Instrument Description

MATISSE in the VLTI context

The instrument is currently largely stand-alone from the infrastructure of the VLTI. As such it does not use external fringe-tracking, and only acts as a coherencer, meaning it will only center the entire fringe package after a number of exposures, but does not keep the individual fringes in position. A coherent adding up of indivdual exposures is therefore only possible if the fringes can be centered in each exposure, or if assumptions on the fringe movements are made. The use of GRAVITY as an external fringe tracker for MATISSE is being investigated.

The CIAO adaptive optics system for the UTs is currently not offered for MATISSE.

MATISSE Optical Elements

The following is a general description of the optical path, for more detail and numerical information see the instrument manuals.

On the warm optics bench, the beams pass through two commuting devices that can interchange the input beams 1 vs. 2 and 3 vs. 4, respectively. They are always used, and any observation consists of at least one BCD-IN and one BCD-OUT measurement. Comparing the BCD-IN with the BCD-OUT data enables correction of instrument-level chromatic effects on the measured interferometric phases; the correction of similar telescope-level chromatic effects is obtained by comparing science targets to calibrators.

After the LM- and N-bands have been separated inside the warm optics part of the instrument, the beams enter the LM or N-band cryostat. The internal setup of each cryostat is comparable, so this general description is applicable to both cryostats. In an intermediate focus the beams pass through a spatial filter, either a slit or a pinhole, with defined size of the order of the point-spread function. ESO has chosen a standard setup for those, but expert users in visitor mode can make use of the alternative options.

When the photo-interferometric splitter is inserted into the beam, source photometry and fringes are observed simultaneously (SIPHOT mode), otherwise they have to be taken sequentially (HISENS mode). Currently the only offered option is to observe L-band in SIPHOT and N-band in HISENS, which is the so-called HYBRID mode.

Users experienced with N-band interferometry might wish to use the correlated flux measurement provided by MATISSE instead of the full visibility information. Since a correlated flux measurement does not require that source photometry is obtained, the execution time per OB is shorter, and correlated fluxes can be obtained for fainter sources than full visibility measurements. However, in that case the user must have information about at least the calibrator flux from an external source, and preferably the science target as well. Correlated flux measurements are not possible in the LM-band.

The dispersive optics provides a choice of several resolving powers. The detector pixels oversample the optical resolution. For this reason, the pipeline enables binning from spectral pixel to spectral channels, and the performance values are given per spectral channel. This improves the performance limits without sacrificing any recorded astrophysical information. Binning any further would destroy such information. The oversampling (i.e., binning) factors are as follows:

          R        pixel/DIT       pixel/channel        
LR L-band 34 65 5
MR L-band 506 118 5
HR L-band 959 118 5
LR N-band 30 120 7
HR N-band 218 819 7

The "pixels per DIT" are the number of pixels read in spectral direction within the default DIT values.

Filters to reduce the total background are inserted according to the chosen spectral resolution and wavelength range. The polarizing filters are not available for scientific observations.

Calibrators and Calibration Strategy

MATISSE if offered with two observing sequences, either CAL-SCI (or SCI-CAL) or CAL-SCI-CAL.

Calibrator stars for N-band and L-band can be found, for instance, with the SearchCal tool provided by the JMMC or through the catalogue published by the MATISSE consortium (see tools). Finding a star that is suitable for both bands at the same time can be tricky. Users should make sure already at phase 1, i.e. for proposal submission, that their chosen calibration strategy is possible and suitable calibrators are available. In case no good L+N calibrators (so called "hybrid" calibrators) are available, the user should consider to use the CAL-SCI-CAL sequence with one calibrator for L-band and one for N-band, this is even more important in case of imaging programs aiming at producing images in both L- and N-band.  

The following recommendations should be considered when choosing a calibrator star:

  • Mid-IR fluxes are hard to measure. It is recommended to compute median fluxes in L and N bands using for example the VizieR Photometry viewer provided by CDS-SIMBAD intead of relying on a single source.
  • Good calibrators should follow the Rayleigh-Jeans approximation, i.e., avoid IR excess. L/N-band flux ratio (at 3.3 and 8.5 microns) should best be within a range of 7 to 9.
  • A SIMBAD search on the calibrator should show "star" as object type.


MATISSE is offered in period 112 as stand-alone instrument on all the standard ATs configurations and on the UTs. It is also offered with GRAVITY fringe tracker (GRA4MAT mode) on the ATs, and for the first time on the UTs without chopping. The user should be aware that when using GRA4MAT with the UTs only N-band correlated fluxes are guaranteed, the LM band limiting fluxes will be verified during upcoming commissioning. The limiting fluxes currently provided for LM band and GRA4MAT on the UTs are extrapolations from the ATs. The use of GRA4MAT is recommended as soon as fringe tracking is possible on a MATISSE science target.

Please find below more details of the modes offered.

MATISSE stand-alone

  • Hybrid is the only observing mode offered. The hybrid mode consists of the following two steps: 1. acquisition of fringes in SiPhot mode LM + HiSens mode N; 2. SiPhot with chopping in LM + Photometry in N. Users interested only in closure and differential phases can skip step 2. All the other users are asked to include step 2 (i.e. the additional time needed to record the N-band photometry, see table "Execution Times" and Overheads)
  • L-band observations are offered with low (R=34), medium (R=506), and high (R=959). 
  • M-band observations are offered with low (R = 34) and medium (R = 506) resolution. M-band measurements need the photometry step.
  • N-band observations are offered with low resolution (R=30) and high resolution (R=218).
  • The DIT values are fixed. The chosen DIT for LM-bands enables to observe a spectral window of 0.08 micrometer in high resolution, and 0.16 micrometer in medium resolution, which the user can center following the table below. In visitor mode the user can center the spectral window freely. The central wavelengths available in low spectral resolution in service mode are 3.5 and 4.1 micrometers.
Setup Central wavelength Spectral feature
Medium and high L-band 3.03 Pfund 10-5
  3.05 Water ice
  3.17 C2H2, HCN, CN
  3.3 PAH
  3.4 CH
  3.52 Nano-diamonds
  3.77 C2H2
  3.88 C2H2
  3.95 C2H2
  4.0 SiO
  4.05 Bracket Alpha
Medium M-band 4.65 Pfund 7-5
  4.78 CO


Execution Times

Low L and M-band 20 min 40 min 60 min
Medium L and M-band 20 min 40min 60 min
High and high+* L-band 25 min 50 min 75 min
N-band photometry** +10 minutes +20 minutes +30 minutes

* The High+ mode is only available with GRA4MAT.

** During the N-band photometry exposures, LM-band data (fringes + photometry) are simultaneously recorded with chopping. We strongly reccomend to includes such observations. For details of when such exposures can be skipped, please see the User Manual.

Sensitivity and Errors vs. Observing Conditions


Starting from P105 a new atmospheric turbulence constraint replaces the current seeing constraint. The limiting magnitudes given below are listed for two turbulence regime. However, bad observing conditions do not only diminish the flux. If a science case is critically dependent to achieve the best possible error bars, it is strongly recommended to request good observing conditions regardless of the target brightness. Observations at seeing values worse than 1.2" are not recommended. All flux limits are given in Jansky and they correspond to correlated fluxes (Flux at a given band times Visibility). Details about the commissioning of MATISSE and the derivation of the instrument performances are described in Petrov et al. 2020. Small differences between the paper and the here published data are given by the fact that difference turbulence categories are used. The numbers provided below are the reference to be used for the call for proposal.

The extended AT-array configuration relies on additional optics to increase the delay line stroke and thus sky coverage. For MATISSE, these additional reflections mean that all limiting fluxes below have to be considered as to be 30% worse for observations on the extended configuration.

Absolute measurements

For closure and differential phase as well as visibility measurements the following limits will allow to achieve a precision per spectral channel at least as good as:

  • For visibilities a precision of 0.1
  • For coherent flux (N-band only) of a SNR of 10
  • 4 deg for the differential phase
  • 5 deg for the closure phase

The limit in low resolution is mostly caused by uncertainty of the photometry and independent of the binning. This limits the precision of the visibilities, not of the detection of fringes themselves. For specific science cases, usable information such as variation of some measurements with wavelength or baseline can be obtained at substantially lower magnitudes. Please contact the instrument scientist for such limits. 

Setup AT UT

T≤ 30%

Seeing  0.8"

τ0 > 4.1 ms

T≤ 70%

Seeing  1.15"

τ0 > 2.2 ms

T≤ 30%

Seeing  0.8"

τ0 > 4.1 ms

T≤ 70%

Seeing  1.15"

τ0 > 2.2 ms

Low L-band  1 Jy 1.5 Jy 0.06 Jy 0.08 Jy
Low M-band  2 Jy 5 Jy 0.5 Jy 0.7 Jy
Medium L-band 8.5 Jy 10.5 Jy 0.6 Jy 0.7 Jy
Medium M-band  13 Jy 16 Jy 1.3 Jy 1.6 Jy
Low N-band* 17 Jy 17 Jy 0.9 Jy 0.9 Jy

* Values are for the N1 sub-band, centered on 8.5microns. If a target is exremely red or quality of masurement is critical at longer wavelengths, see text. Also, since the quality of photometriy a limiting factor for the N-band visbility, these values are for the total flux, not the correlated fluxes as for the other entries.

It was noticed back in Aug 2020 that stand-alone N-band measurements observations of faint sources (N bands fluxes fainter than about 6-8 Jy at 8 and 11 micron respectively) are affected by a bias introduced by the piston correction. Such bias is removed if the GRA4MAT mode is used.

The N-band can be subdivided into N1 (8.5+-0.4micron), N2 (10.5+-0.5 micron), and N3 (11.5+-0.5 micron). If the object is extremely red, or the above defined precision is required in N2 or N3 bands, the following limits apply: For ATs: N2 brighter than 32 Jy, and N3 brighter than 45 Jy. For UTs: N2 brighter than 1.5 Jy, and N3 brighter than 2 Jy.

Correlated flux measurements

For weak targets, photometric measurements become too noisy to normalise in real time the raw correlated flux measurements to form the visibilities. Instead, direct measurement of the correlated fluxes (=absolute fringe amplitudes) is possible. Since correlated flux measurements are, in essence, spectrophotometric measurements on faint targets, they are only offered for good observing conditions and low spectral resolutions. Absolute calibration of correlated fluxes requires special calibrators, as not only their sizes need to be known, but also their SED. If you are interested in using the correlated flux mode and have doubts about your calibrators, you can contact the VLTI Expertise Centre at Leiden Observatory for help with selecting appropriate calibrators and data reduction.

Setup AT UT

T≤ 30%

Seeing  0.8"

τ0 > 4.1 ms

Low L-band  0.3 Jy 0.04 Jy
Low M-band  1.1 Jy 0.15 Jy
Low N-band 6 Jy 0.5 Jy


Relative measurements

For medium and high resolutions, additional limits are given for measurements where only the differential phase is considered, since absolute measurements are better obtained with low resolution setting. Since relative measurements are not as susceptible to sky conditions as absolute measurements, they need not be made under good conditions. We give the limiting magnitudes to achieve a precision of 4 degree per spectral channel in differential phase, and an SNR on the coherent flux in N-band (see above for definition of spectral channel):

Setup AT UT

T≤ 70%

Seeing  1.15"

τ0 > 2.2 ms

Low M-band 1.1 Jy 0.15 Jy
Medium L-band 5.6 Jy   0.35 Jy
Medium M-band  8 Jy 0.8 Jy
High L-band 10 Jy 1.2 Jy
High N-band 25 Jy 1.1 Jy


GRA4MAT mode

The use of GRA4MAT is recommended as soon as fringe tracking is possible on a MATISSE science target

Limiting fluxes derived from the last commissioning run and offered in P109 on the ATs:

spectral mode lowest L flux (Jy)   lowest M flux (Jy)   lowest N flux* (Jy)
  corr. Flux & diff phase closure phase corr. Flux & diff phase
closure phase corr.Flux & diff phase closure phase
LOW (R~35) 0.2 0.25 0.7
MED (R~500) 1 1.5 11 16 not offered not offered
HIGH (R~1000) 2 3 not offered not offered 26 (R ~ 220) 30 (R~220)
HIGH+ (R~3300) 20 25 17 25 not offered not offered

The Limiting magnitude for GRAVITY's Fringe Tracker depends on the atmospheric conditions as follows:

Occurence probability 10% 30% 70%
Seeing <0.6" <0.8" <1.15"
Coherence Time >5.2 ms >4.1 ms >2.2 ms
K mag <8.5 <8.0 <7.0


Limiting fluxes derived from the last commissioning run and offered in P109 on the UTs, only the number in bold are guaranteed:

spectral mode lowest L flux (Jy)   lowest M flux (Jy)   lowest N flux* (Jy)
  corr. Flux & diff phase closure phase corr. Flux & diff phase
closure phase corr.Flux & diff phase closure phase
LOW (R~35) 0.02 0.03 0.07
0.1 0.1
MED (R~500) 0.3 0.5 1 1.5 not offered not offered
HIGH (R~1000) 0.7 1 not offered not offered 0.5 (R ~ 220) 0.7 (R~220)
HIGH+ (R~3300) 6 8 6 8 not offered not offered

The Limiting magnitude for GRAVITY's Fringe Tracker depends on the atmospheric conditions as follows:

Occurence probability 10% 30% 70%
Seeing <0.6" <0.8" <1.15"
Coherence Time >5.2 ms >4.1 ms >2.2 ms
K mag 10.5  10  9
  • For GRA4MAT mode in p1 the user should select the turbulence category based on the GRAVITY magnitude of the source.
  • Spectral coverage goes as follow: LOW and MED cover the full L and M-bands. HIGH is limited to L-band, and HIGH+ starts in L at 4.02 microns (it covers the Br-alpha hydrogen line) and covers the M-band. The values just listed are obtained when selecting DIT = 0.111 s in phase 2. For the UTs the user will be also allowed to select DIT = 0.5 s to increase the spectral windows in higher spectral resolution modes (0.6 micron in MR, 0.3 micron in HR, 0.1 micron in HIGH+). However such DIT requires turbulence category 30% (seeing <0.8", coherence time >4.1 ms) and airmass <1.5. If the user plans to use the long DIT should select turbulence category 30% in P1.
  • All targets with correlated flux N-band < 1 Jy (UTs), and N-band fluxes <6-8 Jy (at 8 and 11 micron respectively, ATs) should be observed with GRA4MAT to avoid a bias in the measurement.
  • The execution time for a calibrated OB in GRA4MAT mode follows the same rules as MATISSE stand-alone, and it is given in the table on top of this page. The user will need to select GRA4MAT template already at phase1 and select the photometry on or off depending on the requirements of the program and the telescopes used. Whether a photometry sequence is possible, see below for chopping.
  • Chopping, that delivers N-band photometry and chopped LM-band fringes, is not offered with UTs. It is offered for ATs only for targets which can run the fringe tracker with 1.1ms DIT. This limits chopping to targets with K<7mag.
  • When in doubt on the K-band flux injected into the GRAVITY fibre (as is often the case for extended targets with only integrally measured values), it is recommened to give a conservatively lower K-band magnitude in the OB, to ensure the fringe tracker runs in an appropriate mode with a longer DIT.


Narrow field off-axis observations

From P112 onwards GRA4MAT is offered with the capability to track fringes with GRAVITY, while MATISSE points to a nearby offset position to record fringes. This is useful to observe, e.g., planets or brown dwarf companions to stars that can be tracked by GRAVITY, when the relative position is well known. In principle, the same performance limits as for on-axis observations apply (see  tables above). However, for small offsets the contrast between the central star and the off-axis target might prove problematic and users should take the different size of the point-spread functions of UTs vs. ATs into account when proposing observations. The maximum distance from the central target is 1 arcsecond for UTs and 3 arcseconds for ATs. Offsets can be defined in the normal GRA4MAT observing template. A cumulative sequence of offsets can be given in a single template. Data from the commissioning of this mode is available for β Pictoris and a calibrator.