcm

-20 -10 0 10 20

23

45

67

Displacement μm

PSF σ

μ m

E=2kVcm

477nm [g]870 [z]1015 [Y]

-20 -10 0 10 202

34

56

7

Displacement μm

E=5kVcm

50 100 150 200

25

1020

5010

0

o

o

o

o

o

|

|

|

|

|

Thickness μm

App

lied

volta

ge

o|

FWHM 75μmFWHM 10μm

5kVcm 2kVcm

Full Depletion

the bottom line is that we need higher substrate voltageto achieve the target PSF at 100μm thickness

15 ndash 35

The biggest contribution to PSF at short wavelengths comes from charge diffusion

req

goal

28

LSST site on Cerro Pachon

29

Rendering of completed observatory

30

The LSST camera

31

tower

cryostatfilter changerand shutter

correctorlenses

Camera components