DIY IR-FT spectrometer?

[wxyz]

How hard would it be to make your own IR-FT spectrometer?

 

(Thats an Infrared Fourier Transform spectrometer -- used for

ID'ing organic ompounds.)

 

I checked around and even used ones are several thousand dollars.

It seems possible that one could make one, the hardest part

would be the interferometer I think: would nee to have two

mirrors one moveable and the other precisely adjustable.

The light source would be easy (but does an interferometer

work with a non-coherent light source?). And the IR detector you

would have to buy. The FT is easily done in software.

Does any body have experience these things and know why

one would be hard to build?

[UC]

The moving mirror needs to be moved at a precisely known and stable speed. In order for the fourier transform to work, the location of the moving mirror must be known as precisely as possible at all times. Good instruments use two additional light sources (one a laser) solely to keep track of the moving mirror, IIRC. You also need a beamsplitter, which has to be an IR transparent material polished to obscenely flat and parallel surfaces.

 

There is some discussion on sciencemadness.org about building an FTIR without a beamsplitter. Personally, I think the beamsplitter is a better idea. Less parts than the series of lenses and mirrors you'd need without it.

 

IR and near-IR sources are all non-coherent light. Globars, Nernst glowers. Heck, you can even use tungsten filament bulbs for near-IR. I believe the IR light is passed through a collimating lens though.

[wxyz]

thanks a lot for your reply. you mentioned two light sources are used in addition to the IR source. The laser I understand, as it will allow you to keep track of the position of the moving mirror to within less than a wavelength of the laser light. What is the other light source?

 

Also I have heard that KBr or NaCl crystals are used for IR beam splitters. It seems like the moisture in the air would attack these crystals so the whole interferometer would have to be sealed from the atmosphere. Is that true? Or maybe they coat crystals with something to protect them.

[lisanke]

Surprising there wasn't more feedback on this question. Aside from some of the precision parts, the most important component in FTIR is signal analysis done entirely in software. I'm certain some other issues (besides position control, like thermodynamics) are important. For ATR FTIR we used liquid-N2 to cool apparatus which may be a general requirement of FTIR. This question prompts me to investigate further.

Perkin-Elmer sells replacement KBr beam splitter for $3950 so I begin to see the problem with purchasing the optics separately.

[Enthalpy]

Some spectrometers use a transparent wedge to make intereferences whose spacing depends on wavelength. Fringes are observed by a linear CCD. Software de-fourierizes the signal to tell what the light components were. Looks easier for DIY than the moving mirror, as only the wedge is difficult.

 

A different approach would use a piezo transparent material instead of moving a mirror.

 

But why not a grating? A CD sector could be enough. Long ago, amateur astronomers made their own gratings.

[John Cuthber]

CDs will absorb some of the IR very strongly, so would a piezo (and I'd be interested in seeing a piezo driver with enough range of movement).

A detector cooled by liquid N2 will improve the signal to noise ratio, but that isn't so important unless you want fast scanning.

 

A grating would work, but you would need order sorting filters which make things difficult.

There are well documented advantages to an FT system, rather than a scanned one (prism or grating).

 

The third light source referred to is a white light source used to establish the point where the optical path difference is exactly zero.

The speed of the moving mirror is not important, but you need to know exactly where it is and that's a lot easier if it is moving at a constant speed.

 

IR systems generally use mirrors to collimate light, rather than lenses, partly to avoid absorption of the IR but also to avoid chromatic aberration.

Edited July 11, 2013 by John Cuthber

[Enthalpy]

Piezo drivers exist with varied range, depending on the need. Their active element can work by bending for instance, in which case it amplifies the movement as a bimetallic actuator does. Some piezo materials are transparent to the IR; quartz is to 3µm. The piezo could also be outside light's path, but then its index change won't act.

 

Cd-roms have tracks spaced by 2.5µm so they won't need an order sorting filter; the tracks could even be too close.

 

I certainly agree that FT has advantages, but we're speaking about do-it-yourself, which favours a grating.

 

Maybe one process to make with a lathe a grating free of transmission losses:

- Turn flat a piece of aluminium, preferably 6000 or 1000 series alloy. Get 0.3µm smoothness with little skill, much better is easy.

- Leave the part in place, take a sharp 45° angled HSS tool, choose the desired feed according to the wavelength range. 10µm is a small but common value, less is available. Turn "flat" to make a groove.

- Anodize the face black.

- Grind away the anodization from the hills, leave it in the grooves.

(Turning it away would need to anodize within the lathe; processes exist for big parts without a bath)

 

The grating is circular, but this could be an advantage. It results in a lens whose focal length varies according to the wavelength. The lens has spherical aberration, but maybe a sector of it can approximate an ellipsoid if the source and the detector are offset in the proper direction, as in a Jolo telescope. This would focus light on a CCD line detector without materials transparent to IR.

[John Cuthber]

Quartz is transparent to IR in the very real sense that you can use the IR absorption of quartz to measure it.

http://www.hse.gov.uk/pubns/mdhs/pdfs/mdhs101.pdf

 

Do you understand the relationship between the change in path length and the resolution of the spectrometer?

 

You do need an order sorting filter.

The diffraction angle will be the same for 2 µm as it is for the second order diffraction of 1µm and this sort of thing happens regardless of the grating spacing.

A grating is much easier to make, but you do need order sorting filters, whether you like it or not.

 

If I was using a lathe I'd cut a fine screw thread on a metal tube and then cut it along its length and open it out.

However, actually,. I'd just buy a reflective grating on eBay.

But the use of a zone plate made by turning a flat face on a lathe is interesting.

You could also use an old LP record.

[Enthalpy]

Second order diffraction happens, but may be out of the range of the instrument, depending on the grating's spacing and the possible incoming wavelengths. With near infra-red and 2.5µm spacing, things look rather good. It limits the instrument's bandpass, though.

[lisanke]

Thank you both for the excellent direction(s) to proceed. While duplicating a traditional mechanism was my 1st thought, I understand measuring interfere of waves in the right separate is the basis of the spectrometer. It was explained to me (NCSU Advanced Measurements course) that the continuous mirror (causing a continuous circular variation of the interference) was the basis for reverse Fourier transforms conversion from time to frequency domain (spectrum). If a static diffraction grating is used to produce interference wouldn't this only produce interference at one wave separation... does that prevent a continuous spectrum range measurement.

[John Cuthber]

Second order diffraction happens, but may be out of the range of the instrument, depending on the grating's spacing and the possible incoming wavelengths. With near infra-red and 2.5µm spacing, things look rather good. It limits the instrument's bandpass, though.

Without order sorting the bandwidth is limited to an octave at best.

However the OP asked (albeit a long time ago) about "ID'ing organic compounds." and to do that you need a wider range- typically 400 to 4000 cm^-1 which is more than 3 octaves.

 

Lisanke, I think you would benefit from looking at something like this

http://en.wikipedia.org/wiki/Monochromator#Czerny-Turner_monochromator

[Enthalpy]

At 25µm wavelength, CCD detectors would be unaffordable. Looks like gratings are out.

[AbeMichelson]

I've been toying around with the idea myself. I have the luxury of being able to scavange parts from old instruments (have several parabolic mirrors), but have been looking at mirror free ideas. Some comments/thoughts:

1. The source light is usually introduced through an aperture. A hole of around 50-100mm in a piece of Al would work fine.

2. I have found a component source, but haven't found a price http://www.axetris.com/infrared_sources_products.htm

3. I recently found out that the motion sensors (IR) actually use a mid-band detector, LiTaO3. Some IR manufacturers are offering this as a low cost (re:not great) option. The sensitivity is not phenonmenal, but is extremely affordable on a DIY budget.

4. You do not need a CCD for the grating method, the LTO detector would work. That being said the gratings will determine the resolution, each step in the grating will correspond with the measureable wavelength. For good resolution, you will need to be able to have steps less than a micron.

5. You do not neccessarily need a HeNe for mirror positioning, it helps with stability of signal however. You may be able to use a high end solid state. I wonder if the laser/detector components of a CD player would work?

6. Beamsplitter is the highest cost component of this. You are correct in that it degrades over time in air. I have seen one melt! You could use ZeSe, but will lose a lot of throughput or a Ge-coated which will still degrade eventually and costs much more. DO NOT BUY A USED BEAMSPLITTER!

7. Grating may be the way to go for DIY. Here is one I found real quick for under $200 http://www.lightsmyth.com/products/nanopatterned_diffraction_gratings/linear_groove.php . I would shop around, however.

I imagine that it is possible to build a system under $1000.

[445nm]

Hi all, new to the forum this is my first post. I, too, have an interest in a DIY NIR spectrophotometer. I've been doing a fair bit of homework on the properties of diffraction gratings and overall spectrometer design. My thoughts are not to do FT-NIR, but rather, just a NIR spectrometer that is as low-budget as possible (MODS: If I need to start my own thread, just let me know). My basic ideas include the following points:

1) Eliminate optics by using either (A) a Rowland Circle configuration with a constant focal length equal to the diameter of the Rowland circle as is described here: http://www.physics.arizona.edu/~haar/ADV_LAB/ROWLAND.pdf, or else use (B) a holographic abberation-corrected grating designed for a constant-deviation monochromater.
2) Use a relatively inexpensive InGaAs photodiode for the detector such as can be found here: http://www.thorlabs.us/newgrouppage9.cfm?objectgroup_id=285

3) If I am doing 1(A) above, then perhaps I can simply curve some CD-ROM material by sandwiching it between two pieces of some 6" ABS pipe with a window cut out of the inner (concave) piece to expose the grating.
4) Use as simple a configuration as possible of toothed pulleys, a timing belt, a tensioner, and a stepper motor to control the rotation of the moving parts.
5) Use an Arduino to handle the A/D conversion of the output of the photodiode, and also to control the stepper motor.

Am I on the right path here? Does anybody see any potential fatal flaws in my fundamental plan?

Edited April 17, 2014 by 445nm