VIMOS: ESO’s Multi-Object Spectrograph
- VIMOS is a wide-field optical spectrograph with a four-channel design and custom slit masks, serving as the backbone for surveys like VIPERS, VUDS, and GOODS/VIMOS.
- Its flexible configurations support multi-object slit and integral-field modes over near-UV to red optical wavelengths with resolutions ranging from ~180 to 580.
- The instrument’s legacy includes advancements in redshift surveys, cosmic web reconstruction, and innovative calibration pipelines addressing slit geometry and angular incompleteness.
VIMOS, the Visible Multi-Object Spectrograph on ESO’s Very Large Telescope Unit Telescope 3 (“Melipal”), was a wide-field optical spectrograph built around high-multiplex slit spectroscopy and, in a distinct configuration, integral-field spectroscopy. In the literature represented here, it appears as the instrumental backbone of several major extragalactic programs—notably VVDS, GOODS/VIMOS, VIPERS, and VUDS—because it combined a four-channel focal-plane geometry, custom slit-mask operation, and multiple grism configurations with wavelength coverage extending from the near-UV to the red optical. Its scientific role was correspondingly broad: large redshift surveys at intermediate redshift, ultra-deep spectroscopy of galaxies at $2
1. Instrument architecture and observing modes
VIMOS was a four-channel imaging spectrograph whose focal plane was divided into independent quadrants. Across the survey papers, each quadrant is described as covering roughly arcmin, with a single pointing spanning about $218$–$224$ arcmin; the quadrants were separated by an unobserved cross-shaped gap about $2$ arcmin wide (Scodeggio et al., 2016, Balestra et al., 2010, Micheletti et al., 2014). In slit-mask mode, the entrance focal plane accepted custom masks, allowing simultaneous spectroscopy of tens to over a hundred targets per quadrant, and in maximum multiplex configurations the instrument could deliver up to about $1000$ spectra in one exposure (Scodeggio et al., 2016, Garilli et al., 2013).
The dominant use of VIMOS in the cited work is multi-object slit spectroscopy. Slit masks were prepared from direct imaging with the ESO mask-preparation software, and several programs adopted a $1$ arcsec slit width as a standard operating point (Garilli et al., 2013, Scodeggio et al., 2016, Balestra et al., 2010). VIPERS exploited an aggressive short-slit strategy to raise the median number of slits to about $87$ per quadrant, or about 0 per pointing (Garilli et al., 2013). VUDS likewise optimized slit lengths down to 1 arcsec to maximize multiplex (Fevre et al., 2014).
A separate VIMOS integral-field mode is documented for nearby luminous and ultraluminous infrared galaxies. In that configuration the IFU field was 2 arcsec with 3 spectra per pointing, sampled at 4 arcsec per fiber, and combined through a four-point dither into a 5 “super-cube” containing 6 spectra (Zaurin et al., 2010). This established VIMOS not only as a survey spectrograph but also as a spatially resolved optical spectroscopy instrument.
2. Spectroscopic configurations and performance envelope
The instrument was used with several grisms and filters, and its scientific behavior was therefore configuration-dependent rather than fixed by a single resolution or wavelength range. The standard resolving-power definition used in the survey literature is 7, while redshifts follow the usual relation 8 (Guzzo et al., 2013, Franzetti et al., 2014).
A concise view of the principal configurations documented in these papers is given below.
| Configuration / program | Wavelength coverage | Resolution or related metric |
|---|---|---|
| VIPERS LR-Red | 9–0 Å | 1–2 |
| VUDS LRBLUE + LRRED | 3–4 Å | 5 |
| GOODS LR-Blue | 6–7 Å | 8 |
| GOODS MR orange / Fornax MR | 9–$218$0 Å | $218$1 or $218$2 |
| VIMOS IFU HR-Orange | $218$3–$218$4 Å | FWHM $218$5 Å |
In VIPERS, VIMOS was configured with the LR-Red grism, $218$6 arcsec slits, and wavelength coverage $218$7–$218$8 Å; PDR-2 quotes $218$9, while the main survey and filament-analysis papers quote $224$0 and $224$1, respectively (Scodeggio et al., 2016, Guzzo et al., 2013, Malavasi et al., 2016). In VUDS, the low-resolution LRBLUE and LRRED settings were combined to cover $224$2–$224$3 Å at $224$4, and one ECDFS pointing used the MR grating at $224$5 over $224$6–$224$7 Å (Fevre et al., 2014). GOODS/VIMOS split its program between LR-Blue and MR orange, targeting complementary redshift regimes (Balestra et al., 2010). The Fornax compact-stellar-system survey used the MR grating with the CG475 filter over $224$8–$224$9 Å at 0 and 1 Å/pixel (Pota et al., 2018).
Exposure times were likewise survey-specific. VIPERS used 2 s per pointing, split into five 3-minute exposures (Garilli et al., 2013). GOODS/VIMOS used 4 h per mask in both LR-Blue and MR campaigns (Balestra et al., 2010). VUDS pushed to approximately 5 h per grism and per pointing, with the overlap region effectively receiving about 6 h in the combined spectrum (Fevre et al., 2014). This depth enabled continuum sensitivity around 7 mag and detections of weak diagnostic lines in faint galaxies (Amorín et al., 2014, Fevre et al., 2014).
Detector state affected performance materially. During the 2010 upgrade, VIMOS received red-sensitive thicker CCDs and improved flexure compensation; VIPERS reports an increase in red efficiency by up to a factor of about 8 and a reduction in fringing, while VUDS notes that sensitivity at 9 Å increased by about a factor of $2$0 and fringing above $2$1 Å was significantly reduced (Scodeggio et al., 2016, Fevre et al., 2014).
3. Major survey implementations
VIMOS’s place in extragalactic astronomy is clearest through the surveys built around it. VIPERS used VIMOS to construct a magnitude-limited spectroscopic survey to $2$2, with colour pre-selection to suppress $2$3 galaxies, over the CFHTLS-Wide W1 and W4 fields. The completed public release spans about $2$4 deg$2$5, with $2$6 VIMOS pointings and a complete spectroscopic sample of $2$7 galaxies plus $2$8 other objects; the average effective sampling rate is about $2$9 over $1000$0 (Scodeggio et al., 2016). Earlier VIPERS public releases already distributed $1000$1 spectroscopic measurements and then the associated calibrated 1D and 2D spectra (Garilli et al., 2013, Franzetti et al., 2014).
VUDS used VIMOS for a different regime: ultra-deep spectroscopy of very faint galaxies at early cosmic epochs. The survey targeted about $1000$2 galaxies over about $1000$3 deg$1000$4 in COSMOS, ECDFS, and VVDS-02h, with a main spectroscopic sample peaking at $1000$5–$1000$6 and extending to reliable redshifts beyond $1000$7 (Fevre et al., 2014). Its first CANDELS-area public release distributed spectra and spectroscopic redshifts for $1000$8 objects, including about $1000$9 galaxies at $1$0 and $1$1 at $1$2 (Tasca et al., 2016).
GOODS/VIMOS used the instrument in a more targeted field-survey mode. The completed campaign comprised $1$3 masks and yielded $1$4 spectra, with $1$5 LR-Blue redshifts and $1$6 MR redshifts, complementing the FORS2 ESO/GOODS spectroscopy and extending spectroscopic coverage in the CDFS/GOODS-S region (Balestra et al., 2010).
Earlier VVDS work demonstrates the same instrumental logic at somewhat lower sampling. In the VVDS-02h Deep field, VIMOS underpinned a purely flux-limited survey $1$7, from which a group catalogue of $1$8 systems at $1$9 was constructed (0911.3740).
Outside the canonical galaxy-redshift surveys, VIMOS also supported targeted wide-field programs. In the Fornax cluster core, $87$0 masks produced about $87$1 low-resolution spectra and a final catalogue of $87$2 globular clusters and $87$3 ultra-compact dwarfs (Pota et al., 2018). In IFU mode, it produced a spatial atlas of continuum and H$87$4 emission for $87$5 local luminous and ultraluminous infrared systems (Zaurin et al., 2010).
4. Calibration, extraction, and public data products
The reduction environments associated with VIMOS were themselves an important part of its scientific infrastructure. VIPERS reductions were automated within a pipeline derived from VIPGI and managed in the EasyLife environment; redshifts were measured with EZ and then visually validated (Garilli et al., 2013, Scodeggio et al., 2016). GOODS/VIMOS also used VIPGI, with visual feature identification supplemented by EZ template cross-correlation in ambiguous cases (Balestra et al., 2010). VUDS DR1 used VIPGI together with EZ and extensive visual inspection of both 1D and 2D spectra (Tasca et al., 2016).
Calibration chains were survey-specific but structurally similar. VIPERS PDR-1 describes tracing of 2D spectra, first-pass sky subtraction, wavelength calibration from arc lamps, rigid offsets based on sky lines when needed, co-addition, optimal extraction, and flux calibration through the instrument sensitivity function, followed by normalization to the CFHTLS $87$6-band magnitude (Garilli et al., 2013). The later public releases delivered wavelength- and flux-calibrated 1D spectra, optional 2D spectra, and the ancillary arrays required for downstream analysis: wavelength, cleaned flux, flux uncertainty, sky intensity, original flux, and a cleaning mask (Scodeggio et al., 2016).
The first VIPERS public spectra release is especially explicit about per-object products. Each FITS binary table includes columns labeled WAVES, FLUXES, EDIT, NOISE, SKY, and MASK; FLUXES is the wavelength-calibrated, sensitivity-corrected 1D spectrum normalized to CFHTLS $87$7-band photometry, while EDIT is a cleaned version in which fringing residuals, poor subtraction of strong sky lines, and zero orders are mitigated by manual edits plus an automatic PCA-based reconstruction (Franzetti et al., 2014). FITS headers include the VIPERS ID, sky coordinates, selection magnitude, spectroscopic redshift and reliability flag, normalization, and exposure time (Franzetti et al., 2014).
VUDS DR1 added a different calibration emphasis because of its blue coverage and very long integrations. The LRBLUE and LRRED spectra were resampled and joined through the $87$8–$87$9 Å overlap region; atmospheric refraction and extinction were corrected by comparison with broadband photometry, with slit-loss estimates based on a refraction pseudo-spectrum and an elliptical model of each galaxy image through the 00 arcsec slit. After correction, spectroscopic fluxes agreed with broadband photometry to better than 01 rms for 02 Å and better than 03 rms for 04 Å (Tasca et al., 2016).
Several releases also documented noise models and flexure handling. VIPERS PDR-1 writes the sky-residual and noise estimators explicitly and notes wavelength-calibration residuals with a mode of 05 Å (Garilli et al., 2013). The Fornax survey found that time-dependent flexure could shift the wavelength zero-point across exposures; it therefore measured residual sky-line offsets after stacking and applied multiplex-by-multiplex corrections (Pota et al., 2018).
5. Cosmological and environmental measurements enabled by VIMOS
The combination of VIMOS multiplexing and VIPERS sampling density made intermediate-redshift large-scale-structure studies possible at a level previously associated mainly with local surveys. A prominent example is the reconstruction of the filamentary cosmic web at 06 using DisPerSE, where VIPERS yielded the first quantitative detection of large-scale filamentary structure at that redshift and showed that massive or quiescent galaxies lie closer to filament axes than less massive or active galaxies (Malavasi et al., 2016).
Void science followed a parallel track. Using VIPERS, empty-sphere algorithms identified and characterized voids at 07, producing a catalogue of 08 maximal spheres with radii broadly spanning 09–10 Mpc and a void–galaxy cross-correlation function showing line-of-sight anisotropy consistent with outflow velocities (Micheletti et al., 2014). A subsequent completed-VIPERS analysis measured the redshift-space distortion parameter around voids as 11 and inferred 12 at 13 (Hawken et al., 2016).
VIMOS sampling geometry also motivated methodological work. The four-quadrant footprint, slit-collision constraints, TSR/SSR variations, and the unobserved cross imprint a non-trivial angular selection function on galaxy catalogues. VIPERS analyses addressed this in several ways: Bayesian Wiener-filter reconstruction of the redshift-space density field jointly inferred the anisotropic power spectrum, luminosity function, bias, and density field while modeling the inhomogeneous selection function, yielding 14 with 15 uncertainty at redshift 16 (Granett et al., 2015); count-in-cell studies showed that at low sampling the underlying one-point density PDF is better represented by a Gamma expansion than by a Skewed Log-Normal, and explicitly corrected the “angular selection effect of the VIMOS instrument” (Bel et al., 2015); and pairwise inverse probability weighting was shown to recover unbiased clustering from VIPERS slit assignment, confirming that earlier approximate corrections were already sufficient within VIPERS statistical errors (Mohammad et al., 2018).
Earlier VVDS work shows that these cosmological applications emerged from a longer VIMOS lineage. The VVDS group catalogue trained the Voronoi-Delaunay Method on Millennium-based mocks to recover group counts and velocity dispersions under VIMOS-like sampling, finding 17 groups at 18 with global completeness 19 and purity 20 (0911.3740).
6. Galaxy populations, diagnostics, and instrument-driven systematics
VIMOS datasets were also used to construct physically organized views of galaxy populations. In VIPERS, a PCA classification of 21 secure spectra in 22 used an iterative masked repair of noisy or gappy spectra and found that the first three eigenspectra contain about 23 of the total variance, enabling a two-parameter classification space that separates early, intermediate, late, and starburst galaxies (Marchetti et al., 2012). This suggests that, despite low-resolution and imperfect spectra, a large VIMOS survey could support homogeneous spectral typing at scale.
At higher redshift, VUDS used the depth of VIMOS to identify 24 extremely compact, metal-poor, star-forming dwarf galaxies at 25, with 26 kpc, 27, high ionization conditions, and 28 between 29 and 30 (Amorín et al., 2014). Another VUDS analysis classified the Ly31 line morphology of 32 Ly33 emitters from a parent sample of 34 star-forming galaxies at 35, finding about 36 coincident, 37 extended, and 38 offset emitters relative to the UV continuum (Ribeiro et al., 2020).
VIPERS spectroscopy also supported line-diagnostic studies of AGN and quenching. A survey of [NeV] 39 emitters at 40 used the VIMOS wavelength range to isolate AGN in composite galaxies and found younger stellar ages and higher [OII] luminosities for active galaxies in the green valley and blue cloud than in matched inactive controls (Vergani et al., 2017).
These uses make the instrument’s trade-offs explicit. VIMOS frequently operated at low resolution—41–42 in VIPERS and VUDS—which was adequate for robust redshift determination and large statistical studies but limited detailed line-profile diagnostics (Guzzo et al., 2013, Fevre et al., 2014). Fringing, residual sky features, zero-order contamination, and slit-based incompleteness were persistent concerns, especially before the 2010 detector upgrade; the literature responded with PCA-based spectral cleaning, explicit TSR/SSR/CSR weights, mock-catalog validation, and gap-aware topology or void-finding methods rather than assuming ideal completeness (Franzetti et al., 2014, Scodeggio et al., 2016, Micheletti et al., 2014). A plausible implication is that VIMOS’s scientific legacy is inseparable from the correction frameworks developed around it: the instrument did not merely generate large spectroscopic samples, but also forced precise treatments of mask geometry, angular incompleteness, and redshift reliability that became part of the measurement itself.