Looking inside the box Upgraded MIRSI on IRTF

MIRSI - IRTF's Mid-Infrared Imager and Spectrometer


MIRSI is a 2-20 micron camera and grism spectrograph built by Boston University. MIRSI was recently upgraded by IR Labs with a closed-cycle cooler to replace its liquid nitrogen and liquid helium cryostat, allowing MIRSI to be used interchangeably with any other IRTF instrument. It is currently being further refurbished by IRTF with a new array controller and an optical channel called MOC (similar to MORIS on SpeX). MOC will allow simultaneous visible light imaging with MIRSI and serve as a visible light guider for MIRSI.

Instrument Properties


The capabilities for 2022B are listed here.

MIRSI Upgrade

MIRSI is in the process of being upgraded. The overall goal of the upgrade is to improve reliability and usefulness of MIRSI, while reducing operating cost. The changes to MIRSI are:

  • Replacing the LHe cryogen can with a Sumitomo closed cycle cooler. With the LHe cooling, MIRSI would only be cooled for a roughly week long observing run, then allowed to warm until its next run months later. With the close cycler cooler, MIRSI will be kept cold at all times, and be able to be used interchangeably with any other IRTF instrument.
  • New array controller. The previous controller died, and the PCI card was a pain to deal with. Both have been replaced.
  • New chopper. The legacy chopper electronics were approaching their 40th birthday, and were becoming difficult to support. A new chopper actuator, based on a piezo actuator, and modern control electronics, are being developed on the existing chopper hardware.
  • MOC: Visible light CCD camera that allows guiding or simultaneous visible light and mid-IR observations. To support MOC, a cold dichroic was added to the 'nose' of MIRSI.
  • The entrance window was changed from KRS-5 to ZnSe (see note below), due to the higher transmission and superior durability of ZnSe.


    Observers publishing results obtained with MIRSI are requested to reference the following papers:

    MIRSI, A MId-Infrared Spectrometer and Imager: Performance Results from the IRTF
    M. Kassis, J. Adams, J. Hora, L. Deutsch, E. Tollestrup, 2008, PASP, 120, 1271

    MIRSI: A MId-Infrared Spectrometer and Imager
    L. Deutsch, J. Hora, J. Adams, M. Kassis, 2003, SPIE, 4841, 106

    The original Boston University web page for MIRSI


    The filters in MIRSI are listed in the following table. The center wavelength, short wavelength, long wavelength, bandwidth, and peak transmission were measured in 2018 at IfA.
    MIRSI FIlter Table


    The current sensitivity of MIRSI was measured by observing mu UMa on the nights of Feb 1, 2024, and are provided by Mike Connelley. Please note that the 1 sigma sensitivity in the last two columns is for 1 minutes of integration time, and does not include readout and telescope offset overheads. These estimates should help observers writing propsals to better understand what MIRSI can currently do in a given amount of time. Despite installing a science grade detector, the sensitivity of MIRSI does not seem to have improved, and we are investigating other avenues for improvement.

    SNR vs. Coadds

    We recently found that the S/N ratio in a given amount of wall clock time greatly depends on the number of coadds used. Coadds=300 seems to be about optimal. At lower coadds, more time is lost to telescope beam switches. At higher coadds, the S/N ratio does not increase as sqrt(coadds). However, the S/N ratio does increase as the sqrt of the number of images that are stacked.

    Maximum Exposure Time

    The exposure times in the second column are the maximum exposure times that we could use without exceeding the well depth on the sky. Currently, using exposure times greater than 22 ms causes the top ~1/4 of the detector to reset, so if possible observers should use exposure times under 22 ms.

    Table by msc

    Sensitivity with Blind Stacking

    Joe Hora reports that he has detected a 424 mJy source in the broad N-band filter in 23 minutes of observing time with a 1-sigma sensitivity of 13 mJy. This was with 500 coadds, and nodding/offsetting on the array. He notes that this is somewhat better than the table above suggests, but he used 10 ms frame time rather than 5 ms reported in the able.

    On a different object on a different night, he detected a 250 mJy source with a 1 sigma sensitivity of 20 mJy in 40 minutes, also through the broad N-band filter, 10 ms exposure time with 1000 coadds. This sensitivity is more consistent with the values in the table above.

    For an explaination of blind stacking, see the description below in the section on MOC.

    Detector Properties

    This is Table 3 from Kassis et al. (2008)
    Table 3 from Kassis et al. (2008)

    MIRSI Optical Camera (MOC)

    MOC is a visible light CCD camera that is co-mounted with MIRSI. The cold dichroic that was added to MIRSI reflects the IR beam into MIRSI, while the visible light passes through to MOC mounted below MIRSI. It uses the same model Andor EMCCD camera as MORIS, and thus its use and sensitivity will be very similar. Like MORIS, it will have a 1 arcminute field of view.

    MOC enables:
  • Simultaneous visible light and mid-IR observations of a given target. This combination is especially useful to determine the albeidos of asteroids.
  • Visible light guiding on the science target or a field star. This will help to put a target onto MIRSI's slit, and to keep it there. For imaging faint targets, we expect that observers will be able to use the MOC images to determine how to align their MIRSI images. This will allow 'blind' stacking of MIRSI images; the target would not need to be visible in each chop/nod quad and thus fainter targets could be observed.

    MOC's Estimated Sensitivity (50 sigma in 1 minute)
  • g'= 19.8
  • r'= 20.3
  • i'= 19.7
  • z'= 18.2

    What is Blind Stacking

    If MIRSI observes a target fainter than ~1 Jy, then the target is unlikely to be detected in a single A-B image pair. In this case, the observer would need to "blindly" align and stack the MIRSI images to see the target. Previously, the telescope offsetting was not accurate enough to allow the images to be aligned. However, with MOC guiding MIRSI, this is no longer the case. Since MOC is guiding the telescope in each of the offset positions, the offsets of the images in MIRSI will be precise. Knowing your dither pattern and the image scale of MIRSI, you can then calculate the offsets to align and stack your images.

    Entrance Window

    Before MIRSI came off of the telescope, the entrance window was made from KRS-5. The two windows were badly damaged by exposure to moisture, and the effort to repair them was not successful. While KRS-5 is transmissive across a wide spectral range, it is highly vulnerable to moisture damage, and this is almost inevitable in the dome environment. KRS-5 also has about 30% emissivity across the mid-IR, significantly adding to the sky background.

    For these reasons, we chose to make new zinc selenide entrance windows. The advantages are that ZnSe is transparent in more of the visible (helpful for MOC), is easier to polish, has higher transmission (and lower emissivity) at N-band, and is much more weather resistant. The disadvantage is that its transmission drops at 18 microns, and is very low at 20 microns, and essentially opaque beyond 20 microns. This strongly affects observing in the Q-band, but we felt that the trade off was worth it. As such, we expect that the sensitivity of MIRSI in the Q-band will be worse than stated in Table 5. The sensitivity of the narrow 18.5 micron filter may not be strongly affected, but the other Q-band filters and grism would be.
    Back to IRTF Homepage

    Last modified Feb 26, 2024
    Questions to Mike Connelley msc@ifa.hawaii.edu