The table below summarises the NIFS detector properties. Additional important information on the detector array is given below.
|NIFS Detector Basic Parameters
|Rockwell HAWAII-2RG HgCdTe
|2048×2048 pixels (2040×2040 active)
|Physical Pixel Size
|0.043" along slit
|NIFS Detector Read-out Modes
|No. of Samples
The NIFS spectrograph detector is a Rockwell HAWAII-2RG (H2RG) device with 2048x2048 18μm pixels. The outer four pixels on each side are not illuminated, and so act as reference pixels. This leaves an active area of 2040x2040 pixels.
The HAWAII-2RG detector is sensitive to light out to 2.6μm, and uses a HgCdTe detector layer. It uses four output amplifiers that simultaneously read out 512x2048 pixels in around 5 seconds. Rather than using a shutter to control the duration of a given exposure, the recorded image is actually the difference in signal between two read-outs of the detector: one at the start of the integration, and one at the end. For this reason, the minimum permitted exposure time is determined by the read-out time.
There are several settings used to read the NIFS detector, depending on the source brightness and noise characteristics. Unlike charge-coupled devices (CCDs) commonly used at optical wavelengths, the charge stored on a particular pixel can be read multiple times without destroying the charge. Each read of the detector has some uncertainty (read noise), which can be reduced by averaging over several reads (a technique known as 'Fowler sampling', Fowler & Gatley 1990, ApJL, 353, L33). The reduction in read noise comes at the cost of larger overheads. Details of the three read-out modes currently implimented for NIFS (Bright, Medium and Faint Object) are given in the table above.
Coadds are not often used with NIFS. There is little overhead advantage, and having separate frames allows cosmic ray removal if needed. It is usually better not to use coadds unless strictly necessary. Note, the coadd function in NIFS results in averaged counts, not summed.
the NIFS detector is sensitive to cosmic rays, which appear as diffuse patterns that can extend over tens of pixels. They typically do not saturate the detector, adding only a few hundred electrons to the pixel signal. However their diffuse nature makes them rather destructive to the data signal. For this reason, multiple shorter exposures are preferred over few longer exposures, as far as signal-to-noise requirements will allow. Exposures longer than 20 minutes should be avoided, which is in any case a similar timescale to the variations in the night-sky background.
Dark Current and Bad Pixels
The dark current level in the NIFS detector is very low and stable, and has been measured as less than 0.01e-/s/pix. Hot pixels and 'debonded' cool pixels affect a few percent of detector pixels. Hot pixels have a characteristic 'plus sign' shape due to capacitance coupling of neighbouring pixels, causing a leakage of the hot pixel charge to its neighbors.
Bad pixels in science data can be corrected by subtracting an exposure of equal integration time, either in the form of an associated sky observation, or a combined stack of daytime dark frames, each having the same integration time as the science exposure. Daytime darks are particularly recommended if the sky emission lines are to be used for wavelength calibration. Daytime darks should be requested by the PI in their Phase II submission.
The NIFS detector suffers from image persistence, which causes saturated images to appear as faint artifacts in subsequent exposures. These ghost images can persist for several hours. This is particularly noticeable when the NIFS grating setting is changed, since bright sky emission lines from the previous setting can remain as artifacts in subsequent exposures.
Care should be taken to avoid saturation of the detector, and to plan observing strategies in ways that reduce the frequency with which the grating position is changed.
Saturation of the NIFS detector occurs at about 48,000 ADU, which will also result in strong persistence effects, affecting subsequent images for up to several hours. Furthermore, the detector response becomes >6% nonlinear at ∼38,000 ADU. It is recommended to keep the counts below ∼35,000 ADU (= 98,000 e-) per coadd. Saturation can be checked using the ITC for a single central IFU element.
Below is a table of central wavelengths and spectral coverage of the four filters available for NIFS. The Gemini filter IDs are given in the table and transmission curves (and electronic transmission data) can be displayed or downloaded. In practice the wavelength coverage of NIFS data also depends on the grating efficiency and the wavelength cut-off of the detector.
|Central Wavelength (microns)
In addition to these filters, two neutral density (ND) filters are available to avoid saturation of bright targets. They are located within the Focal Plane Unit (same mechanism as the occulting discs used for coronagraphy). Their properties are summarized in the following table. The transmission behavior varies little with wavelength.
|ND Filter Name
Note that the ND filters are not included in the current integration time calculator (ITC), so these factors should be taken into account manually when estimating exposure times.
NIFS has four gratings physically mounted in the dewar which are briefly described in the tables below, the first table gives the standard grating properties. It also states the defaultconfigurations of grating choice which is used in conjunction with a dual blocking filter, one ofZ-J, J-H, or H-K for Z and J, H, and K gratings, respectively.
It is possible to tune the central wavelengths of the NIFS gratings to reach the inter-band regions where telluric absorption is worse. The limits on the central wavelength tuning range for each grating are given in the second table. Delivered central wavelengths are accurate to within 10%.
|Central Wavelength (microns)
|Velocity Resolution (km/s)
|Short Wavelength Limit (microns)
|Long Wavelength Limit (microns)
|Short Wavelength Filter
|Long Wavelength Filter
NIFS offsers ZJHK coronagraphy with 0".2 and 0".5 occulting masks. While an 0".1 occulting mask is available it is not currently offered.