(Pictures)

POINTS
Precision Optical INTerferometer in Space

POINTS is a principal research topic of the SAO Precision Astronomy Group. The POINTS group includes teams at Itek Optical Systems and the Jet Propulsion Laboratory.

History

POINTS is a dual optical astrometric interferometer with a 2m baseline, which can measure the angular separation between two stars with a nominal accuracy of 5 micro-arcseconds (about 25 picoradians) in 2.5 minutes. How small is that? What's the catch? The nominal angle between the two interferometers is 90 deg, but this angle can be adjusted by +/- 3 deg. Thus the difficult problem of measuring this large angle is separated into two simpler measurements: the angle between interferometers and the small angular offset of each star from the axis of the interferometer observing it. The instrument detects a dispersed fringe (channelled spectrum) in a simple spectrometer and therefore can tolerate large pointing offsets, especially during fringe acquisition.

Through the use of high-precision laser metrology of the starlight optics, stable materials, measurement redundancy, and 360 degree closure information, the reduced data give unbiased angular separations of all pairs of stars (even those which the instrument could not have observed together) and a full set of instrument bias parameters. Analysis of quarterly observations of about 300 grid stars from a nominal 10 year mission yields position (0.6 microarcsec), parallax (0.4 microarcsec), and proper motion (0.2 microarcsec/year) for each star with respect to the rest of the grid. This would allow

and many other investigations.

POINTS is presently under consideration by two divisions of NASA-OSSA. It will be a powerful new multi-disciplinary tool for astronomical research. If chosen as the ASEPS-1 instrument by the Solar-System Exploration Division, it will perform a definitive search for extra-solar planetary systems, either finding and characterizing a large number of them or showing that they are far less numerous than now believed. If chosen as the AIM by the Astrophysics Division, POINTS will open new areas of astrophysical research and change the nature of the questions being asked in some old areas. In either case, it will be the first of a new class of powerful instruments in space and will prove the technology for the larger members of that class to follow. Based on a preliminary indication of the observational needs of the two missions, we find that a single POINTS mission will meet the science objectives of both ASEPS-1 and AIM.

What? Could you give that in English, please?
Renderings and diagrams (pretty pictures)

Abstracts of selected papers
For more information
Other interferometry sites

To report problems with this WWW document, send email to
rbabcock@cfa.harvard.edu
(Bob Babcock). Try clicking here.

History


POINTS was one of several advanced concepts suggested by Irwin I. Shapiro in 1974. R.D. Reasenberg became involved a few years later, and later teamed with K. Soosaar et al. of the C.S. Draper Laboratory (Cambridge, MA) for an early design study. The original baseline of 20-25 m shrank steadily over the first 10 years (chiefly as a cost-saving measure), finally resting at 2 m. One guiding emphasis has been on substituting state-of-the-art metrology in place of the advantages of dimension wherever possible. Go back

Footnotes:

Dual:
POINTS uses two interferometers observing two stars simultaneously, and measuring their separation; this is the most plausible way to get a meaningful measurement with 5 microarcsecond final accuracy, since no known gyroscope could be used for measurements at this extreme accuracy. Go back
Optical:
The nominal range of observation wavelengths is 250-900 nanometers; the human eye is typically sensitive from about 380 nm (violet) to 720 nm (deep red).
Astrometric:
The instrument measures the angle between the centers (centroids) of the stellar disks. It does not produce images, except in the sense of measuring the positions of a small number of point sources. An optical interferometer capable of making images with 5 microarcsecond resolution would be a very different beast: it would have about 20-30 apertures spaced by as much as 20 kilometers (12 miles). This size is determined by the approximate formula (wavelength) / (angular resolution). Go back
Interferometer:
Light from a single star enters two sub-apertures in one of the interferometers. A collection of mirrors brings these two samples of the starlight to meet at a 50-50 beamsplitter. Emerging from this beamsplitter are two combined beams; the visible-light waves in each combined beam are the sum and difference of the visible-light waves from each aperture. If you turn on two lamps in your living room, the intensities of the light from each bulb just add together, like two streams of ping-pong balls. But in an interferometer, we carefully arrange for these two samples of the same starlight to arrive at the beamsplitter and be laid on top of each other exactly; if the overlapping beams have travelled exactly the same distance through the optics (within a few microns) and are travelling in exactly the same direction, then their waves can either add or subtract.

The behavior of these two beams is very sensitive to the optical path difference (OPD), which is the difference in the distances the starlight must travel to reach the beamsplitter via each aperture. If the OPD is exactly zero, then the intensities of the two combined beams are equal (by symmetry). But if the OPD is not zero, the delay of one beam causes the two waves to add imperfectly. Indeed, if the OPD is exactly 1/4 wavelength, the waves from each aperture will subtract and tend to cancel in one of the combined beams; at the same time, the waves will reinforce each other in the other combined beam. This means all of the light from both apertures of the interferometer comes out one side of the beamsplitter. If the delay is 3/4 wavelength, the situation is reversed: essentially all of the light comes out the other side of the beamsplitter. If it is one full wavelength, the waves overlap just as they did at zero OPD -- despite the extra cycle of the starlight wave in one side -- and the intensities of the combined beams are again equal. A typical wavelength of light is about one half micron; thus the observed intensities in the two combined beams are extremely sensitive to the positions of the optics and to the direction to the star. POINTS takes advantage of this high sensitivity to determine the orientation of each of its two sets optics with respect to the corresponding star's wavefronts.

Spectrometer:
That description only works for a single color at a time. If the OPD is not zero, then the waves of some colors may add while others subtract. As the OPD gets larger and larger, the shortest wavelengths go through more and more complete cycles of this variation, while the longer wavelengths go through these cycles more slowly. Analysis of the intensities at each observed wavelength (e.g., in POINTS's spectrometer) provides a sensitive measure of the OPD, which in turn gives the angle between the instrument's optical axis and the direction to the star. Each spectrometer (there are two in each interferometer) is a simple design using a prism (for high throughput), a single aspheric reflector, and a CCD.

The technique of interferometry has been used in radio astronomy for almost three decades, but only recently has it become technically feasible for the short wavelengths of infrared and even visible light. Putting the instrument in space overcomes many of the remaining difficulties, and allows much higher astrometric resolution than in ground-based interferometers. Go back

How small is it?
POINTS's 5 microarcsecond measurement accuracy for the angular separation of two stellar disks is 
the width of		as seen at the distance
---------------------------------------------------------------
a pencil (1 cm)		from earth to the moon (400,000 km)
a human hair (0.1 mm)	from Los Angeles to New York (4,000 km)
an atom (1 angstrom)	across a room (4 meters)
... but each stellar disk could be as big as 
the width of		as seen at the distance
---------------------------------------------------------------
a football field (100m)	from earth to the moon 
a table (4 ft)		from Los Angeles to New York
a pencil		from Boulder to Denver (about 40 km)
a human hair		across 4 football fields (400m or 1/4 mile)
an atom 		through 5 sheets of paper
Go back
What's the catch?
Q: Nobody can measure that small, right? You've only counted photon-counting noise, haven't you, and left out all the systematic errors? Doesn't this violate the Heisenberg Uncertainty Principle?
A: No, there's no catch. The single-measurement integration time of about 2.5 minutes is calculated from splitting 5 microarcseconds up into lots of little bitty pieces: a budget for how much noise or systematic uncertainty we think we can handle from each of the error sources we can think of. That includes photon-counting noise as the largest contributor (by design); but as an example, we only allocate about 1-2 picometers (trillionths of a meter) tolerance for the resolution and stability of each laser gauge. We haven't left anything out of the budget that we know of. And picometer resolution for several minutes of observation time doesn't even come close to violating the uncertainty principle for mirrors and other objects that are big enough to see and grasp. Go back
90 deg angle:
The choice of a 90 deg mounting angle has several advantages: Go back
Measurement redundancy:
When there are enough stars in the "grid" (our collection of primary science targets), we can make sure that the number of measurements we make (of star-pairs with a separation that falls within the 90+/-3 degree range of our instrument) is, say, 5 times larger than the number of stars. When we have "extra" measurements like this, there is enough information in the full data set to allow unambiguous determination of the true bias-free star-pair separations along with a large set of instrument bias parameters. If we increase the redundancy of the data set, we may fit more bias parameters without increasing the star-pair separation uncertainties, or we get lower uncertainties with a fixed number of bias parameters. Go back
360 deg closure:
If four stars are spread around a great circle on the sky, with neighboring stars separated by about 90 deg angles, we know the measurements of these four angular separations should add up to 360 degrees. The real stars won't lie exactly in a plane like this (a 2-dimensional cartoon); but we can formulate a similar 3-dimensional constraint, and use it in the data analysis. We have shown that this works through a broad set of mission simulation studies. Go back
Kiloparsec (kpc):
1 kiloparsec (kpc) is 1000 times the distance that gives a 1 arcsecond parallax measurement. This is 3260 light-years, 30 million billion km, 20 million billion miles, or 200 million times the distance from earth to the sun. The distance from New York to Los Angeles fits into a kiloparsec as many times as a wavelength of green light (500 nm) fits into the NY-LA distance. The distance to the center of the Milky Way is about 8 kpc, and to the Small and Large Magellanic Clouds is roughly 50-60 kpc. Go back
ASEPS-1:
ASEPS: Astronomical Studies of Extra-solar Planetary Systems, formerly TOPS, Toward Other Planetary Systems.
AIM:
AIM: Astrometric Interferometry Mission, the only new mission recommended in the report of the Astronomy and Astrophysics Survey Committee of the National Research Council, chaired by John Bahcall (National Academy of Sciences, 1991).