Instrument Feasibility

The current (feasibility-level) MICHI design represents an evolution of MIRES, lead by A. Tokunaga and J. Elias. MIRES was a highly optimized MIR (7.5-26 μm) spectrometer, highly leveraged from a MIR adaptive optics (MIRAO) system. Submitted to the TMT Project Office in 2006 in response to the first-light instrument call for proposals, it helped to establish TIR as an important scientific capability for the TMT but was ultimately not selected as a first-light instrument. The TMT SAC feedback concentrated on two key areas, (1) to include a ‘blue’ 3-5 μm arm and (2) to broaden the capabilities to appeal to a larger TMT community. Several members of the MIRES team are core to the MICHI team, and through our scientific and technical explorations we have independently converged on a feasibility-level design that comprehensively addresses the SAC points. MIRAO (§6.1) and MICHI (§6.2) designs are (very) briefly introduced below.

MIRAO Feasibility-Level Design

At TIR wavelengths, AO is required to deliver high-throughput high-Strehl observations whilst keeping thermal emissivity low and system throughput high. Thus, key requirements differ greatly from the first generation NIR TMT AO system, NFIRAOS. The TIR AO system is driven to a simple implementation with no transmissive elements to the science instrument entrance window. At TIR, the implementation of an ASM (adaptive secondary mirror) is an ideal partial solution. We present a short description of a TIR optimized AO system centered around an optical relay and deformable mirror (DM) but that naturally upgrades to use an ASM. Note that the ASM replaces three mirrors in the optical relay – two off-axis parabolas and the deformable mirror; the rest of the AO system, in particular all wavefront sensors and all mechanical interfaces, are still required for the ASM and the instruments it feeds. In the MIRAO concept all of these components are deployed initially with the optical relay and subsequently with the ASM.

The MIRAO design has evolved since the original concept was developed during feasibility studies in 2006 (Chun et al. 2006). The original MIRAO concept assumed a single TIR instrument (e.g. MIRES) working from 7-14 μm, with a goal of 5-28 μm, a small field of view, and that an ASM would not be available at first light but might be a future upgrade path (Fig. 5). A simple optical relay (two off-axis parabolas, a DM, and a single fold mirror) plus a set of three sodium laser-guide star wavefront sensors provides the laser-tomography AO correction (LTAO). Inside MIRES we placed the required natural guide star wavefront sensors (both tip/tilt/focus as well as a place holder for a NIR high-order wavefront sensor should an appropriate detector become available). Provisions for MIRAO’s LTAO constellation of LGS are designed into the TMT facility LGS system. With the evolution to MICHI, key science requirements for MIRAO have changed. Of particular note, the required science wavelength is now 3-14 μm with a goal of 3-28 μm and the field of view has increased to up to ~30”. Additionally, there is interest in MIRAO feeding multiple science instruments. The original MIRAO concept supports these changes but the relay optics and the order of the AO correction (e.g. the number of actuators) need to be updated accordingly. Progress and the availability of new AO components (e.g. WFS detectors and DMs) makes these changes straight forward. The larger field and shorter wavelengths may require modifications to the LGS asterism but this can be accommodated in the current LGS facility. Details need to be worked out during a future design study.

Figure 5. MIRAO schematic design before ASM deployment (left) and after (right).

In several respects our current approach is easier to implement compared to our 2006 study. Advances in NIR detectors and wavefront sensors enable a high-order natural guide star NIR pyramid wavefront sensor for use on brighter science objects while advances in commercially available DMs reduce cost, risk, and size of the pre-ASM MIRAO optical relay. It is now reasonable to implement MIRAO in a phased approach - early science with a low-cost, low-risk natural guide star system followed by an expansion of the science capabilities with a LTAO system that affords increases both sky coverage and science capabilities. An early science version could be built using existing, off-the-shelf technology.

We have also made significant progress on the feasibility of daytime adaptive optics with MIRAO. The motivation is to enable AO-corrected observations at TIR wavelengths during the first 1-2 hours of each morning, before seeing conditions deteriorate. At TIR wavelengths the sky background is dominated by atmosphere and telescope thermal emission and is largely unchanged between day and night. The challenge is how to operate the AO system during the day. If this can be solved, then it opens a path to a 10-20% increase in the available science time on the telescope. Hart et al. (2016) showed that a sodium laser guide star can be imaged through a magneto-optical filter during the day at contrasts similar to night time observations. These extremely narrow bandpass filters effectively suppress all of the daytime sky but have serious engineering considerations for their implementation in a wavefront sensor. Dungee & Chun (2018, in prep.) show that a simpler but broader filter is sufficient if you trade its increased daytime sky background against the number of subapertures in the wavefront sensor. For a system like MIRAO the trade is favorable and results in an AO system that provides good performance at 5-25 μm. These design trades have not been optimized and there are impacts to the wavefront sensor design that need to be considered during the next design phase. Additionally, Dungee & Chun (2018) show that daytime AO observing can also be done with a NIR Pyramid wavefront sensor albeit at a significantly reduced limiting magnitude (K~8). However, at this magnitude there are science targets of interest (e.g. exoplanets and star formation regions).

Finally, we note the high-throughput and simplicity of MIRAO make it well suited to feed other types of instruments. In particular, the high-contrast Planetary Systems Imager (PSI) will require a first stage AO system. We are in close contact with the PSI AO team, and intend to make detailed shared studies should these WPs be approved. MIRAO could be the AO feed for other future TIR instruments that require AO. It is important to ensure that MIRAO meets the requirements of these systems and this will be studied during the design phase.

Table 1. Summary of selected MIRAO optimized requirements.
Requirement Value Requirement Value Requirement Value
Operational λ range L-Q Corrected field of view 10" with 1' goal Wavefront quality rms phase <350 nm
Sky coverage "All sky", limited by natural tip-tilt stars Delivered Strehl ratio

L&M Bands ~60%

N & Q Band > 90%


MICHI Feasibility-Level Design

In 2008 we evolved the MIRES design to MICHI through (1) enhancing imaging capabilities, (2) adding low-spectral resolution, and (3) adding an IFU, all of which operate at 7.5-26 μm. We also made early investigations of polarimetry and coronography but have not made a final choice regarding their incorporation at the time of this white paper. Work was leveraged from a preliminary optical design using the science requirements at that time. We added feasibility-level discussions of other instrument aspects, documented in a 136p reference document (Okamoto, Packham, & Tokunaga, 2008). From this report, we presented the MICHI design and science cases in SPIE meetings by Okamoto et al. (2008), Tokunaga et al. (2008), and Packham et al. (2012). However, due to a combination of evolving science drivers and technical opportunities, combined with new collaborations, we evolved MICHI to be optimized for 3-14 μm, but are planning to make no technical decision that excludes the Q band (16-25 μm). This scientific choice was dominantly made as (1) the 3-5 μm science cases are more numerous and higher ranked than those in the Q band, (2) the exoplanet community desire for >2 μm capabilities, (3) high-spectral resolution cases are especially strong at 3-5 μm compared to Q-band; technical drivers include (4) the advent of significantly improved TIR detectors at 3-14 μm, (5) the MIRAO 2006 design provides good correction to ~3 μm, and could exploit technical advances for excellent Strehl (~60%) correction to 3 μm, and (6) using immersion gratings from 3-14 μm affords a continuous optical-IR capability to the TMT community.

The instrument is an all-reflective design (except for entrance windows, and possible polarimetric, coronographic, and immersion grating optics, Fig. 6), modular for optimized alignment and/or phased deployment, and compact to fit within the MIRES space envelope. A selection of key instrument requirements used for our design optimization is tabulated below. MICHI dominantly employs off-the-shelf components, with other components in advanced development or characterization phases. The design leverages experience from other similar TIR instruments (i.e. T-ReCS, COMICS, CanariCam, MIMIZUKU, TEXES, EXES, WINERED, VINROUGE, MIRSIS, etc.) of which team members had leading or significant roles in their development, as well as from other on-going projects. The instrument is of relatively low cost and low technical risk. As MICHI is described in detail in the referenced papers, has been presented many times at TMT meetings, on the MICHI web site (, and for space reasons, the interested reader is recommended to those resources for more design information, full team listing, etc.

Figure 6. Feasibility-level MICHI design showing the integral modularity (left) and schematic design (right).

Table 2. Summary of selected MICHI optimized requirements.
Requirement Value Requirement Value Requirement Value
Operational λ range L (3.4-4.1 m),
M (4.6-4.8m),
N (7.3-13.8m)
Imager field of view 24.4x24.4" at L&M,
28.1x28.1" at N
Imager plate scale

11.9 mas pixel-1 at L&M,

27.5 mas pixel-1 at N

Long-slit spectrometer plate scale

11.9 mas pixel-1 at L&M

27.5 mas pixel-1 at N

Long-slit spectrometer resolution R~600 at L, M, &N bands Long-slit spectrometer slit length 28.1 length
High-res. spectrometer plate scale

11.9 mas pixel-1 at L&M

27.5 mas pixel-1 at N

High-res. spectrometer resolution

R~120000 at L&N,
R~100000 at M
Slit length 2" length (but highy dependent on array(see §7)
IFU spectrometer (baseline) N band (only)
10 spaxels
IFU spectrometer resolution R~1000 IFU spectrometer field of view ~0.175" (length) x
~0.07" (width): 35.0
mas per spaxel
Polarimetry (baseline) L,M,N Polarimetry modes Imaging & long-slit spectrometry    

Synergies With JWST, ELT, & ALMA

Space-based observations at TIR enjoys far superior sensitivity compared to the ground, and the 6.5 m JWST will open completely new vistas, ushering in a revolution in TIR astronomy. It is difficult to conceive of all the new vistas to be opened but TMT/MICHI will be ideal to exploit these observations to the fullest and explore new/independent research areas. JWST spatial resolution will be inferior to that of existing ground based TIR capabilities, and inferior to TMT by ~4.6 times (at similar Strehl ratios). Further, JWST spectral resolutions are constrained to a few 103, compared to the 105 planned for MICHI. Thus TMT/MICHI affords (a) the excellent sensitivity and high spatial resolution (inner working angle) essential for exoplanet direct imaging, disk work, and AGN, and (b) the excellent sensitivity and high spectral/spatial resolution essential for exoplanet atmosphere & disk characterization, offering strong science areas that JWST cannot compete with. To follow-up and fully exploit JWST observations, the combination of sensitivity, spatial & spectral resolution is required. TMT/MICHI on MKO is arguably the optimal follow-up combination.

ESO/ELT plans for METIS, a TIR instrument, as one of only three first generation instruments on the ELT. In many respects MICHI and METIS are similar, but with the notable exceptions of METIS not currently planning in their baseline design to offer LTAO capability nor a high-spectral resolution capability in the N band. Further, there are no plans for daytime observing, a capability that seems likely to be possible for MICHI. Despite the differences, there are multiple areas of commonality that can be exploited, as tentatively discussed between the teams, where discussions are leveraged from past and current science and instrument collaborations. For example, we formed a joint array consortium to characterize and purchase new TIR arrays to reduce risk/costs. This affords the possibility to jointly develop read-out electronics, again to reduce risk and/or costs. Collaborations could include common data reduction/interpretation tools, cold chopping mirror, and exchange of knowledge, experience, and personnel especially during assembly, integration, and verification (AIV) stages.

Finally, we note the strong connection between TMT/MICHI and ALMA. ALMA’s similar spatial resolution with TMT/MICHI ensures their synergy of wavelength coverage, and an ALMA/MICHI combination will be especially powerful for the science cases discussed above. Specifically, in the case of disks TIR observations provide dust composition information (e.g. PAHs, silicates, h3O ices) through IR solid state features unavailable from ALMA. For AGN, while ALMA can probe molecular gas and cool dust in the outer torus, TMT/MICHI is indispensable to understand hot/warm dust in the inner torus, polar dust, and outflowing ionized gas (evidence of AGN feedback) with a similar spatial resolution (0.02-0.03”), providing a complete picture of physics around a mass-accreting SMBH in an AGN.