Echelle Spectrograph

The core idea: trade order for groove count

A diffraction grating spreads light because each groove sends out a wavelet, and those wavelets interfere constructively only at angles where the path difference is a whole number of wavelengths. The grating equation is m·λ = d·(sin α + sin β), where m is the diffraction order, d the groove spacing, and α, β the incidence and diffraction angles. The resolving power of any grating is simply R = m·N — the order number times the number of illuminated grooves.

An ordinary spectrograph grating crams thousands of fine grooves per millimetre and works in 1st or 2nd order. An echelle grating does the opposite: it has coarse, widely-spaced grooves (often only 30–80 lines/mm) cut into a staircase profile, and it is used at a steep blaze angle in very high orders — order 30, 80, even 120. Because m is huge, R climbs to 100,000 with a compact, easily-ruled grating. That is the whole bet: get resolving power from the order number instead of from an impossibly fine groove pattern.

The overlap problem the cross-disperser solves

High orders come with a built-in headache: they overlap. Order m at wavelength λ diffracts to the same angle as order m+1 at λ·m/(m+1). At order 80, blue light from order 81 lands right on top of red light from order 80. The free spectral range of a single order — the slice of wavelength it can carry before the next order steps on it — is only Δλ = λ/m, just a few nanometres at optical wavelengths. A single echelle order is therefore a short, dense ribbon of spectrum, useless on its own because a dozen neighbouring ribbons pile on top of it.

The fix is elegant: add a second, low-dispersion disperser — a prism or a coarse grating — with its dispersion axis rotated 90° to the echelle. This cross-disperser nudges each order sideways by a different amount, peeling the stacked ribbons apart into a parallel ladder of curved strips. Read top to bottom you climb through orders; read left to right along any strip you scan wavelength within that order. A 3,800–7,900 Å spectrum at R = 100,000 — which would be metres long if laid out in a single straight line — folds neatly onto a CCD a few centimetres across. The échelle (French for "ladder") earns its name twice: from the staircase grooves and from the ladder of orders on the detector.

Anatomy of a real echelle spectrograph

Light from the telescope enters through a narrow slit or, in modern designs, the polished end of an optical fibre. A collimator turns it into a parallel beam that strikes the echelle. The dispersed beam returns through (or past) the cross-disperser, and a camera lens images the orders onto the detector. Many high-precision instruments are fibre-fed and sealed in a vacuum tank at constant temperature, because a 1 m/s radial-velocity shift corresponds to moving the spectrum by about one-thousandth of a pixel — a thermal expansion of microns would swamp it.

Selected astronomical echelle spectrographs

Instrument	Telescope	Resolving power R	Notable use
HARPS	ESO 3.6 m (La Silla)	≈ 115,000	RV exoplanets, ~1 m/s
ESPRESSO	VLT 8.2 m (combined 16 m)	up to ≈ 190,000	RV, ≲ 0.3 m/s, Earth-mass planets
HIRES	Keck I 10 m	≈ 25,000–85,000	Abundances, RV, IGM
UVES	VLT 8.2 m	≈ 40,000–110,000	Stellar & ISM spectroscopy
NEID	WIYN 3.5 m	≈ 110,000	RV, laser-comb calibrated

Calibration: turning pixels into velocities

An echelle's power is wasted without a precise wavelength scale. The classical reference is a thorium-argon hollow-cathode lamp, whose thousands of sharp emission lines scatter across every order and pin the dispersion solution. For the sub-metre-per-second regime needed to find small planets, two techniques go further. An iodine cell placed in the starlight stamps a forest of molecular absorption lines directly onto the science spectrum, providing a velocity ruler that drifts exactly with the instrument. The state of the art is the laser frequency comb: a train of perfectly evenly-spaced laser lines, each known to better than one part in 10¹¹, blanketing the whole spectrum like a ruler with millions of identical ticks. The comb removes instrument drift and is what lets ESPRESSO chase the ~9 cm/s wobble an Earth-mass planet imposes on a Sun-like star.

Echelle versus a conventional grating spectrograph

Design trade-offs

Property	Conventional grating	Echelle
Order used	1st–2nd	~30–120
Groove density	600–2400 lines/mm (fine)	30–80 lines/mm (coarse)
Resolving power R	~1,000–20,000	~50,000–190,000
Order overlap	Negligible	Severe — needs cross-disperser
Detector layout	Single 1-D strip	2-D ladder of orders
Simultaneous coverage	One narrow band	Whole optical band at high R

Why echelle spectrographs matter

• Exoplanet hunting. The Doppler radial-velocity method that found 51 Pegasi b and thousands since relies on echelle resolution and stability.
• Stellar chemistry. Individual narrow absorption lines reveal abundances of iron, lithium, carbon and dozens of elements.
• Interstellar medium. Sharp foreground absorption lines trace gas clouds along the sightline.
• Asteroseismology & rotation. Line shapes encode pulsations, rotation and magnetic activity.
• Wide coverage at once. Capturing the whole optical band per exposure beats scanning one band at a time.

Common misconceptions

• The cross-disperser adds resolution. No — the echelle supplies the resolving power; the cross-disperser only separates overlapping orders.
• Finer grooves always mean higher R. Echelles deliberately use coarse grooves and win R from the high order number instead.
• Each order shows a different colour band. Every order spans the full wavelength range; they just overlap until separated.
• The orders are straight lines. Cross-dispersion and the camera optics curve them into gently arced strips on the detector.
• An echelle measures velocity directly. It measures position; calibration lamps, iodine cells, or laser combs convert pixel shifts into velocities.
• More orders means more total light. Light is divided among orders; the blaze angle concentrates throughput into the orders you actually use.