HPF Goes On Sky!

The HPF team poses in front of the instrument after a successful delivery to HET.


Things have been quiet here on the HPF blog for the last few months, but that has certainly not been the case in our lab!  Our team has been working diligently, making the final push to deliver HPF to HET and commission the instrument.  In this post, we are excited to announce that HPF has been installed at HET, and is now acquiring spectra on sky!  Here are some highlights from delivery and commissioning.

Delivery and Installation

HPF was delivered from Penn State to McDonald Observatory on a temperature-controlled, air ride truck that was routed to avoid as many potholes and speed bumps as possible.  After a three-day trip, it was carefully offloaded at HET and placed in the telescope basement.  We made sure to capture all the action from multiple camera angles:

Once we had the instrument placed in the basement, we opened the vacuum chamber to verify that the optics had made the trip intact.  While we could only admit a few team members for this task, rest assured that everyone else was nervously pacing the floor just outside!

Once the instrument was placed safely in its permanent home, the team spent approximately three weeks connecting and testing the various subsystems.  These subsystems include the environment control system, the fiber optic feed, the calibration suite, and the near-infrared detector, which we have detailed in previous posts here.

Of course, the new wrinkle this time is that HPF is now attached to an actual telescope!  Ensuring the whole system operated properly required sending a few brave team members above the primary mirror on a lift.

Outfitted with the proper personal protective equipment, HPF team members Chad Bender, Suvrath Mahadevan, and Sam Halverson accompanied HET Mechanical Engineer Emily Mrozinski to the telescope’s prime focus structure.

To put a long story short, HPF is now happy and healthy in the HET basement, and it is now time to get some spectra.

First Light

On the night of November 29-30 2017, HPF acquired first light on the M dwarf GJ 3470, which hosts a transiting exoplanet that astronomers believe has a blue sky.  Here is what the star’s spectrum looks like to HPF:

HPF’s first-light spectrum of GJ 3470.  Each set of three lines is a spectral order–a small section of the full spectrum–with the star’s spectrum in the middle, the spectrum of the night sky above it, and our wavelength calibration source below.

The first light spectrum wouldn’t have happened without great support from all of the talented staff at the Hobby-Eberly Telescope:

During first light observations on the night of Nov 29th: telescope operator Emily Bevins and resident astronomer Steve Odewahn hard at work to ensure HPF records high-quality spectra !

The following day, a subsection of the HPF team celebrated first light at the HET control room, alongside HET telescope operator Justen Pautzke, resident astronomer Sergey Rostopchin, Professor Gary Hill—and senior research scientist Phillip MacQueen who gladly took the photo for us !

Celebrating first light at the Hobby-Eberly Telescope control room!

HPF’s science commissioning has just started, and we will provide more updates as we fully characterize the system’s performance.  For now, though, it is an exciting milestone to get starlight on the detector.  With tantalizing new exoplanets being discovered all the time, and TESS just around the corner, it is a good time to be on sky!

Posted in HPF Hardware, Project Development | Leave a comment

The Camera of HPF

Introduction: Spectrometer Cameras

A spectrometer, as the name implies, records a ‘spectrum’ of an object. This spectrum, in its most basic form, is just a series of images of the instrument entrance aperture (whether it be a star, slit, optical fiber, or otherwise) that has been spatially separated based on the intrinsic color distribution of the source. The separation of these monochromatic (single-wavelength) images depends on the dispersion properties of the spectrometer.

Simulated star and wavelength calibration source spectra, as recorded by the HPF spectrometer. The calibration source spectrum (center channel), in this case a highly specialized laser source which only emits at discrete wavelengths, produces a spatially separated set of images of the spectrometer entrance slit. On the other hand, starlight (bottom channel), contains flux across a broad range of frequencies. This produces a continuous spectral image across each diffraction order.

The optic at the heart of the spectrometer, the high dispersion echelle grating, provides the chromatic dispersion required to separate the individual colors encoded in the incoming starlight, but it is the spectrometer camera that ultimately produces our final image of the stellar spectrum.

Cartoon overview of the HPF spectrometer optical train. Light from the telescope exits the fiber, gets dispersed by a pair of diffraction gratings (first the high dispersion Echelle grating, then the volume phase holographic grating), and is finally imaged onto our infrared detector array by the HPF camera. The camera assembly provides the needed focusing mechanism to deliver our final spectral image.

The spectral resolution of the spectrometer, which tells us how well we can separate individual colors, is entirely set by the properties of the entrance aperture and diffraction grating(s). However, the spectrometer camera must also have pristine image quality, or else our spectral resolution will suffer. Any blurring of the image of the spectrum due to uncorrected aberrations within the camera will degrade our ability to separate individual colors of the star.

Example of image quality degrading spectral resolution. On the left shows a near-diffraction-limited image of our spectrometer slit, while the right shows a highly aberrated image. The effective spectral resolution of the aberrated slit image (right) is nearly 5% lower.

The HPF Camera

Our spectrometer resolution directly affects how well we can measure minute stellar radial velocity shifts, so typical high resolution Doppler spectrometers require specially designed cameras that maintain excellent image quality across all of the diffraction orders. This high level of image quality is also needed to lower our sensitivity to illumination changes coming from our optical fiber delivery system.

Image quality of the HPF camera across the wavelengths that fall onto our infrared detector. The grey boxes represent individual detector pixels, magnified by a factor of 1000x. The colored points show the aberration distribution produced by the camera and upstream optics (a ‘perfect’ imaging system would have no geometric aberrations). As you can see, the expected aberrations in the system are, for the most part, far below what we’d be able to measure in single pixel. This means that the camera will produce a sharp, well-corrected image of our entire spectrum.

The HPF camera system, designed by team member Christian Schwab and fabricated by New England Optical Systems and Optimax Systems, was recently delivered to PSU and is now undergoing vacuum and cold testing.

The delivered HPF camera system! The camera is made of specialized lenses, which are coated with custom anti-reflective coatings. These complex coatings (appearing as green-yellow tints in the images) ensure that we capture every last photon from our targets. The mechanical assembly is unique in its own right; as the camera cools to the HPF operating temperature, the mounts compensate for thermal expansion differences between the various glasses and the aluminum housing. This allows us to test the camera both at room temperature and at the 180K operating temperature of HPF.

With the camera now delivered to Penn State, we’ve received almost all of the optical components for HPF. We’re now knee-deep, for the first time, in our alignment and testing of the entire optical system. Stay tuned!

Posted in HPF Hardware, Project Development | Leave a comment

Stellar Activity in the Near-Infrared: We Need a New Ruler!


We talk about stellar activity a lot on this blog.  Once HPF gets on sky, radial velocity noise from stellar activity will likely be the biggest impediment to finding exoplanets.  Thus, if we want HPF’s chief scientific mission of discovering low-mass exoplanets to be successful, we have to be prepared to deal with stellar noise from Day 1 of our upcoming survey.  In today’s post, we will look in detail at how the HPF team verified a set of near-infrared absorption lines from calcium as a first indicator of stellar magnetic activity for the HPF exoplanet survey.

Research led by HPF’s own Robert Marchwinski suggests that Doppler noise from stellar activity should actually be reduced at the near-infrared relative to the noise levels observed at visible-light wavelengths.  However, “less” does not mean “none,” and we certainly do not want to find any unpleasant surprises!  Following the examples of previous exoplanet surveys, we intend to track the magnetic activity of our target stars by monitoring the variability of atomic lines in the stellar spectra that change as magnetic activity increases and decreases on the stellar surface.

For HPF, this is potentially easier said than done.  The near-infrared wavelength coverage of our instrument, while ideal for targeting the smallest, coolest nearby stars, excludes all of the well-characterized atomic lines that previous Doppler surveys have relied on to provide information about stellar activity.  This is illustrated in the spectrum below.

A near-infrared spectrum of Gliese 411, an M2 dwarf that is the sixth closest hydrogen-burning star to the Sun. Wavelength ranges for HPF's coverage and the most commonly used activity indicators are shown. Adapted from Figure 2 of Kirkpatrick et al. (1993).

A near-infrared spectrum of Gliese 411, an M2 dwarf that is the sixth closest hydrogen-burning star to the Sun. Wavelength ranges for HPF’s coverage and the most commonly used activity indicators are shown. Adapted from Figure 2 of Kirkpatrick et al. (1993).

Therefore, if HPF is to confidently identify nearby exoplanets, we will have to establish at least one new atomic-line index to help us trace stellar magnetic activity.  It would be nice to have this done before HPF goes on sky, but such an experiment is potentially a major undertaking.  Ideally, one would like to use a high-resolution near-infrared spectrograph (of which there are few–hence we are building HPF!) to make many observations of one or more M dwarf stars (so we can see the activity changing).

Proxima to the Rescue

As it turns out, we can use some data that are already available to get started!  More specifically, many observations have already been taken of our nearest stellar neighbor, Proxima Centauri.  As we discussed on the last entry of the HPF blog, Proxima Centauri (Proxima for short) is a mid-M star that is about 4 light-years away from the Sun.  Recently, Proxima was found to host a low-mass planet in its liquid-water habitable zone, making it an exciting target for future efforts to characterize the atmospheres of potentially Earthlike planets and to search for signatures of extraterrestrial life.

Proxima is also relatively active for an M star of its age.  Indicators of magnetic activity in stars (brightness, spectral line variability, X-ray emission) show that Proxima varies on many timescales, from less than a day to several years.  A team of astronomers using the MOST spacecraft to search for a transit of Proxima b have come up empty so far, but in doing so found that Proxima exhibits stellar flares upwards of 60 times a day!  On the long-timescale end, other astronomers have found evidence that Proxima may have a long-term magnetic cycle like that of our own Sun.

In 2009, a team of astronomers used the high-resolution UVES spectrograph and the XMM-Newton spacecraft to observe Proxima intensely for three nights at both X-ray and optical/near-infrared wavelengths.  Their goal was to obtain information about Proxima’s atmosphere and magnetic field at different levels of activity.  Their data set includes hundreds of spectra of Proxima, and covers a significant amount of magnetic variability.

As it turns out, these observations are a great resource for us to study activity indicators for HPF!  In particular, the UVES observations were made with a special setup of the spectrograph, where the resulting spectra started at the hydrogen alpha (H-alpha or \textrm{H}\alpha for short) line, and extended redward into the blue half of HPF’s wavelength coverage.  As you may recall, \textrm{H}\alpha is a well known and characterized indicator of magnetic activity in M dwarf stars, and was instrumental in helping our team separate real exoplanets from activity-induced false positives in the Gliese 581 system.  So with these data, we can measure the variability of the \textrm{H}\alpha line as the magnetic activity stimulates its emission, and look at how the near-infrared lines respond to the same activity.  The most sensitive lines will be our activity tracers for HPF!  The results of this experiment were just published by the HPF team in a research article, but let’s take a look at the main points.


The UVES spectra include 8 near-infrared atomic lines that previous studies have shown are at least somewhat sensitive to stellar activity.  These lines are produced by hydrogen, potassium, sodium, and calcium.  For each of the 562 spectra, we measured the flux in these lines, and compared them to the flux in the \textrm{H}\alpha line.

Examples of the near-infrared lines we considered as candidate indicators of stellar activity, with <img src='http://s.wordpress.com/latex.php?latex=%5Ctextrm%7BH%7D%5Calpha&bg=ffffff&fg=000000&s=0' alt='\textrm{H}\alpha' title='\textrm{H}\alpha' class='latex' /> at top left for comparison. Each line is shown in its average state, and at periods of high (red) and low (blue) activity. The more activity-sensitive lines show a greater difference between high- and low-activity states.

Examples of the near-infrared lines we considered as candidate indicators of stellar activity, with [latex]\textrm{H}\alpha[/latex] at top left for comparison. Each line is shown in its average state, and at periods of high (red) and low (blue) activity. The more activity-sensitive lines show a greater difference between high- and low-activity states.

In this experiment, the analysis was actually relatively simple.  Two of the three calcium lines, and both of the potassium lines, matched the behavior of \textrm{H}\alpha almost perfectly, while the sodium and hydrogen lines we considered were not nearly as useful.  The third calcium line appears to be as sensitive as the first two, but it is contaminated by a nearby iron line, making its measurement less reliable.

The two bluest calcium lines will be excellent activity indicators for HPF, and will allow us to start separating signals from planets and activity from the very beginning of our survey.  The potassium lines also look like great proxies for magnetic activity, but are just a bit too blue to be seen by HPF.  However, there are other instruments in development that use what is sometimes called the “red-optical” wavelength range to look for exoplanets around slightly hotter M stars.  These instruments include MAROON-X and Veloce, and the potassium lines may prove to be valuable tools for their surveys.

Concluding Thoughts

This study was an important reminder to keep an open mind when performing scientific experiments.  At the outset, we were convinced that the near-infrared sodium lines would turn out to be the best activity indicators for HPF.  The sodium lines are deep lines that are easily identified and measured for all M dwarfs, even in measurements of poor quality.  Furthermore, the sodium lines at optical wavelengths are extremely sensitive to activity for M stars, leading us to expect similar results for their near-infrared counterparts.  On the other hand, the near-infrared potassium lines are contaminated by absorption from Earth’s atmosphere, while two of the three calcium lines are much more difficult to identify than the other lines we considered.  We were thus unconvinced that those lines would be of much use.  As it turns out, the answer was exactly the opposite of what we expected!  In sports, one would say “that’s why they play the game.”  For us, perhaps the equivalent is “that’s why we do the experiment.”

There is certainly more work to be done for HPF.  Our wavelength coverage will include wavelengths considerably redder than UVES can access, and there are additional lines we must investigate for activity sensitivity.  Furthermore, it will be interesting to see how all of these candidate tracers behave for a variety of stars and magnetic phenomena.  Nevertheless, it is reassuring to know that our toolbox will be at least partially filled when the survey starts.

Posted in Exoplanet Science, HPF Science, Stellar Science | Leave a comment