In 2009 there are several spectroscopes on the market that are available to amateur astronomers, but they tend to cost between about $2000 and $4000 and are typically optimized for obtaining spectra of the stars and planets rather than lunar spectra. The absorption bands of lunar mineral spectra are very broad (absorption bands are usually between about 150 and 350 nm wide) rather than being discrete lines as in the case of stellar spectra. For this reason, low resolution spectroscopy works well. The simple spectroscope discussed here can be assembled quickly by almost anyone for a very small fraction of the cost of the models I have seen advertised. The finished spectroscope is very flexible since additional higher resolution gratings and slits of different widths can always be substituted if desired. The only problem here is that higher resolution gratings produce spectra that tend to go beyond the field of view of most 640 x 480 pixel cameras (but much larger format cameras are available). But to study mineral composition, the camera must have some sensitivity between about 900 nm and 1100 nm. Most cameras available to amateurs at reasonable cost fail dismally relative to sensitivity beyond about 1000 nm. The "ideal" camera would have the largest pixel format and the highest sensitivity between about 900 and at least 1100 nm. The spectroscope can also be used for planets and for most stars (if you have an autoguider). The spectroscope will not work with all telescopes, however, because the telescope must be able to bring an image to focus that is at least 22 cm (about 8.5 inches) beyond the point of insertion of the spectroscope into the focuser of the telescope. For example, the spectroscope works with my Celestron 9.25" SCT and with my Orion Optics OMC300 telescope, but my Takahashi Mewlon 250 just does not have enough available back focus for this instrument.
Lunar spectra superficially appear to be identical to solar spectra. The same narrow Fraunhoffer absorption bands appear in both. Non-mafic materials such as anorthosite do not have readily detectable absorption bands. Although anorthosite can contain a small amount of iron, it does not display a significant iron absorption band. This is due both to the small amount of iron present and to the effect of impact shock affecting much of the materials of the lunar highlands. However, mafic materials on the moon do have broad absorption bands which facilitate their identification. These absorption bands are due to the presence of higher amounts of Fe+2 in the crystal lattice of these mafic materials . The band centers vary as other metals such as magnesium substitute for iron in the crystal lattice. Pyroxenes absorb broadly at both 1000 nm and 2000 nm, while olivine absorbs only at 1000 nm but not at 2000 nm. Among the pyroxenes, low calcium pyroxene (orthopyroxene) has an absorption band center closer to 900 nm while high calcium pyroxene (clinopyroxene) has an absorption band center closer to 1000 nm. Mixtures of these have band centers in between.
Ideally spectral studies designed to identify lunar minerals would be conducted at both 1000 nm and 2000 nm. However, amateur astronomers do not yet have access to reasonably priced cameras capable of sensitive imaging at 2000 nm. Even if they did, prominent water vapor absorption lines near 1400 and 2000 would make it difficult or impossible to obtain useful spectra at 2000 nm from an Earth based telescope at sea level. Therefore, studies of mafic minerals would be largely limited to the region between 900 nm and 1100 nm which contains both olivine and pyroxene signatures. However, another atmospheric water vapor absorption band is also present between 900 and 1000 nm which must be removed during image processing by calibrating by division of the spectral image of the feature of interest by the spectrum of the Apollo 16 site taken through the same setup within a few minutes of each other. My experience thus far has been that careful calibration of spectra produces useful data between 900 and 1100 nm using a small telescope at sea level. In addition to this principal absorption trough, pyroxenes also show minor absorption troughs at visible wavelengths between 450 and 550 nm. Minor troughs have been described at 450-490 nm, 495 nm, 506 nm, 524 nm, 550 nm. However, it appears that the possible contribution of soil maturation and feldspar concentration has not been completely worked out for these minor troughs. Much of the work at these wavelengths has centered on laboratory mineral specimens and work on asteroids, particularly Vesta and its relatives and telescopic studies of lunar pyroxenes at these wavelengths are sparse in the literature.
Returning to the principal peak, low calcium pyroxenes will absorb closer to 900 nm, high calcium pyroxenes will absorb closer to 1000 nm while olivines will typically absorb between 1000 nm and 1100 nm. Most silicon based imaging cameras have little sensitivity beyond 900 nm. However, some silicon based cameras have enhanced near infrared capabilities. The Hitachi KPF2A camera has good sensitivity between 900 nm and 1200 nm and it is also sensitive to visible light at wavelengths greater than 400 nm. The Hitachi KPF2A is an analog progressive scan camera and requires a special frame grabber board for image digitization. I chose the PCI-1410 board from National Instruments. Other cameras based on non-silicon chips have better extended near infrared capabilities but are enormously expensive. The most common are based on Indium-Gallium-Arsenide chips.
To study lunar spectra it is necessary 1) to obtain crisp detailed spectra which include the important 900 to 1100 nm wavelengths useful for identification of mafic materials 2) to know the precise geographic location from which the spectra have been recorded and 3) perform a radiometric and photometric spectral calibration on the data. Elimination of the confounding effects of water vapor and oxygen bands during the calibration process is necessary. It is therefore necessary to obtain reference spectra from the Apollo 16 landing site during the course of each spectral imaging run. To eliminate telluric effects, feature spectra must be divided by this reference spectrum and then multiplied by an Apollo 16 landing site spectrum in which imager sensitivity has been normalized to a flat spectral response across all wavelengths. The latter is available at the USGS PDS website (see http://pds-geosciences.wustl.edu/missions/lunarspec/index.htm). Initial work has been very encouraging and initial studies of several craters has yielded results that closely match professional observations made with observatory instruments high above sea level. More information will be posted in the near future as data is collected and analyzed. If any of the above seems confusing, then refer to a more complete description of the calibration process here: http://digidownload.libero.it/glrgroup/selenologytoday7.pdf
A simple Lunar Spectrograph
A Simple Home-Made Spectrograph for Extended Objects:
The spectrograph with camera attached as shown above measures 8.5 x 7 x 2 inches and weights about 1.5 lbs. For comparison purposes, my Canon Rebel EOS DSLR camera with 50mm lens attached weighs noticably more than the combined weight of the spectrograph and the Lumenera camera with its 35mm CCTV objective lens attached. The spectrograph is stable and simple to assemble and disassemble. For use, a star diagonal is inserted into the visual back of the telescope and the spectrograph is inserted into the star diagonal. This arrangement minimizes spectrograph tube flexure.
The spectrograph can accomodate four round slit assemblies of different micron width with one filter holder space left empty. The slits are mounted in True Technology 1.25" clear glass mounts which screw into the filter slots on the holder.
Similarly, the upper Lumicon 5 filter holder can accomodate 4 different 1.25" diameter transmission gratings which screw directly into the filter holder slots, leaving one filter holder slot empty.
The spectrograph is assembled by screwing the Lumicon 0.4x compressor lens into an empty eyepiece barrel to form a 1.25" compressor lens assembly. This assembly is then inserted into the lower Lumicon 5 filter holder loaded with up to 4 separate slit assemblies. Next a second Lumicon 5 filter holder (loaded with up to 4 transmission gratings) is inverted, placed over the compressor eyepiece, and the securing screw tightened. Next the barrel from a Meade eyepiece projection coupler is placed over the exposed end of the upper Lumicon 5 filter holder. A few windings of masking tape around the exposed end of the upper Lumicon filter holder will allow a tight friction junction with the Meade coupler barrel. Finally a CCTV objective lens of the desired focal length (35 mm works well) is attached to the camera and its barrel is inserted into the Meade coupler barrel. The diameter of the CCTV lens barrel should be under 30 mm to fit within the coupler. A couple of set screws can be added to the coupler to secure the objective lens in place if the fit of the lens is too sloppy.
Now it is necessary to achieve focus of the object being imaged (use the telescope focus knob for this) and focus of the slit assembly (move the upper Lumicon filter holder up and down on the collimating lens for this). First put a slit in place but use no grating. Release the set screw holding the upper Lumicon filter holder to the compressor lens and move the filter holder up and down until the slit assembly is sharply in focus. Next the blank filter holder slots in both upper and lower filter holders and focus the desired object using the telescope focus knob. Finally, select the desired slit and the desired transmission grating and record spectra of the object. The grating will have to be rotated until the spectra are perpendicular to the slit and free of distortion. If either the slit or spectra are not crisply focused then the upper assembly must be moved slightly upwards or downwards (don't touch the telescope focus knob). The proper slit/grating combination must be selected to ensure that both the slit and the entire spectrum are both present simultaneously within the camera imaging field. This is very important.
The ability to go seamlessly from no slit and no grating to any desired slit/grating combination is an invaluable asset of this design. In order to determine the geography imaged by the slit, it is necessary to first take an orientation photograph with no slit or grating in place. When the spectra are imaged care must be taken to ensure that both the zero order (slit image) and 1st order spectra are both within the imaging field of the camera. Sometimes the slit must be moved slightly to the right or left to accomodate the entire 1st order spectrum within the camera field. Be aware that moving the slit will change the geography that the spectra relate to. When the slit image and 1st order spectra are sharply in focus, an image is taken. It is superimposed on the geographic image taken without use of slit or grating in order to determine the exact geography that the spectra relate to. This is easily done using Photoshop.
It is very important that the slit apparatus be very well light sealed. It is equally important that a dark hood be used around the Lumicon filter holder apparatus during acquisition of spectra to avoid extraneous light (as from the computer laptop) entering the camera independently of the slit.
The photographs below shows my lunar spectroscopy imaging setup with a Hitachi KPF2A extended NIR camera attached to the spectroscope. The telescope is a 9.25 inch F10 SCT. The computer contains a National Instruments PCI-1410 frame grabber board which converts the 60 Hz progressive scan analog output of the Hitachi camera into either tiff images or into an avi file which can be used to stack multiple images of the spectra to increase the signal to noise ratio.
Initial Spectra Recorded with the Simple Spectrograph
The first step in analyzing lunar spectra was to become familiar with the solar Fraunhoffer lines present in the solar spectrum and to use them for wavelength calibration. A solar spectrum showing these absorption lines taken with the spectrograph and a description of the Fraunhoffer lines present is shown below.
Many of these Fraunhoffer lines are visible in the lunar spectrum taken with spectrograph which is shown here:
This slit location for this spectrum is given by superimposing the slit image on a geographic image of the moon:
Thus, the bright spectral band in the image of the lunar spectrum is seen to represent the crater rim of Copernicus.
Spectra of Lunar Features
I have obtained spectra for a number of luar craters and maria with the spectrograph. The image below is of the crater Tycho. The spectra is a mixutre of central peak, crater wall, and crater floor.
The prominent trough between 900 and 1000 nm is typical of high calcium pyroxene mafic material. Crater spectra are calibrated against spectra of the Apollo 16 landing site. This is done by simply dividing the spectra of the area of interest by the spectra of the Apollo 16 landing site take through the same setup at as close to the same time as practically possible which means within minutes of each other (because weather conditions may vary over longer intervals). After first carefully co-aligning the absorption bands of the spectral image of the feature of interest and that of the Apollo 16 site in Photoshop, the images are imported into ImageJ and converted to 16 bit format. A pixel division of the recorded spectral image of the feature of interest by that of the Apollo 16 site is performed using the Process>Image Calculator submenu in ImageJ, selecting the spectral image of the feature of interest as image 1, division as the process, and the spectral image of the Apollo 16 site spectra as image 2. As an example, co-aligned spectra of Copernicus (upper) and the Apollo 16 site (lower) are shown below:
The first curve of Tycho below has been smoothed using a Savitzky-Golay spline. The first, second and fourth derivative curves follow. Derivative spectral curves can better define the band center of peaks and accentuate peak and trough transitions especially when mineral mixtures with overlapping peaks or troughs are present. Often hidden unsuspected minor peaks are revealed. Derivative curves at 2nd derivative or higher are best for this purpose. The 5th derivative is felt to be the most accurate, but could not be reproduced with the software used here. The curves was generated with TableCurve2D (see www.systat.com).
Application of Smoothing Spline to Tycho Reflectance vs Wavelength Data:
1st Derivative Spectral Curve for Tycho Reflectance vs Wavelength Data:
2nd Derivative Spectral Curve for Tycho Reflectance vs Wavelength Data (in yellow):
4th Derivative Spectral Curve for Tycho Reflectance vs Wavelength Data (in yellow):
Mare spectra are generally calibrated against the MS-2 area in southeastern Mare Serenitatis. The image below is a spectral ratio of an area within Mare Tranquillitatis calibrated against the MS-2 site. The slope of the ratio image has been used to compare the titanium content of the various maria but this is a rather complex area of research even at present.
Mare Tranquillitatis/MS-2 Site
A more detailed description of spectra that I have obtained of lunar features with the spectrograph will be included in an article I wrote for the 7th issue of the free online journal Selenology Today. The anticipated release date for this issue is mid to late May 2007.
