My journey into spectroscopy

Calculating Spectral Response

Sun, 20 Jul 2008 19:19:26 -0500

In my previous articles, I have discussed the basic equipment and the technique to capture data and to calibrate the wavelength of the spectrum. This month, we will take the calibration one more step by computing the Spectral Response of the optical system.
So what is a Spectral Response? Many astrophotographers may be familiar with Quantum Efficiency. Quantum Efficiency is essentially a measure of the probability to detect a photon at a specific wavelength. The Spectral Response is almost the same measurement. Instead of measuring the probability of a photon detection at a specific wavelength, the Spectral Response is simply the measurement of current as a function of the photon wavelength. For most people, these two measurements are essentially equivalent in that they show how sensitive their camera is to different wavelengths of light.
In order to compare spectra collected from two systems, you need to compensate for the difference in the sensitivity of the measurement system across the spectrum. Some cameras are very sensitive to the red wavelengths while other cameras are sensitive to the blue wavelengths. Furthermore, different telescope optics have differing abilities to transmit light. This is particularly true when talking about refractors since glass impacts the overall transmission of light. Thus, it is important to measure the Spectral Response of the optical system. Once you have achieved this, you can divide the raw spectra response of a star (or DSO) by the Spectral Response in order to get the true intensity measurement. This process is sometimes called “Flux Calibration”.
The calculation of the Spectral Response of a system is actually quite simple. One of the SW tools I use, VSPEC (which is free), contains a library of 24 spectra of bright, easy to find stars. The library covers all four seasons so you are assured to be able to find at least one star high in the sky at any point of time.
The process of measuring the Spectral Response is thus quite simple. You first need to select a star that is high in the sky. The reason you want it to be high in the sky is to minimize the blue extinction effects of the atmosphere. (The atmosphere does not do a good job of transmitting blue light). For my response, I choose Epsilon Orionis as my reference star. You collect the data the same way as you normally would. Use a low resolution grading and slit width so that you minimize the effects of air absorption lines. Process the spectrum just as you would for other measurements. Once completed, all you have to do is select the “Compute flux of reference star” command to perform the calibration.
For most optical systems, the largest variable in the Spectral Response is the camera. Thus you would expect the Spectral Response to look similar to the quantum efficiency of the camera. My measurement using a SBIG ST-10 Camera and a Mewlon 180 Dall-Kirkham telescope is shown below. For comparison, look at the quantum efficiency plot supplied by Kodak. Note how similar they are.
One interesting aspect of measuring Spectral Response is that you can use this technique to compare the performance of different telescopes as well as get a good idea of the performance of various filters. This is the way you would go about doing just such an experiment.
1. Measure the Spectral Response of the telescope (without using filters)
2. Add a filter (like light pollution filter) to the optical path.
3. Measure the Spectral Response of the entire system with filter. Make sure you use the exact same exposure times for each curve calculation.
4. The filter response is the difference between the two curves.
I plan on performing this type of experiment not only on color filters and light pollution filters but also on telescopes in a future article. Until then, have many clear nights.

First Star Spectra

Sun, 13 Apr 2008 10:02:46 -0500

In the January issue of the Spectrum, I had selected the SBIG DSS-7 Spectrograph for first instrument. I also attempted first light on it by measuring the spectrum of a fluorescent lamp. I compared the collected spectrum to the light pollution present in the night time sky in North Texas. From that, we could easily tell that there are a lot of Sodium (Na) and Mercury (Hg) vapor lamps in the Metroplex. Our next step is to put the Spectrograph onto an actual telescope and measure the spectra of a star.
For this exercise, I decided to use my Mewlon 180 mounted on my old Losmandy G11 mount. This way, I could mount a second refractor on the mount so that I could use a SBIG STV autoguider to guide the mount. This would help me keep the star centered in the slit of the spectrograph. I selected Betelgeuse as our first star to image with the spectrograph. I chose this star ,for one, because it’s an easy star to find from my back yard. It was high up in the sky at that time of the year. Betelgeuse is also a red giant and as such, there are several interesting aspects that can be seen in its spectrum.
Here is how I went about capturing the Spectrum for Betelgeuse.
★ I setup the mount, telescope, spectrograph, and SBIG ST-10 CCD Camera. I made sure the telescope was well balanced and I turned on the CCD camera and set the temperature regulation to -20C. (It was a cold night in December).
★ Next, I waited about one hour for the scope and camera to get to thermal equilibrium.
★ Focusing this Spectrograph is real easy. To do so, you flip the diffraction grading to “mode 0”. In this mode, the grading acts as a mirror. Betelgeuse shows up on the camera just like it would if there was no spectrograph. Focus is achieved using the same methods that would be done for a pretty picture.
★ After focus is achieved, the next step is to move the scope and center Betelgeuse to be in the middle of the entrance slit to the spectrograph. You may recall that the DSS has slit sizes of 50, 100, 200, and 400 micron. I created a mask of the slits which I can overlay onto my images. This allows me to know exactly where to place the star. I decided to aim at the 100 micron slit primarily because I was unsure if I would be able to track the star accurately enough to keep it in the 50 mircon slit. Once the star was centered on the slit, I started the autoguider. We are now ready to capture data.
★ Last month, when we took calibration spectra, the spectra filled the entire CCD field of view. For some reason, I was expecting that to happen when I took spectra of Betelgeuse. I was very surprised however to see just a very thin line going horizontally across the CCD. Surely, I did something wrong. After a couple more test photos, I realized my problem. I am supposed to get a line. The star is at pin point focus and will only illuminate a small area. So, I was doing things right.
★ To increase sensitivity, I put the CCD into 1x4 binning mode. This binning mode will combine 4 rows of pixels into one big row of pixels. This increases the sensitivity of the chip while not impacting the horizontal resolution. Since light is being dispersed horizontally, this is important. The spectrum resolution is in the horizontal direction, not in the vertical direction. I next figured out my exposure time by setting for a maximum of 50000 ADUs (Analog to Digital Units). This will prevent saturating the sensor. After making all of these settings, I came up with a 2.5 second exposure time. (well, Betelgeuse is a bright star).
★ I decided to capture 30 frames of Betelgeuse. After capturing the frames, I also captured 10 dark frames and 10 bias frames all using the 1x4 binning.
Now that we have collected some data, we can go inside and analyze it. Processing the FITs files for spectroscopy starts out the same way as you would for pretty pictures. We remove the effects of dark current and read-out noise by doing a dark frame subtraction. Note that we did not do a flat field. This is something I will need to do in the future. I haven’t done it now primarily because it has all types of complications associated with it that go well beyond the problems you have with generating flat fields for “pretty pictures”. For example, what light source do you use? Remember that the dispersion of light is a function of what is emitted and in the case of a spectrograph that will vary from light source to light source.
After each spectra subframe has been dark subtracted, I stacked them using a median combine function. Note that there is no need to register the images. Since the spectrum is not moving with respect to the slit, all subframes are already in alignment.
Note that the bright line is that of Betelgeuse. If you look closely, you can tell that there are some discontinuities in the spectrum. Even at this point, the hydrogen alpha line is pretty apparent. Still, it looks nothing like the spectrum we’ve seen in books. That will take a little more work. Also note that light did come in through the other slits. I took this spectra during a clear night with a waxing gibbous moon. The moonlight had entered the slit and was also being dispersed onto the CCD image. This is actually useful information since that moonlight effects the accuracy of my spectral measurements. If desired, I can use this data to subtract the effects of the moonlight from the spectra of Betelgeuse.
The next step is to actually measure the spectral output of Betelgeuse from the median average of all of the subframes. We will need to use a different tool for this. AIP4WIN has a tool that allows you to do this. Also, there is a program written by Valerie, a French amateur astronomer, called VSpec. For this article, I will use VSpec since it has more functionality. (Note that part of the interface for this program is still in French, so sometimes, you have to guess at what you are doing).
So, to measure the spectrum in Vspec, just load the median image and then select the measuring boundaries. I set the range so that the line from Betelgeuse is completely within the measuring box. I minimize how much other stuff (moon light, etc) is included. The red lines in the picture show the measurement area. After the measurement area has been set, you can command the application to measure the spectrum. At this stage, the program will sum the pixel values of each column and create a plot of the pixel sums as a function of pixel column number. Our next step is to figure out what wavelength to assign to each column. Earlier I had measured the spectrum of a fluorescent lamp. I knew the wavelengths of some of the emission lines on that lamp. Knowing that data, I can develop a simple calibration curve that maps a wavelength to each pixel. Since I am using the exact same optical equipment (without any mechanical changes) for the measurement of Betelgeuse, I can use that calibration curve on the Betelgeuse plot. The end result looks like this.
As indicated at the start of this program, Betelgeuse is an interesting star. It is a large red giant of class M. As stars finish converting Hydrogen to Helium, they start to grow large and turn red. As this happens, the stars cool and you are then able to observe some of the metals in the stars atmosphere. Betelgeuse’s spectrum is dominated by Titanium Oxide (TiO). Many of the large valleys you see in the above plot are a result of absorption lines from Titanium Oxide.
Note that so far, the spectra we’ve looked at are all black and white. Why is this? Isn’t spectra supposed to have the colors of the rainbow? The answer is, yes, the spectra does have the rainbow colors. However, the SBIG camera is a monochrome camera. As such, all of the spectra is captured as black and white. Because we know which wavelength of light belongs to which pixel, we are able to re-create the spectrum in color. The process of doing this is called “Spectral Synthesis”.
The VSPEC software has a pretty good Spectral Synthesis algorithm. The resynthesized color spectrum of Betelguese thus looks like:
Now that I have successfully captured spectra, my next several articles will show various projects that can be done with a spectroscope. We will start in the next article discussing how to compute the Spectral Response (very similar to Quantum Efficiency of a CCD) of the optical system (telescope and camera).

Beginning DSS

Tue, 15 Jan 2008 11:59:20 -0600

In my last article (November, 2007) I introduced the topic of Spectroscopy, discussed the different products on the market for the amateur, and reasoned why the SBIG DSS-7 (Deep Space Spectrograph) was a good choice for my first spectrograph. In this article, I will discuss the DSS in more detail and describe my experiences in getting “first light”.
The DSS-7 Spectrograph was designed to work with a f/10 telescope and a CCD camera. It was originally intended for the ST-7, but will also work with most all of the ST series cameras. My plan is to use my ST-10 camera for the sensor. I will use the spectrograph with my trusty Celestron C-8 and Mewlon 180. Though the Mewlon is f/12, it should still work fairly well with this device. The only possible area of concern is that I may lose the effectiveness of the slits near the edge of the detector (due to more narrow light cone).
The DSS-7 was designed to separate and focus light wavelengths from 4000 to 8000 angstroms across the width of the ST-7 sensor. This covers the range that humans can see between 4500 (violet) to 7000 (red). Though the ST-10 has a larger sensor than the ST-7, SBIG claims that we will not see any significant increase in spectral coverage due to limitations in the DSS-7 design. The dispersion of the spectrograph is about 600 nanometers/mm. For the 6.8 micron pixel on a ST-10 camera, this works out to be about 4.1 angstroms/pixel. The resolution after factoring in a finite slit width is about 16 angstroms or about 4 pixels on the ST-10 camera.
Light enters the spectrograph through a slit (on the telescope side of the spectrograph). This light is folded (using a mirror) to a lens which collimates the light onto a diffraction grating. The grating separates the light into its color components. This is then focused onto the CCD camera using another lens. A small motor is used to move the slit into and out of the light path. This is useful to allow for the focusing of the image. Focus is achieved by moving the diffraction grating (using another motor) into “mode 0”. This mode is a purely reflective mode and allows you to see the image from the telescope.
The slit for the spectrograph actually consists of multiple, different-sized slits. The central slit is 50 microns. It is flanked by a 100 micron slit (above) and a 200 micron slit (below). 400 micron slits lie at the ends. The 50 micron slit is smaller than the width of human hair. So, you might imagine that very little light makes it through the system. Having different sized slits has its advantages. The wider the slit, the more light you are able to collect and thus, the fainter the object you are able to measure. Of course, nothing comes for free. With a wider slit, you will also reduce the resolution of the spectrum.
With all of these moving parts, there is of course, some adjustments that need to be made in order to get a working spectrum. All of these adjustments can be made in the comfort of your home without a telescope. They make for a good project for one of those cloudy nights. The first step is attaching the CCD camera to the spectrograph. The directions supplied by SBIG are very easy to follow. You basically remove the CCD camera color wheel, add the adapter plate, and then attach the camera to the spectrograph. It’s a pretty simple task. You will note that there are some adjustment screws on the adapter plate. These will be used later to make sure that the slit is oriented in the same direction as the camera pixels.
The next task is to focus the slit. This requires that the camera is running and that the spectrograph is pointed at some light source. (I used the kitchen lamp). Control of the spectrograph and camera is best achieved with CCDOps. This is the SBIG software which is supplied with the camera and spectrograph. In CCDOps, you select the DSS “View Slit Mode”. In this mode, the spectrograph is configured with the slit in the light path and the diffraction grading in its reflective mode (mode 0). The result is that you will see the slit projected onto the CCD sensor. You next continuously expose the camera and adjust the focus of the spectrograph until you get the slit perfectly focused. This may take some time and patience. My first attempts at this resulted in a well focused slit, but when I went and tightened the set screws, the slit would go out of focus. The nice thing is that once you have this adjusted for your camera, you don’t need to adjust it much in the future. I have in fact removed my camera from the spectrograph and reattached it and have not needed to adjust the spectrograph focus.
Once you have the slit focused, you need to adjust the slit orientation. The idea is that you get the slit oriented exactly with the pixel structure of the camera. By doing this, an absorption line or emission line will stay within the pixel boundaries. This will also open up the possibility of binning along the vertical axis. Binning will further increase the sensitivity of the spectrograph.
Now that the spectrograph focus is set, and the slit is oriented to the CCD pixels, we are ready to make our first picture of a spectrum. This can be done without a telescope. In fact, I would encourage the beginner to start without the telescope. Just point the spectrograph (and camera) to a lamp. Or, you could go outdoors and point it at the daytime sky. Or, if you live in a very light polluted area, go out at night, and capture the night sky. Just lay the camera on a table (pointing up), and take several exposures.
I opted to start with a lamp. I chose this option since I figured it is probably the most controlled set of circumstances that I can come up with. So, I used a 60W fluorescent lamp that can be purchased at Walmart. Again, use CCDOps to control the spectrograph and camera. In this case, choose the view spectrum mode. Select an exposure length that does not clip any of the pixels (i.e. no pixels reach full well depth). You will note that CCDOps defaults to vertical binning being turned on. Take advantage of this. Even with bright objects, it shortens the exposure time (always a good thing). And you don’t sacrifice any resolution since your slit is aligned with the pixels. I use the default setting of 1x4 for my work. My first results looked something like the picture above.
Now that we have collected the spectra, what do we do next?. Note that since the ST-10 is a monochrome camera, all we have is a black and white image of the spectrum. Also, what do all of those lines represent?
SBIG ships software with the DSS-7. I found this software very difficult to use, however. One problem I ran into is the fact that the software requires the image file to be of the size of 728x20 pixels. This size works well with a ST-7 camera, but it does not work well with my ST-10 camera which has a different pixel size. I could make the appropriate adjustments and make it work, but it would involve manipulating data and possibly throwing away some valuable information. I was able to find some other software on the internet that is free, however. Visual Spec (VSPEC) is written by Valerie Desnaux. The program started because of all of the work being done in this area by the French Amateur Astronomy Community. Most of the instructions are now available in English and I have had no problem figuring out what they mean. At any rate, VSPEC allows me to read the spectrum photograph and convert it into a graph. It even allows me to calibrate the graph.
To extract the data, you need to select a region of interest. Since we collected data in all slits, and the 50 micron slit will give the most accurate data, I decided to select that region. VSPEC then sums each column vertically. This is equivalent to binning along the vertical axis. The sum of each pixel column is then put into a graph. All we need to do now is to identify two peaks in the graph, and then we will be able to identify any peak (or value) in the graph. I was able to find the spectrum of a fluorescent lamp that looked similar to the one I captured. I identified two peaks and used this to calibrate the graph. The result is shown below. Note that Mercury (Hg) is a very active component of this lamp.
Now that we have calibrated the spectrograph, we can capture other spectra. So, I decided to go outside and capture the light pollution in Plano. The result is shown below. Since I did not use a telescope for this and thus captured what was at zenith, this time, I used the 400u slot just so I could reduce the exposure time.. Since no mechanical adjustments were made to the spectrograph, I can use the same pixel to light wavelength mapping as I did in the calibrated lamp example. The resulting spectrum is as follows.
Note that there is quite a bit of energy in the mercury (Hg) and sodium (Na) regions. This is not surprising since many street lamps are either mercury vapor or sodium vapor lamps. Knowing this spectrum allows us to develop filters that can block out this area and perhaps improve our ability to capture images from the Metroplex.
Now that I have the spectrograph operational, the next step is to attach it to a telescope. We will discuss what is involved with that in my next article.

First photons

Thu, 15 Nov 2007 11:33:16 -0600

One of the main reasons why I became interested in astronomy as a child was the opportunity of discovery. I would watch that TV show of the famous star ship with the bold captain that went “where no man had gone before”. I wanted to explore the unknown. Though I eventually finished school with an engineering degree and have been able to successfully apply that degree in industry, I have never really gotten away from that desire to explore the unknown. This is what continues to draw me to astronomy. It is what has motivated me to observe lots of objects. It is what motivates me to learn how to do astrophotography. And it is what has motivated me to learn about spectroscopy. As I learn about spectroscopy, I plan on writing about my adventures. This is the first of several articles on the subject. I hope that these articles help others understand a little more about astronomy and perhaps motivate some to follow me in this exciting field that few amateurs explore.
Why learn about spectroscopy? The answer is really quite simple. Much of what is learned about our universe is done with a spectroscope or spectrograph. With a spectroscope, you can determine the composition of a star. You can make an estimate of what elements may be in a planet’s atmosphere. You can use a spectroscope to determine if an object is moving away or toward you. You can even estimate how fast it is moving. Using similar techniques, you can determine how fast the sun is spinning on its axis. You can even determine how fast the rings of Saturn are moving around the planet. There are limitless amount of knowledge that can be gained through the analysis of Spectra.
So, how does one get started taking data in this field of astronomy? Unlike astrophotography, this field has not been explored as much by the amateur astronomer. I figured that a good place to start the research project is to go on-line and see what I can buy. (that is the fun part of astronomy, what new toy can you buy). I quickly went to my favorite telescope stores (Orion, Meade, and Celestron) but was unable to find anything to purchase. Wow, maybe this is a little more specialized. Oceanside Photo and Telescope is a great Internet shop that has just about any astronomical gadget you can think of. And sure enough, under CCD Cameras & Accessories, there is a section called “Adaptive Optics & Spectrographs”. In this section, there are several products. Rainbow Optics has a “Star Spectroscope for Visual, Photographic, and CCD Use”. Baader Planetarium has a similar device called a “Blazed Grating Spectroscope/Spectrograph”. And SBIG has two Spectroscopes. One is for Deep Space Spectroscopy (called the DSS-7) and the other is a “Self Guided Spectrograph”. Wow, which one to buy? It looked like I needed to first understand how a spectrograph/spectroscope is supposed to work before I decided to spend some money. To do this, I again used my old friend the internet. In the process, I found two sources that I have found very helpful.
A while back, Maurice Gavin wrote an article for Sky & Telescope concerning spectroscopy. The article titled “The Revival of Amateur Spectroscopy” can be found at: http://www.skyandtelescope.com/resources/proamcollab/3307266.html. Though this article does a great job introducing the subject, it does not really fill in the details that I needed. For that, I found a book that explains everything I’ve needed to know so far. Amateur Practical Spectroscopy by Stephen Tonkin gives a much more detailed analysis of the subject.
These two references help explain some of the basics of spectroscopy. As many may know, a spectrum of white light separates the light into its color components. It typically results in a rainbow. This separation can be achieved using a multitude of methods. Most of us are very familiar with a prism. A prism is a triangular piece of glass. Light enters on one side of the prism. Due to the index of refraction, the light will bend at a specific (and predictable angle). This angle is different depending on the wavelength of light. As the light exits the prism, you will see the rainbow of colors. This bending of light principle is the same principle behind the chromatic aberrations that many of us see in the lower end achromat refractors.
Another way to create a spectrum is to use a diffraction grating. A diffraction grating is either a reflective surface with many close parallel (in microns small) lines on it. It can also be a transmissive surface with the lines engraved on it. When you shine light onto the diffraction grating, a spectrum is produced by diffraction and interference of light.
For a pure, white light source, the output spectrum will consist of a continuous gradient of color. The longer wavelength of light is in the red spectrum. As the wavelength shortens, you move into the yellow and green portion of the spectrum. Blue is the shortest wavelength we can really make out with our eye.
But for other more traditional light sources, like stars, and lamps, the spectrum contains a series of dark lines in it. These lines are absorption lines. To understand what these absorption lines are, let’s look at the classical Bohr model of atoms. In this model, you can think of an atom as a kind of mini solar system. In the middle where the sun would be are the neutrons and protons (the nucleus). Electrons orbit the nucleus in standard orbits. These orbits vary in distance from the nucleus depending on the energy level of the electron. Electrons can change energy levels by either absorbing or emitting a wavelength of light. In the case of a stellar object, we will see absorption lines specific to various elements in that object. In the case of a nebula, we will see emission lines due to the ionization of the gases and dust in the nebula.
So now that we know a little about how spectra are generated, we can go back to choosing which tool we want to purchase so that we can create our own spectra. The Baader planetarium blazed grating spectrograph consists of a diffraction grating and an eye piece adapter to allow you to look at the resulting spectrum. You put the diffraction grating in front of the eye-piece and then aim the telescope at a star. As you look through the eyepiece, you will see a line of rainbow colors were the star is. You may also note that the rainbow is not evenly illuminated. Parts of it are darker. Those dark areas are some of the absorption lines from that star. Jeff Barton has a Baader Planetarium diffraction grating and has found ways to attach a DLSR camera to it and take photographs. The photograph below shows the spectra of Vega. You could take this photograph and analyze it. There is probably enough information there to at least classify what type of star you are looking at. But the details of very specific absorption lines appear to be missing. To see those, you need a spectrograph with more detail.
This is where things can become more interesting. I want to take pictures of spectra of dim objects. When you take a picture of a star, all of the light is focused on one pixel (ideally). However, when you take the spectra of the star, you will not spread all of that light energy over several pixels. Thus, the results are going to be even dimmer. Even more interesting is the fact that to get high resolution spectra, we have to spread the light over more pixels. Thus the more resolution you want, the dimmer your photograph. So, there is a trap you can fall into in the hunt for more resolution. What use is there to resolve the minute detail of a spectrum if you are unable to see it? So, a happy medium needs to be found.
The SBIG spectrographs are quite interesting. The self guided spectragraph has the most resolution. It, however, also is the least efficient of the two devices SBIG makes. The DSS7 is more efficient but does not allow you to guide easily. Since my aim is to collect spectra of Deep Sky Objects (DSOs), I am going to opt on purchasing the DSS-7. I figure that, hopefully, I can use an external camera to do my guiding. I don’t know if this will be accurate enough to hold a star in the slit, but we will find out. The nice thing about this device is that it is more efficient. And that will be very helpful when it comes to analyzing spectra from galaxies and nebulae. And given that those objects cover some area and are not pin points of light, I should be able to keep my scope centered on the objects for a period long enough to collect my spectra.
In my next article, I will describe my attempts of first light with this device. If you have any questions about anything on this topic, please feel free to contact me either by e-mail or through the TAS forums. Hope to see you all at a dark site sometime soon.