|
|
|
NASA needs to survey hundreds of weak stars and identify those showing indirect evidence of having planets. These candidates will later be observed by direct imaging using spaceborne interferometers. The survey, called the Doppler survey, is being conducted at Keck, Lick and other world observatories by measuring starlight spectra and searching for periodic variation in the star's velocity, of order 1 to 50 m/s amplitude, caused by gravitational tugging of the star by a planet. The current velocity resolution of 3 m/s is sufficient to detect Jupiter-like planets (12 m/s), but not Saturn-like (3 m/s) or Earth-like (0.1 m/s). The two main issues effecting detectability are instrument accuracy and ability to process a sufficient number of photons. The latter issue relates to spectrometer efficiency and the size of the telescope collecting area to which the spectrometer can couple. Currently, only bright stars are being observed in the survey because of low instrument efficiency and competition for limited time on Keck-sized telescopes. Instrument accuracy is critical because the Doppler shift being sought can be 1000 times smaller than the linewidth of the absorption lines in the star's spectrum, which provide the fiducial features.
Current spectrometers are insufficient to complete the survey at a practical rate. The current method of using a high resolution grating requires many hours of computer time per star to characterize the grating instrument response sufficiently well and deconvolve its characteristics from the data. This step is tedious but necessary because irregularities in the instrument response can easily swamp the small Doppler shift. The grating instrument response is complicated. It has 100's of degrees of freedom, corresponding to the numerous grating grooves. Changes in temperature, coating oxidation, internal air convection, and position of entering light against the slit can produce significant errors. The spectrometers are as large as a kitchen, immobile and highly individual. Since Saturn has a 30 year orbit, detecting Saturn-like planets requires comparison of data through changes in instruments, observatories and scientific personnel. Thus the complicated instrument response of the grating is a disadvantage. Our LLNL spectrometer design introduced here solves this problem by simplifying the instrument response to 3 degrees of freedom, by using an interferometer.
Presently, only bright stars are being surveyed. The key to measuring weak stars is using light collectors having large area. Current spectrometers require a nearly diffraction limited image of the star in order to achieve the required spectral resolution (because the slits must be very narrow.) Presently, the only large area light collectors having small image spot sizes are a few very expensive telescopes, such as Keck. Available time on these telescope is limited, due to competition by the entire community of astronomers for all kinds of projects. The current high resolution spectrometers cannot use blurry sources. Therefore they cannot use inexpensive light collectors called "light buckets", which could have areas as large or larger than Keck at 100x less cost, by using low-quality optics, or by using fiber-connected arrays of smaller collectors. Such light bucket arrays are attractive since they could be dedicated to the planet search project, and other astronomical spectroscopy projects, where tight imaging is not required. The maximum size of the star image (or output fiber diameter) capable of being efficiently used by the spectrometer is defined by the spectrometer's field of view.
Our new LLNL spectrometer design has a 200x larger field of view which allows us to use light bucket collectors. The increased field of view also allows observation of multiple stars simultaneously, further increasing the rate of stars surveyed. Thirdly, the increased field of view allows efficient high resolution spectroscopy on diffuse sources such as nebulae and atom laser plasmas, not possible now. Because the LLNL design uses an angle-independent interferometer for the Doppler resolution, the signal is not subject to source/slit positional errors, as is the current spectrometers. No existing spectrometer can match our combination of arbitrarily high spectral resolution, wide field of view, high efficiency, high signal to noise, portability and accurate instrument response.
A prototypical instrument has already been constructed and benchtop testing is underway on sunlight. (By measuring sunlight over a month's time, we seek to detect the 12 m/s @ 27 day tugging of the moon on Earth. This will verify our ability to measure Jupiter-like planets on other stars, since the magnitude of a Jupiter-on-star effect is similar.) Our objective for FY99 is to add refinements necessary for observing starlight and conduct proof-of-principle demonstrations at an observatory, most likely Lick. Secondly, to extend the bandwidth, which increases the photon flux; and thirdly, install turnkey versions of the instrument at major observatories for use on a continuing basis by the astrophysical community. Fourthly, to explore advantageous use in solar, plasma and atomic physics. The refinements needed for observation of starlight include methods for efficiently converting a round beam cross-section to a slitlike cross-section, and repackaging optics into a portable box.
The instrument, which is called a cross-fringing spectrometer, is a series combination of an angle-independent interferometer and a high throughput low resolution grating (Fig. 1). By factoring the instrument into two parts, the best features of both components are achieved. The interferometer produces arbitrarily high spectral resolution (controlled by delay length) at arbitrarily wide field of view. Its sinusoidal instrument response has only 3 degrees of freedom, and therefore is optimum for detecting Doppler shifts much smaller than an absorption linewidth. The grating improves the net fringe visibility by spectrally dispersing the fringes, thereby preventing component fringes having random phases from blurring together. Since the grating is not responsible for detecting the Doppler shift, it can be optimized for efficiency, small size and cost. Our net instrument will be the size of a TV set, making it readily space-launchable. (In contrast, current high resolution spectrometers are not practical to put aboard the Hubble space telescope, for example.)
The interferometer produces fringes which can be splayed perpendicular to the coarse wavelength axis created by the grating. A minute Doppler shift manifests a vertical movement of the entire fringing pattern. The sensitivity to Doppler shift is set by the interferometer delay. A CCD camera records the pattern. The light can optionally pass through an iodine vapor cell to create reference wavelength fiducials. Data is taken for starlight alone, iodine alone, and starlight with iodine. The day to day change between iodine and starlight spectra yields the change in star velocity caused by a planet's orbit.
Fig. 1. Our cross-fringing spectrometer is a combination of an angle-independent interferometer and a high efficiency low resolution grating. An interferometer by itself produces fringes which blur together. The addition of the grating prevent this blurring, increasing fringe visibility.
Sunlight is now being observed through a basic form of the instrument, residing in our lab at B132/R2719. Example data is shown in Fig. 2a. The predicted Moiré fringing effect between the Fraunhofer lines of the solar spectrum and the interferometer fringe comb are observed. Construction of the computer algorithms to distill a velocity from the CCD image is underway. An excellently qualified astronomer post doc, Jian Ge, has been hired from Prof. Roger Angel's group at U. of Ariz. He joins us this summer. All major hardware needed for proof-of-principle operation on sunlight has been acquired. Visibility and professional dialogues among the astronomical community has been established.
Fig. 2. a) Solar absorption lines seen through our crossfringing spectrometer. Only a subset of the full spectrum is shown. Interferometer creates a fringe comb with a periodicity of ~0.23 Å, which is dispersed horizontally by the spectrometer. Tilting an interferometer cavity mirror creates vertically splayed fringes for monochromatic signals, and a slanted periodic comb for broadband signals. b) The overlap of the absorption lines with the slanted comb creates a Moiré effect. A Doppler shift causes Moiré pattern to shift vertically for all lines monolithically. c) Intensity profile along a vertical slice is sinusoidal and therefore has only 3 degrees of freedom, minimizing errors when measuring small Doppler shifts. In contrast, conventional spectrometers have 100's of degrees of freedom.
This instrument will accelerate the survey of thousands of stars in NASA's Doppler planet search, and make practical observation of weak stars, not now being observed. The more accurate instrument response of the interferometer (only 3 degrees of freedom instead of 100's for a grating) will improve the velocity confidence, allowing detection of Saturn-like or smaller mass planets. The 200x increased field of view will allow use of inexpensive large area light collectors ("light buckets"), such as the Hobby-Eberly telescope in Texas and future fiberoptically connected telescope arrays. These large collecting areas will speed the search by increasing the number of photons. The compact size (TV-size instead of kitchen-sized) makes spaceborne platforms practical for the first time. High resolution spectroscopy can now be performed on wide field of view objects such as nebulae, hydrogen clouds, atom laser clouds, and plasma physics experiments.
For further information about this project or the White Light Velocimetry Project, contact David Erskine at 925-422-9545.
UCRL-MI-142025
For questions regarding this web page, contact H Division Web Team
Last modified on 12/13/00