MAROON-X: High-Precision Exoplanet Spectrograph

• MAROON-X is a high-resolution (R≈85,000) echelle spectrograph optimized for extreme-precision radial velocity measurements, using a dual-arm white-pupil design.
• It features innovative fiber injection, microlens-based pupil slicing, and a vacuum-enclosed, milli-Kelvin controlled environment to ensure sub-meter-per-second stability.
• Targeting late-type M dwarfs, MAROON-X achieves ~30 cm/s photon-limited precision and uses dual calibration (Fabry–Pérot etalon and ThAr lamp) with ensemble offset methods.

MAROON-X is a stabilized, fiber-fed, high-resolution (R ≈ 85,000) echelle spectrograph installed on the 8.1-meter Gemini North telescope at Maunakea, Hawai‘i. It is optimized for extreme-precision radial velocity (RV) measurements, particularly targeting late-type (M dwarf) stars, with the scientific goal of discovering and characterizing low-mass exoplanets—potentially Earth-size—in the habitable zones of nearby stars. MAROON-X combines a dual-arm white-pupil echelle design, a microlens-based pupil slicer and double scrambler, a vacuum-enclosed optomechanical environment with milli-Kelvin stability, and a temperature- and pressure-controlled Fabry–Pérot etalon for drift calibration. The instrument routinely achieves photon-limited RV precision of ∼30 cm s⁻¹ on bright M dwarfs (per exposure, per run), and, when calibrated for long-term drifts, delivers sub-meter-per-second stability over multi-year baselines—enabling the detection of Earth-mass planets at multi-year orbital periods (Basant et al., 20 Feb 2025, Seifahrt et al., 2022, Seifahrt et al., 2021, Brady et al., 16 May 2024).

1. Optical and Mechanical Design

MAROON-X employs a dual-arm, white-pupil, cross-dispersed echelle layout delivering simultaneous coverage from 500–920 nm. The optomechanical layout comprises:

• Fiber Feed and Injection: Starlight is injected via a compact Fiber Injection Unit (FIU) into a 100 μm octagonal science fiber (0.77″ FOV at Gemini), with near- and far-field scrambling to mitigate guiding- and illumination-induced RV errors (Seifahrt et al., 2018, Sutherland et al., 2016, Seifahrt et al., 2016).
• Pupil Slicer and Double Scrambler: A 3× microlens-array (MLA) based pupil slicer, coupled to three 50 × 150 μm rectangular fibers, translates the output fiber’s pupil into stable pseudo-slits feeding the spectrograph. The double scrambler further stabilizes the illumination, decoupling output from input-position variations (Seifahrt et al., 2016).
• Spectrograph Core: The modified KiwiSpec R4-100 design operates with a 100 mm beam, R4 echelle grating in quasi-Littrow, dichroic splitting into blue (500–670 nm) and red (650–900 nm) arms, with VPH cross-dispersers and custom refractive cameras focusing onto 4k×4k STA4850 CCDs. The blue arm uses a 30 μm epi CCD; the red arm employs a 100 μm deep-depletion CCD for extended red sensitivity (Seifahrt et al., 2018, Seifahrt et al., 2021).
• Environmental Control: The main optics reside in a vacuum enclosure (<10⁻⁶ mbar) with temperature stability of 1–5 mK, essential for maintaining long-term spectrograph stability. Camera arms are pressure-sealed (ΔT ≤ 20 mK) with the bench temperature controlled to ±0.01 K (Seifahrt et al., 2018, Seifahrt et al., 2016).

Key technical parameters:

Parameter	Value	Note
Spectral resolution	R ≈ 85,000	±5 % across band
Wavelength range	500–920 nm	55 echelle orders
Fiber feed	100 μm octagonal, 0.77″ on sky	Front-end at Gemini bottom port
Throughput (peak)	≃8 % (measured), goal 15%	End-to-end at 700 nm
Detectors	STA4850 4k×4k (blue & red arms)	15 μm pixels
Main optics environment	Vacuum enclosure	ΔT ≤ 1–5 mK
Simcal source	Fabry–Pérot etalon	NKT supercontinuum

Values quoted from (Seifahrt et al., 2022, Seifahrt et al., 2018, Seifahrt et al., 2021).

The choice of non-circular (octagonal, rectangular) fibers with thick round claddings and low-shrinkage adhesives was based on low focal ratio degradation (FRD) and robust, moderate scrambling gain (⟨SG_min⟩ ≈ 500–1000), as established in systematic laboratory tests (Sutherland et al., 2016).

2. Wavelength Calibration and Drift Control

MAROON-X achieves high-precision wavelength solutions and drift tracking using a dual calibration strategy:

• Fabry–Pérot Etalon: A temperature- and pressure-stabilized FP etalon (FSR ≈ 15 GHz, finesse ≈ 40–50) is illuminated by a broadband supercontinuum laser to produce a dense, unresolved comb of lines. The etalon spectrum is simultaneously recorded with each science exposure in a dedicated calibration fiber (“simcal”), enabling sub-m s⁻¹ drift corrections over nightly and short-term timescales (Basant et al., 20 Feb 2025, Seifahrt et al., 2022).
• Thorium–Argon (ThAr) Lamp: ThAr exposures define the absolute wavelength solution at the start and end of each run and are used to monitor long-term drift of the etalon (e.g., Zerodur spacer aging, ∼2.2–2.5 cm s⁻¹ d⁻¹). Initial absolute orders/mode numbers for the etalon are anchored to ThAr and a theoretical model (Basant et al., 20 Feb 2025).

Drift equation for science exposures:

$$\lambda_{i,j}^{\rm science}(t) = \lambda_{i,j}^{\rm master} + \delta v_j(t)\,\frac{\lambda_{i,j}^{\rm master}}{c}$$

where $i$ indicates pixel, $j$ order, and $\delta v_j(t)$ the spline-fitted drift measured by etalon lines (Seifahrt et al., 2022).

The etalon achieves short-term (nightly) precision of ∼30 cm s⁻¹. However, long-term drift (notably in the FP cavity) at 2.2 cm s⁻¹ d⁻¹ and discrete “jumps” from baseline interruptions (e.g., earthquakes, re-alignment, power cycles) necessitate a more advanced offset calibration for multi-run datasets (Basant et al., 20 Feb 2025).

3. Data Reduction, Radial-Velocity Extraction, and Ensemble Offset Calibration

Pipeline Workflow:

• 1D Extraction & Wavelength Solution: Fiber spectra are extracted order-by-order; etalon and ThAr lines are fitted with combined box+Gaussian profiles. After initial identification, a high-order polynomial fit and cubic spline (∼30 knots/order) define the dispersion relation, with drift corrections applied per exposure using simcal information (Seifahrt et al., 2022).
• RV Measurement: The RVs are extracted using template-matching (SERVAL algorithm or similar), which builds a high-SNR template spectrum from deep co-addition and minimizes χ² over Doppler shifts: $$\chi^2(v) = \sum_{i} \frac{\left[f_i(\lambda_i)-[T(\lambda_i(1+v/c))\otimes \rm IP] \right]^2 }{\sigma_i^2}$$ Typical photon-limited per-exposure uncertainties are 30–50 cm s⁻¹ on sufficiently bright/lined M dwarfs (Seifahrt et al., 2022, Basant et al., 20 Feb 2025).

Ensemble Offset Calibration (Key for Long-Baseline Precision):

To address run-to-run “zero point” RV offsets (arising from baseline perturbations), Basant et al. (Basant et al., 20 Feb 2025) implemented an ensemble method:

1. Pre-processing: Deduct median RV per run and clip outliers (>4σ) in both RV and activity indices.
2. Signal Characterization: Use generalized Lomb–Scargle periodogram (FAP ≲ 0.1%) to detect significant periodicities.
3. Model Fitting: Fit planetary signals with (multi-)Keplerian models (parameters: $$P, K, t_0, h \equiv \sqrt{e} \sin\omega, k \equiv \sqrt{e}\cos\omega$$); activity with GP kernels (Quasi-Periodic, SHO, Double-SHO).
4. Offset Computation: For each star and each run: $$\Delta\mathrm{RV}_{\rm run} = {\rm median}\left[ \mathrm{RV}_{\rm measured}(t) - \mathrm{RV}_{\rm model}(t) \right]$$
5. Iterative Ensemble Combination: Use an initial reference (Barnard’s Star), iteratively align other stars and average differences to derive a global per-run offset and uncertainty (bootstrap across stars per run).

This ensemble calibration reduces multi-run “zero point” uncertainty to ≈0.5 m s⁻¹, an order of magnitude improvement over the prior several m s⁻¹ offsets that would otherwise swamp low-amplitude, long-period planet signals (Basant et al., 20 Feb 2025).

4. Radial Velocity Performance, Science Cases, and Sensitivities

After ensemble offset calibration, MAROON-X achieves ≤70 cm s⁻¹ RMS residuals over ∼2.5-year baselines for bright, quiet targets (e.g., HD 3651), matching the residuals from contemporaneous state-of-the-art EPRV instruments (EXPRES, NEID) (Basant et al., 20 Feb 2025).

Injection–Recovery and Completeness:

• Direct recovery of synthetic 1 m s⁻¹ Keplerians (at 9.5 and 28.5 days) into the HD 3651 time series yields detection at >0.1% FAP, with parameter recovery within 1–2σ of truth.
• Monte Carlo injection–recovery of 10,000 mock planets provides >80% detection probability for 1 m s⁻¹ signals at 10-day periods (combined arms); 79% at 100 days. Sub-m/s detection sensitivity persists out to periods >1,000 days (Basant et al., 20 Feb 2025).

This demonstrates MAROON-X's capacity to probe the habitable zone of mid-M dwarfs for Earth-mass planets across multi-year campaigns.

Science Applications:

• RV mass measurements for small, transiting planets (e.g., TOI-1450Ab, M=1.26±0.13 M⊕) with <10% precision (Brady et al., 16 May 2024).
• Obliquity measurements of faint late-type stars via high-precision Rossiter–McLaughlin observations (e.g., ±18° on TRAPPIST-1, V=18.8) (Brady et al., 2022).
• Planet confirmation and mass/radius/composition studies for temperate, potentially habitable exoplanets; routine few-σ Earth-mass sensitivity in 10–30 min exposures on V=10–12 M dwarfs (Seifahrt et al., 2021, Seifahrt et al., 2022).

5. Systematics, Fiber Optics, and Instrumental Limitations

Identified Limitations and Mitigations:

• Etalon Long-Term Drift: Slow chromatic aging of FP cavity coatings leads to ∼2.2–2.5 cm s⁻¹ d⁻¹ drift; regularly anchored to ThAr exposures and mitigated by planned upgrades to frequency combs (Basant et al., 20 Feb 2025, Seifahrt et al., 2022).
• Run-to-Run Offsets: Discrete multi-m/s RV jumps from flexures, hardware interventions, or major telescope events. Empirical removal via standard-star ensembles; routine offset errors controlled to <0.5 m s⁻¹ (Basant et al., 20 Feb 2025).
• Fiber Reconnection: Disconnected fibers induce 1–3 m s⁻¹ RV steps; minimized by tracking standard stars each run (Brady et al., 16 May 2024, Seifahrt et al., 2022).
• Telluric Contamination: Atmospheric lines, especially in the red, are masked during template-building and RV extraction (Brady et al., 16 May 2024).
• Chromatic Systematics: Weak arm-to-arm RV discrepancies can emerge over years, suggesting subtle uncorrected calibration differentials; pipeline treats each arm independently during modeling (Brady et al., 16 May 2024).
• Thermal & Pressure Stability: Only the main slit and grating reside in vacuum. Remaining optics/cameras are bench/air-mounted, making them more sensitive to ambient changes (1.8–7 m s⁻¹ mK⁻¹). Recent upgrades add improved bench thermalization, pressure control, and software interlocks to limit excursions (Seifahrt et al., 2022, Seifahrt et al., 2021).
• Fiber Optics: Laboratory testing established that thick, round cladding and minimization of connector stress are critical for low FRD in octagonal and rectangular fibers, supporting the goal of sub-m/s RV stability (Sutherland et al., 2016). Active agitation is implemented to reduce speckle-induced modal noise.

6. Technical Innovations and Upgrades

MAROON-X introduces several advances in EPRV instrument engineering:

• MLA-Based Pupil Slicing: Integrated pupil slicer and double scrambler stabilize both near-field and far-field illumination, yielding ≳75% geometric throughput, and RV precision ≲0.5 m s⁻¹ (Seifahrt et al., 2016).
• Vacuum and Milli-Kelvin Thermal Control: Optical stability at the 1–5 mK level is achieved through a combination of vacuum and engineered environmental enclosure, supporting multi-year precision at 30 cm s⁻¹ (Seifahrt et al., 2018, Seifahrt et al., 2016).
• Simultaneous FP Etalon Calibration: Routine exposure-by-exposure drift correction using a stable etalon comb, soon to be replaced by a laser frequency comb, enhances long-term traceability (Seifahrt et al., 2022).
• Pipeline & Analysis Techniques: A hybrid reduction pipeline emphasizes high-order wavelength fits, template-matching RV extraction (SERVAL), and multi-star ensemble drift tracking for multi-run data (Basant et al., 20 Feb 2025, Brady et al., 16 May 2024).
• Planned & In-Progress Upgrades: Installation of frequency-comb calibration sources, improved on-bench thermal control, sky-fiber calibration channels, and a chromatic exposure meter to mitigate remaining sources of long-term RV systematics (Seifahrt et al., 2022).

7. Comparative Context and Future Prospects

MAROON-X occupies a unique role among high-precision RV spectrographs: it is optimized for the red-optical regime (650–900 nm), making it especially suited to observe the faint, late-M dwarfs that are inaccessible to blue-optimized instruments such as HARPS, ESPRESSO, and CARMENES (Brady et al., 2022, Seifahrt et al., 2021). Coupled to Gemini North's 8.1 m aperture, it routinely attains SNR sufficient for 0.6–1.0 m s⁻¹ precision on V=12 M dwarfs in 10–30 min, and has demonstrated sensitivity to Earth-mass exoplanets at semi-major axes comparable to their host’s habitable zones over >1000 days (Basant et al., 20 Feb 2025, Brady et al., 16 May 2024).

Ongoing upgrades—most notably the deployment of a frequency comb for absolute calibration, improved thermal control, and environmental monitoring—are expected to reduce residual long-term systematics to ≲10 cm s⁻¹, opening the possibility of routine characterization of terrestrial planets around the nearest M dwarfs and enabling comprehensive EPRV surveys on an 8 m-class platform (Seifahrt et al., 2022, Seifahrt et al., 2018).