Raman spectroscopy

[benzyme]

I’m going to order some tubes as well, to mount the filters. I don’t know why I got stands, they don’t seem practical for what I’m trying to do.

I opened up the spec to align the optics. It’s very tedious. The first thing I noticed was the fan wasn’t plugged in.
Still working on fine tuning the alignment. The left half of the CCD sensor corresponds to the lower pixel range.

 

[benzyme]

Alignment done.
I ordered a thorlabs patch cable, 5-pk of stackable tubes, retaining rings, SMA-905 adapters, and collimation lenses (generic, chinese). I think my configuration will be complete after all that arrives.

 

[benzyme]

The fiber-adapter plate doesn't fit that laser. I had to wrap it in electrical tape to screw it in, then hot-glued it (like what they did with the optics bench on the spectrometer). I guess when it's that close to the diode outlet, alignment won't matter too much, but it just illustrates the poor precision of cheap, mass-produced chinese goods. Now I'm worried about the collimation lenses..

 

[benzyme]

I didn't post spectra, because obviously, the excitation wavelength would predominate the entire pixel range. This is why a beamsplitter and band reject filter are required. The analyte is lysergol. Notice the raman scattering, a reddish-orange fluorescence.
Just waiting on the tube segments, retaining rings, patch cable, and lenses.

 

[benzyme]

'Some new uses for filtered fiber-optic Raman probes: in situ drug identification and in situ and remote Raman imaging' 1999. S. Michael Angel; J. Chance Carter; Dimitra N. Stratis; Brian J. Marquardt; William E. Brewer

 

[Loveall]

Thanks benzyme, another interesting paper. After reading it, I wonder if we can use Raman to check for HPBCD complexation also. We have the drinkable/sublingual/nebulizable complexed salvinorin project and monitoring complexation rates would be very helpful.

One question about the calibration. As I understand it, adjusting the optics centers the wavelength of interest and focuses (narrows) the peaks. The first mirror needs to align with the slit/fiber and have it in its focal point. The second mirror needs to have the CCD detector in its focal "plane". The diffraction grid receives and emits collimated light and the incident angle on the grid determines what wavelenghts get sent to the CCD detector.

After that, to calibrate the pixel number to a wavelenth (or ramam shift), the program needs to fit a 3rd order polynomial to a know reference spectrum, right? What polynomial coefficients did you get? What did you use as a reference?

Thanks again.

 

[Loveall]

Well... I'm still waiting for that spanner wrench to setup the optical tube.

In the meantime I've looked moar at the spectrometer. It turns out the pixel vs. wavelength formula can be written down as the sum of the sine of some angles, one of witch is an arctangent. I can post that if anyone is interested, the formula gives some insight on how the spectrometer behaves when moving things around in a spreadsheet simulation for example. Not sure why the software uses a generic 3rd order polynomial with unphysical coefficients (?).

Also in the spolier below is a very rough simulation of the spectrometer in the Optical attracted program. It could be fine tuned to more closely resemble our spectrometer and one can play with the mirror angles/position in the program (below is a picture of the simulation).

Now... where is that spanner wrench! :want:

Spoiler

# OpticalRayTracer 9.6
# * OpticalRayTracer Home Page
# 2018.07.18 11:16:00 EDT

program {
antialias = true
askBeforeDeleting = true
beamAngle = 0.000000e+00
beamCount = 5
beamWidth = 1
clipboardGraphicXSize = 1280
colorArrow = -2147483393
colorBaseline = -16760832
colorBeam = -4194304
colorGrid = 1086374080
colorHighBackground = -1
colorLensOutline = 276865279
colorLensSelected = 268484608
colorLightSource = -16776961
colorLowBackground = -16777216
colorTerminator = -16777216
decimalPlaces = 4
defaultWindowHeight = 600
defaultWindowWidth = 900
dispScale = 0.042214
dispersionBeams = 8
divergingSource = false
helpScrollPos = 0
interLensEpsilon = 1.000000e-06
intersectionArrowSize = 0.050000
inverse = false
maxIntersections = 64
selectedComponent = 6
selectedTab = 0
showControls = true
showGrid = true
snapValue = 0.000000e+00
surfEpsilon = 5.000000e-04
tableLineLimit = 500
virtualSpaceSize = 100.000000
windowX = 100
windowY = 100
xBeamRotationPlane = 0.000000e+00
xBeamSourceRefPlane = -30.000000
xOffset = -3.094207
yEndBeamPos = 1.800000
yOffset = -44.420187
yStartBeamPos = -1.800000
}

object {
active = true
angle = 0.000000e+00
centerThickness = 0.100000
dispersion = 59.000000
function = 2
ior = 1.520000
leftCurvature = 3
leftSphereRadius = 10.000000
leftZValue = 20.000000
lensRadius = 9.600000
name = Terminal Plane
rightCurvature = 3
rightSphereRadius = 10.000000
rightZValue = 20.000000
symmetrical = true
thickness = 0.100000
xPos = -19.997900
yPos = -46.114000
}

object {
active = true
angle = -42.500000
centerThickness = 0.100000
dispersion = 59.000000
function = 1
ior = 1.520000
leftCurvature = 3
leftSphereRadius = 6.000000
leftZValue = 20.000000
lensRadius = 4.000000
name = Mirror 1
rightCurvature = 3
rightSphereRadius = 6.000000
rightZValue = 20.000000
symmetrical = true
thickness = 0.100000
xPos = 0.000000e+00
yPos = 0.000000e+00
}

object {
active = true
angle = 95.000000
centerThickness = 0.552385
dispersion = 0.000000e+00
function = 0
ior = 1.520000
leftCurvature = 2
leftSphereRadius = 23.600000
leftZValue = 850.000000
lensRadius = 3.600000
name = Lens 1
rightCurvature = 2
rightSphereRadius = 23.600000
rightZValue = 850.000000
symmetrical = true
thickness = 0.000000e+00
xPos = -1.500000
yPos = -12.500000
}

object {
active = true
angle = 88.000000
centerThickness = 0.400000
dispersion = 59.000000
function = 1
ior = 1.520000
leftCurvature = 1
leftSphereRadius = 36.000000
leftZValue = 20.000000
lensRadius = 1.700000
name = Lens 2
rightCurvature = 1
rightSphereRadius = -36.000000
rightZValue = 20.000000
symmetrical = false
thickness = 0.400000
xPos = -5.260000
yPos = -52.500000
}

object {
active = true
angle = 84.000000
centerThickness = 1.600000
dispersion = -1.330000
function = 0
ior = 1.320000
leftCurvature = 3
leftSphereRadius = -20.700000
leftZValue = 0.000000e+00
lensRadius = 3.100000
name = Lens 3
rightCurvature = 3
rightSphereRadius = 6.000000
rightZValue = 20.000000
symmetrical = false
thickness = 1.600000
xPos = -8.195500
yPos = -36.658600
}

object {
active = true
angle = 67.000000
centerThickness = 0.100000
dispersion = 68.000000
function = 1
ior = 1.520000
leftCurvature = 3
leftSphereRadius = 6.000000
leftZValue = 20.000000
lensRadius = 1.900000
name = Lens 3
rightCurvature = 3
rightSphereRadius = 6.000000
rightZValue = 20.000000
symmetrical = false
thickness = 0.100000
xPos = -8.226700
yPos = -36.605100
}

object {
active = true
angle = 21.000000
centerThickness = 0.459292
dispersion = 59.000000
function = 1
ior = 1.520000
leftCurvature = 1
leftSphereRadius = 37.700000
leftZValue = 20.000000
lensRadius = 5.000000
name = Lens 2
rightCurvature = 1
rightSphereRadius = -45.800000
rightZValue = 20.000000
symmetrical = false
thickness = 0.400000
xPos = 1.982600
yPos = -46.598500
}

 

[benzyme]

sorry, I was on vacation.
I only did a single wavelength tuning, haven't calibrated. I'm still waiting on the tube segments, and collimators, before I do some serious tinkering. Ideally, multiple wavelengths would be used to do the calibration. I just don't have the cash to throw down on a proper HgAr calibration lamp right now

 

[Loveall]

I think we worked out the math/geometry of the spectrometer.

First thing to note is that there is an ideal wavelength range where 100% of the light hits the spectrometer's CCD. This is because the focusing mirror is not big enough to capture all diffracted light outside the range of 45 to 75 degrees (as measured in the first picture).

The formula for the wavelength (λ) at each CCD pixel (p), λ=λ(p), depends on the angle of the collimated light leaving the collimating mirror (γ), the adjustable angle of the diffraction grid (δ) and focusing mirror (α) the pixel positon (p0) opposite the center of the focusing mirror and on the distance between them (f2). These parameters shown in the second picture where they are defined by the dashed black lines which are either parallel or perpendicular to the CCD surface. The formula is,

λ = d { sin(γ-δ) + cos(δ+2α+arctan[(p0-p)/f2]) }​

Where d is the line width of the diffraction grating (for 1800 lines per mm, d = 555.6nm). Note that when fitting the equation the 3 adjustable angles only represent two free parameters. For fitting purposes one can simply use,

λ = d { A + cos(B+arctan[(C-p)/D]) }​

and obtain the constants A, B, C, D from the fit which each have a physical meaning. This should be an alternative to the 3rd order polynomial approximation that the software uses whose fitted coefficients don't have a clear meaning (at least to me).

The third picture is an example of a setup that works well for a 532nm laser. The parameters used to generate the plot are:

γ = 20°, d = 555.6nm, p0 = 1240, f2 = 2700, δ = 3°, α = 13°.​

The last two parameters (diffraction grid and focusing mirror angles) where adjusted while the others were kept constant to find this optimal setting, which I think is how one would adjust a focused spectrometer. Increasing (decreasing) δ moves the chart down (up) and increasing (decreasing) α = 13° moves the chart to the right (left).

Disclaimer: This is all just math at this point and has not been verified experimentally yet. There could be errors, but I've cross-checked the derivation several ways so I'm hoping it is correct. I can also post a working spreadsheet to tinker with the parameters, and/or the formula derivation if anyone is interested in that.

 

[Loveall]

tale of four lasers: 405, 532, 680, and 780nm
power: 100, 200, 5, and 800mW, respectively.
the analyte: lysergol (λmax = ~308nm) , ~2mg/mL in 50:50 MeOH:H2O

Following up on the previous post (wich covered the 532nm laser), I played around with the spectrometer expectatios for these lasers. An effective way to center the spectrum is by rotating the difraction grid. However, for the two lower energy lasers the refraction angles become very acute and the effective wavelenth range small. Better results are obtained when the diffraction grid is changed from 1800 l/mm to 1200 l/mm for the two higher energy lasers (see plots below). I believe this jives with the text in the refurbished spectrometer's vendor website, where the upper range of the 1800 l/mm grid is quoted at 700nm.

Below are suggested diffraction grid setup angle and diffraction grating for the other lasers above. Other parameters are unchanged from the previous post, so when changing the Raman source lasers, one may only need to adjust the spectrometer's diffraction grating.

405nm: δ = 11, d = 555.6nm (1800 l/mm)
532nm: δ = 3, d = 555.6nm (1800 l/mm)
680nm: δ = 8, d = 833.3nm (1200 l/mm)
780nm: δ = 4, d = 833.3nm (1200 l/mm)

Finally, the spectrometer sensitivity is lower at the higher wavelengths (last picture), so the 800mW power for the 780nm laser seems like a good idea.

 

[benzyme]

great info, Loveall. much appreciated :thumb_up:

got the tube segments, and ordered a cage cube to mount the beamsplitter. It has SM1 threading to couple the tube segments, and the baseplate can be mounted to the stand. Waiting on the notch filter, some couplers, and the collimating lenses, and then this app should be complete.

got an Edmunds toy catalog in the mail. like Thorlabs, lots of nice stuff.

 

[Loveall]

I'm still waiting for the spanner wrench so the optical tube can be set up

THORLABS sent me some free snacks. That was surprisingly nice of them. :d

 

[Loveall]

Results :thumb_up:

The last parts finally came in. I put together the optical tube. Don't have everything fully focused and the spectrometer is not calibrated, but I went ahead and tested a simple sample of Toluene. Looks like we have a match :d

As a cross check I ran an empty cuvete and there was no raman spectra, but it there are a couple small peaks I'm calling background. There is also still a lot of green despite the edge filter. Pictures below.

Next steps are to get everything focused/aligned/calibrated, and then see if we can get a spectra for tryptamine goodies instead of bulk toluene.

 

[benzyme]

very impressive sir.
can’t wait to get the rest of my parts this week. I’ll test against tryptamine.

 

[blue.magic]

Keep up the great work!

Hell if only I have time for all this. I just began reading "Applied TLC" :d

But watching the thread closely...

 

[Loveall]

Alright, I added old spice to the toluene. There is MASSIVE fluorescence spetrum . There are small waves in said comtinuum. I hope that by subtracting/analyzing the fluorescence out, the toluene spectrum and (hopefully) a DMT spectrum will arise. Any tips for doing this? For now I'll be reading benzyne's resources he posted and be trying to fit the fluorescence to some kind of function, treating the residual of the fit as the signal.

Below is the image of the fluorescence and raw data in CVS format (along with a calculated raman shift).

On the plus side, I recentered the spectrometer and can see further into the red. The equations from the previous post were useful (I used the toluene peaks, including two new ones that appeared at higher Raman shift to calibrate). I decreased the angle of the diffraction grid and the focusing mirror. The equation fit well and I got 6 degrees and 10 degrees for those angles respectively (was shooting for 3 and 14 but it is difficult to control). With this setup I still have 100% theoretical light up to pixel 1300 or so.

Fell free to analyze the data. Can anyone recover toluene peaks (they should be there)? What about a DMT signal (or oxide DMT)?

I may also consider getting an short-pass filer and check on the anti-stokes radiation, cause this fluorescence is crazy.

Many thanks.

 

[benzyme]

do you use a band reject (“notch”) filter?

I feared this. Toluene is similar to benzene, so it will fluoresce in the 220nm range. Once you get
into conjugated heterocycles like indole derivatives, the wavelength of max absorbance increases beyond 285nm.
it seems that anything that would fluoresce under a blacklight (365nm) would flood our CCD sensors. I think a 632nm module
would work, and they’re relatively cheap.

 

[Loveall]

do you use a band reject (“notch”) filter?

I feared this. Toluene is similar to benzene, so it will fluoresce in the 220nm range. Once you get
into conjugated heterocycles like indole derivatives, the wavelength of max absorbance increases beyond 285nm.
it seems that anything that would fluoresce under a blacklight (365nm) would flood our CCD sensors. I think a 632nm module
would work, and they’re relatively cheap.

I currently have a 540nm long pass filter. Since I don't have a notch filter I would need to buy a new filter (eg short pass) to look at anti Stokes.

If I understand your suggestion, we can try a lower energy laser to avoid fluorescence excitation, right? So I see a few options:

1) Get a higher wavelent laser and a new long pass edge filter. Pros: at higher wavelenth fluorescence will be less of an issue. Cons: less ramman emmision and lower CCD efficiency (can be overcome with a more powerful laser). I have a cheap red laser I can try to check for fuorescence (I don't think I need optical filters for that). Can cross check the test with an equivalent 5mW cheap green laser to see if longer wavelength turns fluorescence off.

2) Get a short pass edge filter and look for anti-stokes radiation. Pros is that fluorescence should be gone (the trend for fluorescence can be seen in the plot and it does indeed dissapear near the excitation energy). Con is finding a reasonably priced filter (shortpass don't seem to be as overstocked). And another Con is a lot smaller Raman signal on the anti Stokes side.

3) Analyze the fluorescence out with math (fluorescence is smooth, Raman is spikey). Data is posted so any nexian interested can give it a shot.

Soo, you are suggesting going down 1), right?

I'll give the "poor man's" fluorescence test a shot and report back. Wife made me take the setup down so it may be a few days.

Many thanks for the explanation and suggestion.

 

[benzyme]

yup 1)

aliphatics and even isoprenes may not be an issue...but it seems when more electronegative moieties are present, fluorescense may be more of an issue. In the case of indole, it’s the imine. I’ve read that 532 is useful for analyzing inorganics and carbon nanotubes. We’d need to go more towards red and NIR to analyze our favorite molecules.

 

[benzyme]

final (optics) configuration.