Laser-Diode based Raman Spectroscopy

Ok, we've all seen the number of articles popping up on the internet on the use of either green or violet laser diodes for Raman Spectroscopy, some of which are extremely interesting in that they use laser pointers (OTC & cheap as piss) and one interesting one where they have methodically replaced the dearest components of a working Raman Spectrometer with cheap, OTC components...

They used a laser pointer with a notch filter, coupled to optical fibre cable, which is then used to take the laser light to the sample, a second optical fibre cable, with a notch filter included, is then used to take the Stokes/Anti-Stokes radiation to the spectrometer, where it is analyzed and converted for graphing purposes.

Now, the spectrometer is the expensive bit in the equation - whereas it really shouldn't have to be, it should be possible to use a CCD, which converts images, by reference to their specific wavelength & intensity, into electronic signals which can then be decoded by a number of devices...

What I am suggesting is that if we could find out how to harness the ACTUAL raw data from a CCD chip, we wouldn't need much more than that (well, apart from a lens & grating), to get ALL the information we need for a Raman Spectrum. We then utilize a CCD chip directly at the vision port of the spectrometer.

The programming side of it should interest someone, after all, it would use the same numbers/letters as are assigned in jpeg/png type files, only this time instead of reconstructing the entire picture, we'd only want to count the precise number of pixels of each specific color, and assign a wavelength to that color. That way we can map wavelength v intensity, giving a Raman spectra

PS Here are two J.Chem.Ed articles (uploaded by Polverone in another thread):

Wakabayashi & Hamada, A DVD Spectroscope: A Simple, High-Resolution Classroom Spectroscope J.Chem.Ed 83(1) 2006 pp.56-58

Wakabayashi, Resolving Spectral Lines with a Periscope-Type DVD Spectroscope, J.Chem.Ed 85(6) 2008 pp.849-853

And here is the site the author's of the articles just cited, mentioned. Of particular note is their description of the spectral emission of laser diodes as being effectively monochromatic, that is another area where the big savings lie, both in time & money.

Although, looking at the second article again, they aren't just using it as a spectroscope, but as a veritable UV-Vis Absorption Spectrometer... Interesting.

So we could have UV & Raman capabilities with fuck all outlay, can anyone access the supporting information for both articles? I'd dearly love to see the pattern for the latter version of the spectroscope

Because what I am thinking of, is if the software to digitize the image from a digital camera is truly opensource, then it would digitize it regardless of whether it was from a UV-Vis adsorption spectrum or a Raman spectrum - so 1 camera, one DVD and one cuvette, with two light sources, could effectively be used to generate both the UV-Vis adsorption spectra and the Raman spectra of whatever is in the cuvette. That should eliminate a LOT of guesswork I'm betting

PS Anyone here with programming experience? There are actually open-source image manipulation tools online and one or more of these could be modified to fit with this concept, which would be an absolute boon to amateur scientists and underfunded educational institutions to boot.

PPS Here is a good step by step - they use a grating, but the idea is the same - extremely simple - just imagine a Y junction with the grating in the middle - with two light sources at the other ends (etc. one a laser, one a mercury vapor lamp), turn one light source on, record the spectrum, turn it off, turn the other on, record the spectrum and we're done...

Edit 30-odd IIRC

Now, I have sussed it out a little - what would probably be the easiest solution would be to convert the image to a greyscale image (it's a spectra still, so provided it is properly aligned, the greyscale will accurately represent the original spectra), then make a histogram - X=Shade & Y=Intensity (or the number of units of that shade).

Now Adobe Photoshop has this ability, but it shouldn't be the case that the image conversion software costs more than the fucking hardware. I will now go and scout around for some freeware/opensource applications that can translate a greyscale image to a histogram, once that is to hand, then building a spectrometer would be the next step.

[Edited on 29-1-2010 by unome]

The best option I can see from a quick perusal of the internet information, is the RAW Format - ie. allows for the receipt of unmodified RAW data from the CCD (nb open source) and collating that data so as to build a histogram directly therefrom. The RAW data format group have already got a shitload of information & more importantly code which should be able to be made to work with this concept, after all, we aren't doing any image modification, simply polling the data to build the histogram.

The only bit that really worries me is the calibration of the spectra - which wavelengths = what part of the histogram... Of course in a perfect world, the spectra would be in order - based upon the concept of planning for the worst & hoping for the best, it may be necessary to program in the various swaps/changes to positions in order that the histogram correlate to the required spectral analysis.

[Edited on 29-1-2010 by unome]

From what I can see, the hardware, particularly for the UV-adsorption version, is pretty simple.... The hardware for the Raman variant is going to be a little harder, optical fibre isn't exactly given away and the bandwidth & notch filters are going to be sticking points. The diodes, blue diodes particularly, have improved out of sight since the blueray thing took off, the prices have also dropped too.

I'd really like to try and build something that could be attached to a USB port, take its power from the USB port, and send its information back through the same port.

Now, what would be nice is if anyone has the necessary know-how to design or explain the design of the bandpass filter and the notch filter - seems to me they are nothing but colored glass/plastic and not worth the $3-400 price tags, especially given we'd be using laser diodes, which by definition should be near enough to monochromatic anyway.

[EDIT]

This article (attached) is very enlightening, it suggests (to me at least) that the only filters one would need would be a laser-line filter (lower cost than bandpass) and a cut-on filter that corresponds closely to the laser line filters wavelength. That would make for a narrow bandwidth laser, which lies just under the steep cut line for the cut-on filter, which would allow us to avoid using notch filters (which cost a fucking fortune) and just use relatively cheap cut-on filters.

Of course, with the sort of outlays that would be required, it would make sense to invest in a cheap holographic grating to match, but yeah, should be able to keep the prices reasonable

But, insofar as the UV-Vis absorption spectrometer, I cannot imagine that many people on this forum cannot access a DVD, some cardboard, some aluminium sheeting, a halogen light and a digital camera...

The graphics manipulation might be somewhat daunting (although there are freeware programs that should be able to handle it, not just photoshop), but if it were made easier, then it would be an EXCEPTIONAL development for amateur chemistry. Everyone who wanted one, could build one and have it to hand in their lab/garage/etc. Sure would beat the hell out of TLC for measuring the progress of a reaction.

Attachment: Erdogan.Mizrahi.Thin.Film.Filters.Raman.Spectroscopy.pdf (159kB)
This file has been downloaded 1660 times

I just found this image (screencapture) at this site

Looking at the screencapture, it would appear that they have simply taken a cross section of the ccd generated picture, aligned it with the known spectra and then worked out the brightness via the histogram/graph... Sounds simple when said quickly, anyone here any good at programming?

I think if it were possible to generate such graphs from the simple adsorption (UV-Vis) spectrometer (using the DVD as the grating), then and only then, would it be worthwhile spending the couple of hundred bucks needed to try for the Raman




I just found another article with some bearing on the problem, I haven't read it all yet, but maybe some here could make use of it

[Edited on 30-1-2010 by unome]

Attachment: Petrov.FilterBased.Program.Freeware.pdf (328kB)
This file has been downloaded 1601 times

Yeah, I just made a wooden version of the one from the first article - be aware you have to trim the bloody DVD to make it fit (there is NO way a full DVD will fit in the box specified - half a DVD fits well), but there is visible spetra - which are obviously & visibly different under normal lightbulb/halogen/sunlight.*

I think it might be time to look around for some cheap holographic gratings to try and get to grips with this...

PS I was thinking instead of using a mirror/periscope arrangement in the second article, what about carefully removing the shiny side from a dvd, the lines are still there...?

PPS No, I meant as in learning to do a repetitive set of manual tasks, using various different image processing tools (that various members may or may not have), each one requiring the steps to be (a) undertaken in a different order; or (b) done altogether differently.

* Pics to come - what is cool is that on top of the visible spectra through the viewing window - is the reflected spectra which is thrown onto the counter/benchtop in front of the unit. I suspect that that would be intercepted by the CCD and provide a nice picture - just have to play around with the angles.

[Edited on 2-2-2010 by unome]

As noted, the woman (I assume her name is Mary) from maryspectrogra.org, had serious problems with the initial setup of the Raman Spectograph because of the off-wavelength excitation of the sample, due to the fact that she used a laser-pointer, which although monochromatic enough for most purposes, was a LONG way from monochromatic enough for this one (where the Rayleigh scattered light is the main thing that has to be excluded from the spectra). Thus, above where I suggested she would probably have been better off getting a laser-line filter - thereby narrowing the possible wavelength (with a (+/-)2nm filter, she would have had 530-534nM laser light) to an acceptable level, especially given that the off-the-shelf edge-filters are either going to cut-off/cut-off steeply @ 500/550nM respectively. Thus the Rayleigh scattered light from the narrowed laser would NOT get through to the spectroscope (which is probably why so many of the companies advertising 'cheap' (used relatively speaking) Raman Filter sets suggest using a laser-line filter and a longpass edge filter, such as here, here and here - although to be honest, the later only supports the use of the narrow laser-line bandpass filter, it suggests using the much more expensive notch filter).

Incidentally, using only "normal" mirrors, the set-up should be a hell of a lot easier... We aren't trying to direct light in two different directions at once using the one mirror. That makes alignment a shitload easier and the use of camera parts I'm yet to understand (thankfully for me). Personally I'd probably use lenses for a microscope to get the laser light into the sample and the scattered light out. I'd be inclined to try and bounce the laser light around the sample-area for as long as possible, the more it bounces around the more it will elicit responses and focusing those responses back to the collection point only makes sense.*

* Is it possible/feasible to put the cut-on/cut-off filter below the collection lens? Theoretically that would bounce the laser light back into the sample would it not?

PS I honestly think the amount of machining is overrepresented in the PDF's off maryspectra.org - honestly, with a little less haste and a lot more forethought (including the ability to build on her mistakes), it should be possible to build this a whole lot more easily than she did. The first and foremost tip - don't go machining metal until you can get a plastic/wood one to work. Get prototyping & testing out of the way before trying to get things to "LOOK" or "FEEL" right. I cringed when I read that she had only aligned the laser in one axis (horizontal) and that it was misaligned in the vertical axis. Working out a way to hold the sample should have occurred to her WAY before it did.

She does get the kudos for doing the job and actually making the thing, I have GREAT respect for her in that regard. I just think it could possibly have been a little better thought out (this from the person typing with fingers covered with skin from his thighs due to a poorly thought out way of heating glassware). I honestly think some of the techies from this forum could contribute GREATLY to the idea and on the way, assist the astronomers to build BETTER spectroscopes for their purposes as well (that thought might well encourage them to be helpful with software design).

[Edited on 5-2-2010 by unome]

[Edited on 5-2-2010 by unome]

Thanks not_important, was kinda wondering when I'd hear some sage advice from you...

So if the Raman light is going to be 550-up, then the edge filters would be the way to go (steep edge, cuts on @ 550nM)?

I'm hoarding funds now to purchase the line-filter and the edge filter, what about with the 405nM blue/violet lasers? How bad are they - I mean, how far off-wavelength can we expect the light to be? I ask because I cannot find a reasonably priced line-filter for that wavelength.

What size/line spacing grating would you recommend? There are shitloads of options and a wealth of contrary opinions available...

Nice stuff coming in But let's take a step back, look at the difference in performance this person got when using a CD-RW with the reflective coating removed, as a transmission grating, which is also used by this to make an even better spectrometer (this design is actually the design that is being sold on the internet as being complete with a scale, I'm unsure if any benefit accrues to the designer).

Now, given that with a transmission grating we can avoid having multiple lenses and mirrors, also given that we can determine the actual spectral alignment with mathematics, I propose starting first with a Vis-type spectroscope using a CD-RW, then once I get that working, I'll try with a denuded DVD-R (although that will entail working out the formulae for where to put the various components, especially the scale.

I also have a sneaky suspicion, given that spectrometers can be used as a modular component, that I could use the same transmission grating with the Raman signal...

[Edited on 12-2-2010 by un0me2]

I'll have to double post this, as I have to mention that I have been trying my fucking damndest to remove the aluminium backing off some CD-R's and have had no real success, I also checked the protector disc in the spindle (the blank one) and couldn't see any discernable spectra, thus leading me to conclude that it did not in fact, have the requisite lines/mm. That left me with trying out the idea in this (thanks to Jokull) paper, which does in fact work. I will next try it out from the word go, first I have to clean up a shitload of mess from trying to scrape the fucking aluminium off (with the attendant scratches, etc. that that would leave on the non-grating side of the transmission grating, which presumably would have serious consequences for any spectra done with the artifacts in place).

I will try to carefully, with a scalpel-type blade, work my way around the extreme edges of the coating, then try and lift it in one go with reinforced tape. I'll then attempt to use a toilet roll tube/paper towel tube to set up one of these as proof of concept. That done, we'll proceed

PS Although, thinking about it, if one were to cut the circles into an intact CD



Then attach the tape to the shiny side, presumably the edges of the circle that is cut would be sufficient to remove the need to carefully dislodge the hold of that layer from the inside and outside extremities of the CD-R (presumptively, if the cut on the inner diameter and outer diameter, allow the entire thing to come off, then it is only really adhering on the inner and outer diameters, not through the middle). Kinda hard to explain, does anyone follow my logic?

PPS I was thinking, based upon the article above where the proper camera angle, relative to the grating is 19.2', that if the grating were inset in the middle part of the tube, such as to allow the camera to be placed directly over/inside the lower end of the tube (upper end is the slit), then that should give the best results for our purposes, no?



I am also seriously wondering if I could cut round 'transmission gratings' from a DVD-R, would it be easier to separate the two layers (actually there are apparently 3 sandwiched into it), allowing me to utilize the much higher line/mm resolution for this project

[Edited on 13-2-2010 by unome]

Cutting the circle out, THEN removing the aluminium layer works like a fucking dream. Now, I am in the process (sans protractor, there is NOT one in the fucking house:mad of determining the notch I have to take out of a second tube in order to put the camera at the most advantageous angle of 19.2'.

I also have to work out the most advantageous angle for DVD-R's as the same technique works a charm with them too*

*Personally, I find it easier to draw the circle, then cut the piece out of the DVD-R/CD-R then trim it, it avoids a lot of stray scratches and possible damage to the flatness of the transmission grating.

PS Given I have no protractor, I have to work it out - I'm guessing here (well, not quite, but it has been a LONG time), but tan 19.2 = O/4, so tan 19.2 being 0.365531827, then that x 4 = O, ie. O=1.46cm, as I am using a decent ruler, 14.5mm should be easily doable.

PPS I am currently reinstalling the drivers for a Microsoft LifeCam VX-1000. It is about the cheapest digicam I can think of, with absolutely crap resolution, etc. But it should suffice for this part of the project. We'll cross the other bridges when we come to them

FUCKING HELL

Anyone got a fix for those fucking cameras? I have two sitting here, neither of which I can get to work

BTW I have just attached the threaded upper neck section of cordial bottles (that are the right diameter) to both ends of a toilet roll tube with tape, then spray painted the whole lot, inside & out, black. I cut out the center of on of the lids and mounted a circular 'transmission' grating in it with araldite. In the morning, both the Araldite and the spraypaint should be dry and I can start putting the whole thing together (I'm going to cut a small rectangle out of the other lid then use two pieces of metal (straight edges) to form the slit over the same.

I just thought that threading everything would be a worthwhile idea, given the purpose of the plan, to make a modular spectroscope.

[Edited on 13-2-2010 by unome]

Finally managed to get the bloody camera(s) (found another one, a VX3000 - which has much better resolution and gives better results) working and the pics are attached (from my wooden version of the JChemEd article (the first one). Next, I'll grab that online image tool (that is mentioned in one of the articles) and try and convert them to recognisable spectra.



[Edited on 14-2-2010 by unome]

Nother double post, but I thought I'd post this - it is the primary spectrum from a new, tungsten filament halogen bulb resolved over a reflective DVD grating (the wooden one I built earlier).

I cleaned the pic up (ie. took a short section out of it that contained the cleanest spectrum, removed noise, adjusted the brightness, etc.).

Now I just have to work out how to get that program so I can get a histogram of it

Yeah those kids gratings look alright - for the moment I'm looking into RGB ==> Wavelength conversion (ie.nm), the DVD diffraction grating seems to be working quite well enough for the present, when I start getting the spectra from the spectrometer popping itself up (without additional interaction from the user) then I'll start looking into the use of proper diffraction gratings - I'm thinking of the gel-type on a aluminium background (pretty much what I'm using with the CD/DVD's now, but straight lines).

I'm interested to see that the instructables model has the same (or essentially the same) spectra from a halogen desk lamp, which means the RGB section is working, now to convert that to wavelengths programatically (not so trivial - see the attached jpeg - shamelessly stolen from http://www.philiplaven.com/p19.html)

There are ways to do it, it is just deciding which is going to work best with logic chips and the drivers involved. In terms of what is necessary, we know the steps that have to be taken and in what order. One of the issues is that fortran is the program of choice (and I have no knowledge of it).

[Edited on 16-2-2010 by un0me2]

I didn't read all the thread but here is what I know. I may repeat what others have said so excuses

Since I'm doing some holography I know what a good laser means.

For a regular green laser as a pointer or a module, not only you may have lots of residual IR from the pumping diode, but there are lots of mode hoppings too (changes in the frequency of the laser) and power levels. That means your laser is not purely 532 but there are residual lines as 532.01 and so; they come directly from the IR pumping diode and will generate "parasitic" Raman lines too. So we have to get rid of these.

There are a few lasers on the market for 532 nm that are "single mode". Such a laser is required by Raman spectroscopy, confocal microscopy, citology (devices for cells counting), DNA sequencing etc. Such a laser used to cost a fortune (at least 15-20 k$ new) and they still cost about that much. The reason is that they replace the gas lasers; at 3000-5000 hours an Ar ion laser needs a tube refill, which is very costly (thousands of bucks) while these lasers are supposed to last at least 7-10000 hours with the same power levels. The power supply requirements are also very different; an Ar ion laser would suck some 4-10 kW/h, while a solid state laser less than 0.1 Kw (depending on the type).

The good news is that these expensive solid state lasers started to pop as a surplus since 2005 or so, and they are not very difficult to find. Moreover, the price is way lower than what they used to cost.

I don't know anything about the power requirements for a homemade Raman spectroscope, but I feel like 5-10 mW would do it (the first ever documented Raman spectroscope was made by a (very smart) girl and she used a 5 mW green laserpoiter only to see that the lines she got were "multiple" due to the multimode low quality laser - the link may have been already posted).

Nowadays (year 2010) a 100-150 mW green stabilized laser would be like 1000-2000 bucks, while a 50 mW can go down to 400$. There are even 10 mW versions for much cheaper as here: http://cgi.ebay.com/Coherent-215M-laser-10-Mw-532nm-excelent...


These lasers require very precise alignment of the optics, an extremely high quality TEC (thermoelectric controller) that stabilises the laser head within 0.001 degrees C, special components, ring resonator design etc. To give you an idea, the warmp time for these lasers is about two minutes; after that single line operating mode is achieved.

As about the 405 nm blueray diodes - they are far worse than a measly red laser pointer in terms of spectral stability; they barely have a coherence of 1-2 mm so they mod hopp like a bunny shot in the balls. I really doubt they can be used for Raman spectroscopy in an efficient mode.

Quote: Originally posted by a_bab

Since I'm doing some holography I know what a good laser means. For a regular green laser as a pointer or a module, not only you may have lots of residual IR from the pumping diode, but there are lots of mode hoppings too (changes in the frequency of the laser) and power levels. That means your laser is not purely 532 but there are residual lines as 532.01 and so; they come directly from the IR pumping diode and will generate "parasitic" Raman lines too. So we have to get rid of these. There are a few lasers on the market for 532 nm that are "single mode". Such a laser is required by Raman spectroscopy, confocal microscopy, citology (devices for cells counting), DNA sequencing etc. Such a laser used to cost a fortune (at least 15-20 k$ new) and they still cost about that much. The reason is that they replace the gas lasers; at 3000-5000 hours an Ar ion laser needs a tube refill, which is very costly (thousands of bucks) while these lasers are supposed to last at least 7-10000 hours with the same power levels. The power supply requirements are also very different; an Ar ion laser would suck some 4-10 kW/h, while a solid state laser less than 0.1 Kw (depending on the type).

But Sir CV Raman did NOT have access to photodiode/LEDs and laser-line filters which could narrow the wavelength of the incident, collimated light to just over 1nm (1.4nm, eg. 532nm (+/-)0.2nm).... If he had have access to these, I strongly suspect he'd have tried it. Fucking sight easier than generating enough light through a monochromator.

I suspect (but dont KNOW) that it whould have been done with a gas discharge lamp. and one or two pin holes isolating a single mode. the then week single beam of light whould have been split appart on a grating and another pinhole used to isolate the required wavelength the very week light whould have been then passed through the sample and off another grating. onto..... a photographic plate and exposed for quite posibly a very very long time where the relly relly relly relly week signal might have appeared.

getting any modest amount of power at any wavelength into a single mode is hard.

getting white light into a single mode with enough power to do anything ? thats borderline inposible

comercialy packaged laser diodes are oftern fiber coupled by simply gluing (or soldering) a cleaved fiber directly onto the diode substrate and hoping for the best you will notice that many manufacturers will start getting a bit fuzzy when it comes to specifying the power output of a single mode fiber coupled laser diode becuse it varies from one device to another.


comercialy when people want any serious amount of light down a single fiber they will get a bunch of bigarsed IR laserdiodes and "pump" the multimode light down a Double clad fiber the fiber itself has two cores a outer multimode core that the pump light bounces around in. and a inner single mode core that has been doped with some rare earth ions that are able to lase.

so basicly you pump multimode light in one end and it causes the fiber itself to lase and 'new' single mode light comes out the other end.

http://en.wikipedia.org/wiki/Fiber_laser



and it is quite important for analitical spectroscipy becuse multiple modes (light traveling sightly non paralell or not 'Spacialy coherant' http://en.wikipedia.org/wiki/Coherent_light#Spatial_coherenc... ) will cause the spectrum to 'blur' like an image being out of focus.


as for LEDs.. im sorry its reeealy not worth the effort. you just cant couple more than a nano or microwatt down a single mode fiber and evan then the spectrum of the led whould be so broad that you whould have to throw away most of that if you wanted monocromatic light.


on another note if you are looking for optical components i whould recommend

http://www.thorlabs.com/ <- holographic gratings under $80

and if they dont have what your after try

http://www.edmundoptics.com/

Yeah I know what their gratings are worth - I also know what their laser-line/bandpass filters are worth too. The availability of those are what this is based upon.

Yes, for analytical spectroscopy where the actual spatially coherent, collimated light goes into the device, certainly, lasers are the way, for Raman - where the light you shine on it is the first thing removed by the filter prior to allowing it to enter the spectroscope? Why would it matter? I'm not just being a smartarse here, I really do want to know, that is why I am reading the papers on the subject by Raman himself, he did get it to work. If what you suggest is true, then I see NO reason why filtering the light to a single bandwidth, and passing that light into the solution, would not have the same effect that we seek, after all it is the photons that we are after and I'm not at all certain that they'll discriminate about the source of their excitation.

EDIT

In the attached paper, the Author states the following:



Attachment: Koga.Simulation.of.Visible.Spectra.from.Digital.Color.Images.pdf (53kB)
This file has been downloaded 1644 times




I added some files I found online - but seriously, looking at the problem we have a spectral range of what - 400 (say 420) to 900 at best? So with a shitty 300K resolution (640x480px) camera, we have 1px per wavelength in the visible spectrum?

So, without allowing for metamers* (other RGB values that give the same perceived color), it would not take a genius to simply sit down and feed a logarithm to the computer that each specific RGB value for that one pixel is either 000000 or whatever it is supposed to be, up to FFFFFF. That frees up that part of the system and would allow for rather simple calibration (even with higher resolution cameras, just multiply the number of pixels by the resolution along the baseline/480).

* With metamers, if they exist, they either equal the DB color for that pixel or they are noise and left out (or they = 000000/black). By averaging the entire 640 height along that line, we either get a colored pixel, or a black pixel @ that wavelength.

Intensity is another issue altogether, it can be best obtained by the looks of things by using greyscale and histograms. Or by adding the number of colors = the DB color for that pixel (including metamers that either do or they don't), and use that as a measure of intensity... That does away with a whole lot of fucking around, and if one were to pull the IR filter off the front of the CCD sensor, then use the long axis to go from 400-1040, or below 360-1000nm whichever fits better, it still will be only one value per pixel regardless.

Attachment: Smits.An.RGB.to.Spectrum.Conversion.for.Reflectances.pdf (197kB)
This file has been downloaded 1207 times

Attachment: Ward.EydelbergVilshin.Picture.Perfect.RGB.Rendering.using.Spectral.Prefiltering.and.Sharp.Color.Primaries.pdf (1.1MB)
This file has been downloaded 1330 times

Attachment: Wilkie.Larboulette.Purgathofer.Spectral.Color.Systems.and.Appearance.Metrics.for.Fluorescent.Solid.Colors.pdf (583kB)
This file has been downloaded 923 times

Attachment: Li.etal.An.Improved.Method.to.Estimate.Reflectance.Parameters.for.High.Dynamic.Range.Imaging.pdf (777kB)
This file has been downloaded 1001 times

[Edited on 19-2-2010 by unome]

Righto, I just hacked fuck out of a poor defenceless little VX1000, I prised off the little rubber grommets holding on the mounting & then unscrewed that, then removed the single philips head screw from the rear of the camera, then had to remove the inner grommets from where the mounting attached... I then unscrewed the lens and removed the small clear-ish piece of glass which is the IR filter. I have now reattached the front fact (and the lens - although I took the opportunity to remove the fucking annoying call button from on top - it slides out forwards with the front of the camera off).

I'm going to build a smaller spectroscope this time, with the digicam built in, after the ever elusive VIS-NIR spectra

EDIT - FUCK this goes a LOOOOOONG way into the red

I am using it on the old spectroscope and getting several distinct red bands, fuck all blue....

Now, bar the putting together stage - I've built a wee spectroscope that will fit quite snugly over the front part of both the modified VX-1000 and the unmodified VX-3000 and once the glue finishes setting, should be able to get some decent spectra and start sorting out some issues I'll be able to get exactly the same resolution with both (640x480 is the max resolution of the VX-1000 and the minimum resolution of the VX-3000), so it will be interesting to compare the spectra.

[Edited on 21-2-2010 by unome]

[Edited on 21-2-2010 by unome]

I need to play with this a little more, but that is the SAME lamp as the previous spectra - the red has resolved into at least 3 distinct bands - a hell of a lot wider than before (dwarfing the blue-green) and that pinkish area at the end of the red spectra I'm assuming is the NIR range?

Righto, I've started to get the slit narrowed down and am getting useable spectra

The latest file is a 640x480 shot from the modified VX-1000. What is intriguing is that it shows the full spectrum - and the spectra I got earlier with a 1.3mp shot only makes up about 1/2 of that - on the lower resolution shot, we've lost some detail in the bottom part of the spectrum, the yellow line and some detail from the green, but we have several interesting areas in the newly revealed red, pink and violet half of the spectrum and I'm seriously wondering how wide this spectrum now is in nm, because it is fully twice the width it was before, or at least appears to be.

I'm going to have to pull the filter out of the VX-3000 just to find out if, with the higher resolution, the detail of the lower end of the spectrum will still be intact.







PS The spectroscope I used was my own design - it has a wedge of a DVD inset into a hollowed out section of Foam-board (Black foam filled cardboard sheet). This is set at a 60' angle to the camera. It is attached at the top and bottom to the front and back of the setup - both of which are also made of foam-board - the front one has a 3cm hole in it to admit the lens of the VX- series cameras. The top, sides and base are all thick black cardboard and the whole thing is 5.5cm wide, 5cm high & 4cm wide. The slit is 4cm wide (it really ought to be 3cm) and the exposed portion of the DVD-R (with no bubbles from the cutting - they are covered by the mask) is 4cmx3cm (3cm being the width of the lens and 4cm being the approximate height of the lens offset at 60').

This latest image is a cropped image - I have spectra going vertically the entire height of the picture, but cannot get it to extend the full width of the picture... Anyway, I reduced the exposure time significantly and the number of bands appearing now are amazing.

[Edited on 22-2-2010 by unome]

I just worked it out - the black segment, taking up about 1/3 of the horizontal segment of the spectrum (ie. before the start of the Visible Radiation) is the invisible UV radiation which neither I nor the camera can pick up...

I attached the uncleaned spectra (compare to the one above), if I increase the size of the DVD segment and shift the UV spectra out of the view field, then I'll be able to see all the NIR spectra (which you can only see a part of in those shots).

I also attached a quick design of the spectrometer I am using at the moment (if anyone wants to make it - it is pretty much to scale).



Attachment: spectroscope013.pdf (1.6MB)
This file has been downloaded 924 times

Since we can now measure that which we cannot see, and assign to it the wavelengths we know it corresponds to, then we can also assign the wavelengths of what we can see, based upon the measurement of what we can't, can't we? Based solely upon which, using the wavelengths of 0-400 as the invisible half of the spectrum above, then the other half should correspond to 400-800 (or thereabouts), with the additional 700-800, the NIR range, adding on to the visible spectrum.

[Edited on 22-2-2010 by unome]

yes it is often possible to see 800nm I can see down to about 840nm it appears just as faint red.

never tried UV

I am currently getting the drivers on this laptop to work and then I'll finally be able to get the spectra from the streetlights and whatever else I can find of interest...

I'd dearly love to be able to call it a spectrometer, but until it can be used as a means of measuring the wavelengths, its a spectroscope and that is all

PS Check out the ACTUAL size in these pics



Here is the spectra of the laptop screen

[Edited on 24-2-2010 by un0me2]



and here is the best shot I could get of the streetlamp (I could SEE far better spectral lines by eye, but fucked if I could get them sharp and thin like I could see myself).

[Edited on 24-2-2010 by un0me2]

are you pointing the camera at a screen or do you have the light bounce straight off the grating onto the ccd ?

depending on your grating and your desired bandwidth. you may find you can get a higher resolution and sensitivity buy illuminating the ccd directly.


allso for each pixle of the image you are only interested in the absolute intensity not the colour value (spectrometer ccd's are monocrome and some spectrometers use a physicly rotating mirror grating to scan the spectrum past a single photodetector and detect the spectrum as a function of time )

This means that you will have to add the red green and blue channels of the camera together to try to compensate for the varying sensitivity of each channel to various wavelengths. a photoedditing program may do this when you convert from 24bit colour to 8 big grey scale.

another thing that may help is saving the data from the camera as a RAW file or a BMP to try to avoid any compression of the image. most compression will throw away some data not visible to the eye. and some of this information may be what you want !

an interesting thing that you should be able to do with your setup that dosnt require any fancy light sources or optics is oximetry.

http://en.wikipedia.org/wiki/Pulse_oximetry

basicly pass regular incandescent light through your finger and you will get a spectrum that looks like this spectrum of my finger.

the little sholder labled a is the absorbsion by the hemoglobin in your blood and the area marked b is another absorbsion point for hemogobin by looking at the ratio between these two points and comparing it to a empirical table you can tell the oxygen content in the blood.

but more amusingly if you can see the spectrum in real time. the lobe marked A will bob up and down in time with your pulse !

I've tried using DVD/CD's as transmission gratings and had less than impressive results... I'm out in the street with the laptop and the camera trying to hold onto the Camera/spectroscope and orientate it to where I'm getting the best spectral response through the slit (including trying to modify the slit width etc.) while also trying to fuck aroudn with the control panel for the camera, playing with exposure, gamma, etc.

When I remove the camera (or more appropriately, when the camera removed itself) and look through the viewport I'm seeing extremely narrow, very well resolved lines (such as I've come to expect), but capturing them on the CCD is rather more difficult than I had expected. I'll try again

I've got to get a couple of small bulbs, wonder if those little 12V lamps (breakdown ones) they sell at supacheap would work?

Quote: Originally posted by un0me2

[...] while also trying to fuck aroudn with the control panel for the camera, playing with exposure, gamma, etc. When I remove the camera (or more appropriately, when the camera removed itself) and look through the viewport I'm seeing extremely narrow, very well resolved lines (such as I've come to expect), but capturing them on the CCD is rather more difficult than I had expected.

Here's one using the other camera with a long-life, smart-bulb



For the Raman, I am seriously considering taking one of those super-bright Luxeon diodes (Royal Blue, 700mW) and then buying one of the diode to fiber couplers, then using the bought one as the basis for making a cobalt-blue one in acrylic.

That will (1) colimate the light and (2) remove everything but a small bandwidth from getting through. One filter on the other side and boom, we'll know whether or not it has any functionality at all.

[Edited on 27-2-2010 by unome]

Same bulb, but using a transmission grating - ~70' front to back (ie. front is closer & @ the bottom), slit is about 40mm long, about 1mm thick.



Another one, starting to get the resolution I'm after FINALLY



That last one actually splits the two reds and then splits the blue into about 3-4 parts, so the resolution would have to be up there wouldn't it?

[Edited on 1-3-2010 by unome]

Right, I took this narrow slice out of a transmission grating shot - notice the resolution? There are some very narrow, VERY finely resolved lines, even the green is resolved into two in the 8-bit version (greyscale).





And here is the 3D image of it from ImageJ

Check out that doublet in the green



Woo-fucking-hoo

Now, if I lay that slice horizontally and increase it's width to match the graph (just stretch the image), plus swap it around so that the 400nm region is on the left (instead of the right), then we have something recognisable as a graph of wavelength v intensity. Now all I need to do is get some decent, fairly low power LED's (high powered white LED's are too fucking bright for this little baby, they wash out the spectra) and play around a little so I can start getting absorption spectra from colored solutions, then we can play...

After that, try and get a laser to excite the fuck out of organics and get the Stokes/Anti-Stokes (fairly high-powered NIR diode lasers are easier to get than blue/green and a whole lot cheaper - plus the IR-filter on the CCD should really block the fucking laser line if one chose wisely).

[Edited on 5-3-2010 by unome]

Quote:

After that, try and get a laser to excite the fuck out of organics and get the Stokes/Anti-Stokes (fairly high-powered NIR diode lasers are easier to get than blue/green and a whole lot cheaper - plus the IR-filter on the CCD should really block the fucking laser line if one chose wisely).

Nice to hear from you not_important...

Yeah, I've been looking at the little neon bulbs that can be slotted into circuits (some even come on boards, just connect 'em), for calibration purposes, they draw fuck all current and aren't going to wash out the spectrum in the minispectrometer...

You say that the Luxeon LED's are too wide a wavelength, but the datasheets seem to suggest that at most there is a half-width of about 10nm and that at full-power, the dominant emission is fairly narrow. Given the massive (compared to 5mW laser pointers) amount of light emitted from the 0.7W Blue LED's (dominant wavelength ~440-460nm @ 700mA), I still suspect it is worth a try... Put that through the relevant filter and if the specs are telling the truth then I'd be expecting at least half of the total emission to fall within the necessary waveband, if it doesn't work - nothing ventured = nothing gained

If I can narrow the output band from the LED's down with filters, I will, but I'm working backwards from what Sir C.V. Raman did, a blue filter to get only blue light (obviously He concentrated a shitload more light than I'm going to, but if 5mW lasers can do it, I cannot see why this cannot). The main purpose behind trying to utilize LED's rather than laser diodes, is the reported problems with them for this application and the costs involved in trying to source 'proper' lasers... For proof of concept, ANY Stokes excitation that can be measured will suffice, if it is recognisable that is a bonus. If that point can be reached using low-budget, low-tech gadgetry and some workarounds, then that gives me hope (in much the same way that being able to produce a half-decent graph of the wavelength v intensity, using freeware and way out of date, cheap as shit digital cameras did above).

I am sorry to hear that about the NIR lasers, I was hoping that I could utilize the insensibility of the CCD sensor to the NIR emissions to remove the laser line and the Stokes-excitation (which is increased wavelength from the laser line if I have it right, ie. NIR too), leaving only the Anti-Stokes (lower wavelength than the laser line - ie. visible), which given sufficient power in the excitation laser, I presumed would be able to be utilized (pretty much using the weaknesses/nature of the accessible OTC equipment to improve the capability of an OTC sensor). Is there any way to improve the utility of the 200mW NIR (<800) Laser's?

I was thinking if the Stokes radiation from a 5mW laser is measurable, there has to be some hope that the Anti-Stokes radiation, from a much more powerful excitation laser would be at least measurable (and thus useable). The IR filters on the cameras don't have that sharp an edge, but if the laser is above it, then it wouldn't register would it? Neither would the Rayleigh Scattered or Stokes Excitation? Leaving only the Anti-Stokes to contend with? Thus capitalizing on the inherent weakness of the chosen sensor?

Righto, I've worked out what I want to do to try and prototype this and see if it works - I can (by the looks of a local site), get a small kit to run one/two of those high-powered blue-LED's off a standard 9V battery, the same place sells neon bulbs and "white" LEDs (apparently, by the data sheet not yellow phosphor on blue), which I intend to wire up to run off battery power too.

That saves me fucking around trying to set up breadboards/PCB's while prototyping.

I have some ideas for the filter(s), a dark blue filter will restrict the wavelength going into the solution and a yellow filter will allow everything but blue out of the solution, perfect for my needs.

I'm experimenting here, it is far from a foregone conclusion it will work I'm basing the concept upon what I've read (standing high and dry on the shoulders of some MAJOR giants). When I finally get all this shit together I'll post the details and probably some pictures of the setup and the results (good or bad).

Quote: Originally posted by unome

... You say that the Luxeon LED's are too wide a wavelength, but the datasheets seem to suggest that at most there is a half-width of about 10nm and that at full-power, the dominant emission is fairly narrow. Given the massive (compared to 5mW laser pointers) amount of light emitted from the 0.7W Blue LED's (dominant wavelength ~440-460nm @ 700mA), I still suspect it is worth a try... Put that through the relevant filter and if the specs are telling the truth then I'd be expecting at least half of the total emission to fall within the necessary waveband, if it doesn't work - nothing ventured = nothing gained

So sorry, been busy (got work being done on the house & sundry other productions)...

First and foremost, I want to show that it can be done... Following that, I intend to work something out as an analytical tool, but first I would like to demonstrate (mainly to myself more than anything) that it can in fact be done and from there, we'll take it as it comes.

What about the cheaper bandpass filters (half-width of 5nm, ie. full width of 10nm), that would take in the majority of the light from the diodes cited wouldn't it? At a price of losing resolution, but yeah... I want to get something working, it can be be remade/remodeled later, I need a prototype

There are LED optics (fucking cheap too) that are designed to work specifically with the Luxeon and/or Cree LEDs, they take the gaussian emission and narrow it down to a 10' arc or even collimate it for fiber optics. I intend trying both, a 10' arc, through a microscope objective and then fiber optics.

I was just thinking, given that slit width maketh the width of the spectrum, I could mount spectrometer so it is divided into three compartments with only one grating, but three slits. All three spectra would show on the CCD image (sideways - so the long axis would pick up the spectra of all 3). Build a neon bulb into the middle and each and every picture would be automatically able to be alligned (alignment with every use - that removes all the mirrors, fancy tricks and everything - plus no moving parts), based upon the known peaks of the neon bulb.

Anyway, with the filters available now (or even just including a simple monochromator) I could utilize a 12V automotive spotlight bulb (tungsten halide IIRC) and just take the wavelengths I want from that - given that they are 100-250W, that would give several watts of light in a given wavelength, but I'm trying to be gentle The monochromator would be nothing more nor less than a blue filter, which would then be narrowed further by filtration.

As it has been demonstrated that miniscule amounts of light are capable of causing measurable Stokes-shifted excitation, then that too would work (probably cheaper than what I have outlined into the bargain). As the light source would not be operating continually for any real length of time, heat would be manageable.

But not_important, I am merely trying to do what I think is possible, I appreciate your assistance and am not trying to argue with you, your knowledge on this point surpasses mine by so great a margin as to be disturbing. All I am doing is showing that (a) lasers aren't NEEDED; and (b) that this will be possible at home, sooner rather than later.

My reason for aiming at the bottom end of the visible scale is simple, that gives me the greatest chance of capturing the Stokes-shifted radiation within the known limitations of the CCD/CMOS sensors. My reason for trying to avoid laser-diodes, is that they eat too much power and according to you (and others) are of doubtful utility for this application.

[Edited on 8-3-2010 by unome]

Very little, to maybe something.

DVD diodes are ~650 nm, CD drives use ~780 nm. The absorption spectrum for the 0.5% doping level is attached. The weak bands become stronger at higher doping levels, so in part the answer depends on the doping level, and in part on the size of the Nd:YVO4. You can pump big slabs off the peaks, that way the pump energy can penetrate deep enough into the slab. But with a bit a few mm in cross section much of the pump energy will just go through it.

The optical drive lasers are a few mw for the read laser, and around 100 mw for the write laser. I beleave the pump laser in a 5 mw green pointers is around a quarter watt, certainly at least 100 mw ( ~80% up conversion efficiency, maybe 10% doubling)




[Edited on 10-3-2010 by not_important]

So I'm going to need to find a decent source of the 1W Sony 808nm diodes?* OK, trouble is our nanny state has banned anything serious laser wise here (goons were pointing them at pilots on approach at the main airports), so I'd (1) have to find a supplier and (2) get one who would write on the package - electronic parts assorted.

It is illegal to purchase anything over 1mW without licensing and a whole lot of paperwork in terms of finished pointers, diodes are something else again, they are electro-optics and appear to be exempt (at present). I'm still awaiting replies from various suppliers of the Vanadate:KTP coupled-crystals, although given that the epoxy(??) that is used to join them doesn't like high pump power, I may have to go with separate crystals in-line.

Now, what, in your opinion, would be the 'BEST' strength of 532nm light for amateur Raman Spectroscopy?

... I don't mean the lowest possible and I don't need beast-design, I mean something that is going to give reasonable results, time and time again, with the least issues (from crystal heating, diode breakdown, etc.). Remember this is amateur, I'd like to resolve as many of the spectral characteristics as possible, doesn't mean I'll be able to, but that is the goal.

Now, the other question would be, how will building these solid-state mitigate the mode-hopping behavior?

* If anyone has a decent source of these things please share, I'm getting overwhelmed with the sheer volume of so-called honest salesmen that I'm dealing with at present, many of whom seem to speak a bizarre blend of pigin english and pseudo-tech geek

Ok, I have ordered the suggested diode-laser modules - any suggestions on where to get the safety goggles?

I mean, given the numbers, you are saying that we would be dealing with Class IV (>500mW) levels of the 808/1064 and IIIb (5-500mW) levels of the 532nm? Surely the better idea would be to prevent the escape of the 808/1064nm light altogether? Presumptively there are filters that WILL remove them from the scene (the 1064 is a real hazard, given it is non-visible).

On that subject, if something like this were built (from the attached paper) - using straight out CD/DVD type reflective gratings, we could (provided we could work out a way to absorb the stray wavelengths) narrow the beam considerably, perhaps even more than with the laser-line filters?

From what I can make of the design, the reflective grating acts as a monochromator - effectively only transmitting (provided the stray wavelengths are absorbed) the line that it is calculated to reflect on that angle (dependent upon grating and the size of the mirror/hole...).. While this looks somewhat advanced for what I am trying to do, a modified version, with a moveable reflective grating, totally enclosed with the ONLY vision being via CCD camera (I have no real wish for cataracts or permanent blindness), should allow for fairly accurate analysis (especially when put through a transmission grating).

Given the reflective indices for CD/DVD's are known (or can be worked out), then it should be possible to narrow the wavelength down dramatically. Worth a shot anyway, unless you have objections to the scheme?

PS EXACTLY the same schematic is used to describe the narrowing of a diode beam in the second PDF.



Attachment: Turner.etal.Frequency.Noise.Characterisation.of.Narrow.Wavelength.Diode.Lasers.pdf (219kB)
This file has been downloaded 1603 times

[Edited on 13-3-2010 by unome]

Attachment: Hawthorn.Weber.Scholten.Littrow.Configuration.Tuneable.External.Cavity.Diode.Laser.with.Fixed.Direction.Output.Beam.pdf (111kB)
This file has been downloaded 1384 times

Quote: Originally posted by unome

Ok, I have ordered the suggested diode-laser modules - any suggestions on where to get the safety goggles?

Well fuck me... If I haven't stumbled onto something, this is apparently PRECISELY what the big boys do... Diode onto a holographic grating, then the required wavelength is diffracted onto a mirror, they call it line-locking / laser-locking / wavelength-stabilization

Presumably the other, stray and/or unwanted wavelengths are absorbed into some material, but that is the basic system - ie. no need for precision narrow-bandpass filters, ONLY the required wavelength can escape, thus even if it mode-skips, only the intensity will change, not the wavelength.

Attachment: Shine.Day.Focus.Tunable.Diode.Lasers.pdf (62kB)
This file has been downloaded 1284 times

Attachment: Volodin.Wavelength.Stabilization.and.Spectrum.Narrowing.of.High.Powered.Multimode.Laser.Diodes.and.Arrays.by.Use.of.Volu (882kB)
This file has been downloaded 1031 times

The latest paper compares Volume Holographic Gratings and Volume Bragg Gratings (way too expensive) with heatsinks and achieves wavelength widths of .3nm which is fucking spectacular and way beyond our requirements...

Attachment: Kohler.etal.Wavelength.Stabilized.High.Power.Diode.Laser.Modules.pdf (497kB)
This file has been downloaded 1478 times

Let's go back to Mr Koppen's site, and look at the mathematics (which I kinda suck at, but anyhow)... The numbers of interest here are already semi-established, 550nm comes off a CD @ ~20' & off a DVD @ ~48', whereas the equivalent numbers for 500nm are 18.5' & 42.5', so right away we see that for this to work with 532nm the angle of the DVD grating is going to have to be around about 45', which works wonderfully (I'll work on my maths & get an absolute angle).

The actual angle is going to be worked out according to His equation "sin(a) = l/D", with D=0.74um, l=532nm, which allows us to solve for sin(a) = 0.74/532 or some such.... If the number it spits out is between 44 & 46.5 it should be about right...

Trouble is, when I solve for sin(45)=0.850903525 and 0.74/532=0.00139097744, thus there is not going to be a sleep-filled night ahead of me

Alternatively, we could use an indexed screw to perform fine adjustment (when we get the thinnest green line possible, that is probably it) starting from the point given by the 'virtually' linear progression of the wavelengths we can see that by dividing the difference between 500 and 550 (5.5') /50 and then multiplying that by 32 = 3.52 (then adding that to the angle for 500nm) - ie. 42.5 + 3.52 = 46.02' we should be able to get close)

What would we use to absorb the stray light-radiation? Any ideas?

The other good thing is, the fact that this is Raman Spectroscopy and we want the laser to go in from the side of the sample, while we collect the filtered light from right-angles thereto (IIRC), so we can forgo the mirror and just put the light straight into the sample through a slit/lens, which simplifies fabrication somewhat.

[Edited on 14-3-2010 by unome]

Beam dumps are generally used when you want to absorb a beam (duh). The fundamental frequency beam emerges from the NLO at a difference angle than the generated overtone, so you position a beam dump to suck up that unwanted energy (part of the reason experimental setups are large, to give space for the beams to separate).

In this case for safety reasons, and considering the relatively low power in terms of heating stuff, it's likely better to use simple filters to drop the IR beam levels to to safer levels, then a grating to fan out everything followed by the slit to isolate, and lots of absorbers to keep stray light down. It would be a lot of work to make enough beam dump stacks, and the power level just isn't there.

There's the 808 amd 1064 nm beams, the LED-like radiation of the diode laser than can extend down into the green, and Raman shifted light from all three laser frequencies generated in any material they pass through including filters, optical fibers, and so on. Light is Rayleigh scattered, as well as by inhomogeneous areas of the optics, surface imperfections in the grating and other optics, and dust. It doesn't appear as a neat spot offset from the desired light after dispersion, but from the entire optical path at all angles plus reflections.

I don't think a deep slit really helps, else you would see it used a lot. Instead double or triple monochromators are used, each stage giving roughly a reduction in unwanted light to roughly 1/1000 the incoming level. That's with good replica gratings and other optics and using collimating mirrors rather than lenses, I'd not expect quite so good results with DVD gratings.

Also remember that the CD/DVD grating is not truly parallel lines, but rather a portion of a spiral and thus an array of curved lines.

Bulk silicon is not a good absorber of visible light. It appears grey, semimetallic if the finish is smooth. Bulk carbon is difficult to work with, the cheap forms dust off badly, the hard forms have some transparency. Carbon black in a carrier is OK. Micro-rough surfaces are better than smooth ones, thus dyes and thin paints that don't fill the spaces between fibers in paper products give better results than thick paint.

Optics Communications
Volume 270, Issue 2, 15 February 2007, Pages 262-272
doi:10.1016/j.optcom.2006.08.038
Common black coatings – reflectance and ageing characteristics in the 0.32–14.3 μm wavelength range

black fabric is pretty good, running about 96-98 percent.




[Edited on 15-3-2010 by not_important]

Sorry, I've been unable to reply for a couple of days, and I'm still awaiting delivery of my laser modules...

That said, has anyone got any ideas on how to attach (adhesive obviously, but what type?) the grating film from Edmund Optics (they sell it through Anchor Optics too) onto an acrylic/glass substrate and maybe even onto an aluminium adhesive tape/ aluminium sheet (for the mirrored reflective gratings)...

I was just wondering, it would seem to allow for much more reproduceable results, with a much better 'feel' (almost like it was being done properly)... The holographic grating sheet is so cheap it is scandalous, if we can utilise a fairly common adhesive (maybe methacrylate?) to bond it to a clear/reflective backing and then a clear front, it would be bloody useful and very difficult to damage too easily...

It would also allow for the gratings to be made, not just improvised, which would allow us to design Littrow-configuration setups with 1, 2, 3 or even more stages of monochromation in order to get what we want. Personally, I'd prefer a filter that just cut off everything over 800nm (the two main sources of stray light), then try and narrow the actual beam-width and stabilize the wavelength of our module.

The blaze-angle, plus the choice of 500-1000l/mm would allow for some interesting designs (from a physics perspective - almost echelle grating like) of that "decent" film, which should allow us to narrow the beam dramatically with some thought, and a little ingenuity.

Quote: Originally posted by unome

That said, has anyone got any ideas on how to attach (adhesive obviously, but what type?) the grating film from Edmund Optics (they sell it through Anchor Optics too) onto an acrylic/glass substrate and maybe even onto an aluminium adhesive tape/ aluminium sheet (for the mirrored reflective gratings)...

I'm posting again because I found a truly interesting article on the subject under discussion from the fbi.gov site (stranger things have happened)

I'm in the midst of reading it at present, I'll have to look into the epoxies, is there any specific reason why I'd need them rather than the methyl methacrylate polymer that is used in joining the component parts of DVD's (multiple parts methyacrylate polymeric base, a colored dye (to allow for writing to it), a grooved substrate, a grooved top then a false layer IIRC, I was looking at a diagram of one just last night).

If they can perform the job, but we are only dodging them due to the curved grating, I suspect we could also use the same adhesive...

Attachment: Eckenrode.etal.Portable.Raman.Spectroscopy.Systems.for.Field.Analysis.pdf (236kB)
This file has been downloaded 1157 times

PS Looked for optical adhesives and got this which is used for right through the VIS-NIR range and is even available in Oz for a decent price

PPS I was just thinking, if I got a decent optical density piece of glass in yellow (transmits >500nm) and some mid-orange (transmits<550), then that would remove most of the stray light wouldn't it? Prior to the dual monochromator - really, just two reflective gratings on the opposite side of a slit that can be tuned to be on a certain angle (both on the same angle, but opposite)

EDIT

I am sorry to put this here (but I felt strongly that it had to somewhere), but I am truly humbled by the amount of time and effort some VERY intelligent people, some of whom are very senior members of this forum, have expended in helping me to comprehend the difficulties of this project and the multifarious challenges involved therein. That said, I would like to thank everybody who has taken the time and effort to help, not just me, but everyone who stands to benefit from this if it can only be made to work. I'm not in the habit of 'pissing in pockets' and I suck badly at brown-nosing, which makes it hard for me to express my thanks. Please, if you are one of the people involved, take it as it meant, wholeheartedly and with absolute honesty.

[Edited on 19-3-2010 by unome]

I'm in the process of building the software and then cobbling it together with a TWAIN interface so that I can acquire the image.

I'm still awaiting arrival of the lasers, hopefully I'm not going to have to argue them through customs, although I am fully prepared to so do.

I've previously pulled the IR filter out of a CCD camera, there really is nothing to it, just undo the optics, to expose the CCD itself, look inside the optics and there is a tiny square of colored glass (very light blue). Remove it, reattach the optics (lens assembly and focus etc.).

Insofar as the color - if I use a neon lamp (they are cheap and the drive-electronics are simple), the spectra I want is going to have to be accessed via iterating through the pixels (as an array) and converting each pixel from an RGB color to 8-bit greyscale which is done using an algorithm, which can be modified to give the lumen at each pixel position. As each picture is several hundred pixels high, getting the average luminosity/intensity for the column (@ each pixel position, ie. x, y[1:200]) should enable me to draw a line, pixel position v average intensity (which will be fucking useful for noise reduction).

With Raman & a >550nm cutoff, if we use the long-side for the spectra (assuming we'll only want the visual) we have 500 pixels (out of 640), therefore we can move it back or forwards to line up with the known spectra of the neon bulb, plus have 2 pixels/nm, so 0.5nm resolution is possible (although not without some serious work).

EDIT

PS Why not use transmission gratings to clean up the laser light? I've seen several dozen articles on using such designs (the holographic grating in most of them 'is' (sic - how many years of schooling & I write that) formed by laser interference with a photoreactive substrate wedged between two layers of non-reactive clear outer layers, volume phase holographic gratings IIRC), and was thinking if the filters were utilized as the outer layers of the wedge (glass/grating/glass), then that potentially kills many birds with one stone, while protecting the rather fragile grating film from damage by effectively making it into a slide-type arrangement, which would protect the gratings from fingerprints, dust, etc. and also remove the need for adhesive, plus, quite probably allowing us to use less monochromatic stages...

NB I've actually got some articles on how it is done, they use lasers to react with 1,2-naphthoquininone/methacrylate between two sheets of methacrylate. The naphthaquinone is photoreactive and forms an opaque line where the laser hits it. It is actually quite interesting and may be an answer in itself later on.

[Edited on 20-3-2010 by unome]

Hell, notice they used EXACTLY the same OG550 Schott-glass bandpass filter I was talking about earlier? They also got a useable Raman spectra from a 4mW 532nm Laser, which means if I use a beamsplitter (to split the 532nm beam into two beams - reference (solvent) and analyte+reference (solvent+whatever) raman spectra are within reach).

Can anyone tell me why they bothered with the super-expensive notch filter when they already had a >550nm filter that would exclude the Rayleigh peaks anyway?



Also notice, they did not use any filter on the 532nm laser source whatsoever (from what they've written themselves).

The microscope objective would be an interesting idea - use one for each of the laser inputs and outputs - although to save fucking around, I'd suggest putting the laser in @90' to the output, that way you can place a mirror on the other side to the orange OG550 Schott glass filter >550nm (everything under that is cutoff).

Of course, given the diode laser modules I have are KNOWN to have no IR filter, we'd need to use that at least (I've included a PDF file, which is essentially just a cleaned up webpage describing the module laser's under discussion - I also added the schematic it cited (and linked to)).

[Edited on 23-3-2010 by unome]

Attachment: Section5.20mW.532nm.sku26887.Green.Laser.Module.Surgery.pdf (106kB)
This file has been downloaded 957 times

Quote: Originally posted by Polverone

I just spied this article from last month's JCE: Inexpensive Raman Spectrometer for Undergraduate and Graduate Experiments and Research. They're using a commercial visible spectrometer for data acquisition, but the illumination is with a green laser pointer. Supplementary information

Quote: Originally posted by unome

Hell, notice they used EXACTLY the same OG550 Schott-glass bandpass filter I was talking about earlier? ...Can anyone tell me why they bothered with the super-expensive notch filter when they already had a >550nm filter that would exclude the Rayleigh peaks anyway? Also notice, they did not use any filter on the 532nm laser source whatsoever (from what they've written themselves).

Below 1000 cm^-1 includes most of C-C ring breathing range as well as the C-halogen, C-S, S-S, C-O-C bands among others. See the table below and the attached image for common compounds that have bands below 1000 cm^-1.

4-Acetamido-phenol
shift calc. wavelength
213.3 538.1
329.2 541.5
390.9 543.3
465.1 545.5
504 546.7
651.6 551.1
710.8 552.9
797.2 555.6
834.5 556.7
857.9 557.4
968.7 560.9
---------------------------------------------
1105.5 565.2
1168.5 567.3
1236.8 569.5
1278.5 570.8
1323.9 572.3
1371.5 573.9
1515.1 578.6
1561.5 580.2
1648.4 583.1
2931.1 630.3
3064.6 635.6
3102.4 637.2
3326.6 646.4

Quote: Originally posted by Polverone

Accept the security certificate. It is a self-signed cert designed to protect your traffic from snooping by third parties, not to prepare and validate this site for commerce.

Quote: Originally posted by JohnWW

Quote: Originally posted by Polverone   Accept the security certificate. It is a self-signed cert designed to protect your traffic from snooping by third parties, not to prepare and validate this site for commerce. Unfortunately, that security certificate is not capable of being "accepted" by a downloader, and I cannot see anything about it being "self-signed".

Schott OG-550 Glass Filters do NOT fall into that range the transmittance at 550nm is down around 0.5% admittedly, but at 560nm it is up to 78%.





As to the secondary harmonics, I suspect (Strongly) that the article using the 4nm pointer wasn't really interested in mode-hopping and all the rest. That said, with a single transmission filter I can line lock it and remove the extraneous 808 & 1064 nm bands with a filter.

Apart from that, it should be only too easy to adjust the program to take into account the percent transmission and adjust it so that the relevant spectral image is corrected. In terms of noise reduction, on top of averaging the column luma, I was also thinking of finding the median and those numbers lying outside an 80% range of the median/mean (can't think which would be better to use) could then be transposed with the mode integer, then the mean is taken.

I was also thinking - we wouldn't have to iterate through the entire pixel range for each sub-picture - we could convert EVERY pixel in the sub-picture to a luma value, then simply work with the column array, average that and get a string of x, y(integer). Which we can correct for both the known transmission curve of the filter and for noise. Use of the trigonometric classes within C++ (for example) would allow us to assign absolute wavelenths from the grating, once we can compare known wavelengths with pixel position.

For example, the grayscale image is purely intensity - nothing more, so if we average out the columns in the grayscale image, we get an average intensity for that column/pixel position. The more rows, the better the average. If we identify the most repeated value (the mode), then identify the median, we can determine the central point of our column's intensity values, if we then replace anything falling outside a predetermined range of the central point (ie. the median) with the mode value, that will, in conjunction with the high number of rows averaged, give us some very useful information (ie the mean), with noise and dud pixels pretty well removed from the equation(s).

This will give a single value for each column, which we could adjust for the variable transmission of each wavelength through the OG-550 filter, in order to give a true representation of the Raman spectra if the filter gave 100% transmission purely mathematically. In order to do so however, we would have to find the peaks of the neon spectra (from the central image), it looks like this:



(NB Taken from Wikipedia)

Now, we can assign the peaks from that spectra (which is our central sub-picture), from that, we can backtrack (whether we are using a transmission or a reflective grating, to determine the angle of each spectral peak from the incident light source. That gives us the coefficient for the grating & our sensor, we then utilise that to determine and assign wavelengths to EVERY pixel.

If needed we can then modify the column value for various pixels, by the known transmission of that wavelength through the OG-550 filter, in order to model all spectral data as if transmission at EVERY wavelength were even.

No small task, not at all... In fact it is every bit as challenging as building the bloody spectrometer. Not only do I have to work out what bits to buy, what angles they are needed at, where the fuck I am going to get a decent, cheap, beamsplitter, as well as everything else, I also have to work within my USB 2.0 Power budget.

The neon bulbs are a doddle to buy, they are super cheap and require fuck all power, the laser takes 250mA/3V (which is more than I have available without pretending to be a mobile phone battery).

I need some help here... Anyone want to help work out the circuit diagram(s)? I have contacted the OEM Company that supplied the original iPhone 2MP CMOS Unit and have asked for the specs (but had to sign a confidentiality agreement to even request them). I badly need to work out what connector I need to access the data therefrom.


[Edited on 24-3-2010 by unome]

Ok, back on this because I've been thinking about it, there are two real options that I can see

Either purchase the 808nm 1W Diodes (not too bad pricewise) and some NdYAG / KTP Crystals and make a properly designed and properly aligned 532nm laser (with reflective mirrors on the first and second stages - reflecting the feedback back into the lasing medium the diode in stage 1 and the KTP/NdYAG crystal in stage 2), with luck that will phase/line lock it. Then use a 70-90 l/mm laser beam splitter to remove the stray wavelengths and get the whole down to at most a 1nm line. Making those gratings should be a cinch, get some UV curable polymer sheet (the PCB one would do), or a UV cure epoxy resin (preferred option) and use a cad program to draw 70-90 lines /mm then print it out on a high quality printer directly onto a transparency sheet. Etch the non-cured epoxy off the glass/acrylic sheet and bam, one low resolution grating.

The same could be done, with much more ease, using one of the 200mW 650nm Diodes (but I can find fuck all on them for Raman). What would the relevant shifts be for that? Has anyone got any resource (or link thereto) that will enable schmucks like me to work out the extent/percentage shift? Is it that easy? Or is there some other trick?

Personally I like the idea of locking the 650nm Diode (as the designs evolved, they have basically returned to where they started, the interior of the external cavity is separated from the diode cavity by a filter and a small aperture. As the filter is reflective to every bit of light that doesn't get there on the right angle, if we built a pseudo etalon inside the first external cavity with the entirety of the returned/reflected light being directed DIRECTLY at the diode, then the feedback should be solved (so the device will lock). A mirrored surface behind the diode would then direct the remainder of the light back at the diode too (from behind - ie. a mirrored surface each end = an etalon - Here is the thesis I got it from.



The phase/line locked diode could then be line-narrowed by running it through one of the laser line splitting transmission gratings (mentioned above).

But we come back to precisely where the *)(&)&()* raman shifted chemical groups will be, is there some formula? I really HATE not knowing this shit & I can't even find it anywhere I've looked.

[Edited on 25-10-2010 by aliced25]

I found some of them (in the attached paper), but in order for the "Raman Shifts" to make sense, one has to get the concept of "wavenumber" and "wavelengths" straight. It took me a little while. I'll attach an image from www.semrock.com, which shows the problem:



Notice how few wavenumbers there (comparatively) between 800nm-1000nm? Only 2,500cm^-1? But there are 30,000cm^-1 between 200nm-500nm?

This shows why Raman Spectroscopy is usually done anywhere but the lower half of the spectrum (UV-NIR). The wavenumbers are so numerous that the shifts would be impossible to measure/discern. Whereas at the upper half of the spectrum, the closer we get to the 800-1,000nm section the better (although it needs a stronger laser & a camera without an IR Filter, because imaging with a CMOS/CCD Image Sensor between 800-1,000nm (invisible) requires a fairly big spike to be noticeable.

I'm actually busily transcribing the shifts in the paper I've attached, when I'm done transcribing the shifts, what I'll do is work out the shifts in nm for each group listed from a 405nm, 532nm, 650nm and an 808nm laser diode, where we would expect to see the Raman Spectral result for those groups using each laser.

PS The relevant tables are on pages 4-7.

Attachment: Tables.Wavenumber.Shift.Frequencies.pdf (738kB)
This file has been downloaded 1689 times

Quote: Originally posted by aliced25

... But we come back to precisely where the *)(&)&()* raman shifted chemical groups will be, is there some formula? I really HATE not knowing this shit & I can't even find it anywhere I've looked.

Quote: Originally posted by aliced25

Notice how few wavenumbers there (comparatively) between 800nm-1000nm? Only 2,500cm<sup>-1</sup>? But there are 30,000cm<sup>-1</sup> between 200nm-500nm?

not_important,

Thanks for that spreadsheet - I have looked at it several times, but until I could work out the shift for each group and the wavenumbers themselves, it didn't make sense. I dislike not knowing HOW something works, for that I needed 2 pieces of the puzzle, the shift in wavenumbers, and the calculation of the shift in nm based upon the excitation laser is a whole lot easier if the reasons why it works are known.

I didn't see the calculation in that spreadsheet sorry, dumb of me, but hey... Ah hang on, I'm not getting it to work - the numbers are just sitting there, no calculation. I'll try downloading it again, but even if there is an issue with the download, if the calculation was there, surely it would show up regardless? I mean, even if the calculation wasn't carried out, the formula would still be there?


Here, I did a pdf with the solutions for 405, 532, 650 & 808nm & I also attached the Workbook (Open Office Version). Just type the Excitation Laser wavelength into the box at the top of the first page.

[Edited on 1-11-2010 by aliced25]

Attachment: RAMAN.TABLE.405.532.650.808.pdf (325kB)
This file has been downloaded 5686 times

Attachment: Raman.Solver.Chart.ods (16kB)
This file has been downloaded 636 times