NIRC2 Instrument and Data Format

NIRC2 Instrument and Data Format*

1. Introduction

NIRC2 is a near-infrared instrument designed to take full advantage of the adaptive optics on the Keck II telescope of The W. M. Keck Observatory on Mauna Kea, Hawaii.The Co-PIs are Keith Matthews and Tom Soifer, both of Caltech. The instrument was built by Keith Matthews and engineer Sean Lin of Caltech, with help from James Larkin, Ian McLean, and others at UCLA (detector electronics and related software), and Al Conrad, Bob Goodrich, and Allan Honey at Keck Observatory (software). Support in Waimea was provided by Jim Bell, Randy Campbell, and Drew Medeiros. It was delivered to Hawaii in the spring of 2001, seeing first light in the summer of 2001.

NIRC2 is positioned behind the AO bench on the Left Nasmyth Platform of Keck II. The instrument operates from 1 to 5 µm, providing three selectable cameras to cover image sizes between 10"x10" and 40"x40". Two filter wheels with 18 positions each provide a variety of filters and/or grisms, while a focal plane mechanism provides slits and occulting spots for coronography. A dedicated slide carries larger grisms for spectroscopy. Six selectable pupil masks are available to reduce background noise sources; four of these rotate in concert with the telescope pupil and one is specific to spectroscopy. The detector is a 1024x1024 Aladdin-3 InSb array with four-quadrant readout into 32 channels.

2. Optical path

The light provided by the AO system enters the NIRC2 dewar through a CaF2 window. It then passes through two preslit blades which provide the first level of baffling. These movable blades change position depending on which camera is in use and whether or not a slit is in the beam. After the preslits the light is collimated. The beam comes to a focus where there are two slide mechanisms. The slit slide carries several slits of varying widths. A mask slide beneath it selects which slit provides light to the rest of the instrument. The slit slide also contains a set of coronographic spots. A third use of the slit and mask slides is to provide a focal plane baffle for imaging through the three cameras.

Beneath the slit is a wheel which provides a selection of pupil masks, including one for spectroscopy, and two rotating masks designed to block light from the spiders as well as the secondary obstruction. Next comes two filter wheels with 18 positions each. Below the filter wheels is a relatively slow shutter, which can be used to close off external light and may also be used to stop an exposure if, for example, the AO system loses lock on a target. The beam next passes through a grism slide, with three slots plus an open position.

Last in the optical path is one of three selectable cameras. These have nominal scales of 0.01, 0.02, and 0.04 arcsec/pixel, matching the expected diffraction performance of the telescope and AO system at different wavelengths (J, K, and M bands, roughly). The cameras focus onto the InSb detector, described in more detail below.

3. Mechanical description

All moving parts in NIRC2 are inside the dewar. The filter wheels and pupil mask are on rotating wheels, while other mechanisms (preslits, slits, coronagraphic spots, grisms, the shutter, and the cameras) are on slide mechanisms.

Cooling is provided by a closed-cycle system similar to that installed on NIRSPEC. Two cold heads provide three stages of cooling, controllable via software. Temperature sensors are provided on the detector head, optical bench, cold shield, and other places in the dewar.

4. Detector and Electronics

The detector is a 1024x1024 Aladdin-3 array with 27 µm pixels. The detector QE is ~ 80% at 1.7 microns.

The electronics provide three readout modes: single sample, CDS (correlated double sampling), and MCDS (multiple correlated double sampling) modes. The pixel sampling rate is nominally 250 kHz, but can be changed up to ~350kHz. Each quadrant is read out in eight channels (32 channels total). The gain of the detector is 4 electrons/DN.

The read noise in a CDS readout is 38 electrons. For MCDS mode with up to 64 reads, readout noise drops as the square root of the number of reads. Beyond that the noise continues to drop, although not as rapidly as the square root. It continues to drop to 512 reads, beyond which it has not been tested. Section 6.2 shows the read noise as a function of number of MCDS reads.

The detector is linear to about 10,000 DN. This is the result of a compromise between linearity and dark current.

5. Observation modes

NIRC2 has three basic modes.

5.1 Imaging

Narrow- and broad-band filters are provided. The filter complement is given in the filter page, together with sensitivity and background values to aid in signal-to-noise estimates.

Three different cameras provide scales of 0.01, 0.02, and 0.04 arcsec/pixel. These provide good matches to the Keck diffraction size at J, K, and M bands.

The main concerns in imaging mode are image scale or field of view, filters, exposure times, and dither patterns. The "camera" command is used to select one of the three cameras in NIRC2.

Table 1: Camera Options

Camera Name(s)	Scale (arcsec/pixel)	Field of view (arcsec x arcsec)	Band in which diffraction size matches 2 pixels
wide cam04 0.04	0.04	40x40	M
medium cam02 0.02	0.02	20x20	K
narrow cam01 0.01	0.01	10x10	J

Note that the camera command will take different names; it might be easier to remember the pixel scale rather than the relative field size, for example.

Also note that the medium-field camera has significant field curvature, hence image quality across the entire field suffers. It is possible to achieve high image quality for point sources, however.

For imaging, when a new camera is selected, the appropriate masks in the slit and slitmask slides are automatically put into place.

5.2 Coronagraphy

Occulting spots can be placed in the focal plane using the slit slide. The spots are opaque areas deposited on a quartz substrate. The spot transmissions are not zero, but have been measured to be 0.12% in both J and H, and 0.22% in K. (These numbers are relatively uncertain). A variety of spot sizes are available. The following lists their names, with the number after "corona" being the spot size in milliarcseconds:

corona100
corona150
corona200
corona300
corona400

corona600
corona800
corona1000
corona1500
corona2000

5.3 Spectroscopy

Grisms may be installed either in the filter wheel or in a dedicated grism slide. Currently there are two grisms, both in the grism slide. The tables below list various useful parameters for NIRC-2 spectroscopy. The columns are:

Camera
Filter
Order
Central wavelength of the NIRC-2 filter (microns)
Dispersion (Angstroms/pixel; remember that 1 Angstrom is 0.0001 microns)
Wavelength range of the filter (microns)
The number of pixels needed to cover the entire wavelength range. Bold red letters indicate that the entire filter passband will fit on the detector.
The resolving power per pixel. If you use a three-pixel wide slit, e.g., your resolving power would be 1/3 that listed.

Table 2: LOWRES grism

Camera	Filter	Order	Central wavelength (microns)	Dispersion (Ang./pix)	Wavelength range of filter (microns)	# pixels to cover range	Resolving power per pixel
Narrow	J	6	1.248	1.092	0.163	1490	11,430
	J	5	1.248	1.308	0.163	1250	9,540
	H	4	1.633	1.628	0.300	1840	10,030
	K	3	2.196	2.144	0.336	1570	10,240
	Kp	3	2.124	2.144	0.336	1640	9,910
Medium	J	6	1.248	2.190	0.163	740	5,700
	J	5	1.248	2.616	0.163	620	4,770
	H	4	1.633	3.255	0.300	920	5,020
	K	3	2.196	4.113	0.336	820	5,340
	Kp	3	2.124	4.113	0.336	850	5,160
Wide	J	6	1.248	4.373	0.163	370	2,850
	J	5	1.248	5.236	0.163	310	2,380
	H	4	1.633	6.519	0.300	460	2,500
	K	3	2.196	8.226	0.336	410	2,670
	Kp	3	2.124	8.226	0.336	430	2,580

Table 3: MEDRES grism

Camera	Filter	Order	Central wavelength (microns)	Dispersion (Ang./pix)	Wavelength range of filter (microns)	# pixels to cover range	Resolving power per pixel
Narrow	J	6	1.248	0.644	0.163	2530	19,380
	J	5	1.248	0.704	0.163	2320	17,730
	H	4	1.633	0.929	0.300	3230	17,580
	K	3	2.196	1.267	0.336	2650	17,330
	Kp	3	2.124	1.267	0.336	2770	16,760
Medium	J	6	1.248	1.288	0.163	1270	9,690
	J	5	1.248	1.407	0.163	1160	8,870
	H	4	1.633	1.858	0.300	1610	8,790
	K	3	2.196	2.534	0.336	1330	8,670
	Kp	3	2.124	2.534	0.336	1390	8,380
Wide	J	6	1.248	2.576	0.163	630	4,840
	J	5	1.248	2.797	0.163	580	4,460
	H	4	1.633	3.420	0.300	880	4,770
	K	3	2.196	5.078	0.336	660	4,320
	Kp	3	2.124	5.078	0.336	690	4,180

Slits are also available in a variety of widths. A list is given below, where the number after "slit" gives the width in milliarcseconds.

slit10
slit20
slit30
slit40

slit60
slit80
slit120
slit160

Note that since the slits and coronagraphic spots are on the same slide, it is not possible to do coronagraphic spectroscopy.

6. Observing with NIRC2

6.1 Sensitivities/backgrounds

Below is a table listing all filters available with NIRC2, including zeropoint and sky estimates (when measured), and the maximum exposure time expected in the narrow-field camera (before the sky saturates). See the NIRC2 webpage for the latest suite of filters.

Table 4: Filters, sensitivities, and background rates
(Clicking on a filter name will bring up a plot of that filter's transmission curve.)

Filter	Central wavelength (µm)	Bandpass width (µm)	Cut-On Wavelength (µm)	Cut-Off Wavelength (µm)	Photometric zeropoint (Note: This is for Strehl = 1!!)	Sky (mag./ sq. arcsec)	T(max) (sec)
Broad Band
z	1.0311	0.0481	1.007	1.0551
Y	1.0180	0.0996	0.9682	1.0678
J	1.248	0.163	1.166	1.330	25.35	14.9	10,000
H	1.633	0.296	1.485	1.781	25.44	13.6	2750
K	2.196	0.336	2.028	2.364	24.63	12.6	630
Ks	2.146	0.311	1.991	2.302	24.53	12.2	630
Kp	2.124	0.351	1.948	2.299	24.74	12.2	630
Lw	3.5197	1.3216	2.8589	4.1805
Lp	3.776	0.700	3.426	4.126	23.2	2.91	0.27
Ms	4.670	0.241	4.549	4.790	21.2	-0.12	0.14
Narrow Band
He I A	1.0847	0.0182	1.0756	1.0938
Pa_gamma	1.096	0.016	1.088	1.104
Jcont	1.2132	0.0198	1.2033	1.2231
Pa Beta	1.2903	0.0193	1.2807	1.3000
Hcont	1.5804	0.0232	1.5688	1.5920
CH₄S	1.5923	0.1257	1.5295	1.6552
FeII	1.6455	0.0256	1.6327	1.6583
CH₄L	1.6809	0.1368	1.6125	1.7493
He I B	2.0563	0.0326	2.0400	2.0726
Br_gamma	2.1686	0.0326	2.1523	2.1849
H2 v=1-0	2.1281	0.0342	2.1112	2.1452
H2 v=2-1	2.2622	0.0388	2.2428	2.2816
Kcont	2.2706	0.0296	2.2558	2.2854
CO	2.2891	0.0267	2.2757	2.3024
H₂O	3.0629	0.1549	2.9855	3.1404
PAH	3.2904	0.0555	3.2627	3.3182
Br_alpha_cont	3.987	0.069	3.952	4.021
Br_alpha	4.052	0.068	4.018	4.086

The NIRC2 Zero point values were updated after photometric tests in April 2004.

This table shows the zero-point, Z, and sky brightness, S, through various NIRC2 filters, combined with the narrow camera and the circumscribed pupil (the "open" position of the pupil wheel).

Z is the magnitude of the star + 2.5 log10 (counts/sec) - 2.5 log10(Strehl). Note that the Strehl ratio will vary according to atmospheic conditions (e.g. seeing), the brightness of the AO lock star, and the distance away from the lock star.

The gain, G, is 4.0 electrons/count.

S is the sky brightness in magnitudes per square arcsec. The sky background in counts per pixel is given by:

counts/sec/pixel = (arcsec/pixel)^2 * 10^[0.4*(Z-S)]

where (arcsec/pixel) is the pixel scale, e.g. 0.040 for the wide camera.

T(max) is the maximum exposure time without saturating the background in the narrow camera.

For different pupil masks:

Inscribed circle, subtract 0.4 from the zeropoints above, subtract 0.5 from the sky backgrounds above.
The other pupil masks have not been measured, but they should lie between these two extremes.

For different cameras:

T(max) should be divided by 4 for the medium camera, and 16 for the wide camera.

Notes:

a) The thermal bands have not been corrected for air mass. This is a future measurement.

b) The grisms have not been measured. A rough efficiency (not including slit losses) is 50% for the grism throughputs (slightly better at H and K).

6.2 Noise sources

Noise sources include noise in the electronics, statistical noise from the background, and statistical noise from the target itself. Which of these noise sources dominates depends on your observing mode. Bright targets will often be dominated by their own statistical noise, while faint targets may be dominated by background noise. In the thermal infrared (beyond ~ 2.5 µm) almost all observations are dominated by background.

Read noise depends on the readout mode of the detector. In CDS mode it is 38 electrons/pixel. In MCDS mode it drops roughly as the square root of the number of reads until around 64 reads, then drops more slowly until at least the 512-read level. A plot of the read noise vs. number of MCDS samples is shown below:

Figure 1. Read noise vs. number of MCDS samples.

Statistical noise from the target and background are simply the square root of the number of electrons recorded from the source.

Dark current is also a function of the number of MCDS samples. This is because every time the detector is read the four on-chip amplifiers glow slightly. This glow shows up as an elevated background and rises in the background in the four corners of the image. The figure below shows the dark current (in DN/sec) as a function of the number of MCDS reads for a 200 sec. exposure. The drop in effective dark currrent above 300 samples is not understood.

Fig. 2. Dark current vs. number of MCDS samples (200 sec exposure).

Input f/ ratio	f/15
Wavelength range	0.9-5.3 microns
Field of view	10x10 arcsec 20x20 arcsec 40x40 arcsec
Pixel scale	0.01 arcsec/pixel 0.02 arcsec/pixel 0.04 arcsec/pixel
Filters	Z, Y, J, H, K, Ks, Kp, Lw, Lp, Ms, HeIA, HeIB, Pa_gamma, Jcont, Pa_beta, Hcont, H2, CH4S, CH4L, Fe II, Kcont, CO, H2O, PAH, Br_alpha, Br_alpha_cont
Pupil mask	two circular fixed, four hexagonal, rotating
Slits	10, 20, 30, 40, 60, 80, 120, and 160 milliarcsec
Coronagraphic spots	circular: 100, 150, 200, 300, 400, 600, 800, 1000, 1500, 2000 milliarcsec
Grisms	medres and lowres grisms (see tables 2 & 3), pupil imaging lens, and clear (imaging mode).
Minimum frame time	0.182 sec, full frame
Detector	1024x1024 InSb Aladdin-3 array, 27 micron pixels 32 simultaneous outputs, grouped in 8 pixels across a row in each of 4 quadrants gain ~ 5 electrons/DN subarray capability 35 K operating temperature
Read noise	38 electrons/read 11.5 DN/sqrt(N) for N samples in MCDS mode, up to N = 64, dropping more slowly beyond that until at least N = 512. (Note that N = 2 corresponds to two reads, or CDS mode.)
Dark current	< 0.1 electrons/pixel/sec
QE	80% at 1.7 microns
Well depth	100,000 electrons (at 1% linearity?)