You are here


Content owned by venu.kalari

Instrument capability

GHOST performs echelle spectroscopy between 363 and 950nm, for a single target with spectral resolution (R) greater than 75,000 (in the high resolution mode), and for either one or two targets (separated by at least 102 arcsec) with R of around 50,000 (the standard resolution mode). A description of the capabilities of each mode, and the instrument, follows.

  • Standard resolution mode allows users to obtain single or dual target spectra with resolving power greater than 50,000, depending on the wavelength in question. A series of hexagonal microlens form an input light aperture subtending 1.2 arcsec at the maximum, are sliced into seven constituent microlens increasing the resolution by 3. Dual targets must be separated by at least 102 arcsec, and lie within a circular area of 7.34 arcmin. While sky subtraction is afforded by a dedicated sky IFU, located 3.3 arcsec away in the IFU1 probe arm, one can also adopt the second standard resolution IFU for better sky subtraction, for e.g. in crowded fields, or extended objects. The instrument may be rotated at any position angle.
    This mode is recommended for most science cases needing high resolution, but also the increased throughput if observing faint targets (V-band ≥ 19 mag), or needing the sky subtraction for crowded fields.
  • High resolution mode will allow users to obtain a single spectra with resolving power greater than 75,000, depending on the wavelength in question. The maximum width of the input aperture remains the same as the standard resolution slit, however the final image is sliced by a factor of 5, providing similar resolution to a slit of width of 0.24 arcsec. Dedicated sky fibres located 3 arcsec away provide sky subtraction. This mode is recommended for science cases that desire the extra gain in resolution, at the loss of throughput, particularly at the blue end. Faint targets (with V-band magnitudes > 16 mag) are not recommended for this mode.

The following mode is not yet commissioned and is not ready to be offered to the community. We are working on making them available in future calls. 

  • Enhanced Radial Velocity Precision mode will allow users to obtain high resolution spectra, along with a simultaneous ThXe calibration source, allowing users to reach radial velocity precisions of few tenths m/s. This mode is under active development, and is not currently offered.

Field of view and slit unit

The Cassegrain unit is mounted on the telescope. It contains the two robotic probe arms each carrying one IFU head, which in turn contain 2 or 3 individual IFUs. From the IFUs a 32m fiber cable runs to the spectrograph in the pier laboratory. The total field of view covered by the cassegrain unit probe arms is 7.34 arcmin, with each covering a semicircle of diameter 3.7 arcmin. The minimum separation between dual targets is 102 arcsec.

Field of view of each GHOST probe arm, with probe arm holding IFU 1 (ended by the blue line) pointing towards the east, and IFU2 (green line) pointing towards the west without any position angle. Each IFU covers an a semicircular area with a minimal overlap, and together allow the IFUs to patrol an area of 7.34 arcmins without any vignetting.

The positions, and metrology of each individual probe arm is shown below. Each standard resolution IFU subtends 0.94 arcsec2 on the sky, with a dedicated sky IFU subtending 0.4 arcsec2. The high resolution IFU covers a total aperture of 0.92 arcsec2, with the high resolution sky IFU having an area of 0.34 arcsec2. The IFU areas are not circularly uniform, but are best approximated by the hexagonal area displayed. The position of the dedicated sky IFUs can be seen in the Gemini OT. Each probe arm also has a dedicated atmospheric dispersion corrector (ADC), which are tuned to the conditions at Cerro Pachon. These provide corrections for airmasses up to 2, allowing significantly better blue throughput at lower airmasses.

Labelled microlenses of the two probe arms of IFU1 (left), and IFU2 (right), as seen in the instrument focal plane, with the distance translated to the on-sky scale. Science microlenses are given in black, guide microlenses in red, and sky microlenses in blue. The labelled microlenses given can be corresponded to the psuedo-slit image, and the echelleogram shown latter in these webpages. Guide microlenses are not labelled for clarity.

Since the focal plane of the unit does not allow for movement along the applicate axis, the telescope software corrects for loss of focus across the focal plane, by defocussing each IFU based on distance from the centre of the focal plane. This leads to small slit lossses (typically less than a few percent) when compared to the centre. Due to this effect, it is preferred that single target objects are observed at the centre of the focal plane whenever possible, and dual targets are best observed at radially symmetric positions from the centre.

Position of sky IFUs (blue) when compared to SRIFU1 position. For some objects cases, if sky subtraction is desired (extended, or crowded objects as in this case where the sky fibers fall on a target) it is good to use SRIFU2 to allow for sky subtraction, with the benefits being outweighed against the small loss in throughput when moving away from the focal plane centre, or can be avoided by moving the instrument position angle as demonstrated here. In the topmost figure, the dedicated sky IFU in the SR mode falls on a nearby target. This can be avoided by either changing the position angle as demonstrated in the bottom figure, or using SRIFU2 for sky subtraction. The Gemini OT provides accurately the position of the sky IFU. For most objects although, the sky IFU accurately is able to observe a blank region of sky. 

The integrated FWHM (full width at half maximum) of each IFU is measured when light is passed from the fibre cable to a slit unit, containing two broadband filters having transmissions between 420-600nm, and 600-760nm. This allows measurement on object of the full-width half maximum, which is corrected for zenith distance and wavelength to determine the IQ percentile bin during night time observations. Given the fixed limited aperture, slit losses become significant at IQ greater than IQ85, and poor seeing spectroscopy is mostly viable only for bright objects. The psuedo-slit is essential for obtaining radial information which is scrambled during fiber transmission.

The final image from the IFUs before passing to the slit unit is sliced by the width of each individual microlens, thereby increasing the spectral resolution by a factor of 3 and 5 for the standard and high resolution cases respectively. Each microlens in the probe arm can be identified in the final slit image, and are shown below. The final image formed at the psuedo-slit is in principle not horizontally aligned, but titled by around 9 degrees, and this involves a further correction accounted for in the data reduction.

Top: Cutout of dual IFU (SRIFU1 and SRIFU2 following the instrument nomenclature) standard resolution slit unit flat, taken in 2x2 binning in the standard read mode, where the pixel size is binned. The reconstructed slit with microlenses arranged horizantally is seen. From the left to the right seen are the seven IFU1 microlenses, three sky microlenses, and seven IFU2 microlenses. The top reconstructed slit is passed through the blue filter; while the bottom one is via the red filter. Bottom: Cutout of the single IFU high resolution slit unit flat, taken in 2x2 binning in standard read mode, in binned pixel mode. The slits are reversed for display purposes, and the 19 IFU1 microlenses, and the seven sky microlenses are seen. The final internal arc lamp microlens, number 62 lies after microlens 61 but is separated by an empty microlens.

The final spectra preserves the arrangement of the slit, and is demonstrated here with each individual fiber marked. The final spectral format, and the detector properties can be found in the data reduction and components webpages.

Spectral Range and Resolution

The full GHOST spectral wavelength coverage is 350-1,030nm, but the useful range is 363-950nm. This is limited by not only by the throughput of the detector and also the precision of the wavelength calibration at the bluest wavelengths, but by the range within which the instrument meets the resolution and throughput specifications. GHOST has only one moving part, the slit mask positioner (essentially switching between standard, and high resolution slits), and rests inside an actively controlled and monitored thermal and pressure enclosure. Shown below are spectral orders and the respective central wavelength in echelleogram format in both the red, and blue detectors.

GHOST echellogram in both the blue (left) and red (detectors), with the order numbers, and certain wavelengths marked, illustrating both the range in wavelength, and the order tilt.

The spectral resolution is calculated based on lines of an observed ThAr lamp spectrum. The average resolution element in SR mode was determined to be 2.8 pixels, with a overall resolving power of R=53000 in blue, but slightly higher in red. In the HR mode, 1.8 pixels is the average per resolution element, overall resulting in a minimum resolving power of at least 80,000. 

Spectral resolution as function of wavelength

Binning Options

GHOST offers 8 different on-chip binning options: 1x1, 1x2, 1x4, 1x8, 2x2, 2x4, 2x8, 4x4 (spectral x spatial). While the ideal binning option may differ for different science cases, Table 1 (below) lists the recommended binning options for the major operational modes of GHOST. Note that 8x binning in the spatial direction is only recommended for single-object modes, as extreme spatial binning can merge objects.

Table 1: Recommended binning for main GHOST operational modes
Operational mode Resolution mode On-chip binning Notes
Dual target Standard resolution 2x2
Dual target, faint Standard resolution 2x4
Single target, faint Standard resolution 2x8* *Only suitable for single object since extreme spatial binning merges objects. Trade-off of read noise versus extra sky in object pixels
Single target High resolution 1x2
Single target, faint High resolution 1x4


Example of the throughput calculated for a blue and a red exposure of LTT3218 on January 29 2023. Individual orders are plotted, as are black lines that trace the peak throughput in each order. The y-axis displays throughput, and the x-axis displays wavelength in Angstroms.

The throughput is depending on the atmospheric conditions and a change in just 0.1 arcsec in the FWHM of a stellar image can result in additional slit losses of order 10%


Magnitude corresponding to SNR=30 per resolution element as a function of wavelength for the December 8 2022 observation of LTT2415. Only values at the order centres are relevant. Black points indicate requirements, where the nearest order center should meet or exceed this value. Top panel shows the entire GHOST wavelength range. Bottom panel shows a zoom in at blue wavelengths. In both plots, the y-axis displays the limiting magnitude for SNR=30/resolution element in 1 hour, and the x-axis displays the wavelength in Angstroms.

GHOST achieves a sensitivity of AB mag = 17.5 (450nm), 15.3 (363nm), 16.1 (375nm), 17.0 (550nm), 17.2 (900nm) and 16.2 (950nm) in a 1 hour observation for 30 sigma per resolution element in standard resolution mode in dark time (SB50), clear sky conditions (CC50) and nominal image quality (IQ70).

Guiding Options

The guiding is performed using the Peripheral Wavefront Sensors (PWFS2), with additional fine guiding provided by the instrument microlenses. In general, there must be a Peripheral wave front sensor guide star within 6.75 arcmins for two IFUs, (and for a single target within 9.85 arcmins). This is used to acquire the base position of the focal plane, and center the science IFUs.

The guide fibers of the instrument are used to center before an exposure, correcting for small probe-mapping errors, or flexure with the telescope guiding. This guiding, along with P2WFS is monitored over the duration of the exposure, providing small corrections to the positioners to keep IFU well centered. The probe mapping is currently precise enough that most targets, across the focal plane are captured in both IFUs within pointing errors of the P2WFS (0.2 arcsecs).