HERMES Spectrograph Overview

Updated 10 March 2026

HERMES spectrograph is a high-resolution, fibre-fed optical instrument designed for detailed stellar surveys and time-domain studies, with distinct AAT and Mercator configurations.
AAT HERMES employs a four-channel design with 392 fibres, delivering resolutions up to 40,000–50,000 and enabling large-scale Galactic Archaeology surveys like GALAH.
Mercator HERMES features a single-channel white-pupil echelle design optimized for stability and precision radial velocity work, achieving up to R ≃ 85,000 in high-resolution mode.

The HERMES spectrograph refers to two distinct but independently prominent facility-class, high-resolution optical spectrographs: (1) HERMES at the Anglo-Australian Telescope (AAT), designed primarily for the Galactic Archaeology GALAH survey, and (2) HERMES at the 1.2m Mercator Telescope at La Palma, focused on flexible, high-fidelity spectroscopy for time-domain astrophysics and abundance studies. Both systems employ advanced fibre feeds, white-pupil echelle designs, and precise environmental and calibration controls, but serve different multiplexing, spectral, and survey requirements. They have become strategic instruments for large stellar surveys, stellar radial velocity monitoring, and high-precision spectroscopy. Distinctions between the AAT and Mercator HERMES instruments are critical, since performance, architecture, and primary science use cases are not interchangeable.

1. Instrument Architectures and Optical Design

HERMES at the Anglo-Australian Telescope (AAT)

HERMES at the AAT is a four-channel, fibre-fed optical spectrograph dedicated to high-multiplex, moderate- to high-resolution spectroscopy within the context of Galactic Archaeology (Sheinis et al., 2015). Its design is dictated by integration with the 2dF robotic fibre positioner:

Fibre system: 392 science fibres per plate, 2° patrol field. Fibres are grouped into ten-fibre slitlets, fed by 50 m lengths, with ±10 μm slit-end alignment accuracy. Dual banks (“slit A” and “slit B”) allow simultaneous field reconfiguration/readout.
Collimation and splitting: The output beam is collimated using a 1 m spherical mirror and two large off-axis correctors (620 × 330 mm). Light is separated among four spectral channels (“Blue,” “Green,” “Red,” “IR”) by a series of dichroic beam splitters; each channel is dispersed by a 500 × 200 mm custom VPH grating at ∼67.2° incidence.
Cameras and detectors: Each arm features an f/1.67 refractive camera (380 mm front element), cryostat optics, and a 4k × 4k e2v CCD231-84.
Spectral coverage: Four non-contiguous bands within 370–1000 nm (Blue: 471.8–490.3 nm, Green: 564.9–587.3 nm, Red: 648.1–673.9 nm, IR: 759.0–789.0 nm).
Spectral resolution: Two modes—nominal (R ~ 28,000) and high-resolution “slit-mask” mode (R ~ 40,000–50,000, ∼50% decreased throughput).

HERMES at the Mercator Telescope

The Mercator HERMES is a single-channel, fibre-fed, white-pupil echelle spectrograph optimized for long-term stability, efficiency, and broad spectral coverage in time-series spectroscopy and precision radial velocity (RV) work (Raskin et al., 2010, Raskin et al., 2013, Beck et al., 2015).

Optical train: A fibre-fed (up to two fibres: HRF and LRF), bench-mounted white-pupil echelle incorporating a monolithic echelle grating (52.676 lines/mm, blaze ∼69.7°) and fused silica cross-disperser prisms.
Camera and detector: f/3.1 all-refractive camera delivers a curved field to a 2k × 4.6k (historically) or monolithic 4k × 4k (current) e2v CCD, back-illuminated, with AR coating matched to order curvature.
Spectral format: 377–900 nm (55 contiguous orders except at long-wavelength end), instantaneous coverage.
Modes: HRF (R ≃ 85,000, image-slicer enabled), LRF (R ~ 63,000, double-scrambler for improved illumination stability).
Calibration: Dedicated Thorium-Argon (ThArNe) hollow-cathode lamp feeds, simultaneous reference enabled in LRF or planned for HRF with future upgrades.

2. Spectral Performance, Resolution, and Throughput

Instrument	λ coverage (nm)	R (nominal)	Multiplex	Peak Throughput
HERMES (AAT)	370–1000 (4 bands)	~28k [std], 40–50k [HR]	392	~10% total at 500 nm (Sheinis et al., 2015)
HERMES (Mercator)	377–900	85,000 [HRF]; 63,000 [LRF]	1	~17–28% spectrograph alone; ~7–17.5% system (Raskin et al., 2013, Raskin et al., 2010)

AAT HERMES meets the GALAH survey requirement of S/N ≥ 100 at V = 14 in 1 hr exposures in all channels, with throughput reaching design targets of ~10%. Per-band empirical flux formulas allow direct calculation of S/N as a function of magnitude and integration time:

$S/N = \frac{N_\star}{\sqrt{N_\star + N_{\text{sky}} + N_{\text{dark}} + N_{\text{read}}^2}}$

where $N_\star$ is the stellar source count, with empirical relations provided for each spectral arm. Resolution is characterized as $R = \lambda / \Delta\lambda$ , e.g., at 500 nm, $\Delta\lambda \approx 0.018$ nm ( $R=28,000$ ); at 650 nm, $\Delta\lambda \approx 0.014$ nm ( $R=45,000$ ).

Mercator HERMES achieves high peak efficiency at 550 nm (28% bare spectrograph, 17.5% including telescope), with S/N = 100 attainable for V = 10.4 (HRF) in 1 hr under median seeing. Resolution element FWHM at 550 nm is ∼0.0065 nm (HRF). The instrument profile is Gaussian, spanning 2–3 pixels (Beck et al., 2015, Raskin et al., 2010).

3. Calibration, Stability, and Data Reduction

Both spectrographs emphasize environmental stability and robust calibration chains:

AAT HERMES: Mechanical flexure is negligible (focus drift <0.1 μm/h), and a temperature-controlled enclosure ensures sub-μm focus stability. Instrumental flexure, detector stability, and fibre cross-talk (3–5% for outer fibres) are well characterized and managed by a custom data reduction pipeline.
Mercator HERMES: The spectrograph is sited within a temperature- and pressure-controlled enclosure; the optical bench holds ΔT <0.01°C over months. Recent stabilities: RV repeatability of 2–4 m s⁻¹ (simultaneous ThAr), night-to-night RV scatter 10–20 m s⁻¹ (standard), and long-term scatter in standards at ≲50–60 m s⁻¹ (Raskin et al., 2013, Beck et al., 2015, Merle et al., 2023).
Wavelength calibration for both typically uses in-situ ThAr (and optionally Fabry–Pérot at Mercator) with optional simultaneous lamp exposure.
Data Reduction: Dedicated pipelines (AAO’s 2dfdr for AAT, HermesDRS/Mercator for Mercator) perform bias subtraction, optimal extraction, blaze correction, scattered-light modeling, order merging, wavelength calibration, and correction for instrumental systematics. Pipeline products from HERMES/Mercator achieve 0.5% order-overlap agreement and RV precision at the few m s⁻¹ level (Raskin et al., 2010, Raskin et al., 2013).

4. Operational Performance and Survey Outcomes

AAT HERMES (GALAH Survey):

Simultaneous observation of up to 392 stars in a 2° field.
First light: October 2013. Main survey saw >69,000 stars in first 56 nights; long-term allocation to reach 1 million stars (Sheinis et al., 2015).
Throughput is stable across most fibres; ∼2.5% are persistently >30% below average, typically associated with slitlet/retractor groupings. No evidence for temporal throughput degradation in first two years (Simpson et al., 2016).
Chromatic and plate-position dependent throughput variations are present but do not degrade science outcomes.

Mercator HERMES:

Long-term, stable time-series spectroscopy of bright stars (V ≤ 11 for optimal S/N), with more than 42,000 spectra over about 1250 nights (Raskin et al., 2013).
Radial velocity precision reaches few m s⁻¹ in simultaneous ThAr mode, sub-100 m s⁻¹ over multi-year datasets (Merle et al., 2023).
Applications span asteroseismology, binary orbits, abundance studies, and solar analog monitoring.

5. Scientific Use Cases and Validation

Galactic Archaeology

AAT HERMES was purpose-built for the GALAH project, enabling detailed chemical abundance mapping and kinematics for up to one million Milky Way stars (Sheinis et al., 2015). Multichannel design and field multiplex are essential for chemical tagging and reconstructing the assembly history of the Galaxy.

Stellar Abundance and Time-Domain Studies

Mercator HERMES has produced solar reference atlases (asteroid and moon spectra) matching FTS and HARPS atlases to <0.3% in flux (Beck et al., 2015), and is validated for chromospheric activity, rotation, lithium abundance, and precise RV work on solar analogs (Beck et al., 2016, Salabert et al., 2016). Long-term RV drift control (<60 m s⁻¹), high S/N, and robust continuum preservation underpin high-precision line profile studies, binary mass determinations, and chromospheric activity tracking (Merle et al., 2023).

6. Limitations, Optimizations, and Upgrades

AAT HERMES: Initial channel astigmatism and VPH grating wavefront errors were resolved by repolishing and recoating prior to full survey operation. Minor Littrow ghosts (≤0.03% flux) are removed in the pipeline. Fibre cross-talk and slitlet alignment remain focal areas for data reduction optimization (Sheinis et al., 2015, Simpson et al., 2016).
Mercator HERMES: Planned upgrades include narrow (25 μm) reference fibres for full-resolution, full-efficiency simultaneous calibration, and adoption of octagonal fibres for improved modal scrambling. Simulation and on-sky assessment of S/N and RV stability enhancements precede full upgrade deployment (Raskin et al., 2013).