In the second part of my experiment a small flashlight with the infrared filter taken from the camera placed over its lens was placed where the ultraviolet light was for the first part of the experiment. This illuminated the minerals with visible light but also blocked the infrared so if anything was imaged it was not simply a result of infrared reflection off the mineral into the spectroscope.
In order to test whether the spectrum was really infrared, I designed a visible light blocking filter from two Lee stage lighting filters, numbers 106 Primary Red and 181 Congo Blue. The transmission curves for these filters indicate they eliminate all but a small portion of the red end of the visible spectrum, and pass about 80% of the near-infrared. This filter was placed outside the slit when the spectroscope was used to screen minerals for an infrared spectrum.
How specimens were chosen
My original process for choosing minerals was looking for minerals that fluoresced red under ultraviolet light. I thought that because infrared is just beyond visible red on the electromagnetic scale, minerals that fluoresce red might fluoresce in the infrared as well. In addition to a visual search through the minerals I placed each mineral in front of my spectroscope and illuminated them with ultraviolet light. I placed the visible light filter over the slit to block all visible light from getting to the camera. Therefore if the camera imaged anything it was infrared.
I expanded the search to include minerals that fluoresced in other colors, and finally, I checked minerals that did not fluoresce in visible light at all. In the third phase of the search I illuminated minerals with the modified flashlight described above and checked all of them for invisible fluorescence using the visible light filter once again.
Mineral Samples
Calibration
To calibrate the spectroscope scale I first used a helium gas tube, that when excited gave off light emission lines. By placing the illuminated gas tube in front of my spectroscope I obtained a spectrum of the helium gas. However, I found that every time I took the spectrum of the helium gas there were more lines present than what was published. (I read in the gas tube instruction sheet that this was due to mercury contamination.) I was unable to distinguish which lines belonged to which spectrum. It is also unlikely that the amateur mineralogist has a helium gas tube on hand.
So I used a red laser (wavelength of 633nm) and a green laser (wavelength of 543nm) to calibrate my scale. However, by the same reasoning, it is unlikely that an amateur mineralogist has a red and green laser on hand. So I found and used a fluorescent light tube with known emission lines in its spectra due to mercury. These lines are prominent and well-defined: red = 612 nm, green = 546 nm and violet = 436 nm. I placed this light in front of the slit of my spectroscope to obtain its spectrum. The green absorption line was lined up with the eleven on my scale. Then by using simple ratios I confirmed that the other known lines in the spectrum were where the ratios predicted them to be on my scale.
To calibrate my graphing scale I used ratios obtained from the spectrum of the fluorescent light. From these ratios I found that there were 1035 nanometers in 80 millimeters, the entire length of my scale. By using the Set Scale command in Scion Image I found that there were 472 pixels in 80 millimeters. Therefore, 1035 nanometers covers a distance of 472 pixels. This ratio reduced to 2.19 nanometers for every pixel. Since every point on a graph corresponds to one pixel I added 2.19 recursively to my lower visible bound, 400 nanometers. The scale for the entire visible and infrared spectra was from 400 nm to 1000 nm. The visible red cut off was at 700 nm.
Taking and processing the spectra:
To obtain the spectra of the minerals I set up my spectroscope and positioned the camera so that it could see the slit and the scale. I then calibrated it using a fluorescent light. After the spectroscope was calibrated I proceeded to place minerals, one at a time, in front of the slit and under the ultraviolet light. While doing so, I wore special UV blocking safety goggles to protect against the harmful effects of short wave ultraviolet.
If infrared spectra were present, I took three photographs of each mineral’s spectrum in total darkness, first under short wave light (254 nm) and then under long wave UV (360 nm). The webcam software settings were adjusted for high resolution and low light levels.
Although the minerals were screened by using a visible light filter, this filter was removed when the spectra were imaged. Repeated testing both ways showed that when the infrared spectra were taken without their visible portion, they were extremely dim and difficult to extract from the random noise of the camera’s background levels. This may be due to several factors. First, the cutoff filter removed some of the infrared light as well as visible. Second, when the camera ”auto-adjusts” to gain the maximum contrast for a dim image, the background noise levels become so high they mask the faint infrared spectra.
I then opened each photograph in Adobe Photoshop and converted the color .jpg spectral images into grayscale images. I also inverted them and saved them as .tifs. I then opened each image in Scion Image. I used the Plot Profile command in this program to give me numerical data for the image. I then opened each set of numbers and combined them all into one spreadsheet in Microsoft Excel. Here I averaged the three images of each mineral’s spectrum. I then took the averages of each mineral and put them into one spreadsheet. I opened this spreadsheet in PSIPlot. This graphing program allowed me to smooth out the continuum curve of each averaged mineral by using the Smooth RMS command. After doing this I was able to graph each mineral’s spectra. I then analyzed the graphs of the minerals to find any similarities in their continuum curves by creating combined graphs.
Controls and Error
There are several sources of error in this experiment. One error could occur if the wavelength values of the spectra were measured incorrectly. This was avoided by using reference values for the wavelengths of the emission lines in the fluorescent light. Since the camera, the slit and the scale were not moved while I was collecting data, I used the slit for a scale reference point for the conversion of the images to graphs. The range of error in this was +/- 1 millimeter. According to my scale 1 millimeter is equal to 12.94 nanometers, so the approximated range of error is +/- 13 nm.
A second possible error to consider is second order spectra. Second order spectra, as with first order spectra, are caused by the constructive interference of light waves. The diffraction grating acts as a wall with many windows in it. Each window acts as an individual source of light, which emits its own light waves. These waves travel out through space in a circular pattern. Where the waves of each light source intersect to cause constructive interference is where the spectrum of the original light can be seen. A filter that blocked all visible light, but let infrared pass through was placed in front of the slit. To find where second order spectra occurred I placed a flashlight in front of the slit and took a photograph of its spectrum with and without the color filter over the slit. It was determined that in this set-up, the second order visible light did not overlap into the first order infrared light. To further guard against second order UV contamination I taped an ultraviolet blocking filter over the slit that came from an extra pair of the safety goggles I wore to protect my eyes.
Long wave and short wave spectra of the ultraviolet light were imaged to determine if any infrared radiation was emitted by the lamp. The spectra were processed, graphed and smoothed. Even if the UV lamp is shone directly into the camera, the amount of infrared in short wave is minimal, about equal to the noise levels generated by the camera. But there is a small amount of infrared distributed in a band emitted by the long wave lamp. The band extends from 700-770 nm. This spectrum was recorded to check against any possible contamination by reflection into the spectroscope. Fortunately, most of the spectra were made using the short wave lamp.
Six “dark frames” were taken of the totally dark shoebox to determine the pattern and level of noise in the camera. These were averaged, graphed, smoothed and characterized so they could be compared against the spectra in order to determine if they were truly infrared or just an artifact of noise. The camera noise typically varies by an uneven undulation of 25 nm and its intensity averages +/- 2 of the maximum natural range of the y axis on the graphs.
The experimental set-up provided for a constant level of illumination by using controlled distances and an AC powered UV light. That way, battery life could not affect the brightness of the UV lamp.
Every sample was imaged three times and the spectra were averaged. The averaged spectra were further smoothed using a “root mean squared” technique before analysis.
The minerals used in this study came from my school’s collection. Approximately 350 samples were screened for infrared spectra under the ultraviolet light (both wavelengths). They represented a wide variety: silicates, carbonates, sulfates, ores and halides. Of those, 119 showed some trace of infrared spectra. This group was narrowed down to 35 specimens that showed an unmistakable infrared spectrum.
For part two of the investigation, approximately 400 samples were checked under the “visible minus IR” light source. This included the above types of minerals plus native elements and sulphides.
I researched background information about the samples and compiled it into a spreadsheet. This included such information as its chemical composition, place collected, any unusual or extra information known about the sample, and possible activators.
While taking the spectra, I recorded the fluorescent color and UV wavelength.
The contents of the Data Book are as follows:
Section 1: The spreadsheets with the mineral information.
A.
Averages- contains the numerical data for all images of the minerals and their averages.
B.
Rare Earths- contains the numerical data for all minerals with rare earth elements as suspected fluorescence activators.
C.
Scapolites and Franklin- contains the numerical data for all minerals in the scapolite category and the two Franklin Calcites.
D.
Pegmatites- contains the numerical data for all minerals that form in pegmatites.
E.
Calcite- contains the numerical data for all minerals in the calcite category.
F.
Feldspars- contains the numerical data for all minerals in the feldspar category.
Section 2: Mineral spectra graphs
A.
Controls- the graph of the ultraviolet light, long wave and short wave. This also contains the noise profile (dark field) for the spectroscope and the graph of the fluorescent light used for calibration.
B.
Curvesheet.PDW- the graph of the mineral’s complete visible and infrared spectrum. The y-axis is labeled with the mineral number. The x-axis is measured in Nanometers. The graph’s scale ranges from 400 nm to 1100 nm.
C.
Ircurvesheet.PDW- the graph of the infra red portion of the mineral’s spectrum. The y-axis is labeled with the mineral number. The x-axis is measured in Nanometers. The graph’s scale ranges from 700 nm to 1100 nm.
D.
Spreadsheet- Mineral Composite Graphs: lists the numbers of the minerals on each graph. Includes the rationale for choosing that comparison, and the result of the comparison
Spreadsheet- Mineral Groupings and Suspected Activators: lists the mineral number, color of fluorescence, suspected activators, and the group; pegmatite, feldspar, carbonate, or rare earth.
E.
Mineral Composite Graphs- composite graphs of minerals with similar chemistries, locations, environments, colors and wavelengths.
Out of the 35 minerals that showed a definite infrared signature there were two main groups, silicates and carbonates. The silicates consisted of feldspar, muscovite, quartz, spodumene, and beryl. The carbonates consisted of calcite, tufa, and scapolite. With few exceptions, the infrared fluorescence was stimulated by short-wave ultraviolet radiation.
The two main groups of mineral samples were further divided into categories for analysis such as growth environment, chemical composition, and suspected activators.
Growth environment is where and under what conditions the mineral was formed. Pegmatites were the most common growth environment of the samples. This is when igneous rock cools slowly, causing large crystals to form. Activators are impurities or substitutions in the mineral that cause deformities in the crystal structure. These deformities cause the mineral to fluoresce.
Composite graphs were made of all pegmatites, feldspars, minerals with rare Earth elements as suspected activators in them, calcites and all scapolites. These graphs were too jumbled to analyze so I made composite graphs of pairs of minerals with similar chemistries and locations.
The calcites that fluoresced orange had manganese as a suspected activator and showed similar graphs. Both samples of Franklin calcite had very similar spectra, with sharp small peaks that may be indicative of a rare Earth element presence. Out of 8 calcites that fluoresced under short wave ultraviolet only two graphs did not have similarities with the others. There were no similarities in the graphs of both calcites that fluoresced under long wave ultraviolet. However the two short wave calcites that did not match the other 6 matched each other. The graphs of both samples of tufa (another form of calcite) were exact matches under long and short wave. There were spikes at 800, 825, 920 nm on the graphs of the tufas. This may be due to a rare Earth element activator.
There were four types of feldspar that fluoresced: microcline, orthoclase, albite and Cleavelandite. All three albites (including the two Cleavelandite varieties) showed similarities in their graphs. There was a pattern in the four microclines, but it is not as obvious. There is a shift in color in two of these, so that the graphs appear to be stretched out relative to the other two. This may possibly be due to manganese substitution for the iron activator. Manganese and iron both fluoresce red, but iron has the deeper red of the two. There was no apparent pattern in the nine spectra of the orthoclase group, with the exception of two minerals that match closely.
The difference in the orthoclase may be due to the different levels of sodium in the minerals. Unlike calcite, which has a fairly regular composition, the feldspar group has a lot of ionic substitutions. Orthoclase, for example, can contain up to 10% sodium as a replacement for some of the potassium. The change in atoms causes changes in the crystal and the chemistry of the mineral.
The graphs of the radioactive granite and the quartz showed striking similarities. Both of these minerals are suspected to have rare Earth elements in them, and they also show some small sharp peaks.
All four scapolites showed similar spectra, both the samples of (common) scapolite and wernerite. This may be due to their common chemistries. Within each sub group the similarities are much higher. This is probably due to the fact that they have different activators. Scapolites are a type of carbonate, so one was compared to a calcite. It was a very good match. The carbonates appear to be very consistent in their spectra.
Although most of the pegmatite grou p are feldspars, there are other minerals in this group that were compared to see if the same environment of formation would be responsible for similar spectra. They are: beryl, scapolite, muscovite and spodumene. These were all compared to each other and to the feldspars.
Beryl and scapolite matched well. This is interesting because both of these minerals occur in pegmatites and in marble deposits. Microcline and beryl had some similarities, but the match was not as clear. However beryl and spodumene matched well. There is no suggested activator for either of these, but an interesting observation is that they both fluoresce yellow. When compared to wernerite, which also fluoresces yellow, the match is not as clear between either beryl or spodumene.
There were no similarities between the graphs of the scapolites, spodumene, and feldspar.
The particular piece of beryl in this study is unusual because it fluoreseces yellow. This narrows its likely origin to New York state or at least the eastern U.S. The fact that its spectrum matches with scapolite links it to the Grenville marble formation of Canada and also to the famous Franklin marble deposits. So it was compared with a piece of Franklin calcite. It was a very good match, including a small sharp spike at 777 nm, which may indicate a rare earth activator. The Franklin Calcite was also compared with tufa, which also contains small spikes. The spike at 777 nm appeared in this sample as well.
The muscovite was the only mineral that did not fluoresce in the visible but did fluoresce in the infrared. Its spectrum was compared to the spectrum of the ultraviolet short-wave to rule out contamination. No contamination was detected.
This experiment has shown that it is possible to detect infrared spectra of minerals using inexpensive equipment. It has also shown that the results are consistent. This experiment has also shown that some minerals do fluoresce in the infrared and not in the visible. There are two main groups that show fluorescence in the near infrared, carbonates and silicates. Of the carbonates calcite was the most abundant. Of the silicates feldspar was the most abundant. In some cases, such as calcite, there is a consistent pattern. In other cases, such as with the feldspar, there is not a consistent pattern. This may be because of the extreme variations in the chemistry and crystal structure of the feldspars.
One mineral, muscovite, was found that fluoresced in the infrared and not visibly. This proves that fluorescence in the infrared and not in the visible is possible and that it can be detected with this type of equipment.
This experiment attempted to obtain infrared fluorescence with visible light stimulation. Although no such spectra were observed, it may be possible to record them with more sensitive equipment.
Extensions
There were minerals that showed a small sharp spike somewhere in its spectrum. These small spikes may be signatures of rare earth elements. I tried to look up and find a match for theses lines in the Handbook of Chemistry, but was unable to find any. This could be due to the fact that the Handbook of Chemistry only lists the lines of a gas not of transitions in a solid state. A student could investigate if and how these lines shift when moving from a gas to a solid.
·
Bancroft Area Minerals. Gibson, Shirley. Ministry of Natural Resources. Bancroft, Ontario. 1981
·
Electromagnetic Spectrum. imagers.gsfc.nasa.gov/ems/infrared.html
·
Fluorescence. Robbins, Manuel. Geoscience Press, Inc. Phoenix, 1994.
·
Handbook of Chemistry and Physics, 1st Student Edition. Weast, Robert C., Ed. CRC Press, Boca Raton FL, 1988.
·
Mineralogy. Berry, L.G. and Mason, Brian. W.H. Freeman and Company, San Francisco, 1959.
·
New, Mid and Far. www.ipac.caltech.edu/outreach/edu/regions/irregions.html
·
The Story of Fluorescence. Wain, Harry C. Raytech, Middlefield CT, 1965.
·
www.uvminerals.org. Fluorescent Minerals Society.
