Direction-detection spectrometer concepts the CCAT

Matt Bradford + others24 October 2006, in progress

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 2

850 micron counts (stolen from A. Benson talk)

Recall:modified IMF and star formation timescale included to reproduce 850 micron counts

Redshift Distribution from GALFORM model -- similar to Chapman

350 um window13% of sources

450 um window:21% of sources

Models provide approach to CCAT population z distribution: Apply to C+

350 & 450 microns window are likely to access 31% of the 850 micron population in C+

C+ 158 microns redshifted for CCAT spectrograph

Recall: a 1 mJy 850 micron source is a ULIRG, independent of redshift

Source Countsredshift

Arp 220 L 1.12E+12 0.1 0.5 1 1.5 1.83 2 3 4.3Arp 220 D 0.077 d_L (Gpc)

Use 850 microns to get started Arp 220 flux 6.00E-12 0.454 2.82 6.634 11 14.095 15.7 25.841 39.81( values taken from Benson presentation) 850 restframe 772.7 ## 425.0 340.0 300.4 ### 212.5 160.4

nu Fnu / Fbol 2.00E-02 2.00E-01Arp220 nu Fnu 2.61E-14 1.20E-13 5.64E-13 1.27E-12

log Snu Snu log N>Snu N>Snu Power law nu Fnu luminosities of galaxies below, assuming Arp220 like spectrum-1 0.10 4.85 7.1E+04 3.5269E-19 1.12E+11 1.10E+11 7.89E+10 8.31E+10

-0.5 0.32 4.53 3.4E+04 -0.05929 1.1153E-18 3.55E+11 3.49E+11 2.49E+11 2.63E+110 1.00 4.07 1.2E+04 -0.09301 3.5269E-18 1.12E+12 1.10E+12 7.89E+11 8.31E+11

0.5 3.16 3.15 1.4E+03 -0.22257 1.1153E-17 3.55E+12 3.49E+12 2.49E+12 2.63E+120.75 5.62 2.55 3.5E+02 -0.36708 1.9833E-17 6.32E+12 6.20E+12 4.44E+12 4.68E+12

1 10.00 1.95 8.9E+01 -0.46602 3.5269E-17 1.12E+13 1.10E+13 7.89E+12 8.31E+121.25 17.78 1 1.0E+01 -1.16014 6.2719E-17 2.00E+13 1.96E+13 1.40E+13 1.48E+13

63.1 OI obs 69.4 95 126.2 158 178.573 189 252.4 334.43158 CII obs 174 237 316 395 447.14 474 632 837.4

Line FractionULIRG line flux 1.00E-03 ### ## 7.33E-19 ### 1.62E-19 ### 4.83E-20 2.04E-20

CCAT spectroscopic sensitivities: 5 sig, 1h W/m2: 1.8e-19 (350)1.2e-19 (450)

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 5

Options for far-IR through mm spectrometers Grating spectrometer is the best choice for point sources

1st order octave of instantaneous bandwidth Good efficiency But only moderate resolution

Fabry-Perot naturally accommodates spectral mapping But scanning time results in sensitivity penalty, esp for

searching Fourier transform spectrometer (FTS) couples the full band to a

single detector Sensitivity penalty

Heterodyne receivers provide the highest spectral resolution But suffer from quantum noise NEPQN ~ h [1/2 vs. NEPBG ~ h [n (n+1) 1/2 Also offer limited bandwidth:

10 GHz IF bandwidth at 1 THz gives ~ 100

C+ 158 microns redshifted for CCAT spectrograph

Source densities: 1 mJy (850) source densities: 1.2e4 per square degree -> means one per 65-100 (@ 450,350) CCAT beams

Based on 1 hour sensitivities 1 mJy population is likely to be candidates for spectroscopy

Wideband spectroscopic follow-upFP: 256 x 256 beams

gives 67-100 sources per fieldhave to scan 220 resolution elements

sky (sqdeg) N sources scan time / source350 4.77E-02 560 220 3.92E-01450 7.88E-02 926 253 2.73E-01

Grating slit - 1 X 256 beams randomly positionedN sources scan time/source

350 1.86E-04 2.2 1 4.6E-01450 3.08E-04 3.6 1 2.8E-01

Grating MO -- 20 sources assembled onto slitN sources scan time/source

350 20 1 5.0E-02450 20 1 5.0E-02

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 7

350, 450 m windows w/ R~1000-1500

Examples of submillimeter-wave broadband systems: ZEUS for the JCMT / APEX

Cornell -- Stacey et al.

Grating

Detector Array

LP Filter 1

LP Filter 2

BP Filter Wheel

M2

M3

M4

M6

4He Cold Finger

Entrance Beam

f/12

Scatter Filter

3He Dual Stage Refrigerator

M1

4He Cryostat

M5: Primary

Entrance slit

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 8

A new R~1000 echelle spectrometer for CCATAMULE -- Atacama MUltiband Longslit Echelle

Design: Grating 816 micron pitch

Tilt λi λax BW R sλit #pixosλit

58eg 439 485 9.7% 80054eg 418 462 9.3% 822 5.8 2.0862eg 456 504 11.0% 903

57eg 330 356 7.1% 1100 4.3 1.4463eg 350 377 8.2% 1245

56.5eg 221.3 232.7 5.0% 1646 2.7 0.96

60.5eg 198.7 207.4 4.3% 1920 2.7 0.96

3rorer

4thorer

6thorer

7thorer

Assuming 128 spectral element array-- e.g. 0.86 mm pixels -- f/2.5 spectrometer, slightly oversampled

Angular deviation off the grating 18 deg total.collimator must be oversized by 12 cm !--> 30 cm diameter collimator --> grating 30 cm by 40 cm, to accommodate spatial throughput

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 9

AMULE is large

So grating and collimator large fraction of 1 meter in all dimensions -- 1.5-2 times larger than ZEUS

Reimaging optics size will depend on the size of the slit, but also grows relative to ZEUS:--scales as telescope f# x #of beams: Relative to ZEUS, AMULE will have 8/12 x 128/32 = 2.7 times larger reimaging optics. • Requires 35 cm (+ overhead) window if reimaged from telescope focus inside cryostat (but can be shaped like a slit)

Optics envelope inside cryostat approaching 1 meter in all dimensions.

Large but doable.

30 cm

40 cm

35 cm

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 10

How about an imaging Fabry-Perot(BIG Imaging Fabry-Perot Interferometer (BIFI))

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

SPIFI demonstrates concept, at JCMT & the South Pole

5x5 spatial array, two scanning FPs provide R up to 10,000 at 200-500 microns

60 mK ADR-cooled focal plane

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 11

CO 3-2 M. Dumke et al. 2001,

ALMA will resolve out extended emission in nearby galaxies

CO 7-6 Bradford et al. 2003,

ALMA Primary beam at 810 GHz

And Herschel under-resolves it

Herschel beam at 810 GHz 16x16 array on CCAT

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 12

How about an imaging Fabry-Perot (BIG Imaging Fabry-Perot Interferometer (BIFI))

BIFI will be much larger than SPIFI due to the large throughputLimitation is beam divergence in the high-res FP.Dcol ~ 1.5 λ (R x nbeams)1/2

1-D field size for 20 cm beam

High-order FP spacing(mm) w/ F=60

Order-sorter also requires collimated 2.2cm (or slow) beam

Field size (1-D) driven by 20 cm beam

3 min x 3 min field

wavelength array col. Bm. array spacing d locol220.0 128 15.2 44 18.33 2.23330.0 97 19.9 20 27.50 2.26370.0 78 20.0 16 30.83 2.27430.0 58 20.1 12 35.83 2.28490.0 42 19.4 8 40.83 2.12650.0 25 19.9 5 54.17 2.23850.0 14 19.5 3 70.83 2.251200.0 7 19.4 1.5 100.00 2.25

R=2000 R=10000

CCAT instrumentation workshop: Caltech 15-16 Mar 2005 Matt Bradford 13

BIFI will be much larger than SPIFI due to the huge throughput

8 x 20 cm = 160 cm collimator focus

Full field at f/8:20 cm window!

Collimated beam + overheads:25 cm dia(and etalon must be near pupil)

Etalon spacing is modest: few cm even for 650 m

Faster final focal ratio (2-3) to accommodate large array

Array is as large as 10 cm

Factor of two in all dimensions of the optical train

WaFIRS is an ideal broad-band point source architecture. -- possibility for multi-object spectroscopy

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Currently 4-5 cm delta zper module

But could be smaller, far from limits: -- stiffness of plates -- detector illumination -- feed (f lambda from the telescope)

WaFIRS is an ideal broad-band point source architecture. -- possibility for multi-object spectroscopy

50 cm for> 10 modules