≡ Menu

First Light for ESPRESSO

What great news that ESPRESSO, the Echelle SPectrograph for Rocky Exoplanet and Stable Spectroscopic Observations, has just achieved ‘first light.’ The spectrograph is installed on the European Southern Observatory’s Very Large Telescope at the Paranal Observatory in northern Chile and its powers are prodigious. For ESPRESSO makes it possible, for the first time, to combine the light of all four telescopes at the VLT. This creates an instrument with the light collecting power of a 16-meter telescope, a major enhancement to the exoplanet hunt.

Image: The room where the light beams coming from the four VLT Unit Telescopes are brought together and fed into fibres, which in turn deliver the light to the spectrograph itself in another room. One of the points where the light enters the room appears at the back of this picture. Credit: ESO/P. Horálek.

Thus the enthusiasm of lead scientist Francesco Pepe (University of Geneva):

ESPRESSO isn’t just the evolution of our previous instruments like HARPS, but it will be transformational, with its higher resolution and higher precision. And unlike earlier instruments it can exploit the VLT’s full collecting power — it can be used with all four of the VLT Unit Telescopes at the same time to simulate a 16-metre telescope. ESPRESSO will be unsurpassed for at least a decade — now I am just impatient to find our first rocky planet!”

Image: This colorful image shows spectral data from the First Light of the ESPRESSO instrument on ESO’s Very Large Telescope in Chile. The light from a star has been dispersed into its component colours. This view has been colourised to indicate how the wavelengths change across the image, but these are not exactly the colours that would be seen visually. Close inspection shows many dark spectral lines in the stellar spectra and also the regular double spots from a calibration light source. The dark gaps are features of how the data is taken, and are not real. Credit: ESO/ESPRESSO team.

We’re going to be hearing a lot from ESPRESSO because it will greatly improve our powers of radial velocity observation. Remember what we are doing when we use these techniques. Radial velocity involves extracting the tiny Doppler signature of star motion as the star is pulled first one way, then another, by the planets around it. Compared to the size of the star, the movements are small but we can trace them in the star’s light spectrum.

Repeating changes to the spectrum as it shifts toward red, toward blue, back toward red, give us the data we need to identify a planet, and until the Kepler mission came along, radial velocity was the primary means we used to find such worlds. 51 Pegasi’s planet, the first found to orbit a main-sequence star, was found using radial velocity methods in 1995. Now we use a mixture of methods including transit studies, direct imaging and gravitational microlensing.

Image: The Echelle SPectrograph for Rocky Exoplanet and Stable Spectroscopic Observations (ESPRESSO) successfully made its first observations in November 2017. Installed on ESO’s Very Large Telescope (VLT) in Chile, ESPRESSO will search for exoplanets with unprecedented precision by looking at the minuscule changes in the properties of light coming from their host stars. This view shows the inside of one of the ESPRESSO front-ends where all the active components of the spectrograph are located. Credit: Giorgio Calderone, INAF Trieste.

Combining methods can be hugely useful, for while radial velocity lets us measure the mass and orbit of the planet (I won’t, for the purposes of this post, get into the problem that RV mass measurements can produce only minimum mass estimates), transits can help us deduce its density. But for those planets that do not transit, we’d like to move further and further down the scale, making detections of ever smaller worlds, terrestrial-class planets perhaps like our own.

Image: the moment of first light, with the team jubilant in the VLT control room. Credit: Giorgio Calderone, INAF Trieste.

From an ESO fact sheet on ESPRESSO:

The radial velocity technique has been so far the most productive in terms of extra-solar planet detections. Low mass planets (one to few Earth masses) are especially interesting because according to formation models they could represent the bulk of the planet population. However they are more elusive and require extremely stable instruments. The HARPS instrument, with a precision better than 1m/s, has discovered up to now the vast majority of planets with masses smaller than Neptune, giving an invaluable experience in view of the realization of more precise instruments. With a radial velocity precision better than 10cm/s, an Earth mass planet in the habitable zone of a low mass star can be detected.

Exactly so. ESPRESSO picks up where HARPS left off. While HARPS could achieve a precision of one meter per second, ESPRESSO gets us down below 10 centimeters per second. The upscaling is the result of ESPRESSO’s placement to tap the four VLT telescopes as well as advances in spectroscopic technology. The benefit will be in characterizing much less massive planets unavailable for our scrutiny through transits or direct imaging, further bulking up the exoplanet catalog and deepening our statistical analysis of planets near us in the galaxy.


Comments on this entry are closed.

  • Tom Mazanec December 6, 2017, 13:02

    Does the turbulence of the star’s photosphere cause problems with such velocity measuring? If so, how is it solved?

    • Ashley Baldwin December 6, 2017, 14:13

      It does indeed . Very much so. The 10cm/s precision mooted would only be practicable for studying the most stable of stars . To be fair that is going to be the sort of Sun like G dwarfs that might harbour an “eta Earth “. That said it does look as if the Sun is an unusually quiescent star even by most standards .

      As an multi purpose device, in the shorter term the ESPRESSO RV strategy appears to be built around the targeting of “quiet” stars . It will not initially employ the sort of photospheric modelling algorithms planned for use with the bespoke RV EXPRES spectrograph soon to be deployed on the 4.2m Discovery telescope and central to the “100Earths” project. This will look at a far more diverse sample of nearby stars including much more active K and M dwarfs .

      • Ronald December 7, 2017, 6:59

        When you mention ‘the sort of Sun like G dwarfs’ and ‘targeting of “quiet” stars ‘, in connection with ESPRESSO, do you also have the solar twin sample in mind that is now the target in the Solar Twin Planet Search project, using HARPS?
        In other words, will ESPRESSO indeed be used ‘to pick up where HARPS left off’ for this solar twin search as well?
        See for instance: “The Solar Twin Planet Search I; Fundamental parameters of the stellar sample”, in which 88 solar twins are mentioned.

        • Ashley Baldwin December 7, 2017, 14:16

          All that I’ve read from the ESO doesn’t amount to much more than just targeting quiet stars . I can’t see it being as simple as just picking up from HARPS though . At absolute best it had a sensitivity of 50cm/s and even then struggled with more active stars . ( the sort of late M dwarfs where this much lower sensitivity could theoretically locate terrestrial planets in close “hab zone ” orbits )

          I suppose that ESPRESSO could target a small number of less optimal but high profile neighbouring stars and do coincidental near continuous photometry with say the Las Cumbres telescope global network. This was the highly bespoke strategy adopted by the Pale Red Dot campaign in order to pick the Proxima b RV signal out of its parent star’s active photosphere .( and now planned for other neighbouring red dwarfs like Barnard’s Star) . Effective but labour intensive however and not suitable for a high return observation run like “100Earths”.

          • Andrew December 31, 2017, 10:27

            Ashley Baldwin obviously has some kind of grudge against, or is envious of ESO. Pale Red Dot completely relied on ESO’s HARPS telescope to help confirm Proxima b. Of course, VLT ESPRESSO is optimised for detecting planets around quiet stars but it will still be the most powerful tool for detecting planets around any star within 40 light years. And, though HIRES will be deployed on the ELT, CODEX is definitely going to happen; no one is even seriously suggesting that it will not be ready on schedule.

    • DJ Kaplan December 6, 2017, 15:00

      From what I understand, the RV measurement on this instrument depends on the red shift behavior of the star’s overall light.

  • andy December 6, 2017, 13:20

    I wonder what’s going to win out between finding smaller planets and a more detailed view of stellar jitter.

    • Ashley Baldwin December 7, 2017, 14:18

      Absolutely . But I suppose over time the technology and strategy will be refined to deliver . The Jitter Bug eh ?

  • DJ Kaplan December 6, 2017, 13:22

    Great news! I like to tell people that, with the technology we formerly possessed, it would not be possible to detect the Earth from the point of view of Alpha Centauri. Now, that may be within our reach.

  • Ronald December 6, 2017, 14:36

    Just wow! Finally.
    22 years since 51 Pegasi’s planet, we have come a long way.
    10 cm/s is actually down to Earth-sized around Sun-sized.
    Noise may become a problem, but there seems to be this ‘comb’.
    Now waiting for ELT and CODEX….

    • Ashley Baldwin December 6, 2017, 19:10

      CODEX is not going to happen. The plan now is to replace it with the wider bandwidth ( 0.35-2.5 micron ) and much more versatile HIRES spectrograph as a first phase instrument instead . Still has a resolution of at least 100000 though and will be more than capable of precision RV on even dim stars . Unlike CODEX , it will also eventfully it will link up with the XO adaptive optics and polarimetric imager originally intended for EPIC to allow direct imaging and high dispersion spectroscopy over a biosignature rich bandwidth . Not first phase though given significant technological development needed and high cost . The end result will likely be the most potent exoplanet characterising instrument on the ground or in space for several decades .

      Given the capability of combining all 4 Unit telescopes EXPRESSO VLT is likely to be THE RV instrument for the foreseeble future , even after the advent of the ELTs . It’s just a pity there are no plans ( or thanks to the E-ELT, funds either ) to combine the light of the scopes at a coherent focus to effectively create a partially filled aperture style interferometer.

      • Ronald December 7, 2017, 4:10

        Thanks for the update!
        Will the precision of the foreseen HIRES like spectrograph be below 10 cm/s?

        • Ashley Baldwin December 7, 2017, 14:31

          A spectroscopic resolution of over 100000 ( the HIRES working group haven’t determined the final instrument performance yet ) I’m sure could deliver theoretical sensitivity on a par with CODEX. But with its bandwidth extending into the NIR ,where there is generally less stellar photospheric noise , HIRES may offer much more than the VIS only bandwidth of CODEX. Unlike CODEX it is much more of an all round spectrograph though ( hence its preference ) so it’s unclear how much time it can devote to exoplanet velocimetry. I suspect that HIRES’ role in exoplanet science will be via direct imaging and high dispersal spectroscopy however , with future RV work left to instruments like ESPRESSO and EXPRES on smaller but still capable telescopes with less of a premium on time than the E-ELT.

          • Ronald December 12, 2017, 7:56

            I just read two recent publications on ESPRESSO (https://arxiv.org/abs/1401.5918, https://arxiv.org/abs/1711.05250) and understood that 10 cm/s will be the lowest detection limit foreseen for this instrument.
            This is approx. the RV for Earth around the Sun. But since most planets will not orbit in a plane exactly along the line of sight, there is also the m sin i correction (in other words, by far most Earth mass planets will show a smaller RV).
            And then there is stellar noise (jitter).
            All together, this makes me realize that even with ESPRESSO it will be very hard to detect an Earth-sized planet in an Earth-like orbit around a Sun-like star, and by far most of these will still remain undetected.

  • Project Studio December 6, 2017, 21:57

    How does the spectroscopic capability apply to the radial velocity analysis? Is it just that it is easier to detect small shifts in the higher frequency spectral lines? Also, considering the wide effective aperture, is there a potential for wider-field surveys of regions rather than individual stars?

  • Jean Schneider December 7, 2017, 5:40

    What is also interesting with ESPRESSO is that it will be coupled to
    the SPHERE camera to improve the detection by direct imaging,
    after an idea proposed in 2007:
    For the implementation see http://adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2016arXiv160903082L&db_key=PRE&link_type=ABSTRACT&high=57d693ab9502994

    • Ashley Baldwin December 7, 2017, 14:40

      This has been posited by a number of astronomers more recently, especially after the the discovery of Proxima b. Such a combination on an 8m scope should just be able to image the planet and perhaps a few others around nearby M dwarfs. A small return though and METIS on the E-ELT will have similar but much greater sensitivity on the E-ELT as a first phase instrument too. The issue with SPHERE-ESPRESSO as ever is cost , at a time when the ESO are prioritising their limited budget for the E-ELT. The technology of SPHERE will form the basis of the future imager planned to link with the E-ELT HIRES spectrograph.

  • Harry R Ray December 7, 2017, 10:28

    There are three scenarios for measuring the masses of the TRAPPIST-1 planets with ESPRESSO. Scenario number one: Able to do it with the VLT> Scenario number two: Due to the extreme FAINTNESS of TRAPPIST-1, UNABLE to do it with the VLT, and therefore will have to wait for the E-ELT to come on line. Scenario number three: Due to the extreme REDNESS of TRAPPIST-1, unable to do it with any OPTICAL telescope, no matter how powerful. If any reader(OR Paul Gilster)knows which scenario it is, PLEASE POST IT HERE!

    • Harry R Ray December 7, 2017, 11:32

      One of the FIRST TARGETS for ESPRESSO should be K2-18b, a 2.2 Rearth planet with a Teff CLOSER TO Earth’s Teff than ANY OTHER Earth-sized or Super-Earth planet! A mass range has just been determined, with the MOST LIKELY mass at 8.0 Mearth, and a MOST LIKELY density at 3.7 gcm3. However, the RANGE is plus or minus 1.9 Mearth, meaning that this planet can still be either predominantly ROCKY(@9.9Mearth), a “water world”, or predominantly GASSY(@6.1Mearth). If ESPRESSO determines that the mass is closer to 9.9 Mearth than 8.0 Mearth, it could have STAGGERING IMPLICATIONS for planets like Kepler 22b, which is too far NORTH for ESPRESSO to CURRENTLY observe. ALSO: K2-18 is only the SECOND system(Kepler 20 being the FIRST) where a transiting planet orbits OUTSIDE of a NON-transiting one!

    • Ashley Baldwin December 7, 2017, 14:44

      No need either. There are further transit observations planned with both Hubble and Spitzer over the next few years and these will significantly refine the ephemerides of the TRAPPIST planets . This will in turn allow accurate constraint of their masses via the alternative technique of transit timing variations . No worries about the orbital inclination uncertainty then either.

      • Harry R Ray December 8, 2017, 10:34

        Not so fast! Google ArXiv: 1711.05691 and click on the FIRST item on the list page(The Transit Light Source Problem:False Spectral Features and…). I NEVER bought into the ABSURDLY LOW densities derived for some of the TRAPPIST-1 planets from SUPPOSED small TTV’s in the K2 light-curve data! NEITHER did Gillon et al, who still use their own preliminary Spitzer data on their website. According to this paper, BOTH of these data sets may be incorrect.

    • Harry R Ray December 29, 2017, 10:43

      Regardless of which of THESE scenarios is the correct one, the masses of the TRAPPIST-1 planets WILL be revealed via radial velocity by the GIANO-B high resolution infrared spectrograph when it comes on line. “A Microphotonic Astrocomb.” by E. Obrzud, M. Rainer, A. Harutyunyan, M. H. Anderson, M. Geiselmann, B. Chazelas, S. Kundersmann, S. Lecompte, M. Cessoni, A. Ghedina, E. Molinari, E. Pepe, F. Wildi, F. Bouchey, T. I. Kippenberg, T. Herr

  • Spaceman December 7, 2017, 11:30

    So, when noise like stellar jitter is taken into account, how much of an improvement will this exciting new instrument be over its predecessor HARPS?

    • Ashley Baldwin December 7, 2017, 14:46

      Unfortunately stellar jitter very much needs to be taken into account and its noise makes the 10cm/s ESPRESSO stretch goal very much theoretical in all but the most optimal cases.

  • DJ Kaplan December 7, 2017, 13:26

    Maybe now we can begin to move away from our fixation with red dwarf stars (nicknamed the Galactic Hanging Fruit) and begin to find truly Earth-like worlds…

  • Michael December 7, 2017, 15:17

    Speaking of VLT, a few days ago ESO published a set of images of four main belt asteroids (incl. Pallas): http://www.eso.org/public/images/potw1749a/?lang

  • Michael Fidler December 8, 2017, 22:35

    Looking with great anticipation for the results from ESPRESSO, SPHERE and RV EXPRES “100Earths” project. Wondering what the other currently operating great observatories are using and planning for the next decade. (Keck, Large Binocular, Gemini, Subaru, Magellan, Etc.)

    This just in,
    NASA Hosts Media Teleconference to Announce Latest Kepler Discovery.

    “NASA will host a media teleconference at 1 p.m. EST Thursday, Dec. 14, to announce the latest discovery made by its planet-hunting Kepler space telescope. The discovery was made by researchers using machine learning from Google. Machine learning is an approach to artificial intelligence, and demonstrates new ways of analyzing Kepler data.”


    AI is Now Helping Kepler Find Planets.


    This also has some good links to articles related to it, especially this one;

    Now if you have $3,000.00 to spare, you can do this on your own:

    NVIDIA’s ‘most powerful GPU’ ever is built for AI.



    ‘First rule in government spending; why build one, when you can have two at twice the price?’
    “The Titan V is available today and is limited to two per customer.”

    • Project Studio December 9, 2017, 0:25

      Any pattern recognition through machine learning will depend on training the algorithm with examples of positive features – we will still need to guess what sorts of ‘signal’ represent a positive sign of alien METI so the AI can learn to recognise it. Astronomers have recently used machine learning to search astronomical data for features that could indicate the types of optical distortions created by gravitational lenses – this works because samples of actual gravitational lenses are used to train the algorithm.

      Its always been the case that we need to decide in advance what to look for – whether that is periodic blips in the quite Hydrogen absorption band or some other type of signal. We’re still far from General AI at least in a practical application sense.

    • Harry R Ray December 10, 2017, 15:10

      This will obviously be regarding (a)CONFIRMED planet(s) and not just “candidates”, but; if you remember, earlier this year, machine learning turned up 20 COMPLETELY NEW habitable zone planet candidates, and since then, HST and SST have been following up on SOME of them! My guess is that this has something to do with KOI7923.01.

      • Harry R Ray December 14, 2017, 10:44

        Proxima Centauri MAY have already stolen Kepler’s thunder today!!! ArXiv: 1712.04483. “A Candidate Transit Event around Proxima Centauri.” by Yeting Li, Gudimundur Stefansson, Paul Robertson, Andrew Monson, Caleb Canas, and Suvrath Mahadevan. A POSSIBLE hot(mos likely TOO hot to be habitable)”water world”(i.e. LESS than 0.4 Mearth)in a 2-4 day orbit! The derived radius is ~1Rearth, SIGNIFIGANTLY less than the POSSIBLE Proxima b CANDIDATE transits(from BOTH of the Kipping/MOST and the indipendant Liu et al observations(~1.25Rearth)!

  • ljk December 18, 2017, 14:49

    A new approach for detecting planets in the Alpha Centauri system

    By Jim Shelton

    December 18, 2017


    • Harry R Ray December 18, 2017, 18:29

      What is very exciting is that the authors mention the possibility of MULTIPLE <0.5 Earth-mass planets instead of JUST ONE! If the LATEST putative transit(mentioned ABOVE)turns out to be REAL(and I believe it has a FAR BETTER CHANCE of being real than ANY of the putative transits of Proxima b being real)there could be A COUPLE MORE INSIDE the orbit of Proxima b, making the system VERY SIMILAR to that of TRAPPIST-1! It still AMAZES BE that Proxima Centauri was NOT on the target list of the TRAPPIST original campaign!

  • ljk January 4, 2018, 14:16