Dave Harrington Research

The Basic Idea:

My main interests are in new instrumentation and it's applications. There's a lot of neat work to be done near the boundary between science and engineering. By using new instruments and approaches or by combining techniques (like adaptiv optics, polarimetry and coronography), we can get new perspectives and insights. I've mainly used a high-resolution spectropolarimeter or polarimetry for my work, especially my thesis, but there's a lot more out there. I find building instruments and working with new techniques to be fascinating. The independent perspective a new instrument brings can broaden our understanding of an object or clarify a problem (or create problems). For instance, my thesis was initially on the spectropolarimetry of Herbig Ae/Be stars: intermediate mass young stars. Some previous publications used scattering theories to explain the results at moderate spectral resolution. The story changed at high resolution. There are only two spectropolarimeters on large telescopes (representing a roughly 20-fold improvement from all previous measurements on these stars) and both are in Hawaii. When we observed at these stars, an entirely unanticipated effect was seen. It didn't match any theory or model predictions at the time and is stimulating a lot of interesting work and discussion.... thanks to a new capability.

Instrumentation - Spectropolarimeter:

Spectropolarimetry is a growing field, but there are only a handful of instruments. There are only two high resolution spectropolarimeters on large telescopes (4m) - HiVIS at R=50000 and ESPaDOnS at R=68000. There's one on a 2m (Narval, an ESPaDOnS copy) and a medium resolution instrument on a 4m (ISIS at R=8000). In 2003 our group began design, construction and calibration of a spectropolarimeter for the HiVIS spectrograph (R~15000 to 50000) on the AEOS 3.67m telescope. HiVIS is a long-slit cross-dispersed echelle spectrograph with the polarizing optics just behind the slit. The installation, testing and calibration was mostly complete by mid 2005. The paper describing the basic instrument properties was published in Harrington et al. 2006. A number of upgrades and fixes were done leading up to the 2006-2008 observing season. I've led the writing of a dedicated data-reduction pipeline in IDL to process the data and do the polarization calculations. HiVIS uses a coude path that has many oblique reflections (fold mirrors) before the entrance-slit. This setup introduces very significant telescope polarization effects and a very thorough polarization calibration was done to prove the instrument's usefulness. A PASP paper describing the instrument calibration, polarization properties of the telescope and reduction software was published in Harrington & Kuhn 2008, or read the first few chapters of my dissertation - publications. With this new instrument, we can now do some exciting science.

Stellar Spectropolarimetry:

The environment around a star, especially obscured stars with circumstellar material, is very complex and dynamic. There is evidence of accretion, winds, disks, and jets, with many of these things happening simultaneously, and all variable in time. One major problem in understanding these processes is their small spatial scale - they occur mainly within a few stellar radii of the host star. Imaging with that kind of resolution even for the closest stars is basically impossible this century, and other methods must be developed used. Just like spectroscopy, the high resolution polarized spectra of a star can be used to get information about this region near the star. Models of stellar magnetic fields predict strong circular polarization in spectral lines. Absorption and scattering are thought to cause spectropolarimetric effects from circumstellar disks, stellar winds, and other regions around the star. We've been observing various stars with AEOS and CFHT to see what we can learn from the spectropolarimetric signatures. Some initial observations of Herbig Ae/Be stars showed an unexpected result - these polarization signatures strongly correlated with absorptive effects. This was first published in an ApJ Letter (Harrington & Kuhn 2007). That data inspired a group of people to create a new model based on optical pumping (Kuhn et al. 2008). We then looked at essentally every bright obscured star we could observe and found that this absorptive polarization effect was ubiquitous in obscured stars (Herbig Ae/Be, Be, Post-AGB, RV-Tau etc). It's everywhere. See Harrington & Kuhn 2009.

CfAO Professional Development &
Akamai Maui Short Course
There is a network of interwoven programs designed to train future teachers (grad-students and post-docs) and to give Hawai'i college students the tools to get in to grad school or to get a local high-paying job. There are several organizations doing this. My first encounter with this was through the Professional Development Program (PDP) run by the Center for Adaptive Optics (UC. Santa Cruz). This is essentially a training program for grad students and post-docs. We get taught current educational techniques, how to design and implement inquiry-based labs and activities and how to effectively engage a diverse classroom. There's an intensive 1-week seminar, then we design a lab, activity or course and teach it. I've been involved in the Akamai Maui Short Course - a one week intensive prep-course taught by 3-4 PDP participants (with help) for Hawai'i students teaching basic optics and adaptive optics. These students then do a paid summer research project either with local companies (like Trex, Textron, H-nu photonics) or research organizations (like PDC, UH, IfA). These students do extremely well, with 85% remaining on the 'science pathway' and most geting job offers from their host organizations. I've also participated with the teaching team for the CfAO's week-long adaptive optics summer school.

HiVIS & Curvature Adaptive Optics

Adaptive optics is quickly becoming a critical component of many instruments on large telescopes. By using a deformable mirror to remove the image motion and blurring caused by the atmosphere, the capability of an instrument is greatly increased. AO can also assist other techniques like coronography, polarimetry or spectroscopy. One first has to sense how the incoming light is distorted (with the wavefront sensor), then you have to correct it (with the deformable mirror). This is usually done now with a Shack-Hartmann wavefront sensor and a deformable mirror (typically with push-pull actuators). The incoming wavefront hits a lenslet array and gets broken up into an array of spots. The spots are imaged on a ccd and deviation of these spots from a regular grid is measured. This is fed in to software that reconstructs the wavefront shape. This wavefront shape is then sent to the deformable mirror to correct the beam.

Curvature based systems correct the beam in a moderately different way. The deformable mirror is a bi-morph, not push-pull. A voltage on an actuator doesn't "poke" the mirror, it curves the mirror. The wavefront still hits a lenslet array, but the lenses feed avalanche photo diodes (APD's) through fibers (not a ccd). The lenslet is not at a pupil plane - it's "put" upstream and downstream of the pupil by using an oscillating membrane mirror at an upstream focal plane. It turns out that difference in intensity between upstream and downstream measured by a lenslet is proportional to wavefront curvature. The lenslet array is mapped 1-to-1 with the deformable mirror - one lenslet, one actuator. Since APD's are noiseless, you can read fast and use the right number actuators for your guide star. Since the lenslet signal is directly proportional to curvature, and the deformable mirror is a curvature mirror, the wavefront never needs to be constructed. A 100 actuator system can be run at 2kHZ by a simple laptop (not a supercomputer). Typically these systems do about 3 times better than Shack-Hartmann systems.

The worlds second-largest curvature AO system, Hokupa'a 85, was built by the UH AO group and is being adapted for use with HiVIS. Hokupa'a is the progenitor for the 85-element system on the 8m Gemini South telescope (NICI). Comparing NICI with the competing Shack-Hartmann system on Gemini North shows the power of curvature AO - NICI closes loop on stars 1 magnitude fainter and does a better correction (Strehl) with half the number of actuators.

Spectropolarimetry of Deep Impact:

We observed the Deep Impact event with the HiVIS spectropolarimeter July, 2005. This was a unique opportunity to try long-slit high-resolution spectropolarimetry on a bright comet, Tempel 1, and to be part of the huge ground-based support effort to characterize the ejected material. Basically every major observatory on the planet performed some kind of observation for this very unique event. Karen Meech, wrote a paper in Science describing the ground-based observing effort (Meech et al. 2006). AEOS was one of a few to do polarimetry during the impact. The results were published in an Icarus Deep-Impact special issue (Harrington et al. 2007). The polarization of the coma varies with wavelength and phase-angle (as well as object!). Temple 1 was scattering light at 40 degrees at impact. Most comets at that scattering angle show a few % polarization at 650nm rising to 5-8% by 950nm. Basically, we saw the polarization of the scattered light from the coma start 'blue' get a bit more 'blue' (higher polarization at 650nm than at 950nm). The results were presented at AOGS (Singapore) and Deep Impact - A World Event (Brussels). A very interesting picture of the event has emerged since these initial meetings. There were only a few observations of 'blue' polarization and hardly any models exploring it. Now people are finding that blue polarimetric colors are at least somewhat common and are exploring new models of cometary dust.

Instrumentation - Imaging Polarimeter:

Joe Masiero led a group that built a dual beam imaging polarimeter for the UH 2.2m telescope. He's doing asteroid polarimetry as part of his thesis. This instrument is quite powerful, having a polarimetric accuracy well below .1%. It's simple, has a dedicated data reduction script and is open to the UH community. The design is different than most imaging polarimeters and it's designed specifically for point-sources. A savart plate provided two orthogonally polarized stellar images and two waveplates allow the measurement of the complete polarization state of the object (linear and circular). The polarimeter is mounted just upstream of the ccd. The main calibration is published in Masiero et al. 2007 - PASP 119. I provided some help with the concepts, the dedicated IDL reduction scripts, and initial calibration of some of the optics. Imaging polarimetry is another one of those topics that is becoming more popular becuase of it's power in providing more information about unresolvable spatial-scales, surface properties and object geometries. It's also good for increasing the contrast between polarized and unpolarized objects. You can typically supress unpolarized light by a factor of 100 to 10000, allowing for detection of otherwise-hiden polarized sources (like planets next to stars or circumstellar disks around stars).

IR Spectropolarimeter:

HiVIS is really two spectrographs - one in the visible (450-950nm) and one in the near infra-red (1000-2500nm). The IR spectrograph is also a high-resolution (R~7000-30000) cross-dispersed echelle. It has three cross-disperser settings for J, H and K band. See Thornton's Dissertation or SPIE paper for the capabilities of the IR spectrograph. We've been working with the IR arm to build a spectropolarimeter similar to the visible one. We've designed a cold savart plate (calcite bonding good to 70K in a vacuum & optical modelling) and a cold mount for installation inside the dewar. We've bought waveplates, mounts, and rotation stages for the rest of the polarimeter and are in the testing & construction phase.

	The Basic Idea: My main interests are in new instrumentation and it's applications. There's a lot of neat work to be done near the boundary between science and engineering. By using new instruments and approaches or by combining techniques (like adaptiv optics, polarimetry and coronography), we can get new perspectives and insights. I've mainly used a high-resolution spectropolarimeter or polarimetry for my work, especially my thesis, but there's a lot more out there. I find building instruments and working with new techniques to be fascinating. The independent perspective a new instrument brings can broaden our understanding of an object or clarify a problem (or create problems). For instance, my thesis was initially on the spectropolarimetry of Herbig Ae/Be stars: intermediate mass young stars. Some previous publications used scattering theories to explain the results at moderate spectral resolution. The story changed at high resolution. There are only two spectropolarimeters on large telescopes (representing a roughly 20-fold improvement from all previous measurements on these stars) and both are in Hawaii. When we observed at these stars, an entirely unanticipated effect was seen. It didn't match any theory or model predictions at the time and is stimulating a lot of interesting work and discussion.... thanks to a new capability.
	Instrumentation - Spectropolarimeter: Spectropolarimetry is a growing field, but there are only a handful of instruments. There are only two high resolution spectropolarimeters on large telescopes (4m) - HiVIS at R=50000 and ESPaDOnS at R=68000. There's one on a 2m (Narval, an ESPaDOnS copy) and a medium resolution instrument on a 4m (ISIS at R=8000). In 2003 our group began design, construction and calibration of a spectropolarimeter for the HiVIS spectrograph (R~15000 to 50000) on the AEOS 3.67m telescope. HiVIS is a long-slit cross-dispersed echelle spectrograph with the polarizing optics just behind the slit. The installation, testing and calibration was mostly complete by mid 2005. The paper describing the basic instrument properties was published in Harrington et al. 2006. A number of upgrades and fixes were done leading up to the 2006-2008 observing season. I've led the writing of a dedicated data-reduction pipeline in IDL to process the data and do the polarization calculations. HiVIS uses a coude path that has many oblique reflections (fold mirrors) before the entrance-slit. This setup introduces very significant telescope polarization effects and a very thorough polarization calibration was done to prove the instrument's usefulness. A PASP paper describing the instrument calibration, polarization properties of the telescope and reduction software was published in Harrington & Kuhn 2008, or read the first few chapters of my dissertation - publications. With this new instrument, we can now do some exciting science.
	Stellar Spectropolarimetry: The environment around a star, especially obscured stars with circumstellar material, is very complex and dynamic. There is evidence of accretion, winds, disks, and jets, with many of these things happening simultaneously, and all variable in time. One major problem in understanding these processes is their small spatial scale - they occur mainly within a few stellar radii of the host star. Imaging with that kind of resolution even for the closest stars is basically impossible this century, and other methods must be developed used. Just like spectroscopy, the high resolution polarized spectra of a star can be used to get information about this region near the star. Models of stellar magnetic fields predict strong circular polarization in spectral lines. Absorption and scattering are thought to cause spectropolarimetric effects from circumstellar disks, stellar winds, and other regions around the star. We've been observing various stars with AEOS and CFHT to see what we can learn from the spectropolarimetric signatures. Some initial observations of Herbig Ae/Be stars showed an unexpected result - these polarization signatures strongly correlated with absorptive effects. This was first published in an ApJ Letter (Harrington & Kuhn 2007). That data inspired a group of people to create a new model based on optical pumping (Kuhn et al. 2008). We then looked at essentally every bright obscured star we could observe and found that this absorptive polarization effect was ubiquitous in obscured stars (Herbig Ae/Be, Be, Post-AGB, RV-Tau etc). It's everywhere. See Harrington & Kuhn 2009.
	CfAO Professional Development & Akamai Maui Short Course There is a network of interwoven programs designed to train future teachers (grad-students and post-docs) and to give Hawai'i college students the tools to get in to grad school or to get a local high-paying job. There are several organizations doing this. My first encounter with this was through the Professional Development Program (PDP) run by the Center for Adaptive Optics (UC. Santa Cruz). This is essentially a training program for grad students and post-docs. We get taught current educational techniques, how to design and implement inquiry-based labs and activities and how to effectively engage a diverse classroom. There's an intensive 1-week seminar, then we design a lab, activity or course and teach it. I've been involved in the Akamai Maui Short Course - a one week intensive prep-course taught by 3-4 PDP participants (with help) for Hawai'i students teaching basic optics and adaptive optics. These students then do a paid summer research project either with local companies (like Trex, Textron, H-nu photonics) or research organizations (like PDC, UH, IfA). These students do extremely well, with 85% remaining on the 'science pathway' and most geting job offers from their host organizations. I've also participated with the teaching team for the CfAO's week-long adaptive optics summer school.
	HiVIS & Curvature Adaptive Optics Adaptive optics is quickly becoming a critical component of many instruments on large telescopes. By using a deformable mirror to remove the image motion and blurring caused by the atmosphere, the capability of an instrument is greatly increased. AO can also assist other techniques like coronography, polarimetry or spectroscopy. One first has to sense how the incoming light is distorted (with the wavefront sensor), then you have to correct it (with the deformable mirror). This is usually done now with a Shack-Hartmann wavefront sensor and a deformable mirror (typically with push-pull actuators). The incoming wavefront hits a lenslet array and gets broken up into an array of spots. The spots are imaged on a ccd and deviation of these spots from a regular grid is measured. This is fed in to software that reconstructs the wavefront shape. This wavefront shape is then sent to the deformable mirror to correct the beam. Curvature based systems correct the beam in a moderately different way. The deformable mirror is a bi-morph, not push-pull. A voltage on an actuator doesn't "poke" the mirror, it curves the mirror. The wavefront still hits a lenslet array, but the lenses feed avalanche photo diodes (APD's) through fibers (not a ccd). The lenslet is not at a pupil plane - it's "put" upstream and downstream of the pupil by using an oscillating membrane mirror at an upstream focal plane. It turns out that difference in intensity between upstream and downstream measured by a lenslet is proportional to wavefront curvature. The lenslet array is mapped 1-to-1 with the deformable mirror - one lenslet, one actuator. Since APD's are noiseless, you can read fast and use the right number actuators for your guide star. Since the lenslet signal is directly proportional to curvature, and the deformable mirror is a curvature mirror, the wavefront never needs to be constructed. A 100 actuator system can be run at 2kHZ by a simple laptop (not a supercomputer). Typically these systems do about 3 times better than Shack-Hartmann systems. The worlds second-largest curvature AO system, Hokupa'a 85, was built by the UH AO group and is being adapted for use with HiVIS. Hokupa'a is the progenitor for the 85-element system on the 8m Gemini South telescope (NICI). Comparing NICI with the competing Shack-Hartmann system on Gemini North shows the power of curvature AO - NICI closes loop on stars 1 magnitude fainter and does a better correction (Strehl) with half the number of actuators.
	Spectropolarimetry of Deep Impact: We observed the Deep Impact event with the HiVIS spectropolarimeter July, 2005. This was a unique opportunity to try long-slit high-resolution spectropolarimetry on a bright comet, Tempel 1, and to be part of the huge ground-based support effort to characterize the ejected material. Basically every major observatory on the planet performed some kind of observation for this very unique event. Karen Meech, wrote a paper in Science describing the ground-based observing effort (Meech et al. 2006). AEOS was one of a few to do polarimetry during the impact. The results were published in an Icarus Deep-Impact special issue (Harrington et al. 2007). The polarization of the coma varies with wavelength and phase-angle (as well as object!). Temple 1 was scattering light at 40 degrees at impact. Most comets at that scattering angle show a few % polarization at 650nm rising to 5-8% by 950nm. Basically, we saw the polarization of the scattered light from the coma start 'blue' get a bit more 'blue' (higher polarization at 650nm than at 950nm). The results were presented at AOGS (Singapore) and Deep Impact - A World Event (Brussels). A very interesting picture of the event has emerged since these initial meetings. There were only a few observations of 'blue' polarization and hardly any models exploring it. Now people are finding that blue polarimetric colors are at least somewhat common and are exploring new models of cometary dust.
	Instrumentation - Imaging Polarimeter: Joe Masiero led a group that built a dual beam imaging polarimeter for the UH 2.2m telescope. He's doing asteroid polarimetry as part of his thesis. This instrument is quite powerful, having a polarimetric accuracy well below .1%. It's simple, has a dedicated data reduction script and is open to the UH community. The design is different than most imaging polarimeters and it's designed specifically for point-sources. A savart plate provided two orthogonally polarized stellar images and two waveplates allow the measurement of the complete polarization state of the object (linear and circular). The polarimeter is mounted just upstream of the ccd. The main calibration is published in Masiero et al. 2007 - PASP 119. I provided some help with the concepts, the dedicated IDL reduction scripts, and initial calibration of some of the optics. Imaging polarimetry is another one of those topics that is becoming more popular becuase of it's power in providing more information about unresolvable spatial-scales, surface properties and object geometries. It's also good for increasing the contrast between polarized and unpolarized objects. You can typically supress unpolarized light by a factor of 100 to 10000, allowing for detection of otherwise-hiden polarized sources (like planets next to stars or circumstellar disks around stars).
	IR Spectropolarimeter: HiVIS is really two spectrographs - one in the visible (450-950nm) and one in the near infra-red (1000-2500nm). The IR spectrograph is also a high-resolution (R~7000-30000) cross-dispersed echelle. It has three cross-disperser settings for J, H and K band. See Thornton's Dissertation or SPIE paper for the capabilities of the IR spectrograph. We've been working with the IR arm to build a spectropolarimeter similar to the visible one. We've designed a cold savart plate (calcite bonding good to 70K in a vacuum & optical modelling) and a cold mount for installation inside the dewar. We've bought waveplates, mounts, and rotation stages for the rest of the polarimeter and are in the testing & construction phase.