Research Projects


Our research activities are mainly concentrated on searching for exoplanets and brown dwarfs, observing high reshshift quasar absorbers, and developing innovative techniques and instrumentation for cutting-edge astronomical observations.

OBSERVATIONS

The Dharma Planet Survey

Observational Cosmology and Quasar Dust Absorber Systems

The SDSS-III Multi-object APO Radial Velocity Exoplanet Survey (MARVELS)

The FIRST M Dwarf Planet Survey
High Redshift Gamma Ray Burst Follow-ups
 

TELESCOPE, INSTRUMENTATION AND TECHNOLOGY

The Dharma Endowment Foundation Telescope
The TOU Very High Resolution Optical Spectrograph
Florida IR Silicon immersion grating spectromeTer (FIRST)
EXPERT and LiJET
Silicon Immersion Grating Technology
Michelson Type Dispersed Fixed-delay Interferometers


 

The Dharma Planet Survey

The Dharma Planet Survey (DPS), led by Dr. Jian Ge, aims to detect and characterize close-in low-mass planets around nearby solar type stars (FGK dwarfs) and low mass M dwarfs at the orbital region (~1-450 days)  at very high radial  velocity (RV) precision (~1 m/s, e.g. Tau Ceti, the right figure) with high observation cadence (about 100 RV measurements per target) at the 50-inch fully dedicated automatic telescope, the Dharma Endowment Foundation Telescope (DEFT). It will characterize transiting low-mass planet candidates from NASA TESS mission to determine their orbital parameters, planet masses and densities. The ultimate survey goal is to detect potentially habitable super-Earth planet candidates to independently measure the occurrence rate and distribution of super-Earths, and provide high priority targets for future space direct-imaging missions (such as WFIRST-AFTA and LUVOIR surveyor) FGKM dwarfs (29 late F, 66 G dwarfs, 30 K dwarfs with V<7, and  25 M dwarfs with V<10 and within ~25 pc) in 2016-2020, and then continue to monitor targets with linear trends and observe more targets after 2020.

The DPS survey adopts a totally different strategy than previous low-mass planet surveys using ground-based Doppler spectrographs. Instead of surveying a large number of targets with variable number of measurements (from a few RV data points to ~400 RV data points), the DPS survey will offer a homogeneous high cadence for every survey target, i.e., about 100 RV measurements randomly spread over 450 days (about one measurement every other night for a typical survey target after excluding the Arizona Monsoon season and bad weather).

The automatic nature of the DEFT telescope, and its flexible queue observation schedule are key to realizing this homogenous high cadence. This cadence will minimize the time-aliasing in detecting low-mass planets, especially those in highly eccentric orbits, which can be missed by previous surveys. Because of the homogenous cadence for every survey star, both detections and non-detections from the survey can be reliably used for statistical studies. Thus, the proposed survey strategy, cadence, and schedule will offer the best accuracy to assess the survey completeness, to determine occurrence rates and distributions of low-mass planets. As illustrated in the right figure, our survey strategy allows for probing most of the parameter space in the planet mass-period distribution of nearby bright FGK and M dwarfs to independently verify characteristics (such as distribution, orbital properties, stellar properties, and occurrence rates) of the close-in low-mass planet population identified by the Kepler mission. It also covers the periods ~200-450 days, which are critical for the upcoming WFIRST-AFTA and LUVOIR direct-imaging missions. This survey will offer the best homogeneous data set for constraining formation models of low-mass planets with periods less than 450 days.

This DPS survey strategy provides an efficient way to explore habitable low-mass planets around nearby FGKM dwarfs shown in the above figure with greatly improved survey sensitivity and completeness compared to previous Doppler surveys. The DPS can reach planets masses ~10 times smaller (50% completeness) in the habitable zones around FGK dwarfs than reached by previous RV surveys. Improved measurements of occurrence rate and distributions of low-mass planets of nearby FGKM dwarfs will help establish a global view of habitable worlds in the solar neighborhood.

Survey Target Selection and Predicted Planet Yields: Our survey targets were selected from the following catalogs: Gliese Catalog of Nearby Stars (Spring 1989); ROSAT All-Sky Survey: Nearby Stars (Huensch et al. 1999).  

-    Spectral type: ~F5V-M5V.

-    Brightest stars in each subtype that satisfy the following criteria:

o   No stellar companions within 5

o   vsini < 3 km/s

o   RV jitter < 2 m/s for GK dwarfs,  < 3 m/s for F dwarfs, and <4m/s for M dwarfs if available

o   In active stars, using any available indicators.

            log RHK < -4.85

                X-ray luminosity and bolometric luminosity, Rx < -3.0, if activity level is unknown

o   < 30 RV observations with better than 3 m/s Doppler precision if observed before

o   RAs are equally distributed from 0-360 deg. and DECs are between -20 and 90 deg.

The left two figures shows magnitude and effective temperature distributions of our selected targets. 44 of 100 targets are in the Eta-Earth Survey list and eight planets have been detected among these 44 Eta-Earth survey stars. Some adjustment of targets, including vetting out of target candidates with RV jitters > 2 m/s (except > 3 m/s for F dwarfs, > 4 m/s for M dwarfs), has been made to optimize the survey efficiency and sensitivity. Our survey simulation forecasts that DPS will yield a total of 82 low-mass (less than 36 Earth masses) planets based on current instrument performance and measured planet occurrence rates. It includes ~27 super-Earths, 55 Neptune-mass planets, ~10 habitable super-Earth, and 18 habitable Neptune candidates.

Early On-Sky Performance: Since the first light of science observation at Mt. Lemmon on October 10th 2016, the DPS survey has observed 80 FGKM dwarfs with 52 stars observed more than 30 RV measurements by the end of October 2017. The results show a Doppler precision of 0.7-1.8 m/s (RMS) achieved for all of the RV stable stars. The long-term RV RMS measurement error of Tau Ceti is slightly better than that obtained by HARPS. Two known super-Earths (HD 1461b and HD 190360b) are confirmed. 

Among survey stars with more than 30 RV measurements, 13 show RV RMS larger than 3 times the average single measurement error, indicating possible real signals. To date, seven planet candidates (DPS-1b, DPS-2b, DPS-3b,c, DPS-4b, DPS-5b, DPS-6b) were identified among six stars. The distributions of the single measurement RV precision is shown in the left figure and RV scatters over long-term from the DPS early survey targets are shown the right figure. By combining the high RV precision with the unique high cadence, the DPS will allow for significant improvement in the survey completeness for lower mass planets at short periods and planets with longer periods than those studied with previous high precision RV surveys. Consequently, the DPS survey will offer the community the first high-precision and high-cadence RV measurements of nearby FGKM dwarfs.

 Back to Top


 

Observational Cosmology and Quasar Dust Absorber Systems

Dust grains play a key role in the astrophysical processes in the Universe. The amount of dust attenuation is a key parameter to accurately derive star formation rates from the rest frame ultraviolet (UV) spectra. Measuring the dust extinction curve and making a corresponding dust reddening correction allows the accurate determination of the intrinsic Spectral Energy Distributions (EDs), galaxy, AGN and quasar luminosity functions, photometric redshifts and physical properties of galaxies. In spite of its importance, dust extinction in both low and high redshift galaxies is poorly known.

     The most remarkable and intriguing feature in the dust UV extinction curve is the broad 2175 absorption bump, or called the UV bump. It is observed ubiquitously in the Milky Way (MW) and Large Magellanic Clouds (LMC). The origin of the bump, its dependence on environment, and its evolution have been a mystery for the past 50 years since its discovery.

Dr. Jian Ge, his students and collaborators have identified a new population of quasar absorption line systems (QALs), called quasar 2175 dust absorbers (2DAs), at z~0.7-2.5 in SDSS quasar spectra. These absorbers have a distinguishing broad UV bump at 2175 in the underlying shape of the quasar spectra, similar to that seen in Milky Way diffuse clouds. The right figure shows a typical broad 2175 UV bump in a SDSS quasar spectrum. The first z>1 2DAs were discovered by Dr. Ge's team in 2004 (Wang et al. 2004, ApJ, 609, 589).

Our quasar 2175 absorber identification technique is called the quasar spectra pair method (Jiang et al. 2011, ApJ, 732, 110). The left figure shows a typical quasar spectrum and fits with different extinction curves to extract a UV bump and extinction curve using this method. To date, we have identified a total of 493 quasar 2175 dust absorbers in SDSS DR12 data. These data are being used for statistical studies of the properties of these rare quasar absorbers. The absorption lines associated with these objects allow us to study the interstellar medium in great details to understand their physical and chemical conditions, processes and evolution.

In collaboration of an international collaboration team, including Drs. Jason Prochaska (UCSC), Pasquier Noterdaeme (Institut dAstrophysique de Paris  (IAP), Hongyan Zhou (PRIC), Shaohua Zhang (PRIC), Tuo Ji (PRIC), Peng Jiang (PRIC), Anand Sriannand (Inter-University Centre for Astronomy and Astrophysics), Don York (Chicago), Xiaohui Fan (UA),  Don Schneider (PSU), Fred Hamann (UF/UCR), Niel Brandt (PSU),  and Britt Lundgren (Yale), we have followed up some of the quasar 2DAs at Keck, VLT, GTC, MMT, HST, Palomar 5m, GMRT and Lick 3m telescopes. The above figure shows velocity profiles of metal lines for the 2175 absorption system toward quasar J1459+0024, a representative quasar 2DA. 2DAs on average have higher metallicity (the left figure) and depletion levels ([Fe/Zn]~ −0.6 to−2.1) than DLAs or subDLAs (Ma et al. 2017a,b). Their average dust depletion is about 10 times higher than the mean for DLAs, and has partial overlap with sub-DLAs. The majority of them have dust depletion levels and patterns between that of the cold disk clouds and warm halo clouds in the MW as shown in the right figure. 2DAs are found to simultaneously harbor cold neutral (C I, H I, Cl I) and/or molecular (CO and H2) gas, which may serve as reservoirs for star formation in the host galaxies. All 2DAs show strong C I absorption lines with logN(C I) > 14.0 cm-2. There also appears to be a weak correlation between UV bump strength and dust depletion. Their metallicity is strongly correlated with dust depletion [Fe/Zn] and velocity widths (the left figure). 2DAs show larger velocity widths than DLAs and subDLAs. Based on the mass-metallicity relation, the estimated stellar masses of 2DAs are in the range of ~109 to ~2 1011 Mʘ with a median value of ~2 1010 Mʘ (Ma et al. 2017b).

    Our ultimate goal with this project is to understand the nature of the quasar 2175 dust absorbers. The results of this investigation will address extinction biases associated with quasar absorption line systems and will cast light on the physics of dust grains at different cosmic epochs. This project offers a global perspective on the formation and characteristics of dust in the ISM of high z galaxies. The results will help understand global enrichment of chemicals and stellar mass assembly in the early universe.

Back to Top


 

The FIRST NIR M Dwarf Planet Survey

    The FIRST NIR M dwarf planet survey aims to search around 200 nearby M dwarfs for low-mass planets with the Florida IR Silicon immersion grating spectromeTer (FIRST) NIR spectrograph and the AST 2m robotic telescope at Fairborn Observatory. FIRST was developed by Dr. Ge's team in 2009-2013. This new generation cryogenic IR spectrograph offers broad-band high resolution IR spectroscopy with R=55,000 at 0.9-1.7 mm in a single exposure with a 2kx2k H2RG IR array. The predicted instrument long term Doppler precision is about 3 m/s including measurement errors from photon noise, calibration and telluric absorption and emission lines. Our primary science goal is to look for habitable super-Earths around the late M dwarfs and also to identify transiting systems for follow-up observations with JWST to measure the planetary atmospheric compositions and study their habitability. Our secondary science goal is to detect and characterize a large number of planets around M dwarfs to understand the statistics of planet populations around these low mass stars and constrain planet formation and evolution models. Our survey baseline is expected to detect ~30 exoplanets, including 10 Super Earths, within 100 day periods. About half of the super-Earths are in their habitable zones. The AST, with its robotic control and ease of switching between instruments (in seconds), enables great flexibility and efficiency, and enables an optimal strategy, in terms of schedule and cadence, for this NIR M dwarf planet survey.

    The left figure shows Late M dwarf orbit period vs. stellar host mass to be explored by the FIRST survey. The shaded regions are the HZ, with darker regions corresponding to less constraint for habitability (i.e., 0%, 50% to 100% cloud coverage, respectively). The dashed (dotted) curves show where planets induce a 4 m/s (2 m/s) reflex velocity in the star, assuming a circular orbit. The left most curve is for a 1 Earth-mass planet,  a 5 Earth-mass planet, and a 10 Earth-mass planet. FIRST can detect planets to the left of the curve.

 Back to Top


 

High Redshift Gamma Ray Burst Follow-ups

    In collaboration with Drs. Bing Zhang (UNLV), Antonino Cucchiara (UVI), and Matthew Muterspaugh (TSU), Dr. Ge's team has implemented an Near-IR (NIR, 0.9-1.7 mm) fiber bundle Integral Field Unit (IFU) spectroscopy mode in the FIRST near IR (NIR) high resolution spectrograph for capturing medium-resolution Gamma Ray Burst (GRB) NIR spectra within 10 minutes of the trigger release by NASA's Swift satellite using the TSU AST 2m robotic telescope at Fairborn Observatory in Arizona. The IFU fiber bundle is made of 19 tightly packed circular fibers with 80 micron core diameter (the figure below) and can simultaneously capture sky images within 11 arcsec field of view. The broad simultaneous NIR wavelength coverage (0.9-1.7 mm) will allow the prompt identification of GRBs in the z=6.5-13 redshift range and the acquisition of high signal-to-noise afterglow spectra. Using such high quality data we will be able to investigate the chemical content (e.g. neutral hydrogen, metal enrichment) of the first generation of galaxies during the re-ionization epoch, a task that will be very challenging (and time consuming) even with dedicated Lyman-break galaxy surveys using the new generation of 30m telescope.

    The fast response of the robotic telescope (in ~5 min) and the sufficiently large (~11 arcsec) FOV will always allow to cover the average Swift-XRT or the future SVOM-SXT error circles (~3 and ~10 arcsecs respectively) and to capture a high-z GRB spectrum during its brightest phase. The IFU 3-D imaging spectroscopy nature allows us to reconstruct both broad band and narrow band images of the GRB field, which can provide accurate coordinates with subarcsec accuracy to facilitate follow-up observations at other observatories. GRBs at redshifts between z=6.5-13 will be identified through detecting the unique strong Lyman alpha absorption trough due to the Gunn-Peterson effect. Detection of these high redshift objects will provide one of the most powerful tools for investigating the first generation of star formation and the reionization era in the early universe. IR observations can also help to identify many GRBs deeply embedded in the dusty environment and reveal the nature of the progenitors of GRBs at moderate redshifts. The expected increased number of observed z>6 GRBs by AST+FIRST will help more accurately constrain the high-z GRB rate, shedding light on the fate of massive stars, their role on re-ionization, and cosmic star formation rate at high redshifts.

    Assuming that the GRBs at z =6.5-13 have the same afterglow luminosity as those at z = 0.3 -4.5, FIRST can detect most of them at a spectral resolving power of R ~ 700 within ~1-2 hours after the bursts.  During the ground-based GRB follow-up observations, we will also provide sub arsec precision coordinate information of the GRBs to GRB Coordinates Network (GCN), allowing quick follow-up afterglow observations by other groups all over the world.

Back to Top


 

The 50inch Dharma Endowment Foundation Telescope

    A fully dedicated 50-inch automatic telescope (the right figure) was installed at Mt. Lemmon in Arizona in November 2015 for the Dharma Planet survey of low-mass planets around nearby FGKM  dwarfs and follow-up observation of transiting planet candidates from the NASA TESS mission. The telescope began science operation on October 10th 2016. This telescope is a refurbished one of the retired 50-inch automatic telescope at Morgan-Monroe Observatory of Indiana University (IU).

    This 50-inch telescope was donated to the University of Florida for the exoplanet research and high resolution spectroscopy in July 2014. The telescope (except the mirrors) was moved to Mt. Lemmon in March 2015. The telescope primary and secondary mirrors were re-coated with the diamond Brite broad band reflecting coating (better than 98% reflectivity for most of the wavelengths between 0.38-3 microns) by H.L. Clausing Inc. in May 2015 and were delivered to Mt. Lemmon in June 2015.

    This automatic telescope has average pointing accuracy of about 50 arcsec, slewing time is within 1 min, and guiding accuracy of 0.3 arcsec. A user-friendly remote observing capability is planned to be developed for teaching and training of students.

Back to Top


The TOU Very High Resolution Optical Spectrograph

    TOU (formerly EXPERT-III) is a new generation, fiber-fed, highly stable, optical extremely high resolution, cross-dispersed echelle spectrograph operated in a vacuum chamber (see the figure below). It was developed by an engineering team led by Dr. Ge in 2010-2013. It adopts the HARPS design but with several design refinements to significantly reduce its volume and construction cost while substantially increasing its wavelength coverage (the figure below, 0.38-0.9 micron vs. 0.38-0.69 micron in HARPS) while maintaining a similar spectral resolution (R=100,000 vs. R=115,000) and spectral sampling (3.4 pixels). The extra wavelength coverage at 0.69-0.9 micron compared to HARPS has benefited RV measurements for K and M dwarfs, which have a peak flux around long wavelengths. A commercial R4 (76 deg blaze angle) echelle with a 112x408mm ruled area is used as the main disperser and two prisms made of PBM18Y glass are used as the cross-disperser. The spectra from 0.38-0.9 m are recorded with a 4kx4k Fairchild back-illuminated CCD array. The detector completely covers spectra from 0.38-0.8 m and nearly completely covers 0.8-0.9 m (only the edges of the dispersed spectra are off the detector, see a Tau Ceti spectrum below). The detector is controlled by the Leach controller from Astronomical Research Cameras, Inc. The CCD detector is cooled by one cryotiger with PT-30 gas and runs at -100C to reduce its dark current. The entire instrument bench (1.34x0.8 meter in dimension) is installed inside a vacuum chamber with a volume of 1.46x0.85x0.48 m3.

    TOUs vacuum is kept under 0.01mTorr. This kind of low vacuum level has helped minimize the pressure effect on RV variation and temperature control. The chamber is inside a thermal enclosure where temperatures are precisely controlled to provide a long term thermal stability of the instrument. Small heaters and Lakeshore temperature sensors are placed on the bench (four corners and the central region) and the detector assembly to allow the bench and detector temperature precisely controlled. The entire instrument system is placed inside a temperature-controlled instrument room with monthly temperature variation less than 0.1C  (RMS) . The entire instrument temperatures have been kept to within ~1 mK RMS of a setting point (22C) over a few months. This instrument temperature control has eventually helped us reach ~0.35m/s (RMS) long-term RV stability as shown in the left figure. Like HARPS, TOU adopts the ThAr calibration method. Our current ThAr data processing pipeline has achieved 0.36 m/s instrument drift correction precision over a long term (~60 days, the right figure).

    TOU is fed with one 80 m science diameter fiber from the telescope (3.1 arcsec on sky) at f/4. This science fiber is coupled into four 40 m fibers. The outputs of the four fibers are arranged in a fiber slit head and re-imaged by relay optics to form f/10.2 beams which passing through the spectrograph entrance slit. This configuration allows the instrument to reach R~100,000 in the operating wavelengths. An optical mode scrambler is used in the science fiber inside the chamber. TOU is designed to reach extremely high precision RV measurements (~1 m/s or better) of stars with V~8 within 1 hour exposure.  The table below summarizes TOU instrument parameters.

    The up right figure shows a Tau Ceti spectrum taken in 10 min with TOU at Mt. Lemmon. The 4kx4k CCD detector covers 0.38-0.90 m in a single exposure. The right figure shows part of a reduced spectrum of Tau Cedti around ~530 nm.

Wavelength region

0.38-0.9 m

Focal plane

4096x4096 Fairchild

Pixel size

15 m

Detector QE

~92% 0.38-0.65 m

~85% 0.65-0.75 m

77% at 0.8 m

54% at 0.9 m

Spectral resolution

R=100K

Dispersion per pix

0.017/pix at 0.55 m

Sampling

3 pixels

Fiber diameter

3.1 arcsec (80 m)

Spectral orders

68-161 (0.38-0.90 m)

Single exposure coverage

5000

Design type

White pupil + refractive camera

Collimated beam

98 mm in diameter

Main dispersers

R4, 31.6 l/mm, 76 blaze

Cross-dispersers (prisms)

PBM18Y prisms with 46 and 50 apex angles

Optical bench

1.34x0.8x0.48 meter

Back to Top


 

Florida IR Silicon immersion grating spectromeTer (FIRST)

FIRST is a next generation, fiber-fed, highly stable (temperature controlled), cryogenic, NIR high-resolution, cross-dispersed echelle spectrograph operated in a vacuum chamber. It was developed at UF by Dr. Ge's team in 2009-2013. The FIRST instrument was commissioned at the TSU AST 2m in the fall of 2013. The following figure shows the inside layout. FIRST has two channels: a red channel with a UF made silicon immersion grating (SIG) as the main disperser to simultaneously cover 1.4-1.8 m at R=68K and a blue channel with a commercial R4 echelle to cover 0.9-1.7 m at R=56K. The red channel is primarily used for  the SIG high-resolution spectroscopy technology demonstration while the blue channel is used for  RV measurements of M dwarfs for planet detectiom and characterization. The spectra are recorded with an H2RG 2K2K array from Teledyne Technologies Inc with 2.57 m wavelength cutoff. The detector is controlled by the standard Teledyne ASIC card. The entire instrument bench (1.04x0.45 meter in dimension) is installed inside a vacuum chamber. A 30-layer MALI thermal shield is mounted around the bench to thermally insulate the bench from the chamber. The optical bench is cooled with two cryotigers using standard NF-55 gas and operates at 193K to reduce the thermal background while the detector is cooled by one cryotiger with PT-13 gas and runs at 85K to reduce its dark current. The entire instrument temperature is controlled to within 4 mK over a long term. The instrument is placed inside a temperature-controlled instrument room at Fairborn Observatory with yearly temperature variation within about 1C to further reduce the instrument long-term temperature variations.

FIRST is fed with three 80 m diameter fibers (2.1 arcsec on sky, one science, one sky and one calibration) at f/4. The science beam is sliced into two halves by a Mirror Image Slicer developed from the traditional Bowen-Walraven Image Slicer and passes through the spectrograph entrance slit (half of the calibration and sky fiber beams is blocked to produce the same spectral resolution as the science beam). The FIRST blue channel covers 33 spectral orders (orders 36-68, see the table below) at 0.9-1.7 m while the red channel completely covers 59 spectral orders (orders 195-253) at 1.4-1.8 m. Two broad band sorting filters on a motorized rotation stage are used to select an observation channel. The following table summarizes FIRST instrument parameters.

FIRST is calibrated with a tungsten lamp for flatfielding and an UrNe lamp for wavelength and RV calibration and instrument point spread function modeling. In order to precisely calibrate instrument drift to reach high RV precision (~3 m/s), a separate UrNe spectrum is recorded at the same time as a star spectrum. An UrNe exposure is also taken through the science fiber before and after each star exposure to remove residual drifts not corrected by the simultaneous separate beam calibration. Since the UrNe calibration lamp, like a ThAr lamp, has limited lifetime, frequent replacement of the UrNe calibration lamp requires calibrating offsets for different lamps. This offset calibration is made by shining a second UrNe lamp used occasionally (once per day). Fibers (science and calibration fibers) are mode scrambled by an optical mode scrambler to reduce illumination-induced variations in the RV measurements.

Wavelength region

0.8-1.7 m

Focal plane

2048x2048 H2RG

Pixel size

18 m

Detector QE

81% 0.6-1.0 m

90% 1.0-2.4 m

Spectral resolution

R=56K at 0.9-1.7 m (Blue)

R=68K at 1.4-1.8 m (Red)

Dispersion per pixel

0.10/pix at 1.25 m (Blue)

0.11 /pix at 1.65 m (Red)

Sampling

2.1 pixels

Fiber diameter

2.1 arcsec (80 m)

Spectral orders

36-68 (0.9-1.7 m) (Blue)

195-253 (1.4-1.8 m) (Red)

Single exposure coverage

~0.5 m at 0.9-1.7 um (Blue)

0.4 m at 1.4-1.8 m (Red)

Design type

White pupil + refractive camera

Collimated beam

50 mm in diameter

Main dispersers

R4, 31.6 l/mm, 76 blaze (Blue)

SIG, 16.1 l/mm, 54.74 blaze (Red)

Cross-dispersers

(VPH gratings)

240 l/mm 7.0 blaze (Blue)

310 l/mm, 14.0 blaze (Red)

Optical bench

1.04x0.45 meter

Back to Top


 

EXPERT and LiJET

    The Extremely High Precision Extrasolar Planet Tracker Instruments (EXPERT, the left figure) and the LiJiang Exoplanet Tracker (LiJET) are two second generation Doppler instruments with an identical design developed at UF by Dr. Ge's team in 2008-2011. Collaborators from Yunnan Astronomical Observatory, University of Science and Technology of China and Nanjing University participated in both instrument development and telescope commissioning. These two instruments are designed as part of a global network for hunting for low-mass planets around nearby FGK dwarfs EXPERT is currently located at the TSU AST 2m telescope at Fairborn Observatory while LiJET is located at the 2.4m telescope at the Lijiang station of Yunnan Astronomical Observatory in China.

    EXPERT/LiJET is a combination of a thermally compensated monolithic Michelson interferometer and a cross-dispersed echelle spectrograph (the top right figure) for extremely high precision Doppler measurements for nearby bright stars (e.g., photon limited RV measurement uncertainty of 1m/s for a V=8 solar type star in 15-30 min exposure). The EXPERT/LiJTE instrument bench is inside a pressure controlled chamber. The instrument chamber in installed in a thermally controlled enclosure. The instrument bench includes a thermally compensated monolithic Michelson interferometer system, a double pass cross-dispersed echelle spectrograph, a PMT flux counting system and a 4kx4k CCD camera. The left figure shows the spectral format on the 4kx4k CCD.

     EXPERT/LiJET has R=18,000 with a 72 micron slit and a simultaneous coverage of 390-694 nm (the left figure). The commissioning results show that the instrument has already produced a Doppler precision of about 2-3 m/s for sky measurements with S/N~100 per pixel (the lower right figure). The instrument has reached ~4 mK (P-V) temperature stability, ~1 mpsi pressure stability over a week and a total instrument throughput of ~30% at 550 nm from the fiber input to the detector.

    EXPERT/LiJET also has a direct cross-dispersed echelle spectroscopy mode fed with 50 micron fibers. It has spectral resolution of R=27,000 and a simultaneous wavelength coverage of 390-1000 nm.

    Due to the complicated spectral format with the Dispersed Fixed-Delay Interferometer (DFDI) approach, the data pipeline for processing the EXPERT/LiJET data is still being developed before a large-scale high precision RV survey with these two instruments can be launched.

Back to Top


 

Silicon Immersion Grating Technology

    Silicon immersion diffraction gratings fabricated by photolithography and anisotropic chemical etching techniques developed by a technology team led by Jian Ge will be key dispersing elements for next generation space and ground-based IR spectroscopic instruments. Due to its very high refractive index (n = 3.4 at 2.2 micron), a silicon immersion grating can provide more than three times dispersion of a conventional reflective grating of equal length. Therefore, silicon immersion gratings will enable very compact IR spectroscopic instruments while provide high dispersion power.

    We have successfully developed the world's first silicon grisms in 1999. They have 10x10 mm2 etched grating area and 46 deg wedge angles as shown in the following figure.  The first light of one of the grisms at the Lick 3m telescope with the IRCAL near-IR camera and adaptive optics has demonstrated a diffraction-limited spectral resolution, R = 5,000 at 2.2 micron with a pupil diameter of only 5 mm. This spectral resolution is the highest ever obtained with grisms. Coupled with the Lick AO system, it allows efficient IR spectroscopy at very high spatial resolution at 0.2 arcsec. Combined with a conventionally made CaF2 grism cross-disperser, it allows a complete wavelength coverage in the K band.

    Thanks to the NASA and NSF support, we have developed new etching processes with Tetramethyl ammonium hydroxide (TMAH)  and applied the techniques in making silicon immersion gratings. The TMAH processes, instead of the traditional KOH etching processes, have significantly improved the grating surface quality, e.g., the rms surface roughness is reduced from 30 nm to less than 3 nm, the minimum rms surface roughness is 0.9 nm. We have fabricated silicon immersion gratings with etched areas ranging from 10x10 mm2  to 85x50 mm2, blaze angles ranging from 22 deg (the upper right figure shows 22 deg grisms) to 76 degree, and groove densities from 137 l/mm (7 mm groove size) to 5.4 l/mm (or 185 mm groove size). The 185 mm grooves are 4 times coarser than that available with the commercial echelle gratings. The SIG with  85x50 mm2 etched grating area shown in the left figure has high measured grating efficiency as shown in the right figure. The total integrated scattered light level caused by the micro scale roughness of the etched grating groove surfaces is less than 0.5%

The silicon immersion gratings promise a major impact in IR spectroscopy. The silicon grisms promise a very convenient and inexpensive way to implement intermediate and high spectral resolution in any existing IR camera. The silicon immersion gratings offer high efficiency and very high spectral resolving power (R > 100,000) in the IR for the first time.

Back to Top


Michelson Type Dispersed Fixed-delay Interferometers

    The Dispersed Fixed-delay Interferometer Instrument Principle and History: The approach using a dispersed fixed-delay interferometer (DFDI) for Doppler measurements is completely different from the current echelle approach. Instead of measuring the absorption line centroid shifts in the echelle approach, the radial velocity (RV) is measured through monitoring interference fringe shifts as shown in the right figure. The idea for using a fixed-delay interferometer for high precision Doppler RV measurements was first proposed by physicists in the 1970s and 1980s (Barker & Hollenbach 1972; Gorskii & Lebedev 1977; Beckers & Brown 1978, Kozhevatov 1983). This idea was adopted by the Global Oscillation Net Work (GONG) interferometer (Harvey et al., 1988; Harvey 2002 private communications). This interferometer with a narrow band pass has been successfully used for very high Doppler precision measurements of the Sun (~3 m/s precision for the Sun, Kozhevatov et al. 1995, 1996; sub m/s precision for the GONG measurements, Harvey 2002 private communications).

    The concept of combining a fixed-delay interferometer with a moderate resolution spectrometer for broad band operations for high precision stellar Doppler measurements was proposed by a physisict, David Erskine, at LLNL in 1997. The initial lab experiments and telescope observing with a prototype demonstrated its feasibility (Erskine & Ge 2000; Ge, Erskine and Rushford 2002). A theory for this new instrument concept was developed by Dr. Jian Ge and is described in Ge 2002. For the first time, Dr. Ge proposed to use the dispersed fixed-delay interferometer for multiple object Doppler measurements.

    The DFDI approach offers high throughput and multi-object capability compared to the traditional single object, relatively low throughput echelle approach. The Doppler sensitivity of the DFDI approach weakly depends the spectral resolution compared to the echelle approach (1/2 power of the spectral resolution vs. 3/2 power of the spectral resolution in the echelle method, Ge 2002). This allows the use of a medium resolution but high efficiency first order grating spectrometer for dispersing the fringes to boost the overall detection efficiency, while dramatically reducing the instrument size and cost. The single dispersion order operations allow an implementation of multiple object observations to significantly increase the planet survey speed. Multiple object capability is the most significant advantage for this interferometer approach.

     DFDI instruments: A prototype DFDI instrument was developed  at Penn State in 2000-2001 by Dr. Ge's team , including members Julian van Eyken, Suvrath Mahadevan, Curtis DeWitt, and Jian Liu, in collaboration with Dr. Stuart Shaklan (JPL) and Mr. Mike Rushford at LLNL, and used at the HET 9m and Palomar 5 m telescopes in 2001. It was used again at the KPNO 2.1m telescope in 2002 and routinely provided observations of stars as faint as V = 7.6. The total instrument throughput from the sky to the detector is ~5% (excluding iodine absorption), comparable to or slightly higher than that for current echelle instruments. A new DFDI instrument, called Exoplanet Ttracker (ET) shown in the upper right figure, was developed at Penn State with additional members, Dr. Deqing Ren and Jerry Fridmann under the NSF support in 2002-2003 and commissioned at the KPNO 0.9 meter Coude Feed/2.1 meter telescope in Nov. 2003. The instrument was tested and characterized in Nov. 2003-Nov. 2004. After that, ET was used for a pilot planet survey at Kitt Peak and two external groups for their independent RV studies in 2006-2008.

    A new-generation, multi-object ET instrument, called W.M. Keck Exoplanet Tracker (Keck ET), was developed at UF by Dr. Ge's team in 2005-2006 with the Keck Foundation support. The Keck ET inherits the single object KPNO ET design. The instrument is designed to observe 60 stars simultaneously. In March and April 2006, the Keck ET was commissioned at the SDSS 2.5m telescope. The upper left  figure shows the Keck ET fiber plugging in the SDSS fiber cartridge, the Keck ET instrument setup on an optical bench in a temperature stabilized room and one frame of stellar fringe spectra taken with the instrument. In 2006-2007, a pilot survey of ~420 solar type stars in 8 different fields with V = 8-12 was conducted to detect new planets. A total of 10-14 RV measurements per star were obtained for these survey stars. The RV measurements of two known extrasolar planets (HD 118203b and HD 102195b) are consistent with previous results. The typical photon noise limited errors for stellar data in a 50-min exposure are about 30 m/s for a V=10.5 solar type star (depending on spectral type and vsini), with the best value of 6.9 m/s at V =7.6 in 50 min. The  measured throughput of the Keck ET on different fibers from the telescope to the detector without the iodine cell is 4%-8% (the average is 5.5%) under a ~1.5 arcsec seeing condition.

    The successful demonstration of DFDI technology for single and multiple RV measurements led to development of three survey instruments, the MARVELS multiple object DFDI instrument in 2007-2008 for the SDSS-III MARVELS survey and  high precision single object DFDI instruments, EXPERT LiJET, in 2008-2011.

    Discoveries by the DFDI instruments: To date, DFDI instruments have discovered two planets, HD102195b (ET-1, Ge et al. 2006), and MARVELS-1b (Ma et al. 2016) shown in the upper right figure, 16 brown dwarfs (Fleming et al. 2010; Fleming et al. 2012; Ma et al. 2013; De Lee et al. 2013; Jiang et al. 2013; Ma et al. 2016; Grieves et al. 2017), and over 400 binaries, and confirmed two transit planets (Pepper et al. 2013; Eastman et al. 2016). The MARVELS new brown dwarf distribution appears to confirm the distribution trend in the eccentricity vs. period space discovered by Ma and Ge in 2014, i.e., brown dwarfs with masses less than ~42 Jupiter masses follow the prediction of planet-planet scattering in protoplanetay disks (Fred & Rasio 2008) while brown dwarfs with masses above ~42 Jupiter masses follows the distribution of stellar binaries (Grieves et al. 2017).  In addition, the SDSS-III MARVELS survey has collected over 100,000 RV measurements for a total of over 3300 solar type stars with about 27 RV measurements per star. The survey data will be processed with the latest data pipeline, additional planets, brown dwarfs, and binaries will likely be discovered.

    Technology Spin-offs: The DFDI technology development led to awards of three patents (U.S. patents, 8,570,524(2013), 8,659,845 (2014), and 9,696,138 (2017)).  The UF patented Ultra-stable Monolithic Michelson Interferometer (the U.S., patent, 8,659,845 (2014), Wan & Ge 2010; Ge et al. 2014), or Sine source wavelength calibration method not only has the requirements of reaching ultra-precision RV calibration (better than 0.1 m/s), but its compact size and low cost will allow its implementation into a much wider range of instruments than expensive and large-scale Laser Frequency Combs (LFCs). As displayed in the left figure, the Sine source consists of a temperature controlled monolithic Michelson interferometer fed with a continuum light source through a multiple mode fiber. The instrument creates sinusoidal spectral lines over a wide wavelength range (~380-1350 nm), called a Sine source. The light beam interference between the two interferometer arms produces dense sinusoidal combs, creating ideal evenly spaced spectral features for calibration. The left figure shows Sine source spectra as well as traditional wavelength calibration spectra from a ThAr lamp and iodine cell on the TOU optical cross-dispersed echelle spectrograph (Ge et al. 2012; Zhao & Ge 2012; Ge et al. 2014) at the 2-m Automatic Spectroscopic Telescope (AST; Eaton & Williamson 2004) of Fairborn Observatory in Arizona. It is obvious that the Sine source offers nearly homogeneous high contrast comb features for precision wavelength calibration. 

The left figure displays the broad wavelength coverage (~380-900 nm) of the Sine source with spectra taken with the TOU spectrograph at Fairborn in December 2015 before TOU was relocated to Mt. Lemmon. The right figure shows the real Sine source layout in the UF lab before it was used at the telescope. These spectra were taken with a tungsten lamp as the continuum source, which has a limited calibration bandwidth due to the blackbody radiation. This leads to approximately two orders of magnitude less flux at shorter wavelengths (~400-600 nm) than at longer wavelengths (~700-900 nm). Previous work at the Fairborn Observatory in 2015 required poor SNR (<50 per pixel) at the shorter wavelength region to avoid CCD pixel flux saturation at the longer wavelength region, which was capped around ~40,000. Poor SNR levels in the shorter wavelength region leads to large calibration errors (>1.5 m/s, see the above figure) for these orders.

Although the 2015 Sine source set up at the Fairborn Observatory produced poor SNR levels and large calibration errors (~1.5 m/s) at shorter wavelengths, we plan to implement a commercially available supercontinuum source. The integration of a supercontinuum source will produce high SNR calibration spectra over TOUs entire wavelength range. These supercontinuum sources, such as the compact indigo white light laser from Ultrafast Systems Inc., can cover ~400-2400 nm with flux levels varying only by a factor of a few in the wavelength range of ~400-1000 nm. A supercontinuum source will reduce photon-limited RV calibration error to ~0.3-0.4 m/s for all echelle spectral orders on TOU. The combination of TOUs 92 spectral orders will reduce the overall photon-limited RV calibration error to ~0.04 m/s, allowing RV precision capabilities to detect Earth-like planets around very bright nearby solar type stars with TOU.

Back to Top


To Jian Ge's Home Page


    First created: March 2,2000; Last updated: Dec. 2, 2017