Learning Center What is a mineral?The most common minerals on earth Information for Educators Mindat Articles The Elements The Rock H. Currier Digital Library Geologic Time

Minerals by Properties Minerals by Chemistry Advanced Locality Search Random Mineral Random Locality Search by minID Localities Near Me Search Articles Search Glossary More Search Options

The Mindat Manual Add a New Photo Rate Photos Locality Edit Report Coordinate Completion Report Add Glossary Item

Mining Companies Statistics Users Mineral Museums Clubs & Organizations Mineral Shows & Events The Mindat Directory Device Settings The Mineral Quiz

Photo Search Photo Galleries Search by Color New Photos Today New Photos Yesterday Members' Photo Galleries Past Photo of the Day Gallery Photography

Discussions

💬 Home 🔎 Search 📅 Latest

╳Discussions

💬 Home 🔎 Search 📅 Latest

Groups

Recent Images in Discussions

Techniques for CollectorsAmateur Raman Spectrometer

17th May 2015 00:09 UTCHenry Barwood

Realized this last week that I have most of the parts needed to build a Raman spectrometer. Ordered a 532nm laser diode and a longpass filter from eBay and already have the spectrometer, beamsplitter and objectives along with odd mechanical parts. Has anyone else put one together? If so, I would love to hear from you on technical details and software compatibility issues.

17th May 2015 01:10 UTCKen Doxsee

You never cease to amaze me, Henry! I'm afraid I have no expertise to offer, but just wanted to express my appreciation for your resourcefulness. Please keep us posted as you tackle this new and fascinating project! ---Ken

17th May 2015 07:39 UTCFrank de Wit Manager

Yep. Stefan did!

See http://www.uranglasuren.com/ores/Raman/index.html

17th May 2015 13:28 UTCHenry Barwood

Frank,

Thank you. Very informative site!

17th May 2015 14:05 UTCSpencer Ivan Mather

Hi |Henry, I just wish that I had one, how much would it cost me if you made one for me..?

Spencer..!!!

17th May 2015 15:16 UTCHenry Barwood

Hi Spencer,

All my stuff is made with used/spare parts, so there is no way to tell what it would cost with all new parts. Commercial outlets have units that start around $7,000. We were told a decade ago that they would only be a few hundred dollars by now! Still waiting for that one. If I'm successful, I'll see about a step by step article.

17th May 2015 16:54 UTCRonnie Van Dommelen 🌟 Manager

Henry,

I dreamed of making one too for a few years now. At work we have a green HeNe and some very good spectrometers. The killer for me (besides lack of time) has always been the filter - they are still about $700 for a good one. I recently found a really good website a while back but would have to search again to find it.

I seem to remember reading that green laser diodes are poor sources for this application. I don't recall if the problem was stability or sidemodes, the latter I think. You could probably get it to work, but the results would not be as nice as with other sources.

Good luck! Let us know how it goes.

17th May 2015 17:09 UTCHenry Barwood

Hi Ronnie,

The green is a trade off. The longer wavelengths have less fluorescence, but are more expensive to implement. UV below about 250nm is also useful, but the lasers are very expensive. Notch filters are quite expensive, but you can find long bandpass filters on eBay for under $50 that will work. The rest is just configuring the optical train and getting the data out of your spectrometer.

18th May 2015 17:22 UTCEugene & Sharon Cisneros Expert

Henry,

I am curious as to what spectrometer you are going to use and what its optical resolution is.

Best,

Gene Cisneros

18th May 2015 17:42 UTCHenry Barwood

Gene,

I have two USB 2000 units configured with different resolutions, and old board mount 2000, and a USB 4000. Resolutions range from around 5 nm to down to about .3 nm. Resolution is really not the problem. Software to convert the output to Raman shift is!

17th Jun 2015 00:41 UTCHenry Barwood

Short follow-up to my previous posts.

I've just about secured all the necessary components to start the assembly. Discussions with fellow collectors at the NCMA meeting revealed to me that the 532 and 690nm diodes are indeed frequency doubled and often have strong IR components in them. I've added a spectral clean-up filter to removed the IR in case it is there (some are blocked, some are not). Secured inexpensive long pass filters that should work. Won't know until I test them.

My HR4000 unit is basically the same as the Ocean Optics ES65000 that they use on their Raman units, less thermoelectric cooling. I will have more noise, but it should be OK for my purposes (the 65000 is around $13K). I have an external TEC that I may add to the spectrometer if the noise proves to be a problem.

I've found that my SpectraSuite software is set up for Raman, and I found a beta version of another spectroscopic processing software that will also function, so I should be OK on the software.

I'm designing it around a laser collimator for telescopes and will have the ability to change from 405, 532 and 690 wavelengths with simple change of diodes and filters. I'm also going to add a slide out shutter and a neutral density filter to allow focusing of the laser beam on a specific mineral grain without burning out my eyes!

Hope to have it prototyped within 2-3 weeks. Stay tuned.

17th Jun 2015 21:04 UTCEugene & Sharon Cisneros Expert

Hi Henry,

It was a pleasure indeed to meet and chat with you regarding Raman, at the NCMA. Your presentations were well appreciated by all.

I am glad that you have found a solution to the software issue, as you seemed to be the most concerned about that. I still remain somewhat concerned about the spectral range of the spectrometer, but as we agreed, the price was right.

You mention TEC, but you may be able to get reasonable S/N by averaging several short exposure data sets. This is commonly done in astrophotography with good results.

I look forward to hearing about the results of your first tests.

Best wishes!

Gene Cisneros

17th Jun 2015 21:43 UTCHenry Barwood

Hi Gene,

The OO spectrometers have proven to be quite sensitive for low light levels. Of course, I won't know until I'm actually able to bench test the thing!

SpectraSuite has an averaging subroutine for noise reduction built in, so I can utilize that if I get poor spectra. Again, only time and testing will tell.

I really enjoyed the NCMA meeting, but have been so busy since I returned that I haven't had a change to image much of what I brought back.

Took an emission spectrum and ran it as if it were a Raman spectrum (arbitrarily set the laser as 795nm). Got a laughingly distorted spectra, but it did convert it to Raman shift. Will have to adjust the X-axis for a real Raman spectrum.

11th Jul 2015 17:01 UTCHenry Barwood

Well, I constructed two Raman heads for my microscope. The first one V 1.0 had optical issues and I was unable to get a signal, so I constructed a second head with a different optical path (V 1.1) and obtained a very weak spectrum. I played with the software and adjusted the collimator, but was never able to get a S/N ratio that would allow me to use it to identify minerals. I came close, but no cigar!

13th Jul 2015 22:23 UTCEugene & Sharon Cisneros Expert

Hi Henry,

I am sorry to hear that your tests did not result in more favorable results. This is tricky business and without proper test equipment, it's next to impossible to know where the problems lie. Assuming that your filters are OK, then I would suspect that the laser line is off. Unless you have a stabilized laser, with the line specified and guaranteed, this could be a show stopper.

Gene Cisneros

14th Jul 2015 00:17 UTCHenry Barwood

Gene,

I have analyzed the design and cannot determine the problem (yet). I have ordered some additional components and may give it one more shot after my teaching duties die down a bit. The filters are good, but the degree of collimation appears to be off I haven't completely given up, but definitely had my nose blunted! My spectrometers are actually more sensitive than the ones used in most commercial Raman units, but the optical path appears to be the problem. I was able to get a signal, but very noisy

14th Jul 2015 06:33 UTCMac Kester

I have been working on a raman spectrometer of my own (not working yet, nor finished) design. I am using the common red laser diode in 650 nm, as it can be cheaply cleaned up with a surplus band pass filter. The wavelength was chosen to limit fluorescence, and for the availability of optics. I used a long edge filter, which has the drawback of removing the anti-stokes raman signature. I would love to collect it somehow though, without building an second full optics train.

My system train looks like this:

A 10mw 650 nm red laser is passed through a band filter, polarizer and a collimator lens. This is then fed through the left objective of my microscope, a B&L stereozoom 4, and through a rigidly mounted 25x microscope objective. The other objective has a 80% reduction filter on it, and is simply used for aiming the beam. The working end of the microscope is sealed, and has a white led, IR led - 940nm, and neon bulb in the unit. The white led is used for visually seeing the sample, while the neon and led are used to calibrate the system. The inside of the box is covered with black velvet, which telescope makers tell me is best for absorbing stray light, and as much hardware as I could I coated with it, or black matte enamel, to try and reduce noise. The signal is collected by a 40mm diameter double convex fused quartz lens, coated with Edmund NIR I, which is optimized for 600 - 1050nm. Focal length is 100mm, and at 1/2 the focal length I have the main collimator and the edge filter. The signal then passes to a reflective diffraction grating, which diffracts the signal onto a second fused quartz collection lens, which focuses the signal onto a velvet covered plate, with a slit cut in it. The signal is allowed to expand slightly, and then falls onto a liner photodiode array (1536x1 elements), which is sensitive to 1100 nm. I am planning to attach a peltier unit to the photodide to help cut down on noise with the long sampling times needed. Planning on connecting the photodiode to an atmega system, like an ardrino to process the signal. Output will be text format containing the total signal from each diode, which can then be calibrated to wavelength using neon or an other known signal.

I use the 90 degree style of initial signal collection, due to its simpler construction. I should be able to increase my signal with a mirror 180 degrees from the collection lens. The sample is manipulated with an XY stage.

I have thought about various ways to reduce noise, and I think that having a means to subtract it might be interesting. Two or three extra photodiode arrays in absolute darkness, with their noise signal collected and subtracted from the raman signal maybe? In that case, noise should be a random quantum process, but maybe a pattern will emerge during long sampling times, IE, some artifact from manufacturing. or just focus the signal onto half the array, and use the other half as a control. Just thinking out loud.

I have heard that astronomers use modified webcams with the CCD exposed for long exposure photography and spectroscopy. That might be a line to look down for low cost detector. Astronomers have spectroscopy software available for them as well, simplifying the process.

I would appreciate any feedback on my system! I encourage anyone interested to try and work on this. Its a fun project, and not too hard on the pocketbook, all things considered. I have around $400 into mine, mostly in optics, not counting the hardware, laser or microscope stuff, most of which I already owned, or made myself. Most expensive single part (so far) was the holographic reflective diffraction grating, which was around $100.

Have a good day all,

Mac

14th Jul 2015 13:30 UTCHenry Barwood

Hi Mac,

Your set-up sounds good to me. I use Ocean Optics spectrometers and a 30 mW 532 nm laser diode. I clean up the laser with a narrow bandpass fitter and use a long pass filter for the Raman output. Both filters were surplus from eBay, but they are good filters. I built two heads for my scope, but as I've commented, my optical path was not correctly collimated and the Raman signal was very noisy. I'm working on modifying a vertical illuminator that is designed for my Leitz Pol scope to accept the laser and filters and will build a fiber optic receiver for the eyepiece. This should allow me to use it as a Raman microscope, but I won't know for a few weeks when all the parts get here (I bought an inexpensive vertical illuminator so I wouldn't have to disassemble my research RLM).

Let me know how your unit turns out.

14th Jul 2015 19:32 UTCRonald J. Pellar Expert

Using a control array to try to subtract noise is not effective. Noise is a random process and does not subtract. Whether you subtract or add noise the noise will increase as the square root of the number of additions or subtractions. Control arrays are good for elimination of extraneous signals, stray light, etc., i.e., non-random unwanted signals.

To improve S/N keep averaging and renormalizing multiple measurements. The signal will improve directly with the number of measurements and noise will only increase at the rate of the square root of the number of measurements. Renormalization just scales the signal to a constant maximum with the noise appearing to reduce as the number of measurements increase.

14th Jul 2015 21:29 UTCHenry Barwood

I'm afraid the only way to lower the noise in CCD or CMOS devices is to cool them. Astronomers modify DSLR's all the time to add cooling. Everything from removing the chip and attaching a Peltier cooler to encasing the whole camera in a refrigerated box. If you have a linear array, it should not be too difficult to add a cooler to your set-up

15th Jul 2015 00:34 UTCThomas Lühr Expert

This may be a bit off-topic and I don't know if it's helpful for this certain problem, but some times ago I used (successful) the from Ronald suggested Method for noise reduction of pictures taken by a webcam-based USB-Microscope-Camera.

Therefore I made a number (say 16) of identical images of the same object. This is easily doable with the program "cheese" (runs under Linux based systems). Afterwards I opened all pictures as a stack of layers with GIMP and gave each layer a opacity of 100%*(1/n), where is n the number of the layer and the most lower layer has the number 1. So contributes each layer the same amount to the overall picture, which can generated with the merge-command. A self written plugin makes that work easier.

The result is a (in this case 4 times) better S/R- ratio.

But:If you want to increase the quality in the same level a second time, you need already 256(!) photos/measurements.

Not sure if this helps

Thomas

16th Jul 2015 18:49 UTCRonald J. Pellar Expert

Averaging ten samples will give an increase in S/N of ~3. A hundred samples will improve the S/N to 10 times, i.e., the S/N improves as the square root of the number of samples. One of the commercial Raman spectrometers that I have seen operated does multiple sampling to improve the results.

16th Jul 2015 19:50 UTCStefan Oertel

Hi Henry,

Don't waste time on improving SNR or denoising. Take a strong raman scatterer like sulphur or crocoite and start optimizing the optical path. Make sure to use a high aperture objective to catch as much light as possible and check how much light is passing your filters. I have trouble to trust them having bad experience with ebay filters. How do you couple the light beam into the fiber? Maybe make a drawing of your setup, get some optic guy to run a simulation on it, then start improving. Image improvement is the very last step and overrated. No image -> nothing to improve.....

Cheers,

stefan

16th Jul 2015 19:56 UTCStefan Oertel

Are you sure the laser is really focussed and focussed on the spot where you collect the light? Bournonite easily get burn marks from raman lasers and is good for testing of focus and location of the laser spot without burning your eyes. If you don't get a burned spot after a few seconds, something is wrong with your laser adjustment. My system is very sensitive to that. Use a polished bournonite section for this.

Hope that might help somewhat.

Cheers,

stefan

16th Jul 2015 20:02 UTCHenry Barwood

Hi Stefan,

Thanks for the good advice. One of my problems was determining what would make a test standard for checking the spectrometer. Sulfur and crocoite I have!

I am waiting on a part I ordered and hope to set up a new spectrometer configuration in a week or so.

28th Jul 2015 04:18 UTCHenry Barwood

Follow up:

I constructed V 3.0 and was unable to get a useful spectrum from it. Apparently I just don't have enough spectrometer to make a go of it. I got very weak signals. Scrapped the device and replaced the 532nm laser with a 405nm laser and turned it into a laser spectrometer for my microscope. Here is a spectrum from a grain of sodalite about 20 microns across:

The microscope objective was focused to a point about 10 microns across in the center of the grain. I adjusted the device so that the microscope eyepiece crosshair is centered on the focal spot. This will allow me to take spectra from small mounted microminerals and grains in thin sections and polished mounts.

I also discovered that a small piece of an amber pill bottle works as a primary line suppressor for the laser without blocking everything below around 450nm. Blocks about 80% of the primary 405nm line and second order line.

In spite of my disappointment at not being able to construct a Raman Spectrometer with what I had on hand, the fluorescence device will be quite useful

Mindat.org is an outreach project of the Hudson Institute of Mineralogy, a 501(c)(3) not-for-profit organization.
Copyright © mindat.org and the Hudson Institute of Mineralogy 1993-2024, except where stated. Most political location boundaries are © OpenStreetMap contributors. Mindat.org relies on the contributions of thousands of members and supporters. Founded in 2000 by Jolyon Ralph.
Privacy Policy - Terms & Conditions - Contact Us / DMCA issues - Report a bug/vulnerability Current server date and time: April 25, 2024 13:37:15

Go to top of page